Research On Short-Term Load Forecasting Of Regional Grid

Load forecast is a very important work to power dispatch department. The precision of the load-forecast result is not only related to the safe operation of power system, but also related to the profit of the power enterprises.This paper puts forward a IEM-BP model based on similar day to predict the short-term load. Through the analysis of short-term load characteristics and influence factors, propose a comprehensive similar model to look for similar day. The model taking into account many influence factors, such as the temperature, the rainfall, the date type, the periodicity of load and the nearer data have a greater impact on the predicting value, then use the product of the characteristic quantity and the distance of date to quantify the comprehensive similarity.A new type of intelligent method—Electromagnetism-like mechanism is introduced, and a improved EM algorithm is used to optimize the initial weights and thresholds of BP neural network, then a IEM-BP short-term load forecasting model based on similar days is established. Using respectively the BP model, the EM-BP model, the PSO-BP model and the IEM-BP model to forecast the load on the same day, the simulation results show that the IEM-BP model has obvious advantage in precision. With the IEM-BP model to respectively predict loads of the working days and holidays for one week, the working days’ average relative error in one week is2.60%, the holidays’ average relative error in one week is2.58%, the result shows that the method proposed in this paper is stable and feasible.