Research On Power Load Forecasting Based On Big Data

Power load forecasting is an important basic work that must be done in power network planning and generation planning.With the deepening of smart grid construction and the proposal of the "dual carbon" goal,new energy accounts for an increasing proportion of the total power generation capacity.In order to make the power system operate more stable and reliable,more accurate power load prediction is needed.Power load forecasting is based on a historical data set to predict a future period of load.With the establishment of advanced measurement system,power load data has explosive growth,so the research on power load prediction based on big data technology has very important theoretical and practical significance.Due to the long forecasting period,compared with short-term load forecasting,the result of medium and long-term power load forecasting is subject to more external factors,and the accuracy of prediction is also lower.Therefore,on the one hand,it is necessary to improve the forecasting model,on the other hand,it is necessary to analyze the influencing factors of load forecasting and preprocess load data,so as to improve the accuracy of medium and long term load forecasting.This paper mainly carries out the following aspects of work:Firstly,the K-Shape clustering algorithm is used to cluster load time series data,and load data of different industries are obtained by clustering load data,forming clustering clusters of differentiated electricity consumption patterns.Secondly,Pearson correlation coefficient method and grey correlation coefficient method are used to analyze the influencing factors of electricity consumption cluster.According to the analysis results of strongly correlated influencing factors among electricity consumption cluster,the index database of medium and long term power load influencing factors is constructed,and the influencing factors of different industries are mined by data mining technology.It can overcome the shortcoming of traditional clustering analysis which only focuses on data itself and ignores the time series characteristics of power load,and can reduce the dimension of factors affected by power load.Finally,established based on BP neural network and support vector machine(SVM)of the medium and long term power load forecasting model,combined with the power load characteristics and fully consider the characteristics of time sequence,selection of typical industry is analyzed,according to the clustering results with the corresponding load factors,according to different utilization patterns of clustering cluster and its influencing factors in load forecasting.The results of practical engineering examples show that the prediction results can meet the accuracy requirements of medium and long term power load prediction,which verifies the effectiveness of the proposed method.