Long-term load combination forecasting method considering the periodicity and trend of data

In order to solve the problems of insufficient accuracy long-term power load forecasting and poor applicability model, this paper considers coupling a number macro indicators, such as regional economic development social with time series data load. BP neural network Autoregressive integrated moving average model (ARIMA) are used integrate improve so trend ability annual model. The non parametric function method is forecast periodic in monthly data, combined overall Finally, through comparison grey prediction other models verification MAPE error analysis method, considering combination periodicity significantly improved, which suitable for