The Research Of Power Load Forecasting In Changchun Power Market

The signal load forecasting method which was based on the self-law of time sequence and ignored the external factor effect can not satisfy the precision requirement, meanwhile, the intelligent algorithm with stronger nonlinear fitting ability also provided wide ideas for power load forecasting. Firstly, the single forecasting methods were used to predict and analysis the medium-long power loads, then the grey relational grade method were used to screening factors, the influencing factors sequence and load history value were used as the input vector and output vector in SVM method, the SVM forecasting value can be obtained, the GM(1,1) method was used in sequence growth forecasting, the variance minimization principle was used to decide the combined weights and combined forecasting values were obtained, finally the markov correcting model was used to modify the errors for the final results. The practice results of power load forecasting in Changchun have showed that the prediction accuracy can be increased by this method, and it has certain practical value as well.