Super Short-term Load Forecasting Based on S-LSTM

Abstract

Power load forecasting is affected by uncertain factors such as temperature, humidity, light and seasons, and has strong randomness and volatility, so it is difficult to carry out real-time power load forecasting. This paper proposes a digital twin model oriented to power load forecasting based on a simplified long short-term memory network (S-LSTM, Simplify Long Short-Term Memory). For this reason, the proposed digital twin model realizes the synchronization and real-time update of the physical entities of the power system, and obtains more accurate and faster prediction results than traditional prediction methods. The model uses S-LSTM as the core of the digital twin system, trains the S-LSTM model through RMSProp adaptive learning algorithm, constrains the learning rate, obtains the best parameters, and improves the model's adaptability to different types of data. The simulation results show that compared with other prediction methods, it verifies the effectiveness and feasibility of the method proposed in this paper.