No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Optimal participation of a virtual power plant in electricity market considering renewable energy: A deep learning-based approach
Abstract
Recently, the penetration of renewable energy sources (RESs) and electric vehicles (EVs), has increased significantly in the power system. These resources bring many advantages to the environment; however, due to their small capacity, they cannot participate in the electricity market in a standalone manner. Furthermore, RESs and EVs have high stochastic behavior, that impose considerable uncertainty in RESs generation and EVs behavior profiles. These challenges can be addressed using the virtual power plant (VPP) concept. In this paper, an optimal bidding strategy is presented for VPP to participate in energy and spinning reserve markets. The uncertainties in the stochastic parameters of this system, including those related to the load demand, electricity prices, and wind speed, are handled using a deep learning approach based on bi-directional long short-term memory (BLSTM) networks. A BLSTM network uses the information from complete temporal horizon, accounting for previous and future features, and thus provides an accurate estimate of stochastic parameters. The numerical results confirmed that the BLSTM network outperformed the other methods and can forecast the stochastic parameters with only 3.56% and 3.53% errors in VPP profit when VPP participates in the only day-ahead energy market and both day-ahead energy and spinning reserve markets, respectively.
... The ramp rate of the diesel generator is modelled in Equations (24) and (25). Additionally, Equations (31) and (32) are provided to linearize the DG cost function [24]. ...
... Ensuring that all parameters at each grid point remain within acceptable limits is crucial for optimal load dispatch management, which in turn enhances grid performance and stability. To this end, the study employs linearized AC load dispatch constraints, which are defined in Equations (37)-(43) [24]. Equations (37) ( ...
... Figure 4 displays the MMG system's load demand curve and electricity prices. The specifications of the DGs [24], ESSs [23], EVs [11], and miners are provided in Tables 1-4, respectively. ...
This study presents an innovative energy management framework for multi‐microgrids, integrating the burgeoning domain of cryptocurrency mining. Cryptocurrencies, a novel fusion of encryption technology and financial currency, are witnessing exponential global growth. This expansion correlates with a surge in the prevalence of mining activities, amplifying electricity consumption and necessitating accelerated advancements in urban transmission and distribution infrastructures, coupled with increased financial investments. Despite cryptocurrencies' growth, comprehensive research to capitalize on their potential is scarce. This article introduces an operation cost model for miners in the proposed dual‐stage framework. The first stage is dedicated to day‐ahead scheduling, focusing on peak shaving and valley filling in the electricity demand curve, while concurrently optimizing operational costs. The second stage, updating each 5 min, minimizes imbalances in response to uncertain network conditions. A pivotal feature of this framework is the allocation of revenues generated from mining operations towards enhancing renewable energy resources. Empirical simulations underscore the framework's efficacy, evidenced by a substantial peak shaving of 482.833 kW and valley filling of 4084.42 kW. Furthermore, this approach effectively maintains operational costs within a feasible spectrum. Notably, the demand curve's peak‐to‐valley distance extends to 4 MW, with the revenue from mining activities alone sufficient to offset operational expenditures.
... The alternating direction method of multipliers is developed in [9] for optimal dispatch of a VPP consisting of conventional distributed generations (DGs), i.e., fossil fuel-based units, RESs, ESs, and interruptible load. Optimal participation of a VPP in energy and spinning reserve markets has been conducted in [10] without modeling the impact of uncertainty modeling on the VPP's optimal bidding in the day-ahead (DA) market. In addition, this work is not utilized in the DR program. ...
... The same method is employed in [7] and [11]. In study [10], long short-term memory (LSTM) networks are applied for modeling different stochastic parameters; however, a rough artificial neural network (RANN) is considered for modeling EVs' behaviors without considering the relationship among different features such as arrival time, departure time, and travel distance which increases the number of infeasible samples. In study [23], the authors the authors proposed optimal bidding for EVs and ES aggregators in the day-ahead market (DAM) for the frequency regulation market where the uncertainty in price is modeled by the seasonal autoregressive moving average (SARIMA) model, and load uncertainty is not considered. ...
... Uncertainty Modeling Method Price RES Load EV [8] ⊠ ⊠ Normal, Weibull distribution functions [17,25] ⊠ Point estimate method [5] ⊠ 2-point estimate method [2,4,21] ⊠ Scenario-based method [18] ⊠ ⊠ ⊠ ⊠ - [19] ⊠ ⊠ ⊠ Historical data and Markov [20] ⊠ ⊠ ARMA and adaptive neuro-fuzzy inference system [6] Weibull, Beta, and Normal distribution function [7,11,15] ⊠ ⊠ Weibull, Beta, and Normal distribution function [26,27] ⊠ Weibull, Beta, and Normal distribution function [9] ⊠ ⊠ ⊠ Weibull function [28] ⊠ ⊠ Scenario generation [23] ⊠ ⊠ SARIMA and Copula methods [24] ⊠ ⊠ ⊠ Empirical probability density function [29] ⊠ Monte Carlo simulation [12] ⊠ Robust optimization [14] ⊠ Scenario-based method [3,10,13] RANN and LSTM [16] ⊠ ⊠ ⊠ ⊠ -This paper BLSTM ...
Recently, renewable energy resources (RESs) and electric vehicles (EVs), in addition to other distributed energy resources (DERs), have gained high popularity in power systems applications. These resources bring quite a few advantages for power systems—reducing carbon emission, increasing efficiency, and reducing power loss. However, they also bring some disadvantages for the network because of their intermittent behavior and their high number in the grid which makes the optimal management of the system a tough task. Virtual power plants (VPPs) are introduced as a promising solution to make the most out of these resources by aggregating them as a single entity. On the other hand, VPP’s optimal management depends on its accuracy in modeling stochastic parameters in the VPP body. In this regard, an efficient approach for a VPP is a method that can overcome these intermittent resources. In this paper, a comprehensive study has been investigated for the optimal management of a VPP by modeling different resources—RESs, energy storages, EVs, and distributed generations. In addition, a method based on bi-directional long short-term memory networks is investigated for forecasting various stochastic parameters, wind speed, electricity price, load demand, and EVs’ behavior. The results of this study show the superiority of BLSTM methods for modeling these parameters with an error of 1.47% in comparison with real data. Furthermore, to show the performance of BLSTMs, its results are compared with other benchmark methods such as shallow neural networks, support vector machines, and long short-term memory networks.
... In [4], a VPP consists of a diesel generator, storage systems, electric vehicles, and wind power plants scheduled to participate in DA and spinning reserve markets to maximize its profit. The neural network is utilized to cover uncertainties in electrical load, wind speed, market prices, and the behavior of electric vehicle drivers. ...
... If the car is not available enough in parking to charge above 90% of capacity, the car will be charged to the maximum amount based on the available times in parking (12). The equations for storage, renewables, and CHP units are the same as in References [4] and [22]. ...
... The network data is considered [11], and 330 electric vehicles in PQ buses (Buses that consume electricity) are uniformly considered, including information about the car owner's behavior from the [26]. The statistics for the unit characteristics, cost, and heating load were chosen from [4]. Specifications for TPV are taken from [22]. ...
... Most of previous research works consider a ingle objective function for VPP, while the problem can be optimized from different perspectives using multi-objective optimization methods such as Pareto-based methods [24], weighting method [25], and methods of converting several objective functions into one objective function [26]. In addition, any parameter that is related to the future has uncertainty and must be managed using a proper uncertainty modelling approach such as robust optimization [27], scenario generation [28], and artificial neural networks [29]. Wind speed, solar radiation, load demands, and market prices are the parameters with significant uncertainty [2]. ...
... In this paper, the fuel cell is considered as gas-fired DG. These units generate electricity for the electrical network and, at the same time, are an industrial consumer for the natural gas network [29][30][31]. The operating cost of the DGs is given in Equation (1) as the sum of the production, startup, and shutdown costs, which are defined in Equations (2)-(4), respectively. ...
... An electricity storage system is charged when there is excessive generation and discharged during high electrical load. The degradation cost of ESS is considered in Equation (28) [29]. This unit's minimum and maximum charge and discharge rates are given in Equations (29) and (30). ...
A virtual power plant (VPP) is a solution that brings distributed generation (DG) resources together and allows them to be optimally utilized to meet load demands in the presence of technical and pollution constraints. Electricity, heat, and natural gas are interdependent at the levels of generation, transmission, and consumption, and the interactions of these energy sources need to be considered. This paper presents an optimal model for daily operation of a multi‐energy virtual power plant (MEVPP), including electric, thermal, and natural gas sectors. MEVPP includes small‐scale gas‐fired and non‐gas‐fired DGs, combined heat and power (CHP), power to gas (P2G), boilers, electrical storage, electric vehicles (EV), and thermal storage. Renewable energy resources (RES), including wind turbines (WT), photovoltaic (PV), and PV‐thermal (PVT), also supply P2G technology. Smart grid technologies such as price‐based demand response (PBDR) and incentive‐based demand response (IBDR) are employed for electric loads. The proposed MEVPP model is eligible to participate in day‐ahead electricity, natural gas, heat markets, and electrical spinning reserve market. The scheduling model is multi‐objective to maximize MEVPP profit and minimize carbon dioxide emissions. The Epsilon constraint method is utilized to solve the problem, and the best Pareto point is chosen using the fuzzy satisfying approach.
... The authors of [11] developed a block chain based VPP in the distribution network that can manage its DERs in a decentralized manner. In [12], an optimal bidding strategy model is proposed for a VPP to participate in the DA, spinning reserve, and ancillary service markets. A VPP is modelled in [13], including RES, ESS, CHP, and a boiler in order to reduce environmental pollution by promoting the VPP management. ...
... Finally, the output gate specifies the output information of the LSTM unit. The hidden state (h t ) is the filtered version of the cell state [12,27,28]. The formulation of the LSTM network is presented in Equations (1)- (6): ...
