Selection And Application Of Load Forecasting Method In Binhai Development Area

Load forecasting is the basis of electric power enterprise planning and scheduling operation.Improving the forecasting accuracy is the core target for load forecasting work,and reasonable and normative forecasting mechanism can guarantee the working accuracy.The aim of this paper is to find out a reasonable algorithm selection mechanism,ensuring the accuracy of load forecasting.At first,this paper studies the size and structure of the Binhai Development Area,andexplores all kinds of the load characteristics of the region from the diagram and the form.Through the case study,this paper analyzes several major factors that affect the load of the power grid.Then this paper introduces the mathematical principles and the implementation mechanismwith four kinds of load forecasting algorithms used by the Binhai Development Area power grid in details.The four kinds of load forecasting algorithms are the time series forecasting algorithm,the grey theory algorithm,the fuzzy inference algorithm and the support vector machine.Thirdly,this paper selects 20 working days of the spring and summer as a reference,and compares the accuracy of the four algorithms by making forms.To select an appropriate algorithm,this paper seeks an algorithm selection mechanism that applies to the Binhai Development Area by two methods.One is to compare the average accuracy of the four algorithms for a few days before the forecasting day,and the other is to compare the average accuracy of the four algorithms for the similar day of the forecasting day.Finally,the paper concludes that the similar day method is more valuable than the average accuracy selection method.Besides,this paper also verifies the practicality for the comprehensive forecasting method of the Binhai Development Area power grid and finds out the comprehensive forecasting method can't obtain the higher accuracy than the similar day method and the average accuracy selection method.This paper proposes in the end that,in order to further improve the load forecasting accuracy,therelated management personnel shall be artificially adjusted to the calculation result with the researched load characteristics and the real situation.Toform a complete mechanism of load forecasting,future research shouldcombinethe computer and human power as a whole.