A Research Of Short-Term Load Forecasting Of Changsha Based On Meteorological Factors And Integrated Study

Power Load forecasting is an important task in daily scheduling of power system.Reliable load forecasting is important to ensure electric company daily operations,to improve residents' daily life,to elevate industrial production efficiency,to guarantee the rapid economic development of our country.The stability and sustainable development of social is urgent need of High quality load forecasting model.As the change of power load is affected by various and complicated factors such as social,natural and geographical,load forecasting scheme and model must be designed according to the characteristics of the prediction region'local conditions.As a provincial capital city in central China,Changsha city has obvious seasonal climate change,mainly reflected in the winter and summer,that is continuous hot weather in summer and freezing weather in winter.In recent research and engineering of power load forecasting in changsha city,the effects of climate is less or not considered.By analyzing the correlation between the changes of the power load and changes of the climate,this paper focused on the analysis of the impact of climate change on power load,researched the power load forecasting based on the meteorological factors.Based on the actual power load and meteorological data,this paper puts forward a kind of the power load forecasting of sub sets regression model,by comprehensively analyzing the correlation of the meteorological factor and the power load,extracted the meteorological factors that actually influenced electrical load of changsha city changes greatly,at the same time,considering the power load's temporal variation and holidays' influence to the power load,extracted the temporal features that actually influenced electrical load of changsha greatly,and used the multivariate linear regression algorithm and support vector machine(SVM)regression algorithm to forecast the power load.On this basis,in order to improve the prediction accuracy of SVR regression model furtherly,in view of the error that may result of when the SVR forecasting model selecting data sets randomly,this paper introduced integrated learning technology to strengthen the selection of training set,and presented Bagging-SVR algorithm and Boosting-SVR algorithm which based on the integrated study to improve the effect of the traditional SVR algorithm by weakening the error when choosing samples randomly.We applied it to the changsha load forecasting model.By comparing the experimental results with traditional SVR model,it shows that the accuracy of prediction has been significantly improved.