Dual Gaussian attenuation model of ultrasonic echo and its parameter estimation

Abstract

Signal de-noising and feature extraction, whose performance directly affect the evaluation of non-destructive testing (NDT) results, is essential technology for ultrasonic NDT echo processing. Aiming to solve the problem of nonlinear distortion between measured ultrasonic echo and its mathematical model, which widely used are exponential model (EM) and Gauss model (GM), a dual Gaussian attenuation model (DGAM) of ultrasonic echo signal and its parameter estimation method are proposed in this paper. The proposed dual Gaussian attenuation model parameter estimation (DGAM-PE) method is introduced in three parts: calculating mean square error between measured signal and model, optimizing mean square error by particle swarm optimization, optimum parameters extraction based on optimization result. The simulation and experiment results show that compared with the exponential model and Gaussian model, the proposed dual gaussian attenuation mathematical model of ultrasonic signal in this paper can better simulate the measured ultrasonic echo signal, with a mean square error of 0.0073 and normalized correlation coefficient of 0.9816. Additionally, an improved adaptive particle swarm optimization is proposed in order to enhance the accuracy of parameter estimation results.