Reinforcement learning (RL) is a promising optimal control technique for multi-energy management systems. It does not require a model a priori - reducing the upfront and ongoing project-specific engineering effort and is capable of learning better representations of the underlying system dynamics. However, vanilla RL does not provide constraint satisfaction guarantees - resulting in various potentially unsafe interactions within its environment. In this paper, we present two novel online model-free safe RL methods, namely SafeFallback and GiveSafe, where the safety constraint formulation is decoupled from the RL formulation. These provide hard-constraint satisfaction guarantees both during training and deployment of the (near) optimal policy. This is without the need of solving a mathematical program, resulting in less computational power requirements and more flexible constraint function formulations. In a simulated multi-energy systems case study we have shown that both methods start with a significantly higher utility compared to a vanilla RL benchmark and Optlayer benchmark (94,6% and 82,8% compared to 35,5% and 77,8%) and that the proposed SafeFallback method even can outperform the vanilla RL benchmark (102,9% to 100%). We conclude that both methods are viably safety constraint handling techniques applicable beyond RL, as demonstrated with random policies while still providing hard-constraint guarantees.
The paper describes a simple and efficient control loop decoupler for TITO (two inputs, two outputs) control systems. The inverted decoupling approach is used, and the feedforward filters are just static gains. The feedforward gains are determined in two ways, from a static analysis and from optimization. By investigating several cases found in literature, it is found that the two sets of feedforward gains are very close. The control loop decoupler is implemented in an industrial DCS system and simulation experiments as well as laboratory tests show that a significant reduction of the coupling is accomplished. Automatic tuning of the feedforward gains is derived and implemented in the system as well.
Environmental driven regulations are significantly affecting shipping in recent years, where the shipbuilding industry is required to comply with upcoming restrictions concerning polluting emissions. The all-electric ship (AES) is one of the most promising technologies for complying with the increasingly strict environmental regulations, improving fuel efficiency, and enhancing system dynamic performance. In this study, the DC-distributed power grid of an AES integrated with fuel cells and batteries has been configured using extensive electrification technology, where the system-level shipboard power plant has been modelled with the average modelling method. The model not only incorporates the hybrid power source integration but also the primary and secondary power management as a whole. In addition, a hardware-in-the-Loop (HIL) has been set up to replicate the real-time system behaviour, which is essential for the verification of any optimal power management control algorithms to be developed in future work. Finally, both the mathematical and real-time models are validated against the full-scale hybrid shipboard power system.
Characterized by ultra-fast and arc-free fault isolation, solid state circuit breakers (SSCBs) are getting increasing popularity with the latest development of advanced power semiconductor devices, like the silicon carbide (SiC) MOSFETs. On the other hand, the mature and cost-effective thyristor technology exhibits the beneficial features of low conduction losses, high surge current capability, bidirectional voltage blocking capability, etc. making it a very attractive candidate for SSCB development. However, the conventional thyristors (silicon controlled rectifiers) have no current turn-off capability and need some auxiliary circuits (e.g. LC resonant circuit) or special topology (e.g. Z-source breakers) to be applied in SSCBs. Meanwhile, some active turn-off thyristors (like integrated gate-commutated thyristors, emitter turn-off thyristors, etc.) are specially modified to have current turn off capability, and better suitable for SSCBs. This paper reviews and studies the SSCBs based on the conventional or active turn-off thyristors. The design challenges and performance comparison of SSCBs with different thyristor technologies are also discussed.
Model predictive control (MPC) offers an optimal control technique to establish and ensure that the total operation cost of multi-energy systems remains at a minimum while fulfilling all system constraints. However, this method presumes an adequate model of the underlying system dynamics, which is prone to modelling errors and is not necessarily adaptive. This has an associated initial and ongoing project-specific engineering cost. In this paper, we present an on- and off-policy multi-objective reinforcement learning (RL) approach that does not assume a model a priori, benchmarking this against a linear MPC (LMPC — to reflect current practice, though non-linear MPC performs better) - both derived from the general optimal control problem, highlighting their differences and similarities. In a simple multi-energy system (MES) configuration case study, we show that a twin delayed deep deterministic policy gradient (TD3) RL agent offers the potential to match and outperform the perfect foresight LMPC benchmark (101.5%). This while the realistic LMPC, i.e. imperfect predictions, only achieves 98%. While in a more complex MES system configuration, the RL agent’s performance is generally lower (94.6%), yet still better than the realistic LMPC (88.9%). In both case studies, the RL agents outperformed the realistic LMPC after a training period of 2 years using quarterly interactions with the environment. We conclude that reinforcement learning is a viable optimal control technique for multi-energy systems given adequate constraint handling and pre-training, to avoid unsafe interactions and long training periods, as is proposed in fundamental future work.
