基于VMD-RF算法的小电流接地系统暂时过电压的识别研究OA

In the 10 kV voltage transformer,the focus of the microcomputer harmonic elimination device is to improve the identification accuracy of ferromagnetic resonance.However,the misjudgment examp-les in engineering practice reveal that the existing identification technology still has the problems of insufficient accuracy and modal confusion.Aiming at this problem,this paper comprehensively studies the generation mechanism and fault characteristics of various transient overvoltage signals with similar characteristics,such as ferromagnetic resonance,single-phase grounding fault and arc grounding fault in small current grounding system.On this basis,a new recognition algorithm of artificial intelligence is proposed.The research process is divided into three steps:Firstly,based on theoretical analysis,the basic principles and unique characteristics of different types of overvoltage signals are expounded.Secondly,a more accurate simulation model is used to obtain the data set of various overvoltage signals.The variational mode decomposition(VMD)technique is used to effectively extract the intrinsic mode function reflecting the local characteristics of overvoltage,and the selection of feature subset is optimized by combining energy entropy theory and Pearson correlation coefficient.Finally,through experimental simulation verification,it is proved that the proposed overvoltage identification method based on random forest(RF)algorithm can achieve an accuracy of up to 97.22%,which greatly improves the classification effect of various overvoltage signals,thus ensuring that the harmonic elimination device can respond and act more accurately.