PreprintPDF Available

Abstract and Figures

The increased digitalisation and monitoring of the energy system opens up numerous opportunities % and solutions which can help to decarbonise the energy system. Applications on low voltage (LV), localised networks, such as community energy markets and smart storage will facilitate decarbonisation, but they will require advanced control and management. Reliable forecasting will be a necessary component of many of these systems to anticipate key features and uncertainties. Despite this urgent need, there has not yet been an extensive investigation into the current state-of-the-art of low voltage level forecasts, other than at the smart meter level. This paper aims to provide a comprehensive overview of the landscape, current approaches, core applications, challenges and recommendations. Another aim of this paper is to facilitate the continued improvement and advancement in this area. To this end, the paper also surveys some of the most relevant and promising trends. It establishes an open, community-driven list of the known LV level open datasets to encourage further research and development.
Content may be subject to copyright.
Mon Tue Wed Thu Fri Sat Sun
0
2
4
Day of Week
0
5
10
0
20
40
0
50
0
50
100
0
500
Consumption (kW)
(a) 500
(b) 100
(c) 50
(d) 25
(e) 5
(f) 1
Aggregation
500
100
Count
1962
1973
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
2014
2016
2018
2020
Count
IEEE Transactions on Smart
IEEE Transactions on Power
Energies
Electric Power Systems
Applied Energy
Energy
IET Generation, Transmission
Energy and Buildings
IEEE Access
Sustainable Energy, Grids and
International Journal of Electrical
42%
56%
lpp
p
q
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Management and efficient operations in critical infrastructures such as smart grids take huge advantage of accurate power load forecasting, which, due to its non-linear nature, remains a challenging task. Recently, deep learning has emerged in the machine learning field achieving impressive performance in a vast range of tasks, from image classification to machine translation. Applications of deep learning models to the electric load forecasting problem are gaining interest among researchers as well as the industry, but a comprehensive and sound comparison among different—also traditional—architectures is not yet available in the literature. This work aims at filling the gap by reviewing and experimentally evaluating four real world datasets on the most recent trends in electric load forecasting, by contrasting deep learning architectures on short-term forecast (one-day-ahead prediction). Specifically, the focus is on feedforward and recurrent neural networks, sequence-to-sequence models and temporal convolutional neural networks along with architectural variants, which are known in the signal processing community but are novel to the load forecasting one. © 2021 The Authors. CAAI Transactions on Intelligence Technology published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology and Chongqing University of Technology.
Article
Full-text available
Forecasting has been an essential part of the power and energy industry. Researchers and practitioners have contributed thousands of papers on forecasting electricity demand and prices, and renewable generation (e.g., wind and solar power). This paper offers a brief review of influential energy forecasting papers; summarizes research trends; discusses importance of reproducible research and points out six valuable open data sources; makes recommendations about publishing high-quality research papers; and offers an outlook into the future of energy forecasting.
Article
Full-text available
The daily analysis of loads is one of the most important activities for power utilities in order to be able to meet the energy demand. This analysis not only includes short-term forecasting but it also encompasses the completion of missing load data, known as imputation. In this work we show that adding information of attached or bordering primary substation helps to improve the prediction accuracy in a single substation, since its neighbours may share common weather-related (e.g. temperature, humidity, wind direction, etc.) and human-related (e.g. work-calendar, holidays, cultural consumption patterns, etc.) data. In order to validate these approaches, we test the forecasting and imputation neighboring methodology on a wide variety of datasets. Results confirm that, given a primary substation, the addition of information from surrounding substations does improve the forecasting and imputation errors.
Article
This paper presents a novel extension of the classic nonintrusive load monitoring (NILM) problem from household-appliance level to substation level. A new three-stage regional-NILM method is proposed to deduce the states of different types of loads in a region by disaggregating its substation demand. Three types of loads are considered in this study: (i) traditional loads; (ii) distributed generation such as photovoltaics (PVs); and (iii) flexible loads like electric vehicles (EVs). The proposed method firstly forecasts the traditional load using the long-term historical data and employing spectral analysis to boost the signal-to-noise ratio. Secondly, the PV capacity is deduced by performing peak coincidence analysis between negative residuals and local solar irradiance data. Finally, a novel limited activation matching pursuit method is proposed to estimate the states of the EVs, including the total EV load and number of EVs. The method is assessed on real data collected from 800 substations, 10 PVs and 50 EVs in the UK. Results show the proposed method for estimating the number of EVs outperforms the approaches based on sparse coding, orthogonal matching pursuit and non-negative matching pursuit by 16.5%, 10.2% and 10.0%, respectively. The proposed Regional-NILM solution provides a cost-effective way for distribution network operators to understand the network’s state. It can therefore significantly increase the network visibility without requiring expensive monitoring and avoiding data privacy issues. As such, it can improve the efficiency of demand side management, which is required to accommodate the future large number of distributed energy resources connections.
Article
We present a comparative study of different probabilistic forecasting techniques on the task of predicting the electrical load of secondary substations and cabinets located in a low voltage distribution grid, as well as their aggregated power profile. The methods are evaluated using standard KPIs for deterministic and probabilistic forecasts. We also compare the ability of different hierarchical techniques in improving the bottom level forecasters’ performances. Both the raw and cleaned datasets, including meteorological data, are made publicly available to provide a standard benchmark for evaluating forecasting algorithms for demand-side management applications.
Article
Load forecasting is essential for different activities on power systems, and there is extensive research on approaches for forecasting in different time horizons, from next-hour to decades. However, because of higher uncertainty and variability compared to aggregated or medium and high voltage, the forecasting of the individual household load is a current challenge. This paper presents a load forecasting for multiple households using Bayesian networks. Our model, which is multivariate, uses past consumption, temperature, socioeconomic and electricity usage aspects to forecast the next instant household load value. It was tested using real data from the Irish smart meter project and its performance was compared with other forecasting methods. Results have shown that the proposed approach provides consistent single forecast model for hundreds of households with different consumption patterns, showing a generalisation capability in an efficient manner.
Article
Nowadays, the uncertainty in distribution systems rises, notably due to an increasing share of solar panels and electric vehicles whose power production and consumption are characterized by a high volatility. This poses challenges to distribution system operators to ensure stable and secure operation of their grid. Hence, an optimal integration of these distributed energy resources in real-time control schemes inevitably relies on appropriate forecasts of the near-future system state. This paper investigates the short-term probabilistic state prediction of low-voltage grids for operation purposes. The performance of two quantile forecasting algorithms is evaluated for different levels of distributed energy resources penetration and availability of measurements. Quantile forecasts are finally integrated into the framework of an optimization problem that aims at minimizing the costs associated with overvoltages by suitable solar power curtailment. The advantages of quantile forecasts considering different imbalance prices are demonstrated.
Article
Reducing peak demand is an important cost-saving measure for small and medium enterprises (SMEs) because electricity tariff menus often include a demand charge determined by the yearly highest demand. SMEs are incentivized to reduce the peak demand; thus, information provision services that are suitable for a wide range of SMEs and send alerts about the possibility of exceeding contract demand are needed. We developed a demand forecasting method that incorporated a modified version of support vector regression using only smart meter data and actual weather data as input. We assumed that peak demand alerts are sent to each SME when the forecasted demand exceeds the predefined precaution threshold. The proposed method also has a parameter for intervals of forecasted demand, which controls trade-off between recall and precision of the alerts. Using smart meter data from 273 SMEs, we evaluated the performance of the alerts. Recall was 75.4% for the 1-h-ahead point forecast and 86.9% for the 24-h-ahead interval forecast in one of the best cases.