Research On Load Forecasting Moedl And Application Based On User Behavior And Industry Power Consumption Characteristics

Mid-long term load forecasting is of great guiding value to the development and plan of regional power system.The accuracy of mid-long term load forecasting will directly affect the economy,safety and rationality of regional power system.In contrast with traditional load forecasting,the development of power market reform and new energy brings new challenges to recent load forecasting.This paper is based on user behavior and industry power consumption characteristics,in order to get the situation of load in the future,it fully considers the influence of external factors and the power consumption characteristics.The analysis of user behavior is an important part of the analysis of power system.Firstly,the influence mechanism of user behavior is studied,especially about the connections between external factors and electric quantity.Secondly,based on grey forecasting model GM(1,1),multivariable grey model MGM(l,n)is build in order to fully consider the influence of external factors on consumption behavior.The relevancy between external factors and load is determined by grey relational analysis.In the end,the problem about how to determine the proper number of external factors included in the MGM(1,n)model is discussed.The load forecasting based on industry power consumption characteristics studies load of every separation industry,which is different from the traditional mid-long term load forecasting which only uses load data of full area and ignores the power consumption characteristics of every separation industry.Firstly,by studying the concept and history of classification of industries,a specialized industry classification is designed for load forecasting.Secondly,the power consumption law of every separation industry is studied through fitting load curve.In the end,by accumulating the load forecasting result of every separation industry,the result of the full area is obtained.In the end,a comparison is made between the previous three models about their forecasting accuracy and proper models are selected to build a combined forecasting model based on the objective of minimizing the sum of absolute value of error.