Electricity Load Forecasting Based on CNN-LSTM

With the popularization of smart grid construction, how to predictively provide high-quality long-term stable power supply has become an urgent problem to be solved for the current conflict between supply and demand. Therefore, it puts forward extremely high requirements on how to accurately predict the power load. At the same time, the influencing factors of power load tend to be diversified, resulting in the traditional nonlinear prediction model cannot objectively reflect the actual power load. For this reason, this paper adopts deep learning related theories, introduces holiday influencing factors and temperature influencing factors, calls PyCharm platform for Python programming to compare BP model, LSTM model, BiLSTM model. By doing this, we find that the LSTM model has the best effect, and then adopt convolutional neural narrow network (CNN) and long short-term memory (LSTM) were used to predict the power load data. The results show that the prediction accuracy of CNN-LSTM is 98.94%, and the method provides a new idea for the convenient prediction of power load, and also has positive significance for the multi-factor prediction of power load.