Experimental Study On 3D Acoustic Emission Source Location of Concrete Based On Sparse Least-Squares Support Vector Regression

Abstract

In order to further prove the effectiveness of the sparse least-squares support vector regression (S-LS-SVR) method in damage detection, the authors used the S-LS-SVR model to locate actual damage sources of concrete. The data from acoustic emission testing (AE) are generated and filtered by the pullout test of reinforcement in concrete, and the three-dimensional coordinates of real-time damage sources in the failure process are provided through the model. The S-LS-SVR method is compared with the Newton iterative method and improved exhaustive method for positioning speed, positioning data utilization, and positioning accuracy. The results show that S-LS-SVR is superior to the two other time difference of arrival–based positioning methods in positioning speed, positioning data utilization, and positioning accuracy (data utilization is slightly lower than the improved exhaustive method). The location method based on S-LS-SVR provides the possibility for the application of AE technology in intelligent damage location of bridges, dams, and other service structures.