Efficient wavelet ridge extraction method for asymptotic signal analysis

Abstract

A wavelet ridge extraction method is proposed for efficient estimation of instantaneous frequency and amplitude of the asymptotic signal. Inequalities are derived to provide limitations on scale interval and sampling frequency. The evaluation of optimal scale interval and sampling frequency at any given accuracy can be done by solving these inequalities and minimizing the computational complexity of continuous wavelet transform. After the continuous wavelet transform of the signal is completed with these optimal parameters, a more reliable ridge can be formed using only the discontinuity points on the ridge extracted using the conventional method. The results of a system identification simulation test performed to verify the efficiency of the proposed method indicate that the computational load of the proposed method is about 1/6 of that of the conventional method. When the system identification method is used to measure samples with known moment of inertia, the relative error below 0.08% is achieved.