Short-term Power Load Forecasting of Residential Community Based on GRU Neural Network

Abstract

The power load of residential community has the characteristics of wild fluctuations, complex influence factors and difficult forecasting. To deal with that, a short-term load forecasting (STLF) method for residential community based on gated recurrent unit (GRU) neural network was proposed. The least absolute shrinkage and selection operator (Lasso) and partial correlation analysis are used to analyze the influence of temperature, humidity, rainfall and wind speed on the load. It shows that the average temperature affects the change of the load most among various factors, thus the average temperature is added as an input variable to the load forecasting model based on GRU network. The simulation results show that the proposed method is faster within the similar forecasting accuracy, compared with the long short-term memory (LSTM) network and traditional recurrent neural network (RNN). It's proven to be a more effective residential community short-term load forecasting method.