This paper presents an optimal bidding strategy for a technical and commercial virtual power plant (VPP) in medium-term time horizon. A VPP includes various distributed energy resources (DERs) that can participate in the Pool and Futures markets. Although medium/long-term scheduling provides the opportunity to participate in the futures market, it also raises the possibility of unit failure. In this regard, the impact of distributed generation (DG) units’ failure, as an important challenge in VPP, is incorporated in the proposed model. The model is formulated as a risk-constrained two-stage stochastic problem. The VPP signs futures market contracts in the first stage, and in the second stage, it participates in the day-ahead (DA) market and manages its DERs. Long short-term memory (LSTM) neural network and scenario generation and reduction methods are used to capture the uncertainty parameters of electrical load, DA market prices, wind speed, and solar radiation in the proposed model. The performance of proposed model is investigated in different cases. The obtained results show that the VPP can compensate the losses caused by the DG units’ failure through taking advantage of the arbitrage opportunity.
... α is 3 ($/MW). Emission data are chosen based on [9]. In addition, the profile of different input data is illustrated in Fig. 2. ...
... VPPs load data is given in Fig. A1. DGs emission data are obtained from reference [58] as stated in Table A7. ...
In recent years, the penetration of distributed energy resources (DERs) has increased significantly due to their tremendous effect on network flexibility, economic indicators, and power loss. On the contrary, a diverse assortment of DERs can lead to some challenges in controlling these resources in the power grid. To acquire the maximum benefit of DERs and overcome their challenges the concept of virtual power plants (VPPs) has been suggested. Due to the ability of VPPs to participate in electricity markets and the competition of VPPs to gain more profit we are facing deregulated multi-operator markets, and it is necessary to define VPPs as price maker units. The optimal economic assessment of VPPs in a multi-operator market depends on two folds: modeling inner cooperation between its components and managing external competition with other VPPs. To this end, in this paper, a new framework for optimal economic assessment of a multi-operator VPP system is proposed by considering a combination of non-cooperative and cooperative game theory-based approaches. In the proposed methodology, VPPs compete with other rivals to determine the amount of power exchange and offer prices based on supply function equilibrium. Due to incomplete information of VPPs about other opponents and market construction, a combination of particle swarm optimization and genetic algorithm is proposed to find the Nash equilibrium point. Also, the Shapely value concept is used for fair distribution of shared profit among VPPs components. The effectiveness of the proposed method has been verified in two case studies for a multi-operator VPP with a diverse assortment of DERs. The results show that VPP profit and electricity market prices directly relate to the diversity of resources in VPP. In this regard, the mark-up coefficient of the VPP with a greater number of DERs is about 16% and 32% larger than the two other VPPs which leads to more profit for this VPP and resources in its coalition.
... The application of DL algorithms on power system fault diagnosis, forecasting and classification approaches is also applied by various authors in the previous [30]- [31]. Furthermore, the applications of deep learning algorithms in power systems can be found in [32]- [33]. There are various types of DL methods such as SAE, RNN (Recurrent neural network), CNN, LSTM, etc., in the literature. ...
Islanding detection is a critical task due to safety hazards and technical issues for the operation of microgrids. Deep learning (DL) has been applied for islanding detection and achieved good results due to the ability of automatic feature learning in recent years. Long short term memory (LSTM) and two dimensional (2D) convolutional neural networks (CNN) based DL techniques are implemented and demonstrated well performance on islanding detection. However, one dimensional (1D) CNN is more suitable for real-time implementations since it has relatively low complexity and cost-effective than 2D CNN. In this paper, for the first time, the 1D CNN and the combination of 1D CNN-LSTM are proposed for islanding detection to better exploit the global information of islanding data samples using the strengths of both networks. The proposed methods utilize only voltage and current harmonic measurements as input at the point of common coupling (PCC). About 4000 cases under the modified CERTS microgrid model are simulated to evaluate the performance of the proposed architectures. The simulation results and the presented analysis show that the proposed networks have achieved the maximum accuracy of 100% on the task of islanding detection; especially the proposed CNN-LSTM model outperforms the other approaches. Furthermore, the robustness of the proposed methods is demonstrated with unseen samples under low none detection zone and the expansion of microgrid topology.
... A comparison between supply function approximations allows the reader to understand the importance of using deep learning as a complementary technique to machine learning and to improve the evolution of artificial intelligence each time the VPP feeds the algorithm with the day-to-day data obtained from power systems. In addition, according to [207], it is important to consider that, in order to achieve the optimal bidding strategy of a VPP, the correct collection of data in every element of the system and a robust deep learning-based approach is required. ...
Virtual Power Plants (VPPs) allow the aggregated management of diverse power technologies in systems of different size, complexity and connectivity. The main objectives of a VPP are to ensure the correct integration of these technologies into power systems, to optimize the power production and load demand, and to meet the technical and commercial requirements, considering the stakeholders’ needs. This paper provides a literature review of the most influential factors as regards the management of VPPs. This literature review includes aspects like the evolutionary frequency of the concepts/definitions most used, the VPP operation, planning and interaction with the systems, programming methods, math techniques and software tools used to solve the issues proposed and the international VPP projects as reference for the systems development. Therefore, this work aims to provide the reader the information and concepts that can support for a better understanding of the role of VPPs in power systems, the approach to sustainability and the interaction with the power market as the good way to bring new economic opportunities to small producers, consumers and prosumers of renewable energy.
Solar forecasting is a rapidly evolving field that can substantially contribute to the effective integration of large amounts of solar photovoltaic (PV) capacity into the electricity system. However, newly developed solar forecasting models are rarely tested in an operational context considering the intended application and objective. Besides, models are typically evaluated considering only technical error metrics, disregarding their economic value. This paper proposes an operational bidding strategy that optimizes the participation of a PV power plant in the electricity spot markets. To this end, a novel multistage stochastic optimization method is developed that considers the day-ahead, intraday, and imbalance markets. As the developed method utilizes a scenario generation algorithm, the proposed method can be adopted for a wide variety of related applications. The performance of the developed method is assessed using technical and economic metrics and compared to a reference method. The results demonstrate the effectiveness of the proposed bidding strategy, as it substantially outperforms the reference market bidding strategy. The findings also provide insights into the value of a multistage bidding method, as extending market participation from the day-ahead to the intraday market increases revenues by 22%, while halving the total imbalance. Additionally, the study examines the relationship between the technical and economic performance of solar power forecasting models, revealing a non-linear correlation.
The emergence of electric vehicles is reshaping the energy landscape, requiring the development of innovative energy integration mechanisms to engage prosumers. However, current methods face numerous challenges when actively involving communities. Some key challenges include system uncertainties and time issues, optimizing charging strategies, real-time decision-making needs, privacy concerns, and battery degradation. For instance, the unpredictability of driver behavior and traffic conditions introduces complexities in devising efficient energy integration strategies and economic incentive models. The intricate interplay of these factors necessitates advanced computational techniques, making machine learning an invaluable tool. This paper concisely reviews prominent machine learning algorithms, encompassing supervised and unsupervised learning, focusing on their distinctive capabilities in prediction, clustering, dimensionality reduction, and generative modeling. Additionally, it explores reinforcement learning, emphasizing its aptitude for real-time decision-making. The focus of the study lies in the application of advanced algorithms, specifically examining their effectiveness in various strategic operational frameworks. The aim is to integrate electric vehicles into power systems efficiently. These frameworks include bargaining, contracts, auctions, game theory, and economic incentives such as pricing and cost-profit optimization. Each application includes a concise overview of the general methodology and investigates in-depth discussions regarding the suitability and challenges of deploying machine learning techniques. This paper will guide industry professionals in implementing solutions for electric vehicle dispatching problems and provide valuable insights to academics for further research and development.
The optimal integration of Renewable Energy Sources (RESs) into distribution networks is crucial for sustainable energy generation. However, existing approaches utilized for optimal RES integration, including metaheuristic and analytical methods, exhibit limitations. Analytical methods are simple to implement, computationally efficient, and able to model power system mechanisms, yet they face scalability issues, making them inefficient for large-scale distribution systems. On the other hand, metaheuristic approaches, while flexible for complex problems, suffer from premature convergence problems, limiting their application in real distribution networks. To address these issues, this paper introduces a novel hybrid model that combines the strengths of analytical and metaheuristic techniques. The proposed model employs a deep learning-based method called bi-directional long short-term memory (B-LSTM) network is utilized to handle the uncertainties associated with the Photovoltaic (PV) and Wind Turbine (WT) generation and load demand. Subsequently, a distinctive analytical multi-objective index-based procedure is developed to minimize the real and reactive energy loss and voltage deviation by analyzing the branch current change caused by the RES penetration in distribution networks. The optimization problem is then solved using the binary particle swarm optimization (BPSO) technique. Importantly, the presented method considers the long-term nature of optimal RES integration and reflects the seasonal impacts of the load and RES output powers. The proposed approach is applied to the IEEE 33-bus test system. The simulation results show that the introduced hybrid model, by addressing the limitations of existing approaches, offers a promising solution for efficient and effective RES integration in distribution systems while requiring low computational effort, thereby advancing the transition towards sustainable energy infrastructure.
