Adaptive Chirplet Signal Expansion Based on Transcendental Estimation
基于先验估计的自适应Chirplet信号展开

In this paper, a new time-frequency representation method, adaptive signal expansion algorithm, is presented. The algorithm is based on that essential character of signal space, initial value estimation and precise resolution are obtained simultaneously. Signal is adaptively expanded to a sum of chirplet elementary functions by using match pursuit algorithm. Then, according to expansion coefficients and elementary function parameters, adaptive time frequency distribution is obtained. Its time frequency congregate, noise-reduction and time frequency resolution are not only better than the general time frequency distribution but also better than adaptive time frequency distribution reported and it is able to characterize the signal's nature exactly. The validity of the algorithm and the performance of adaptive time frequency distribution are tested by numerical simulations.