A Bayesian acoustic source localization method using L-shaped sensor clusters in plate structures with local...

Acoustic source localization (ASL) technique holds significant importance for the structural health monitoring of plate-like structures. Most ASL techniques work well for homogeneous plates with uniform thickness. When the plate thickness varies locally, the existing ASL techniques give increased localization errors due to the variations in propagation velocity of guided waves and mode conversion caused by the plate thickness variation. To tackle this challenge, a Bayesian acoustic source localization method using L-shaped sensor clusters (LSSC) is proposed in this paper. This method arranges sensors in an LSSC configuration to mitigate the impact of the wave velocity local variations within the plate. It employs the localization results of the LSSC algorithm as prior knowledge and updates them using the Bayesian inference framework to enhance the accuracy of the final localization results. It employs the cross-correlation algorithm to extract TDOA, thereby avoiding the effects introduced by mode conversion. Through finite element simulations and experimental validation, it is demonstrated that the proposed method localizes acoustic sources more accurately than both traditional Bayesian localization methods and standalone LSSC localization algorithm in plates with local thickness variations.