Data-driven machine learning-based methods have provided immense capabilities, revolutionizing sectors like the Buildings-to-grid (B2G) integrated system. Since the penetration rate of distributed energy resources increases towards a net-zero emissions power system, so does the need for advanced services that ensure B2G-integrated system reliability. The convergence of advancements in machine learning, computational resources at the entire cloud-edge continuum, and large datasets from sensing devices enable the development of these data-driven energy analytics services. This work conducts a systematic text-mining-based literature review to examine the diverse range and trends of machine learning methods used to enhance reliability in B2G-integrated systems. While traditional manual sampling and analysis approaches have limited the effectiveness of previous literature review papers in this field, this systematic literature review work aims to synthesize and summarize the existing body of research more efficiently and effectively. To achieve this, this study collected almost 10,500 papers from Scholar and Scopus databases. It employed text-mining-assisted BERTopic-based topic modelling and statistical trend analysis techniques to uncover semantic patterns and explore the temporal evolution of research themes. A two-dimensional taxonomy was derived to analyze the technical papers from a business and machine learning-related perspective. By quantifying the temporal trends within these topics, the study unveiled insights about the state-of-the-art analytics that ensure reliability in the B2G-integrated system domain while proposing future research directions.
One of the significant environmental issues is global warming, and governments have changed their procedures to reduce carbon emissions. Sustainability is commonly described as having three dimensions: environmental, economic, and social. There are numerous environmental impacts associated with energy systems and the significance of energy for living standards and economic development. Therefore, the movement towards intelligent energy systems and virtual power plants (VPPs) is being pursued more rapidly due to economic and environmental issues. The VPP is one of the technologies used to increase the entire system’s efficiency. Moreover, because of environmental pollution, increased greenhouse gas production, and global warming, countries’ policies have changed towards reducing the use of fossil fuels and increasing the penetration of renewable energy sources (RESs) in distribution networks. However, RESs, such as wind turbines (WT) and photovoltaic (PV) panels, exhibit uncertain behavior. This issue, coupled with their high penetration, poses challenges for network operators in terms of managing the grid. Therefore, the sustainable virtual power plant (SVPP) is a suitable solution to overcome these problems and reduce the emissions in power systems. This study examines the cost of optimal operating of the SVPP and the amount of produced pollution in four different scenarios in the presence of a demand response program (DRP), energy storage system (ESS), etc., and the results are compared. The results indicate that the simultaneous implementation of DRPs and utilization of ESS can lead to a decrease in costs and pollution associated with SVPPs by 1.10% and 29.80%, respectively. Moreover, the operator can resolve the shortage and excess power generation that occurs during some hours.
In recent times, the virtual power plant (VPP) is gaining more attention in power system engineering due to its tremendous potential in enhancing sustainable urbanism, in which, it supplies clean energy from distributed generators. Electricity is deemed a basic requirement for future automotive and ultra-modern technologies. The deficiency of traditional energy resources and their complex generation process make the production cost of electricity increase dramatically. Moreover, traditional power distribution systems are encountering issues in distributing electrical energy to fulfill customer demands. Therefore, this paper proposes a novel power management system named ‘the hybrid genetic local binary knowledge (HGLBK) algorithm’ to manage power distribution in the transmission lines and to optimize the total operation cost of the network. The hybrid optimization algorithm effectively controls the load by supplying the surplus power load to the adjacent feeders thereby optimally selecting and placing the distributed energy resource (DER). The proposed concept is implemented at Kayathar, Tamil Nadu in India, and their real-time data are utilized for modeling the VPP. The proposed VPP concept is implemented in the IEEE-9 bus system and the performance of VPP is simulated using the MATLAB software. The performance of the proposed HGLBK algorithm is assessed by comparing its effectiveness with the existing approaches.
Electric power systems are moving toward more decentralized models, where energy generation is performed by small and distributed power plants, often from renewables. With the gradual phase out from fossil fuels, however, Distribution Energy Resources (DERs) are expected to take over in the provision of all regulation services required to operate the grid. To this purpose, the opening of national Ancillary Service Markets (ASMs) to DERs is considered an essential passage. In order to allow this transition to happen, current opportunities and barriers to market participation of DERs must be clearly identified. In this work, a comprehensive review is provided of the state-of-the-art of research on DER integration into ASMs. The topic at hand is analyzed from different perspectives. First, the current situation and main trends regarding the reformation processes of national ASMs are analyzed to get a clear picture of the evolutions expected and adjustment required in the future, according to the scientific community. Then, the focus is moved to the strategies to be adopted by aggregators for the effective control and coordination of DERs, exploring the challenges posed by the uncertainties affecting the problem. Coordination schemes between transmission and distribution system operators, and the implications on the grid infrastructure operation and planning, are also investigated. Finally, the review deepens the control capabilities required for DER technologies to perform the needed control actions.
Virtual Power Plants (VPPs) have emerged as a modern real-time energy management architecture that seeks to synergistically coordinate an aggregation of renewable and non-renewable generation systems to overcome some of the fundamental limitations of traditional power grids dominated by synchronous machines. In this survey paper, we review the different existing and emerging feedback control mechanisms and architectures used for the real-time operation of VPPs. In contrast to other works that have mostly focused on the optimal dispatch and economical aspects of VPPs in the hourly and daily time scales, in this paper we focus on the dynamic nature of the system during the faster sub-hourly time scales. The virtual (i.e., software-based) component of a VPP, combined with the power plant (i.e., physics-based) components of the power grid, make VPPs prominent examples of cyber-physical systems, where both continuous-time and discrete-time dynamics play critical roles in the stability and transient properties of the system. We elaborate on this interpretation of VPPs as hybrid dynamical systems, and we further discuss open research problems and potential research directions in feedback control systems that could contribute to the safe development and deployment of autonomous VPPs.
Virtual Power Plants (VPPs) are one of the concepts introduced in modern power systems to handle the increasing number of the distributed generation (DG) units. Technical VPPs (TVPPs) consider both financial and technical perspectives of using DGs in the system. Besides, secure and reliable operation of the system is a priority. In this paper, optimal operation of technical virtual power plants in a reconfigurable network is formulated as an optimization problem to resolve the probable contingency problem in the lines of the system. The VPP is assumed to be a multi-carrier energy system including combined heat and power (CHP), renewable DGs and dispatchable DGs beside thermal and electrical storage systems and loads. The uncertainties of renewable based DGs and demand levels are handled using chance constrained programming (CCP). By using CCP in presence of uncertain parameters, the security of the system can be guaranteed in predefined level of probability. Finally, to evaluate the effectiveness, quality and applicability of the proposed methodology, the problem is structured as a mixed-integer nonlinear programming (MINLP) problem which is solved using General Algebraic Modeling System (GAMS) software via Baron solver.
In this paper, a three-stage bi-level stochastic programming approach is proposed for joint energy and reserve scheduling of a virtual power plant (VPP). In this framework, the VPP can provide demand response (DR) services and reserve capacity from external DR providers (DRPs) by participating in a local intraday demand response exchange (IDRX) market and trading with internal load aggregators (LAs). The VPP tries to reach a proper balance between allocating spinning reserve and DR services to reduce the penalty cost resulting from the difference between the day-ahead (DA) scheduled power and the real-time dispatched. To this end, a bi-level problem is formulated, in which at the upper level the objective of the VPP is to maximize its profit, and in the lower level, the LAs maximize their social welfare. The interaction between the VPP and the LAs is modelled as a Stackelberg game; then, by reformulating the lower-level problem using Karush-Kuhn–Tucker optimality conditions, a mathematical programming with equilibrium constraints (MPEC) is achieved. The intended problem is converted into a convex mixed-integer quadratic problem (MIQP) by applying the strong duality theorem. Simulation results demonstrate that providing DR services from the internal LAs and the local IDRX market noticeably affects the VPP’s decisions improving the profit by more than 7% and reducing the imposed imbalance penalty in the balancing market by nearly 50%.
In recent years; owing to the effect of fossil fuels on global warming, the exhaustion of oil fields, and the lucrative impacts of renewable energy resources (RESs), the penetration of RESs has been increasing significantly in power systems. An effective way to benefit from all RESs advantages is by applying them in microgrid systems (MGS). Furthermore, MGS can ease the way for utilizing a large amount of RESs, if its economic-environmental-technical aspects of it are taken into account. In this regard, this paper proposes an optimal solution for the energy management of a microgrid by considering a comprehensive study. In the proposed methodology, different distributed energy resources such as wind turbines generator (WTG), energy storage (ES), combined heat and power (CHP), rubbish burning agent (RBA), and diesel generators (DG) are modeled. In addition, electric vehicles (EVs) are considered a load with uncertainty. The objective function of the proposed method is to minimize the microgrid’s total cost by considering the microgrid’s emission cost and technical constraints. In this study, the microgrid’s technical, environmental, and economic aspects are investigated. In addition, the optimization problem is converted into a mixed-integer linear programming method by using the proper linearization method. In this paper, the increasing effect of wind energy penetration rate on the total price also has been studied. The simulation results show that by increasing the wind energy penetration rate by up to 30% of total power, the total cost will decrease by up to 30.9%.
With the evolution of wind turbine technologies, an immense desire is to integrate this energy into power markets more than ever. On the other hand, wind farms have a limited role in the electricity market due to their uncertainty. The influence of wind farms, electric vehicles, and flexible loads participating in the electricity markets aiming to minimize the deviation of wind power generation capacity from the final amount of cleared power in the market has been explored in this study. Two probabilistic models have been considered for this goal in a day-ahead electricity market. Wind farms, electric vehicles, and flexible loads offer competitive prices in the first scenario by forming a virtual power plant (VPP). In the second scenario, they engage individually in the electrical market. Both generation models incorporate a scenario reduction technique to simulate the uncertainty associated with wind turbine production capacity and a penalty mechanism for violating the wind farm generation from the amount of pledged capacity in the market. The results imply that wind farms, electric cars, and flexible loads benefit from engaging as a virtual power plant scheme component in the electricity market.
