Electric Power Material Demand Forecasting Based on LSTM and GM-BP methods

Electric power material demand forecasting is an important part of power grid planning management, which helps to save power industrial costs and improve capital utilization of power companies. Due to the irregularity of historical data for electric power materials, the characteristic of electric power material demand is complicated and power grid companies lack accurate forecasting methods. By identifying the continuous and intermittent characteristics of electric power material demand, this paper separately designs the Long Short-Term Memory (LSTM) method, and the Grey prediction Model based BP neural network (GM-BP) method to individually forecast the electric power material demand with different demand characteristics. Finally, the proposed methods are tested by the practical data of an electric power enterprise, and simulation results validate that the proposed methods can effectively adapt to different features of electric power material demand as well as predict electric power material demand with good accuracy.