Error analysis and correction of multi-sensor cluster methods for acoustic emission source localization.

The existing multi-sensor cluster acoustic emission (AE) source localization method has good positioning performance, but the parallel assumption in this method could cause positioning error. This paper focused on the analysis of the positioning error of multi-sensor cluster methods with sensor arrangements of isosceles right-angled triangle and triangular pyramid. Meanwhile, two and three-dimensional amendment algorithms for the two sensor arrangements were proposed. Pencil lead break experiments and numerical examples were used to verify the rationality of error source analysis and the accuracy of the amendment algorithms. Results show that the multi-sensor cluster methods can only accurately locate the AE sources in special positions, such as the AE sources satisfying θ = π/4 in the multi-sensor cluster method with a sensor arrangement of isosceles right-angled triangle and the AE sources satisfying cosθ=3/3 in the multi-sensor cluster method with a sensor arrangement of triangular pyramid. The results of pencil lead break experiment show that the two-dimensional amendment algorithm can accurately locate the AE sources in two-dimensional isotropic structure. For the two-dimensional anisotropic structure, the positioning result of the two-dimensional amendment algorithm is 27.8% higher than that of the multi-sensor cluster method with a sensor arrangement of isosceles right-angled triangle. The results of numerical examples show that the positioning errors of the multi-sensor cluster method with a sensor arrangement of triangular pyramid and the three-dimensional amendment algorithm are 24.6 mm and 0, respectively. Due to the correction of the positioning error caused by the parallel assumption, the latter has better positioning performance. Therefore, the amendment algorithms of the multi-sensor cluster methods have certain engineering application value in AE monitoring of two and three-dimensional structures.