Impact Localization using Sparse PWAS Networks and Wavelet Transform

Abstract

In this work, an impact experiment on an aluminum plate is implemented to localize the impact points. A small steel ball is dropped on an Aluminum plate to generate acoustic emission signals acquired by 9 PWAS transducers distributed across the plate. Two possible options exist for determining the time of flight (TOF). The first option is determining the TOF of the starting point of a signal (method 1). The second option is determining the TOF at the location of the maximum signal value (method 2). In our work, both method 1 and method 2 are examined comparatively. Based on the number of PWAS transducers, six networks are created to localize the impact points separately. The first network (Network 1) has four PWAS transducers (the minimum number of PWAS transducers to localize an impact point on a plate with unknown wave velocity). The second one is created by adding the fifth PWAS transducer to the Network 1. The same procedure designs the other networks. The effect of increasing the number of PWAS transducers on the percentage error of the calculated location of an impact point is examined. The continuous wavelet transform (CWT) of the signals are calculated with AGU Vallen wavelet program to determine the TOF of the signals. The results show that the method 1 of determination the TOF has fewer percentage errors than the method 2. The percentage errors of calculated locations of impact points decrease with increasing the number of used PWAS transducers.