Ultrasonic echoes estimation method using genetic algorithm

Liu, Zhongzhu; Bai, Xiaoguang; Pan, Qinxue; Li, Yong; Xu, Chunguang

doi:10.1109/icma.2011.5985731

Cited by 8 publications

(1 citation statement)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Despite the wide use of the GN algorithm, its dependence on the initial guess is a major drawback. A number of attempts have been made to solve the problem of the initial guess for estimating the GM parameters by using the genetic algorithm (GA) [17][18][19], modified artificial bee colony (MABC) [20] and particle swarm optimization (PSO) [21].…”

Section: Introductionmentioning

confidence: 99%

Modeling the ultrasonic testing echoes by a combination of particle swarm optimization and Levenberg–Marquardt algorithms

Gholami

Honarvar

Moghaddam

2017

Meas. Sci. Technol.

View full text Add to dashboard Cite

This paper presents an accurate and easy-to-implement algorithm for estimating the parameters of the asymmetric Gaussian chirplet model (AGCM) used for modeling echoes measured in ultrasonic nondestructive testing (NDT) of materials. The proposed algorithm is a combination of particle swarm optimization (PSO) and Levenberg–Marquardt (LM) algorithms. PSO does not need an accurate initial guess and quickly converges to a reasonable output while LM needs a good initial guess in order to provide an accurate output. In the combined algorithm, PSO is run first to provide a rough estimate of the output and this result is consequently inputted to the LM algorithm for more accurate estimation of parameters. To apply the algorithm to signals with multiple echoes, the space alternating generalized expectation maximization (SAGE) is used. The proposed combined algorithm is robust and accurate. To examine the performance of the proposed algorithm, it is applied to a number of simulated echoes having various signal to noise ratios. The combined algorithm is also applied to a number of experimental ultrasonic signals. The results corroborate the accuracy and reliability of the proposed combined algorithm.

show abstract