Adaptive Parameters Selection Method for Time-Frequency Distribution Series Based on Normalized Entropy

Abstract

To resolve the problem of Gabor transform window width and order selection for Time-Frequency Distribution Series (TFDS), a parameters selection method for TFDS based on normalized entropy has been proposed, especially the adaptive selection method of order. The normalized entropy is used to measure the concentration and cross-terms of TFDS firstly, and then the relation between the order and width of Gabor transform window function and the concentration and cross-terms of TFDS is used to realize adaptive selection of window width and order parameter, which overcomes the subjective selection problem of the order. The simulation results show that the proposed method can effectively select optimal TFDS parameters for simulated and experimental ultrasonic tesing signal, and can get TFDS with good concentration and high resolution.