Virtual power plant (VPP) concept creates additional benefits for different stakeholders, such as system operators, distributed energy resources’ owners, and aggregators. This article presents an auction-based energy management solution from VPP aggregator point of view. To this end, a network-constrained economic dispatch has been developed taking into account a comprehensive resource portfolio, including the conventional and renewable distributed generations, energy storage units, and flexible loads. The VPP aggregator aims at minimizing its expected costs through optimal dispatch of its own resources as well as purchasing and selling electricity from/to upstream network. To this end, a stochastic programming approach has been employed to consider the generation uncertainty of renewable energy resources as well as day-ahead energy market prices using a set of plausible scenarios. The payments to the prosumers are according to an auction-based pricing method that could be an efficient strategy for the VPP aggregator in order to guarantee fairness among consumers and producers within the VPP. In the auction-based pricing scheme, the electricity price at each local node can be calculated according to the demand and generation in each time slot. The proposed framework is a mixed-integer linear programming problem and solved by the CPLEX solver in the GAMS software. It is noteworthy that the effectiveness of the model has been verified, considering several numerical studies.
This study examines the roles of resource consumption accounting and competitive prices in attaining sustainable profitability. The objectives were (1) to determine whether the adoption of resource consumption accounting practices yields significant improvements in competitive strategies in a highly competitive situation where activity-based costing has proved to be insignificant, and (2) to ascertain if the positive relationship between competitive pricing and sustainable profitability is increased by the extent to which resource consumption accounting exerts pressure for sustainability profitability. A PLS-SEM procedure was applied in analysing 129 of the top 30 performing companies’ structured questionnaire responses drawn from five industries in Kurdistan from 2021. The empirical results demonstrated that competitive pricing models involving resource consumption accounting systems provide superior price forecasting, error reduction and profit maximisation capabilities than existing energy models. The study’s outcomes highlight that the extent to which resource consumption accounting exerts pressure on sustainability profitability significantly increases the positive relationship between competitive pricing and sustainable profitability. The results of this study advance construct and item development involving competitive pricing and resource consumption accounting while testing relationships to uncover the moderating role of resource consumption accounting in profit maximisation. Thus, energy and non-energy industrial companies must rely on resource consumption accounting to set competitive prices and enhance and sustain their profitability by considering the overlooked energy pricing stochastic parameters and errors amid rising energy shortages and costs.
The energy market relies on forecasting capabilities of both demand and power generation that need to be kept in dynamic balance. Nowadays, contracts and auctions of renewable energy in a liberalized electricity market heavily rely on forecasting future power generation. The highly intermittent nature of renewable energy sources increases the uncertainty about future power generation. Since point forecast does not account for such uncertainties, it is necessary to rely on probabilistic forecasts.
This work first introduces probabilistic forecasts with deep learning. Then, we show how deep learning models can be used to make probabilistic forecasts of day-ahead power generation from a wind power plant located in Northern Norway. The performance, in terms of the quality of the prediction intervals, is compared to different deep learning models and sets of covariates.
The findings show that the accuracy of the predictions improves by 37% when historical data on measured weather and numerical weather predictions (NWPs) were included as exogenous variables. In particular, historical data allows the model to auto-correct systematic biases in the NWPs. Finally, we observe that when using only NWPs or only measured weather as exogenous variables, worse performances are obtained. The work shows the importance of understanding which variables must be included to improve the prediction performance, which is of fundamental value for the energy market that relies on accurate forecasting capabilities.
Increasing the load demand and penetration of renewable energy sources (RESs) poses real challenges for optimal energy management of distribution networks. Moreover, considering multi-carrier energy systems has increased the efficiency of systems, and provides an opportunity for using the advantages of RESs. In this regard, we adopted a new framework based on the new challenges in the multi-carrier energy micro-grid (MEMG). In the proposed method, a comprehensive MEMG was modeled that benefits from a large assortment of distributed energy resources (DERs), such as micro-turbines, fuel cells, wind turbines, and energy storage. Considering many DERs is necessary, because these resources could cover one another’s disadvantages, which have a great impact on the total cost of the MEMG and decrease the emission impacts of fossil-fuel-based units. Furthermore, waste power plants, inverters, rectifiers, and emission constraints are considered in the proposed method for modeling a practical MEMG. Additionally, for modeling the uncertainty of stochastic parameters, a model based on a multilayer neural network was used in this paper. The results of this study indicate that using a decentralized model, along with stochastic methods for predicting uncertainty, can reduce operational costs in micro-grids and computational complexity compared with optimal centralized programming methods. Finally, the equations and results obtained from the proposed method were evaluated by experiments. View Full-Text
A Virtual power plant is defined as an information and communications technology system with the following primary functionalities: enhancing renewable power generation, aggregating Distributed Energy Resources and monetizing them considering the relevant energy contracts or markets. A virtual power plant also includes secondary functionalities such as forecasting load, market prices and renewable generation, as well as asset management related to the distributed energy ressources. Home energy management systems and building energy management systems have significant overlap with virtual power plants, but these bodies of research are largely separate. Machine learning has recently been applied to realize various functionalities of these systems. This article presents a 3-tier taxonomy of such functionalities. The top tier categories are optimization, forecasting and classification. A scientometric research methodology is used, so that a custom database has been developed to capture metadata from all of the articles that have been included in the taxonomy. Custom algorithms have been developed to generate infographics from the database, to visualize the taxonomy and trends in the research. The paper concludes with a discussion of topics expected to receive a high number of publications in the future, as well as currently unresolved challenges.
With growing shares of renewable power generation, ancillary services such as reactive power become increasingly important to balance power systems. In this work, we provide a framework to assess the suitability of small biomass power plants, i.e., less than 5 MW, for reactive power markets. The proposed methodology consists of two-stages and considers the spatial relationship between the biomass plants’ location and the regional transmission system and tests different plant remuneration scenarios. In the first step, the reactive energy potentials available from biomass plants are quantified and located through spatial analysis, considering locational factors that influence the economic value that these plants bring to the ancillary services markets. Second, a financial analysis is performed to determine how each biomass plant’s electricity bill can be reduced through participation in ancillary services markets. Resulting layers of geographic information provide system operators and policymakers with the available reactive power potential and required financial compensation of biomass plant owners close to the electricity transmission system. The methodology is applied to a case study in São Paulo, Brazil, using data from sugar and alcohol plants installed near the regional transmission system’s 138 kV lines to assess the implementation of a seasonal reactive power market.
Power System operator executes the Unit Commitment (UC) program to determine the optimal generation schedule for a daily or weekly horizon. In smart power systems, cyber-attacks such as False Data Injections (FDI) may resemble a kind of contingency with only manipulating measurements data and without physical initiating events. This paper proposes a network constrained UC model in the presence of cyber-attacks named Cyber-Secured UC (CSUC) model. Normally the operating point obtained by UC program is on its conventional schedule, however in case of cyber-attack risk, it can depart from the conventional schedule towards the cyber secured schedule without violating unit and network constraints during such transition. The CSUC schedule is secured against a type of cyber-attacks called Load Redistribution (LR) attacks. Moving to this new operating point, in case of cyber-attack risk, may incur a small and reasonable additional cost to the cost of conventional generation schedule. In order to guarantee the optimality of solution, the proposed CSUC model is formulated as a Mixed Integer Linear Programming (MILP) model. The Benders Decomposition (BD) method is utilized to solve the model in a master and sub-problem hierarchy. The proposed method is implemented in the IEEE 118-bus test system.
A better understanding of spatiotemporal distribution of PM2.5 (particulate matter with diameter less than 2.5 micrometer) concentrations in a continuous space-time domain is critical for risk assessment and epidemiologic studies. Existing spatiotemporal interpolation algorithms are usually based on strong assumptions by restricting the interpolation models to the ones with explicit and simple mathematical descriptions, thus neglecting plenty of hidden yet critical influencing factors. In this study, we developed a novel deep-learning-based spatiotemporal interpolation model, which includes the bidirectional Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) as the main ingredient. Our model is able to take into account both spatial and temporal hidden influencing factors automatically. To the best of our knowledge, it is the first time of applying the bidirectional LSTM RNN in the spatiotemporal interpolation of air pollutants concentrations. We evaluated our novel method using a dataset that contains daily PM2.5 measurements in 2009 over the contiguous southeast region of the USA. Results demonstrate a good performance of our model. We also conducted simulations to explore the properties of spatiotemporal correlations. In particular, we found the temporal correlation is stronger than the spatial correlation.
Virtual Power Plants (VPPs) represent a novel concept towards flexible integration of distributed energy resources (DERs) into the smart grid. VPPs will eventually help to overcome the stochastic nature of DERs which will result in a more stable and smart power grid. VPPs are constituted of an integration of heterogeneous systems, organizations and entities which collaborate to ensure optimal generation, distribution, storage, and sale of renewable energy in the energy market. By its composition, a VPP forms a kind of collaborative business ecosystem with high degree of interactions and interdependences among stakeholders. This survey, is focused on analysing the trends and also identifying areas of convergence between the collaborative networks (CN) discipline and the VPP concept and development using foreknowledge from the domain of CNs. A systematic literature review method was employed to summarize research evidence, by evaluating and interpreting available research works under specifically defined focus areas, relevant to both domains. The results confirmed a visible convergence between VPPs and CNs although the convergence level manifested differently under each “focus area”. The results showed that within a VPP, various strategic and dynamic collaborative alliances are formed. The forms of these alliances are similar to various CN organizational forms which include goal oriented networks, virtual breeding environments (VBE), grasping opportunity driven networks and continuous production driven networks. Additionally, various underlying functional principles of VPPs were also found to be similar to CN principles which include: virtual organization creation, operation, and dissolution, negotiations, broker services, VBE administrator services, VO planer services, VO coordinator services, and partner search and selection processes. A variety of organizational forms, similar to traditional CN forms or a mix of those could also be found. In terms of implementation level however, most works are focused on simulation or small prototypes.
