Ultrasonic chirplet echo parameter estimation using time-frequency distributions

Abstract

Chirp signals are frequently encountered in ultrasonic non-destructive testing (NDT) and imaging applications. Chirplet echo parameters signify valuable information such as size, shape and orientation of the reflectors as well as the physical properties of propagation path in ultrasonic NDT applications. In most cases, detected echoes are overlapped due to microstructure scattering and minimal spacing of reflectors. This imposes challenges in decomposing the echoes and extracting the echo parameters accurately. Hence, it is very important to develop a methodology to efficiently and accurately estimate the chirplet echo parameters. In this paper, we present a method for estimating the parameters of chirp echo by means of ellipse fitting in time-frequency domain. This study uses two different time-frequency (TF) representations of ultrasonic signal to sequentially estimate the chirplet echo parameters: (i) Wigner-Ville Distribution (WVD), and (ii) Short Time Fourier Transform (STFT). To demonstrate the parameter estimation performance of ellipse fitting method, the signal decomposition and echo estimation algorithm is applied to perform decomposition of ultrasonic experimental signals consisting of many interfering echoes acquired from testing a steel block. The comparison between the reconstructed signal and the experimental signal shows that the decomposition has been successfully performed in the presence of measurement noise and strong interference from microstructure scattering.