Research on Short-Term Electricity Load Forecasting Methods Based on Deep Learning

Because of the increasing proportion of power load in sizeable industrial power industries, this paper proposes a forecasting model based on long short-term memory network LSTM. Short-term power forecasting of power load provides a basis for large industrial power dispatching, maintenance and adjustment and effectively guarantees power consumption. Taking the historical data of power load as training and test data, the output data of the current period model is assembled into the input data of the next period model to dynamically and automatically fine-tune the model parameters to improve the prediction accuracy the mo. The experimental results show that the fitting power load forecasting model reaches 94.2 %, which exceeds SVM and Decision Tree. With the increase of test data, the prediction model can further improve the prediction accuracy.