Improved Multiple Linear Regression And Grey Model Based On Adaboost For Combination Medium-Term Load Forecasting

Medium-term load forecasting is an important link of power special planning, and it is closely related to forecasting accuracy. In addition, the medium- term load forecasting is also extremely important to the security of electricity of network. Taking the timing, conditions and uncertainty characteristics into account, there are not any method to apply to all situations. Therefore, the concre te circumstances is united and concrete analyzed to select an appropriate model.The paper was to study the medium-term load forecasting. First of all, the purpose, significance and the research at home and abroad were reviewed. Then the multiple linear regression algorithm and adaboost algorithm is analyzed. The multiple linear regression, structured by the small sample dates, has the heteroscedasticity impact. With regard to this deficiency, this paper put forward an improved an multiple linear regression algorithm based on adaboost. This algorithm used adaboost algorithm to dynamically adjust the weighting factors corresponding to different samples, coordinated and combined various multiple linear regression models and improved the generalization ability of the algorithm.Then the GM(1,1) algorithm principle was analyzed, Moving average method was raised to revise the deficiency of low accuracy. If the accuracy could not meet the requirements, residual error correction method and markov method were used to set up the residual error GM(1,1) model as the final improved GM(1,1) model. Finally, improved multiple linear regression, improved GM(1,1) model, trend extrapolation model and per capita electricity consumption model were combined by the recursive weight method and ideal point and multiple attribute decision- making method and contrasted to select the optimal combination to overcome the deficiency of single model prediction low accuracy. The improved algorithm and combined model of this paper were taken as an example of load forecasting on the data set of electric power utilization in Jinxian County, which verifies the effectiveness and practical value.