Short-term Power Load Forecasting Based on Particle Swarm Optimization Long Short-term Memory Neural Network

Abstract

Accurate power load prediction at different periods can provide an essential basis for energy consumption reduction and power scheduling. Particle swarm optimization (PSO) and long short-term memory (LSTM) neural networks were introduced into the forecasting method of electric power load. First, aiming at the problem that it is difficult to select the LSTM hyper-parameters, hyper-parameters including the learning rate and the number of neurons were optimized by using the PSO algorithm. Then, the PSO-LSTM based prediction model was established, and the LSTM load prediction model with the optimal hyperparameter group was structured to predict the future change trend of power load. Finally, the historical load data of an electric power company was used to prove the effectiveness of the proposed method. The results demonstrate that, in compare with the traditional method, the load prediction model based on PSO-LSTM has a higher prediction accuracy and stability.