Acoustic sources localization for composite pate using arrival time and BP neural network
Wenfeng Hao, Yingqi Huang, Guoqi Zhao
Abstract
This paper presents a machine learning approach to localizing a five-peak narrow-band modulated sinusoidal signal excitation source within a composite panel. In particular, the Back Propagation (BP) neural network is used. The idea is to use the arrival time of the first wave packet in a five-peak wave signal to locate their source. Specifically, this paper divides the composite material board into multiple regions, designs 8 receiving points to receive the signal from the excitation source, and finds the region where each source is located. The COMSOL numerical simulation platform is used to build a composite plate model and simulate the propagation of five-peak waves to train and test the machine learning network. Correspondingly, carry out experimental verification and use a scanning laser Doppler vibrometer (SLDV) to build a non-contact experimental platform to obtain the wave field information in the composite material plate. The results show that BP neural networks can learn to map signal features to their sources in both contexts.