Research Of Power Generation Forecast Based On The GM-ARIMA Model

Power generation forecasting is the foundation of power system. Currently, the study of generation forecasting has been more thorough. Commonly used methods are:Grey Forecasting Model, Linear Regression Model, Auto Regressive Integrated Moving Average Model and Artificial Neural Network Model, etc. The regression forecasting model and time series model are most widely used. Regression prediction model is used to predict the generating capacity by establishing model between power and other relevant variables. But the complexity of the external factors influencing power generation makes it difficult to analyze their impact on capacity accurately. A more effective method is to research and analyze the historical data of power itself so as to forecast the future capacity.With the development of the electricity market, the accuracy of power generation forecasting is related to the interests of all parties directly. So more and more attention has been attracted to how to improve the prediction accuracy of the models. This paper focuses on studying the method of forecasting the generating capacity and exploring a more accurate and effective method on the basis of the existing models. Firstly, basic theories of the ARIMA model and the GM(1,1) model are introduced. Secondly, grey forecasting theory is used to revise residual of the ARIMA model so as to improve prediction accuracy. Thirdly, ARIMA model is explored which is suitable for the generating capacity of Gansu Province after several fittings, according to five-year production data of Gansu province from2008to2013. Then the non-interval GM(1,1) model is established regarding the residual value of the ARIMA. Finally predictive results of the ARIMA model, GM(1,1) model and the GM-ARIMA model are compared. It is obvious that GM-ARIMA model successfully improved the predictive accuracy of original model.