Damage Identification Method Based on Ultrasonic Guided Wave Sensor Network and Path Optimization Bayesian Fusion Algorithm

Ultrasonic guided wave detection technology is one of the most effective damage identification methods in structural health monitoring (SHM), but external factors will cause errors during identifying damage. Based on ring probability distribution and path optimization, a Bayesian fusion damage localization method is proposed. First, the effective band of healthy signal and damaged signal collected by piezoelectric sensor array is selected and solved correlation. The threshold of the obtained correlation coefficient is screened, to obtain the valid path greatly affected by damage. Then, the time-of-flight (ToF) and damage index (DI) of the valid path are calculated and input into the Bayesian framework as damage characteristics. Finally, the posterior distribution is sampled by the Markov chain Monte Carlo (MCMC) method. The frequency distribution of sample points is Gaussian fit. The peak value of the fit curve is identified as damage position. A finite element simulation model is established to preliminarily verify the feasibility and effectiveness. Some localization experiments were carried out on aluminum plate and carbon fiber composite plate. The results show that the average localizing error is 6.58 mm and the standard deviation of the error is 3.05 mm, which are less than the localizing error using all paths. By this method, the valid path is about half of all paths, which can not only reduce the impact of invalid paths to improve the stability and robustness of localizing results but also reduce the amount of data to be processed to realize the purpose of data lightweight.