Virtual power plants can be regarded as systems that have entered the network after restructure of power systems. In fact, these plants are a set of consumers capable of consuming and generating power. In response to widespread implementation of plug-in hybrid electric vehicles, further investigation of energy management in this type of power plants seems to be of great value. In effect, these vehicles are able to receive and inject power from/into the network. Hence, study of the effects of these vehicles on management of virtual power plants seems to be illuminative. In this paper, management of power consumption/generation in virtual power plants has been investigated in the presence of hybrid electric vehicles. The objective function of virtual power plants problem management is to minimize the overall costs including not only the costs of energy production in power generation units, fuels, and degradation of batteries of vehicles, but also the costs of purchasing electricity from the network. Furthermore, the constraints on the operational of plants, loads and hybrid vehicles, level of penalty for greenhouse gas emissions (CO2 and NOx) produced by power plants and vehicles, and demand response to the immediate price of market have all been attended to in the present study. GAMS/Cplex software system and sample power system have been employed to pursue computer implementation and simulation.
As the power system is facing a transition towards a more intelligent, flexible and interactive system with higher penetration of renewable energy generation, load forecasting, especially short-term load forecasting for individual electric customers plays an increasingly essential role in the future grid planning and operation. Other than aggregated residential load in a large scale, forecasting an electric load of a single energy user is fairly challenging due to the high volatility and uncertainty involved. In this paper, we propose a long short-term memory (LSTM) recurrent neural network (RNN) based framework, which is the latest and one of the most popular techniques of deep learning, to tackle this tricky issue. The proposed framework is tested on a publicly available set of real residential smart meter data, of which the performance is comprehensively compared to various benchmarks including the state-of-the-arts in the field of load forecasting. As a result, the proposed LSTM approach outperforms the other listed rival algorithms in the task of short-term load forecasting for individual residential households.
The storage units decrease the operation cost of active distribution network
considerably if they are managed optimally. In this paper, the short-term optimal
scheduling of stationary batteries is presented. The point estimate method is used
for considering uncertainty of load, wind-based distributed generation and plug-in
electric vehicles as well as their influence on optimal scheduling. The optimal
scheduling consists of minimizing cost objective function under technical
constraints. In this paper, the cost objective function is composed of operation and
reliability costs which are minimized using Tabu search algorithm. The storage
units are used for several objectives, i.e., peak shaving, voltage regulation, and
reliability enhancement. The numerical studies show the advantages of batteries for
energy management in active distribution network, and the impact of uncertainties
on optimal scheduling.
Deep learning allows computational models that are composed of multiple processing layers to learn representations of data with multiple levels of abstraction. These methods have dramatically improved the state-of-the-art in speech recognition, visual object recognition, object detection and many other domains such as drug discovery and genomics. Deep learning discovers intricate structure in large data sets by using the backpropagation algorithm to indicate how a machine should change its internal parameters that are used to compute the representation in each layer from the representation in the previous layer. Deep convolutional nets have brought about breakthroughs in processing images, video, speech and audio, whereas recurrent nets have shone light on sequential data such as text and speech.
This paper addresses the optimal bidding strategy problem of a commercial virtual power plant (CVPP), which comprises of distributed energy resources (DERs), battery storage systems (BSS), and electricity consumers, and participates in the day-ahead (DA) electricity market. The ultimate goal of the CVPP is the maximization of the DA profit in conjunction with the minimization of the anticipated real-time production and the consumption of imbalance charges. A three-stage stochastic bi-level optimization model is formulated, where the uncertainty lies in the DA CVPP DER production and load consumption, as well as in the rivals' offer curves and real-time balancing prices. Demand response schemes are also incorporated into the virtual power plant (VPP) portfolio. The proposed bi-level model consists of an upper level that represents the VPP profit maximization problem and a lower level that represents the independent system operator (ISO) DA market-clearing problem. This bi-level optimization problem is converted into a mixed-integer linear programing model using the Karush-Kuhn-Tucker optimality conditions and the strong duality theory. Finally, the risk associated with the VPP profit variability is explicitly taken into account through the incorporation of the conditional value-at-risk metric. Simulations on the Greek power system demonstrate the applicability and effectiveness of the proposed model.
Uncertainty modeling of Renewable Energy Sources, load demand, electricity price, etc. create a high volume of data in smart grids. Accordingly, in this paper, a precise forecasting method based on a deep learning concept with Micro-clustering (MC) task is presented. The MC method is structured based on hybrid unsupervised and supervised clustering tasks by Kmeans and Gaussian Support Vector Machine, respectively. In the proposed method, the input data sequence is clustered by the MC task, and then the forecasting process is employed. By applying the MC, input data in each hour is categorized into different groups, and a distinctive forecasting unit is allocated to each one. In this way, more clusters and forecasting networks are earmarked for the hours with higher fluctuation rates. The Bi-directional Long Short-Term Memory (B-LSTM), which is one of the newest recurrent artificial neural networks, is proposed as the forecasting unit. The B-LSTM has bidirectional memory—feedforward and feedback loops- that helps us to investigate both previous and future hidden layers data. The optimal number of clusters in each hour is determined based on the Davies-Bouldin index. To evaluate the performance of the proposed method, in this study, three forecasting tasks including the wind speed, load demand, and electricity price are studied in different periods using the Ontario province, Canada data set. The results are compared with other benchmarking methods to verify the robustness and effectiveness of the proposed method. In fact, the proposed method, which is equipped with the MC technique and B-LSTM networks, significantly promotes the forecasting results, especially in spike points.
Growing penetration of Plug-in Electric Vehicles (PEVs) in the transportation fleet and their subsequent charging demands introduce substantial intermittency to the electric load profile which imposes techno-economic challenges on power distribution networks. To address the uncertainty in demand, a novel deep learning-based approach equipped with a hybrid classification task is developed which can take into account the travel characteristics of the PEV owners. The classification structure helps us scrutinize the PEVs demand by allocating a specific forecasting network to each cluster of travel behavior patterns. In our hybrid classification task, first, an unsupervised classifier discerns hidden travel-behavior patterns between the historical PEVs data by clustering them; then, a supervised classifier directs each new PEV data to its appropriate clusterspecific forecasting network. The deep learning-based forecasting and classification networks are constructed based on the Long Short-Term Memory networks to investigate long-and short term features in PEV behaviors. The data-driven structure of our proposed method enables us to observe and preserve the correlation between PEV travel data parameters (departure time, arrival time and traveled distance) and avoid the generation of unrealistic travel samples found in scenario-based approaches. To verify the effectiveness of the proposed method in practical environments, we have studied the impact of the precise forecasting of the PEVs demand in an aggregator’s financial profit in the energy market of the California Independent System Operator market. The numerical results confirm the outstanding performance of our proposed deep learning-based method in forecasting PEVs demand against benchmark approaches in this field such as Monte Carlo, Quasi-Monte Carlo, and Copula with only a 6.77% error in comparison with real data.
Unit Commitment (UC) program is an important study in the operation of power systems which objects to achieve the minimum system operating cost by considering network and unit constraints. In conventional UC, competition between independent units to maximize their profits is not considered. To overcome this shortcoming, in this study, a game theory-based approach has been employed to address the above points. In this regard, we introduce two types of players including power plants and system operator, and solve UC problem with Nash Cournot and Nash Bertrand techniques. Indeed, the main goal of this study is to find a way that maximizes the profit of the system operator by considering the independent units' behavior. To verify the robustness of the proposed method, the introduced UC approach is employed on the IEEE six-bus power system.
In this paper, a multi-modal short-term wind speed prediction framework has been proposed based on Artificial Neural Networks (ANNs). Given the stochastic behavior and high uncertainty of wind speed, in modern power systems with high penetration of wind power, precise wind speed forecasting is necessary for the power utility managers. To ensure that the prediction results are sufficiently precise, in this study, a multi-modal method is designed based on denoising and prediction modules. In denoising module, at first, the optimal configuration of Stacked Denoising Auto-encoders (SDAEs) is determined based on Genetic Algorithm (GA), and, then, SDAEs are applied for pre-training the data to reduce the input data noise. Due to the fact that different data sources have different noise bases, the optimal structure of the denoising module is arranged completely independent of various input data. In the forecasting module, the Sinusoidal Rough-Neural Network (SR-NN) is utilized to predict wind speed. After that, denoising and forecasting modules are stacked together to make the holistic deep structure network. In this study, the Deep Learning (DL) approach has been utilized to extract more robust features from the input data in the denoising process. To handle the high intermittent behavior of wind speed, the rough neurons are considered. Rough neurons are made up of two pairs of conventional neurons that are known to be upper and lower bound neurons. Since the wind speed has a periodic and nonlinear behavior, an ANN, based on the sinusoidal activation function, is more accurate than the typical type of sigmoid activation function. In this paper, weather data from Ahar, Iran, which has a high potential wind power, is considered. In modern power systems, the extra short-term forecasting is also needed; therefore, in this study, wind speed forecasting in one hour and 10-minute intervals are employed. Input data are selected based on Grey Correlation Analysis (GCA), and according to the GCA result, in addition to wind speed data at various heights, environmental data such as humidity and temperature are also considered. To evaluate the efficiency of the proposed method, the simulation results are compared with other structures of ANNs and benchmarking machine learning methods such as Support Vector Machine (SVM) and Autoregressive Moving Average (ARMA) methods in different scenarios. In this comparison, the impact of DL has been thoroughly investigated. In this paper, the comparison of the performance of deep and shallow network structures has been studied.
