The bus-bar net load forecasting based on ARIMA and PSR-BPNN

In order to realize the real-time tracking and forecasting of net load with distributed power bus, a method based on autoregressive moving average (ARIMA) and phase space reconstruction back-propagation neural network (PSR-BPNN) is proposed. Inherent linear and nonlinear components of the net load are considered to increase forecasting accuracy. This method can be divided into three steps: firstly, ARIMA is adapted to forecast the linear components of net load using the historical data; secondly, C-C method is used to reconstruct the nonlinear components; thirdly, the BP neural network forecasting model is established based on the reconstructed series. According to the analysis of forecasting results, the proposed ARIMA and PSR-BPNN forecasting model is more suitable for the net load with distributed PV components.