Research on analog circuit fault feature extraction based on FRFT-KPCA method
Sun Jingjie, Zhao Jianjun, Weimeng Sun · 2011
In the fault diagnosis process of analog circuit, fault features extraction is an important technology. In order to gain effective features of nonstationary and time varying signals, the paper proposed an approach to extract fault features based on fractional Fourier transform (FRFT) and Kernel Principal Component Analysis (KPCA). Particle Swarm Optimization (PSO) is used to determine the optimal value of the fractional order p according to within-class and among-class scatter matrix. And mapping signals in an optimal FRFT domain for separation. Then, KPCA is used to compress the dimension of signal features. The experimental results show that after feature extraction by FRFT-KPCA approach, samples of different signals are well separated in fractional feature space.