Accurate forecasting of the combined loads of electricity, heat, cooling and gas in the integrated energy system is the key to improve the comprehensive efficiency and gain more economic benefits of various types of energy. As an important part of the new generation of energy systems, the integrated energy system contains energy subsystems such as electricity, heat, cooling and gas, and each subsystem employs energy supply, conversion and storage equipment. This form of energy system achieves the coupling of different types of energy in different links. Based on this, this paper firstly combs the coupling relations among different integrated energy subsystems. Secondly, with the help of the weight sharing mechanism in the multi-task learning and the idea of least square support vector machine, a combined forecasting model of electricity, heat, cooling and gas loads based on the multi-task learning and least square support vector machine is constructed. Finally, in order to verify the effectiveness of the forecasting model proposed in this paper, the actual data from the integrated energy system in Suzhou Industrial Park are selected for a case study. The results show that: (1) the combined forecasting model based on multi-task learning and least square support vector machine can accurately predict the electricity, heat, cooling and gas loads of the park integrated energy system. (2) Compared with extreme learning machine and least square support vector machine, the combined forecasting model based on the multi-task learning and least square support vector machine increased the forecasting accuracy of a workday and a weekend by 18.00% and 19.19%, and the average forecasting accuracy increased by 18.60%. (3) Compared with extreme learning machine and least square support vector machine, the combined forecasting model can effectively shorten the training time which is reduced by 58.1% and 35.22%. The results further reflect the application effect of the multi-task learning in energy demand forecasting of the integrated energy system and have a very broad reference value.
With the rapid development of renewable energy, virtual power plant technology has gradually become a key technology to solve the large-scale development of renewable energy. This paper focuses on the stochastic dispatching optimization of gas-electric virtual power plant (GVPP). Based on this, wind power plant, photovoltaic power generation and convention gas turbines are used as the power generation side of GVPP. Power-to-gas (P2G) equipment and gas storage tank can realize the conversion and storage of electricity-gas energy. Price based demand response and incentive based demand response are introduced into the terminal load side to regulate the user’s electricity consumption behavior. GVPP bilaterally connects power network and natural gas network, which realizes the bidirectional flow of electricity-gas energy. Firstly, taking the maximization of economic benefits as the objective function, combined with the constraints of power balance, system reserve and so on, a dispatching optimization model of GVPP participating in multi-energy markets is constructed to determine the operation strategy. Secondly, wind, solar and other clean energy have the characteristics of random and fluctuation, which threaten the stable operation of the system. Therefore, a stochastic dispatching optimization model of GVPP considering wind and solar uncertainty is established based on robust stochastic optimization theory. Thirdly, the evaluation indicators of GVPP operation is determined, which can comprehensively evaluate the economic benefits, environmental benefits and system operation of virtual power plant. Finally, in order to verify the validity and feasibility of the model, a virtual power plant is selected for example analysis. The results show that: (1) After the implementation of price based demand response and incentive based demand response, the system load variance changes from 0.03 to 0.013. Through the comparison of load curves, it is found that demand response can play a role of peak-shaving and valley-filling and smooth the power load curve; (2) Stochastic optimization theory can overcome the uncertainty of wind and solar by setting different robust coefficients Γ which reflects the ability of the system to withstand risks; (3) The optimization effect after introducing the P2G subsystem makes the amount of abandoned clean energy close to zero. The operation risk of system is reduced, and the carbon emissions are reduced by 370 m³ too. The market space is expanded from electricity market mainly to natural gas market and carbon trading market.
Solar energy is becoming one of the most attractive renewable sources. In many cases, due to a wide range of financial or installation limitations, off-grid small scale micro power panels are favoured as modular systems to power lighting in gardens or to be integrated together to power small devices such as mobile phone chargers and distributed smart city facilities and services. Manufacturers and systems’ integrators have a wide range of options of micro-scale photo voltaic panels to choose from. This makes the selection of the right panel a challenging task and risky investment. To address this and to help manufacturers, this paper suggests and evaluates a novel approach based on integrating empirical lab-testing with short-term real data and neural networks to assess the performance of micro-scale photovoltaic panels and their suitability for a specific application in specific environment. The paper outlines the combination of lab testing power output under seasonal and hourly conditions during the year combined with environmental and operating conditions such as temperature, dust accumulation and tilt angle performance. Based on the lab results, a short in-situ experimental work is implemented and the performance over the year in the selected location in Kuwait is evaluated using deep learning neural networks. The findings of this approach are compared with simulation and long-term real data. The results show a maximum error of 23% of the neural network output when compared with the actual data, and a correlation values with previous work within 87.3% and 91.9% which indicate that the proposed approach could provide an experimental rapid and accurate assessment of the expected power output. Hence, supporting the rapid decision-making process for manufacturers and reducing investment risks.
Natural language processing is a technique to process data such as text and speech. Some fundamental research includes named-entity recognition, which recognizes name entities (i.e., persons, companies) from texts; semantic parsing, which is used to convert a natural language utterance to the representation of logical form; and co-reference resolution, which extracts nouns (including pronouns, noun phrases) pointing to the same reference body. In this paper, we mainly focus on the task of mention extraction, which extract and classify overlapping or nested structure mentions. We proposed a neural-encoded mention-hypergraph (NEMH) model to use hypergraph to model overlapping or nested structure mentions and use neural networks to extract features for hypergraph automatically. Unlike the existing approaches, our hypergraph model can effectively capture nested mention entities with unlimited lengths. Also, the proposed model is highly scalable and the time complexity of the proposed model is linear in the number of mention classes and the number of input words. Extensive experiments are conducted on several standard datasets to demonstrate the effectiveness of the proposed model.
Scientific writing is difficult. It is even harder for those for whom English is a second language (ESL learners). Scholars around the world spend a significant amount of time and resources proofreading their work before submitting it for review or publication.
In this paper we present a novel machine learning based application for proper word choice task. Proper word choice is a generalization the lexical substitution (LS) and grammatical error correction (GEC) tasks. We demonstrate and evaluate the usefulness of applying bidirectional Long Short Term Memory (LSTM) tagger, for this task. While state-of-the-art grammatical error correction uses error-specific classifiers and machine translation methods, we demonstrate an unsupervised method that is based solely on a high quality text corpus and does not require manually annotated data. We use a bidirectional Recurrent Neural Network (RNN) with LSTM for learning the proper word choice based on a word’s sentential context. We demonstrate and evaluate our application in various settings, including both a domain-specific (scientific), writing task and a general-purpose writing task. We perform both strict machine and human evaluation. We show that our domain-specific and general-purpose models outperform state-of-the-art general context learning. As an additional contribution of this research, we also share our code, pre-trained models, and a new ESL learner test set with the research community.
An accurate Electricity Price Forecasting (EPF) plays a vital role in the deregulated energy markets and has a specific effect on optimal management of the power system. Considering all the potent factors in determining the electricity prices - some of which have stochastic nature - makes this a cumbersome task. In this paper, first, Grey Correlation Analysis (GCA) is applied to select the effective parameters in EPF problem and eliminate redundant factors based on low correlation grades. Then, a deep neural network with Stacked Denoising Auto-Encoders (SDAEs) has been utilized to denoise data sets from different sources individually. After that, to detect the main features of the input data and putting aside the unnecessary features, Dimension Reduction (DR) process is implemented. Finally, the rough structure Artificial Neural Network (ANN) has been executed to forecast the day-ahead electricity price. The proposed method is implemented on the data of Ontario, Canada, and the forecasted results are compared with different structures of ANN, Support Vector Machine (SVM), Long Short-Term Memory (LSTM) - benchmarking methods in this field- and forecasting data reported by Independent Electricity System Operator (IESO) to evaluate the efficiency of the proposed approach. Furthermore, the results of this study indicate that the proposed method is efficient in terms of reducing error criterion and improves the forecasting error about 5 to 10 percent in comparison with IESO. This is a remarkable achievement in EPF field.
Nowadays, real-time air pollution monitoring has been an important approach for supporting pollution control and reduction. However, due to the high construction cost and limited detection range of monitoring stations, not all the air pollutant concentrations in every corner can be monitored, and a whole picture of the spatial distribution of air pollution is usually lacked for comprehensive spatial analysis and air quality control. To address this problem, satellite remote sensing and spatial interpolation/extrapolation technologies have been commonly used in past research. However, the spatial distribution calculated by remote sensing techniques could be less accurate due to the limited amount of recorded data for testing and adjustments. Performance of traditional spatial interpolation/extrapolation techniques, such as Kriging and IDW, was limited by several subjective assumptions and pre-set formulations that are less suitable for non-linear real-world situations. As an alternative, machine learning and neural network-based methods have been proposed recently. However, most of these methods failed to well consider the long short temporal trend and spatial associations of air pollution simultaneously. To overcome these limitations, this paper proposed a newly designed spatial interpolation/extrapolation methodology namely Geo-LSTM to generate the spatial distribution of air pollutant concentrations. The model was developed based on the Long Short-Term Memory (LSTM) neural network to capture the long-term dependencies of air quality. A geo-layer was designed to integrate the spatial-temporal correlation from other monitoring stations. To evaluate the effectiveness of the proposed methodology, a case study in Washington state was conducted. The experimental results show that Geo-LSTM has a RMSE of 0.0437, and is almost 60.13% better than traditional methods like IDW.
To unlock the potential of flexible resources, a multi-time-scale economic scheduling strategy for the virtual power plant (VPP) to participate in the wholesale energy and reserve market considering large quantity of deferrable loads (DLs) aggregation and disaggregation is proposed in this paper. For the VPP multi-time-scale scheduling including day-ahead bidding and real-time operation, the following models are proposed, namely, DLs aggregation model based on clustering approach, economic scheduling model considering DLs aggregation, and DLs disaggregation model satisfying consumers' requirements, respectively. The proposed strategy can realize the efficient management of massive DLs to reduce the energy management complexity and increase the overall economics with high computation efficiency, which indicate its promising application in the VPP economic scheduling.
