Research on WPD and DBSCAN-L-ISOMAP for circuit fault feature extraction

Abstract To solve the problem of feature extraction in electronic circuits due to nonstationary and nonlinear characteristics fault signals, a method for is proposed, which combines wavelet packet analysis an improved landmark ISOMAP mapping algorithm. The technology used decompose reconstruct signals at multiple levels. extracted entropy construct vector matrix. density-based spatial clustering applications with noise (DBSCAN) algorithm calculate screen points. embed high-dimensional parameter set into low-dimensional eigenspace, extract sensitive subset, apply support machine identify fault. diagnosis experiment three-phase VIENNA rectifier shows that compared principal component method, traditional based on DBSCAN extracts signal equipment more easily.