Reinforcement learning for optimization hyperparameters of Long Short-Term Memory applied to Electricity load forecasting

Accurate electricity load forecasting is a key factor in optimizing the operation of the power system. Moreover, more accurate forecasting helps the better efficiency, reliability, and security of the power system. Uncertain weather and uncertain behavior of consumers is the reason for the uncertain behavior of electricity load. The accuracy forecast model is necessary to research. A new model for short-term electricity load forecast having better accuracy performance is investigated in this paper. The proposed method combines long-short term memory and reinforcement learning for short-term electricity load. Concretely, reinforcement learning is used for the optimization of hyperparameters of Long Short-Term Memory (LSTM). The proposed model is applied to the electricity load forecasting of Vietnam. The experimental results show that the proposed model has better performance than the pure LSTM model.