For the problem of parametric fault diagnosis in nonlinear circuits, an approach based on the comparison of
Volterra frequency-domain models, the wavelet filtering technology and correlation analysis is proposed. Firstly, the output
signals of the fault-free and faulty circuits excited by selected input signals are filtered by wavelet packets. Then, the
coherence functions of the output signal sequences in eight subbands are computed. By the correlation analysis between the
fault-free and faulty circuits, different states of the circuits are identified and the faults are located. The simulation results
show that the proposed method can extract the fault signature effectively and improve the fault diagnosis..