The Alpine ski-resorts are conventionally powered by the electrical distribution network to perform the snow making operations throughout the ski slopes. On one hand, these grids can present a low capacity, on the other the line lengths are not negligible due to natural obstacles and altitude changes. As a consequence, the voltage drops are remarkable when several power loads are fed by the low voltage section. Basing on this aspect, a novel evolution of ski-resorts power grids towards microgrids is pursued to improve the power system quality of service and to foster the implementation of Distributed Energy Resources (DERs). The paper examines the AC electrical distribution system installed in the ski-resort of San Vito di Cadore, Italy. A preliminary power flow analysis identifies bottlenecks on voltage profiles and weights effective solutions to evolve the power grid. Four system configurations are contrasted in terms of voltage drop in order to establish the most convenient redesign. A proper voltage profile along the entire ski-resort is ensured by the last solution which redistribute the loads powering among the five resorts substations. Finally, a techno-economical-environmental analysis explores the possibility of installing DER and storage to feed the resort loads. The study demonstrates the capability of hydro power plant in enhancing microgrid sustainability and environmental friendliness.
Wireless control of modular multilevel converter (MMC) submodules can benefit from different points of view, such as lower converter cost and shorter installation time. In return for the advantages, the stochastic performance of wireless communication networks necessitates an advanced converter control system immune to the losses and delays of the wirelessly transmitted data. This paper proposes an advancement to the distributed control of MMCs to utilize in wireless submodule control. Using the proposed method, the operation of the MMC continues smoothly and uninterruptedly during wireless communication errors. The previously proposed submodule wireless control concept relies on implementing the modulation and individual submodule-capacitor-voltage control in the submodules using the insertion indices transmitted from a central controller. This paper takes the concept as a basis and introduces to synthesize the indices autonomously in the submodules during the communication errors. This new approach allows the MMC continue its operation when one, some, or all submodules suffer from communication errors for a limited time. The proposal is validated experimentally on a laboratory-scale MMC.
Transportation electrification is undergoing a significant transitioning towards the utilization of efficient and reliable energy sources and smart integration schemes, where this transitioning is continuously facing ever-tightening challenges in order to comply with the increased environmental regulations. Among the different means of transportation, the global maritime transport is responsible for 2–3% of global greenhouse gas (GHG) emissions and it is predicted to increase to 17% by 2050 if no changes are adapted. Hence, the international maritime organization (IMO) has targeted to reach a 50% reduction in GHG emissions by 2050 compared to 2008. Hence, alternative energy sources shall be utilized in order to meet these strict GHG emissions reduction targets, where batteries and hydrogen-fed fuel cells can play a vital role in such aspect. Since the output of these two energy sources is unregulated DC voltage, their connection to the whole ship power system can be accomplished in several ways, where each way has its features, in addition to utilizing different power conditioning stages (PCSs), and these features are not well clarified and compared in the literature. Hence, this paper presents an overview of the possible integration schemes that can be utilized in fuel cells and batteries-fed vessels, which is supported with a comparative assessment. This is also presented along with highlighting the state-of-the-art PCSs that are available in the market and can be utilized in these integration schemes within marine vessels. Such overview and comparative assessment are seen to be of significant importance and added value for researchers and developers in both the academic and industrial sides in order to accelerate the adoption of fuel cells in marine systems for zero emission shipping.
Companies are increasingly expanding into overseas operations, often in countries where International Electrotechnical Commission (IEC) standards are common. For the simplicity, commonality, and increased reliability of operations as well as economies of scale in procurement, organizations prefer to buy common products across all operations.
An ageing condition monitoring method for the aluminum electrolytic capacitor (AEC) is presented. The target AEC is used in an ac-dc-ac variable speed drive (VSD) which usually contains an LCR network composed of a dc choke, an AEC and a braking branch. The condition of the AEC is monitored through the capacitance value and the equivalent series resistor (ESR) value. The capacitance value and the ESR value are identified by the LCR network response i.e. magnitude gain and phase shift of the network. The existing components of the VSD are fully utilized instead of adding additional current sensors and active devices. Some potential disturbances in field application such as three-phase voltage unbalances are also considered. Experiments are carried out in a 10-kW general-purpose VSD with different AECs. The maximum errors of capacitance value and ESR values identifications are 1.1% and 5.48%, respectively.
Institution pages aggregate content on ResearchGate related to an institution. The members listed on this page have self-identified as being affiliated with this institution. Publications listed on this page were identified by our algorithms as relating to this institution. This page was not created or approved by the institution. If you represent an institution and have questions about these pages or wish to report inaccurate content, you can contact us here.