Deep Learning Methods Based Short-Term Electrical Load Prediction

Electrical load forecasting plays an important role in power system management and economic development.Due to the complex nonlinear relationship between electrical load and various factors such as economic policy and human activities,it is very difficult to accurately predict the electrical load.In recent years,the deep learning method has been widely used due to its deep network structure and learning method,which can effectively deal with various complex problems.In order to further improve the short-term electrical load forecasting performance,this paper combines the deep learning methods to realize the accurate prediction of electrical load forecasting.Firstly,the domestic and international development statuses of the electrical load forecasting problems are summarized.Then,the classification and characteristics of electrical load forecasting methods are given,and several representative methods of electrical load forecasting are discussed.Secondly,this study combines the data-driven long-short-term memory network(LSTM)and the extreme learning machine(ELM)to present a combined model-based forecasting method for the prediction of short-term electrical load.In this combined model,the LSTM is adopted to extract the deep features of the electrical load while the ELM is used to model the shallow patterns.In order to generate the final forecasting result,the predicted results of the LSTM and ELM are ensembled by the linear regression method.Finally,the proposed method is applied to two real-world electrical load forecasting problems,and detailed experiments are conducted.Thirdly,an integrated prediction model is presented for further improving the electrical load prediction performance.In order to obtain the electrical load characteristics of different depths,the output of each SAE layer is taken as the input of ELM and the output of each ELM model is recorded.Then the results of the different ELM models and the SAE models are linearly regressed to get the final output.The linear regression part is trained by the least square estimation method.In addition,the integrated model is applied to predict two real-world electrical load time series.And,to show the advantages of the proposed forecasting model,detailed comparisons with the SAE,ELM,the back propagation neural network(BPNN),the multiple linear regression(MLR)and the support vector regression(SVR)are done.Finally,the proposed forecasting models are analyzed and summarized,and their shortcomings are discussed.What's more,future research directions are pointed out.