Feature Extraction for Nonlinear Analog Circuit Based on GFRF

A fault diagnosis method for nonlinear analog circuit based on Generalized Frequency Response Function(GFRF) analysis is introduced.Especially,the project of feature extraction in circuit is expounded.The key technology such as designing the stimulus and Volterra frequency-domain kernels measurement are solved in this project.A nonlinear system is identified by applying Volterra frequency-domain kernel estimation.The highest significant order of nonlinear analog circuit is determined and the stimulus is designed.The response of each order is separated by the method of Vandermonele and the Volterra frequency-domain kernels is figured out by the formula.