Saturation Load Forecasting Based on Long Short-Time Memory Network

The number of factors affecting the level of electricity consumption are gradually increasing. To analyze the influence of these factors on the saturation load, a multi-input saturation load forecasting model based on long short-term memory network is proposed in this paper. The model can deal with the uncertainty and time correlation of saturation load forecasting. Historical load information is saved in the long-term memory cell in the model, and the load information is updated according to the short-term factor input. Firstly, this paper takes six factors as the input of the model, and Adam optimization method is selected to do model training. Next, combined with the saturation load criterion, the optimized model is used to forecast the future annual maximum load and electricity consumption, and then the final saturation time and scale can be decided. Finally, an example shows that compared with the traditional prediction model, the model in this paper can do prediction with higher accuracy.