Identification Method of Distribution Network Faults Based on Multilevel SVM
This research constructs the time-frequency matrix of the bus voltage and the main transformer low-voltage side cur-rent waveform,and uses SVD technology to extract the singular spectrum of the waveform,and then obtains key parameters such as the singular spectrum mean that represents the singular value size and the singular entropy that describes the signal complexity as a feature vector.Through simulation and actual testing,it was verified that the identification accuracy of this method exceeds 90%un-der various typical conditions,proving its effectiveness,adaptability and practicability in identifying various fault types.
distribution network faultstime-frequency matrixSVDMultilevel SVM
万方数据
维普
