Short-term Load Forecasting Based on Elman Neural Network

There are lots of challenges facing power grid load forecasting cause by climate change, increasing electricity demand and so on. Meanwhile, it cannot be stored in large quantities for electric energy because it needs to be used immediately in most cases. Therefore, accurate power load forecasting can balance the demand and supply relationship of power supply enterprises effectively, which can decrease the cost of power generation greatly and improve environmental benefits. In this paper, the short-term load forecasting method based on Elman neural network is introduced. Firstly, this method firstly glossy and fill the abnormal and missing data in the historical data, then analyzes the influence of meteorological information and date type on the actual power load, and normalizes the factors with great influence as input parameters of the load model. In addition, the fitting function is used to eliminate the effect of accumulated temperature. Finally, the segmented forecasting method is used to accurately forecast the 15-minute power load within the day.