Search

Traditionally, models pooling flexible demand and generation units into Virtual Power Plants have been solved via separated approaches, decomposing the problem into parts dedicated to market clearing and separate parts dedicated to managing the state-constraints. The reason for this is the high computational complexity of solving dynamic, i.e. multi-stage, problems under competition. Such approaches have the downside of not adequately modeling the direct competition between these agents over the entire considered time period. This paper approximates the decisions of the players via ‘actor networks’ and the assumptions on future realizations of the uncertainties as ‘critic networks’, approaching the tractability issues of multi-period optimization and market clearing at the same time. Mathematical proof of this solution converging to a Nash equilibrium is provided and supported by case studies on the IEEE 30 and 118 bus systems. Utilizing this approach, the framework is able to cope with high uncertainty spaces extending beyond traditional approximations such as scenario trees. In addition, the paper suggests various possibilities of parallelization of the framework in order to increase computational efficiency. Applying this process allows for parallel solution of all time periods and training the approximations in parallel, a problem previously only solved in succession. © 2020 The Author(s)

An accurate solar energy forecast is of utmost importance to allow a higher level of integration of renewable energy into the controls of the existing electricity grid. With the availability of data in unprecedented granularities, there is an opportunity to use data-driven algorithms for improved prediction of solar generation. In this paper, an improved generally applicable stacked ensemble algorithm (DSE-XGB) is proposed utilizing two deep learning algorithms namely artificial neural network (ANN) and long short-term memory (LSTM) as base models for solar energy forecast. The predictions from the base models are integrated using an extreme gradient boosting algorithm to enhance the accuracy of the solar PV generation forecast. The proposed model was evaluated on four different solar generation datasets to provide a comprehensive assessment. Additionally, the shapely additive explanation framework was utilized in this study to provide a deeper insight into the learning mechanism of the algorithm. The performance of the proposed model was evaluated by comparing the prediction results with individual ANN, LSTM, and Bagging. The proposed DSE-XGB method exhibits the best combination of consistency and stability on different case studies irrespective of the weather variations and demonstrates an improvement in R2 value of 10%–12% to other models.

Solar energy constitutes an effective supplement to traditional energy sources. However, photovoltaic power generation (PVPG) is strongly weather-dependent, and thus highly intermittent. High-precision forecasting of PVPG forms the basis of the production, transmission, and distribution of electricity, ensuring the stability and reliability of power systems. In this work, we propose a deep learning based framework for accurate PVPG forecasting. In particular, taking advantage of the long short-term memory (LSTM) network in solving sequential-data based regression problems, this paper considers the specific domain knowledge of PV and proposes a physics-constrained LSTM (PC-LSTM) to forecast the hourly day-ahead PVPG. It aims to overcome the shortcoming of recent machine learning algorithms that are applied based only on massive data, and thus easily producing unreasonable forecasts. Real-life PV datasets are adopted to evaluate the feasibility and effectiveness of the models. Sensitivity analysis is conducted for the selection of input feature variables based on a two-stage hybrid method. The results indicate that the proposed PC-LSTM model possesses stronger forecasting capability than the standard LSTM model. It is more robust against PVPG forecasting, and more suitable for PVPG forecasting with sparse data in practice. The PC-LSTM model also demonstrates superior performance with higher accuracy of PVPG forecasting compared to conventional machine learning and statistical methods.

Recent research has seen several forecasting methods being applied for heat load forecasting of district heating networks. This paper presents two methods that gain significant improvements compared to the previous works. First, an automated way of handling non-linear dependencies in linear models is presented. In this context, the paper implements a new method for feature selection based on [1], resulting in computationally efficient models with higher accuracies. The three main models used here are linear, ridge, and lasso regression. In the second approach, a deep learning method is presented. Although computationally more intensive, the deep learning model provides higher accuracy than the linear models with automated feature selection. Finally, we compare and contrast the proposed methods with earlier work for day-ahead forecasting of heat load in two different district heating networks.