Research On Load Classification Method Based On VMD

Load clustering plays an important role of power system planning,demand-side management and load forecasting.Aiming at the problem of insufficient accuracy of load classification caused by noise and abnormal data,this thesis uses the advantages of irregular data processing,noise reduction and feature extraction of variational mode decomposition(VMD)to analyze load data classification and improve the effectiveness and accuracy of classification.This thesis provides a theoretical foundation for characteristic analysis,load forecasting and scheduling.The research on load clustering method is presented as below.(1)This thesis analyzes the characteristics of power load,introduces load clustering methods and research of load data.(2)The theory of variational modal decomposition algorithm is studied.The processing ability of mixed signal separation and noise robustness are analyzed through simulation,so as to reveal the superiority of VMD.Compared with empirical mode decomposition,it is found that VMD is more advantageous to signal decomposition,and the feasibility of VMD algorithm in load classification analysis is verified.(3)Aiming at the problem of low clustering accuracy of k-means.A load classification method based on VMD and k-means is proposed.The original data is decomposed through VMD.And the obtained clustering results of modes 1 and 2 are carried by the k-means.In the case,compared with k-means clustering results with the results of original load data to realize the more accurate classification of load data clustering,the effectiveness of the proposed method is verified.For the problem that the traditional clustering is not accurate enough,this thesis applies a classification method based on VMD and k-means.The classification results of original load data are further classified by comparing k-means results between decomposed 1 and 2 modes of VMD and the original load.And the effectiveness of the proposed method is verified by an example.(4)Aiming at the problems of high dimension of power load data and insufficient manifestation of sample characteristics,a power load classification method based on variational mode decomposition and singular value energy difference spectrum is proposed.Based on VMD,singular value decomposition and energy difference spectrum,the load curve is converted into energy spectrum curve to reduce the data dimension and reduce the difficulty of load analysis.The effectiveness of the proposed method is verified by example analysis.(5)Aiming at the problem of low clustering accuracy caused by insufficient data characteristics,a power load classification method based on VMD and FCM is proposed.The load characteristics are extracted by VMD.The load curves are converted into intrinsic mode functions(IMF).The characteristic explicit synthetic curves are obtained through data reconstruction,so as to improve the convergence speed and clustering accuracy of FCM clustering function.Finally,the effectiveness of the proposed method is verified by example analysis.