In order to improve the accuracy of short-term load forecasting, a forecasting model based on wavelet analysis, particle swarm optimization (PSO) algorithm, least squares support vector machine (LSSVM) and long-short term memory network (LSTM) was proposed in this paper. In this method, the same length component of the original data was obtained by wavelet decomposition and reconstruction of the power load. The LSSVM prediction model was established for the low-frequency component and the optimal parameters were found by PSO algorithm. The LSTM prediction model was established for the high-frequency component and the final load prediction was realized by combining the prediction results of each component. The experimental results showed that the prediction accuracy of the proposed model is better than that of the traditional LSSVM model, BP neural network model and PSO-LSSVM model, and its feasibility is verified.