Renewable energy sources (RESs) and energy storage systems (ESSs) are the key technologies for smart grid applications and provide great opportunities to de-carbonize urban areas, regulate frequency, voltage deviations, and respond to severe time when the load exceeds the generation. Nevertheless, uncertainty and inherent intermittence of renewable power generation units impose severe stresses on power systems. Energy storage systems such as battery energy storage system enables the power grid to improve acceptability of intermittent renewable energy generation. To do so, a successful coordination between renewable power generation units, ESSs and the grid is required. Nonetheless, with the existing grid architecture, achieving the aforementioned targets is intangible. In this regard, coupling renewable energy systems with different generation characteristics and equipping the power systems with the battery storage systems require a smooth transition from the conventional power system to the smart grid. Indeed, this coordination requires not only robust but also innovative controls and models to promote the implementation of the next-generation grid architecture. In this context, the present research proposes a smart grid architecture depicting a smart grid consisting of the main grid and multiple embedded micro-grids. Moreover, a focus has been given to micro-grid systems by proposing a “Micro-grid Key Elements Model” (MKEM). The proposed model and architecture are tested and validated by virtualization. The implementation of the virtualized system integrates solar power generation units, battery energy storage systems with the proposed grid architecture. The virtualization of the proposed grid architecture addresses issues related to Photovoltaic (PV) penetration, back-feeding, and irregularity of supply. The simulation results show the effect of Renewable Energy (RE) integration into the grid and highlight the role of batteries that maintain the stability of the system.
Nonlinear systems widely exist in the real world. Researches on the synchronization and identification of nonlinear systems have both theoretical and practical interests. However, since the pseudo-random and parameter sensitivity, some nonlinear dynamics, such as chaotic systems, are difficult to achieve the required regression approximation. By means of the fuzzy-neural structure and long short-term memory (LSTM) mechanism, this paper proposes a novel inference structure of the self-evolving interval type-2 fuzzy LSTM-neural network (eIT2FNN-LSTM) for the synchronization and identification of nonlinear dynamics. In order to tackle with the time-dependency data generated by nonlinear systems, recurrent neural fuzzy systems with LSTM structure is introduced into type-2 fuzzy neural networks, where, by means of gate mechanism, a recurrent structure in the time dimension by feeding the rule firing strength of each rule back to itself is implemented. Besides, an online rule generation algorithm based on dynamic density clustering is utilized to achieve structural updates, which can meet the need of high frequency data process in reality. By incorporating direct adaptive interval type-2 LSTM fuzzy control scheme and sliding mode approach, two chaotic systems with external disturbance noise or/and random-varying parameters can be synchronized based on Lyapunov stability criterion, where two methods, i.e., the gradient training and particle swarm optimization (PSO) algorithm are used. In order to further verify the universality of the proposed scheme, besides nonlinear system, real-life datasets are also utilized to verify the effectiveness of the proposed fuzzy-neural inference system.
This paper proposes a stochastic scheduling model to determine optimal operation of generation and storage units of a virtual power plant (VPP) for participating in a joint energy and regulation service (RS) market under uncertainty. Beside electricity, the VPP provides required RSs according to the probability of delivery request in the electricity market. A new model for providing RS is introduced in which the dispatchable generation units are financially compensated with their readiness declarations and will be charged/paid for their real‐time down/up regulations. Besides, the VPP sets up incentive price‐quantity curves to benefit from the potential of demand side management in both energy and RS market. Within the model presented here, the VPP consists of two types of generation units: wind turbine and standby diesel generator; the latter is modeled by considering CO2‐emission penalty costs. The given uncertainties are divided into two parts. Firstly, the uncertainties from the energy market price are simulated using information gap decision theory to evaluate the risk‐based resource scheduling for both risk‐taker and risk‐averse VPP. Other uncertainties affecting decision making such as wind turbine generation, load, regulation up/down calling probabilities, and regulation market prices are modeled via scenario trees. Three typical case studies are implemented to validate the performance and effectiveness of the proposed scheduling approach.
The reduction of global greenhouse gas emissions is one of the key steps towards sustainable development. The integration of Distributed Energy Resources (DERs) in power systems will help with emissions reduction. Virtual Power Plants (VPPs) can overcome barriers to participation of DERs in system operation. In this paper, a model is proposed for the energy management of a VPP including PhotoVoltaic (PV) modules, wind turbines, Electrical Energy Storage (EES) systems, Combined Heat and Power (CHP) units, and heat-only units. The multi-objective operational scheduling of DERs in the VPP focuses on maximizing the expected day-ahead profit of the VPP and minimizing the expected day-ahead emissions. The uncertainty of wind speed, solar radiation, market price, and electrical load is modeled using scenario based approach. Also, two-stage stochastic programming is implemented for modeling the VPP energy management. Three cases have been investigated for evaluating the proposed method: single-objective scheduling of VPP to maximize profit, single-objective scheduling of VPP to minimize emission and multi-objective economic/emission scheduling of VPP. The results indicate the appropriate economic and environmental performance of the proposed method, which provides the possibility of selecting a compromise solution for the VPP operator in accordance with environmental restrictions and economic constraints.
Modern engineered systems generally work under complex operational conditions. However, most of the existing artificial intelligence (AI)-based prognostic methods still lack an effective model that can utilize operational conditions data for remaining useful life (RUL) prediction. This paper develops a novel prognostic method based on Bi-Directional Long Short-Term Memory (BLSTM) networks. The method can integrate multiple sensors data with operational conditions data for RUL prediction of engineered systems. The proposed architecture based on BLSTM networks includes three main parts: 1) One BLSTM network is used to directly extract features hidden in the multiple raw sensors signals; 2) another BLSTM network is employed to learn higher features from operational conditions signals and the learned features from the sensors signals; 3) fully connected layers and a linear regression layer are stacked to generate the target output of the RUL prediction. Unlike other AI-based prognostic methods, the developed method can simultaneously model both sensors data and operational conditions data in a consolidated framework. The proposed approach is demonstrated through a case study on aircraft turbofan engines, and comparisons with other existing state-of-the-art methods are also presented.
The need for frequency regulation capacity increases as the fraction of renewable energy sources grows in the electricity market. An aggregator can provide frequency regulation by controlling its generation and demand. Here we investigate the participation of an aggregator controlling a fleet of electric vehicles (EVs) and an energy storage (ES) in day-ahead regulation and energy markets and determine the optimal size of the aggregator's bids. The problem is formulated as a stochastic mixed integer linear programming model, taking into account the uncertainties regarding energy and frequency regulation prices. The risks associated with the uncertainties are managed using the conditional value-at-risk method. Because most EVs are charged in residential distribution networks, load flow constraints are also taken into account. A formulation based on the rainflow cycle counting algorithm is proposed to include the ES degradation costs incurred from following the frequency regulation signals into the objective function. The problem is studied within the context of a low voltage distribution network adopting the market rules of the California Independent System Operator. The results of the numerical analysis show how joint optimization of EVs and ES can improve the aggregator's profit, and verify that the proposed degradation cost formulation can effectively minimize the degradation costs of the ES.
Energy Storage Scheme (ESS) is of great importance to realize energy management and to optimally utilize Renewable Energy (RE) integration in the electricity system. An increasing exploitation of RE in electricity system raises the concern about the need for Ancillary Services (AS) in a power system. These services are required for maintaining the reliability and security of the supply. This paper proposes a market-based participation of ESS to support large-scale RE penetration for the procurement of energy and AS using Virtual Power Plant (VPP) in a deregulated environment. The proposed VPP consists of a pumped-storage system as one of the recognized ESS and Renewable Power Producers (RPPs). This optimization problem is formulated and solved using an optimal power flow technique which considers network constraints and power flow limits. Spinning Reserve (SR) as one of the main AS is considered in this paper, which is procured under Spinning Reserve Market (SRM). The ability of the proposed approach to provide both energy and SR is tested on 3 case studies and demonstrated by considering a modified IEEE-30 bus test system. Results show that the VPPs can play a significant role in increasing the penetration of RE for the procurement of energy and AS.
This paper proposes a novel model for the day-ahead self-scheduling problem of a virtual power plant trading in both energy and reserve electricity markets. The virtual power plant comprises a conventional power plant, an energy storage facility, a wind power unit, and a flexible demand. This multi-component system participates in energy and reserve electricity markets as a single entity in order to optimize the use of energy resources. As a salient feature, the proposed model considers the uncertainty associated with the virtual power plant being called upon by the system operator to deploy reserves. Additionally, uncertainty in available wind power generation and requests for reserve deployment is modeled using confidence bounds and intervals, respectively, while uncertainty in market prices is modeled using scenarios. The resulting model is thus cast as a stochastic adaptive robust optimization problem which is solved using a column-and-constraint generation algorithm. Results from a case study illustrate the effectiveness of the proposed approach. IEEE
In recent years, due to lack of sufficient quantity of fossil fuel, the need of Renewable Energy Sources (RESs) has become an important matter. In addition to the shortage of fossil fuel, global warming is another concern for many countries and companies. These issues have caused a large number of RESs to be added into modern distribution systems. Nevertheless, the high penetration of RESs beside the intermittent nature of some resources such as Wind Turbines (WT) and Photovoltaic (PV) cause the variable generation and uncertainty in power system. Under this condition, an idea to solve problems due to the variable outputs of these resources is to aggregate them together. A collection of Distributed Generators (DGs), Energy Storage Systems (ESSs) and controllable loads that are aggregated and then are managed by an Energy Management System (EMS) which is called Virtual Power Plant (VPP). The objective of the VPP in this paper is to minimize the total operating cost, considering energy loss cost in a 24h time interval. To solve the problem, Imperialist Competitive Algorithm (ICA), a meta-heuristic optimization algorithm is proposed to determine optimal energy management of a VPP with RESs, Battery Energy Storage (BSS) and load control in a case study.
In this paper, a fully distributed approach is proposed for a class of virtual power plant (VPP) problems. By characterizing two specific VPP problems, we first give a comprehensive VPP formulation that maximizes the economic benefit subjected to the power balance constraint, line transmission limits and local constraints of all distributed energy resources (DERs). Then, utilizing the alternating direction method of multipliers and consensus optimization, a distributed VPP dispatch algorithm is developed for the general VPP problem. In particular, Theorem 1 is derived to show the convergence of the algorithm. The proposed algorithm is completely distributed without requiring a centralized controller, and each DER is regarded as an agent by implementing local computation and only communicates information with its neighbors to cooperatively find the globally optimal solution. The algorithm brings some advantages, such as the privacy protection and more scalability than centralized control methods. Furthermore, a new variant of the algorithm is presented for improving the convergence rate. Finally, several case studies are used to illustrate the efficiency and effectiveness of the proposed algorithms.
A novel distributed optimal dispatch algorithm is proposed for coordinating the operation of multiple micro units in a microgrid, which has incorporated the distributed consensus algorithm in multi-agent systems and the λ-iteration optimization algorithm in the economic dispatch of power systems. Specifically, the proposed algorithm considers the global active power constraint by adding a virtual pinner and it can deal with the optimization problem with any initial states. That is, it can realize the global optimization and avoid the defect of the initial conditions' sensitivity in the optimization problem. On the other hand, the proposed optimization algorithm can either be used for off-line calculation or be utilized for online operation and has the ability to survive single-point failures and shows good robustness in the iteration process. Numerical studies in a seven-bus microgrid demonstrate the effectiveness of the proposed algorithm.
Conditional value at risk (CVaR) and confidence degree theory are introduced to build scheduling model for VPP connecting with wind power plant (WPP), photovoltaic generators (PV), convention gas turbine (CGT), energy storage systems (ESSs) and incentive-based demand response (IBDR). Latin hypercube sampling method and Kantorovich distance are introduced to construct uncertainties analysis method. A risk aversion scheduling model is proposed with minimum CVaR objective considering maximum operation revenue. The IEEE30 bus system is used as simulation system. Results show: (1) Price-based demand response could realize peak load shifting, ESSs and IBDR could increase operation revenue. (2) Threshold α reflects risk attitude of decision maker, which has strong risk tolerant to gain the excess income with low α. (3) In peak period, decision maker would reduce WPP and PV for avoiding power shortage loss. Otherwise, WPP and PV would be called in priority since system reserve capacity is sufficient. (4) When 0.85≤β < 0.95, the decreasing slope of CVaR value is big, decision maker is sensitive on risk. When β≥0.95, VPP scheduling scheme reach the most conservative, net revenue and CVaR value are ¥8995.34 and ¥18834. Therefore, the proposed model could describe VPP risk and provide decision support tool for decision maker.
This paper presents a day-ahead scheduling framework for virtual power plant (VPP) in a joint energy and regulation reserve (RR) markets. The proposed VPP clusters a mix of generation units in term of synchronous distributed generation (SDG) and wind power plant (WPP) as well as storage facilities such as electrical vehicles (EVs) and small pumped storage plant (PSP). It is assumed that VPP provides required RR through its SDG and small PSP based on the delivery request probability of day-ahead market. In order to aggregate EVs, the VPP establishes bilateral incentive contracts with vehicle owners. Moreover, impact of carbon dioxide (CO2) emission of SDG is included by means of penalty cost function. Different uncertain parameters with regard to wind generation, EV owner behaviors, energy and RR market prices and regulation up and down probabilities are considered using a point estimate method (PEM). The case studies are applied to demonstrate the effectiveness of the scheduling model.
Virtual Power Plant (VPP) is introduced as a tool for the integration of distributed generations, energy storages and participation of consumers in demand response programs. In this paper, a probabilistic model using a modified scenario-based decision making method for optimal day ahead scheduling of electrical and thermal energy resources in a VPP is proposed. In the proposed model, energy and reserve is simultaneously scheduled and the presence of energy storage devices and demand response resources are also investigated. Moreover, the market prices, electrical demand and intermittent renewable power generation are considered as uncertain parameters in the model. A modified scenario-based decision making method is developed in order to model the uncertainties in VPP's scheduling problem. The results demonstrated that the optimal scheduling of VPP's resources by the proposed method leads VPP to make optimal decisions in the energy/reserve market and to play a dual role as a demand/generation unit from the perspective of the upstream network in some time periods.
Abstract: Today, traditional networks are changing to active grids due to the burgeoning growth of distributed energy resources (DER), which demands scrupulous attention to technical infrastructures, as well as economic aspects. In this study, from economic point of view, the aggregation of DERs in a distribution network to participate in joint energy and reserve markets is investigated. This approach, which is predicated upon price-based unit commitment method, has considered virtually all the technical data in the proposed model. It is worth to mention that uncertainties of loads and market prices, as an inherent characteristic of the electricity markets, are treated in this study, and their effect on the operation of virtual power plants in energy and reserve markets has been thoroughly discussed. To this end, for both uncertain parameters, a good number of scenarios are generated and using the backward reduction method the number of these scenarios is reduced. The problem is formulated as a MINLP model and is implemented in GAMS software, while its authenticity is validated using particle swarm optimisation method.
To improve the design of the electricity infrastructure and the efficient deployment of distributed and renewable energy sources, a new paradigm for the energy supply chain is emerging, leading to the development of smart grids. There is a need to add intelligence at all levels in the grid, acting over various time horizons. Predicting the behavior of the energy system is crucial to mitigate potential uncertainties. An accurate energy prediction at the customer level will reflect directly in efficiency improvements in the whole system. However, prediction of building energy consumption is complex due to many influencing factors, such as climate, performance of thermal systems, and occupancy patterns. Therefore, current state-of-the-art methods are not able to confine the uncertainty at the building level due to the many fluctuations in influencing variables. As an evolution of artificial neural network (ANN)-based prediction methods, deep learning techniques are expected to increase the prediction accuracy by allowing higher levels of abstraction. In this paper, we investigate two newly developed stochastic models for time series prediction of energy consumption, namely Conditional Restricted Boltzmann Machine (CRBM) and Factored Conditional Restricted Boltzmann Machine (FCRBM). The assessment is made on a benchmark dataset consisting of almost four years of one minute resolution electric power consumption data collected from an individual residential customer. The results show that for the energy prediction problem solved here, FCRBM outperforms ANN, Support Vector Machine (SVM), Recurrent Neural Networks (RNN) and CRBM.
Virtual power plant (VPP) concept was developed to integrate distributed energy resources (DERs) into the grid in order that they are seen as a single power plant by the market and power system operator. Therefore, VPPs are faced with optimal bidding, and identifying arbitrage opportunities in a market environment. In this study, the authors present an arbitrage strategy for VPPs by participating in energy and ancillary service (i.e. spinning reserve and reactive power services) markets. On the basis of a security-constrained price-based unit commitment, their proposed model maximises VPP's profit (revenue minus costs) considering arbitrage opportunities. The supply-demand balancing, transmission network topology and security constraints are considered to ensure reliable operation of VPP. The mathematical model is a mixed-integer non-linear optimisation problem with inter-temporal constraints, and solved by mixed-integer non-linear programming. The result is a single optimal bidding profile and a schedule for managing active and reactive power under participating in the markets. These profile and schedule consider the DERs and network constraints simultaneously, and explore arbitrage opportunities of VPP. Results pertaining to an illustrative example and a case study are discussed.
Distributed energy resources (DERs) can be integrated into a single entity, namely, virtual power plant (VPP). The integration enables them to participate in competitive wholesale electricity markets. According to the different types of uncertainty faced by different DERs, this integration can also act as a risk-hedging mechanism providing a surplus profit. In this paper, using a two-stage stochastic programming approach, risk-averse optimal offering model for a VPP trading in a joint market of energy and spinning reserve service is presented. The conditional value-at-risk (CVaR) is used to control the risk of profit variability. Uncertainties involved in generation of renewables, consumption of loads, calls for reserve service, as well as prices in the day-ahead market, the spinning reserve market (RM) and the balancing (real-time) market (BM), are taken into consideration. This paper assesses how total and surplus profits of VPP are affected by risk-aversion, participation in the RM, and the pricing system in the BM. For this purpose, the role of the integration and the trends of energy and reserve transactions under both single and dual pricing systems in the BM are addressed in detail through a numerical study.
Recurrent Neural Networks can be trained to produce sequences of tokens given
some input, as exemplified by recent results in machine translation and image
captioning. The current approach to training them consists in maximizing the
likelihood of each token in the sequence given the current (recurrent) state
and the previous token. At inference, the unknown previous token is then
replaced by a token generated by the model itself. This discrepancy between
training and inference can yield errors that can accumulate quickly along the
generated sequence. We propose a curriculum learning strategy to gently change
the training process from a fully guided scheme using the true previous token,
towards a less guided scheme which mostly uses the generated token instead.
Experiments on several sequence prediction tasks show that this approach yields
significant improvements.
We introduce Adam, an algorithm for first-order gradient-based optimization
of stochastic objective functions. The method is straightforward to implement
and is based an adaptive estimates of lower-order moments of the gradients. The
method is computationally efficient, has little memory requirements and is well
suited for problems that are large in terms of data and/or parameters. The
method is also ap- propriate for non-stationary objectives and problems with
very noisy and/or sparse gradients. The method exhibits invariance to diagonal
rescaling of the gradients by adapting to the geometry of the objective
function. The hyper-parameters have intuitive interpretations and typically
require little tuning. Some connections to related algorithms, on which Adam
was inspired, are discussed. We also analyze the theoretical convergence
properties of the algorithm and provide a regret bound on the convergence rate
that is comparable to the best known results under the online convex
optimization framework. We demonstrate that Adam works well in practice when
experimentally compared to other stochastic optimization methods.