All-phase fast Fourier transform and multiple cross-correlation analysis based on Geiger iteration for acoustic emission sources localization in complex metallic structures

Abstract

Nowadays, the localization and identification of acoustic emission (AE) source is widely utilized to structural health monitoring (SHM) of complex metallic structures. However, traditional AE source localization methods are generally difficult to localize and characterize AE sources in plate-like structure that has complex geometric features. To alleviate the problem, a novel AE source localization method based on all-phase fast Fourier transform and multiple cross-correlation analysis is proposed in this article. Moreover, least squares and Geiger iteration algorithm are applied to determine the coordinates of AE sources. In addition, an improved Bayesian information criterion (BIC) version named autoregressive BIC (i.e., AR-BIC) is presented to increase the accuracy of source localization. To validate the performance of the proposed approach, the classical pencil lead break tests are carried out on a 316 L stainless steel with 10 laser cladding layers. Experimental waveforms are generated from AE sources near laser cladding layers, the surface of the structure, and on its edges. Additionally, to evaluate the performance of the proposed approach in three-dimensional AE source localization, an industrial storage tank is used to acquire three-dimensional AE sources through manually striking. Finally, to further verify the effectiveness of the proposed approach, comparisons with conventional AE source location methods (i.e., PAC or SAMOS AE acquisition system, Newton’s method, and multiple cross-correlation based on Geiger algorithm) and two representative approaches (i.e., deep learning and Bayesian methodology) for localizing AE sources generated by complex metallic structures are conducted. The comparative results demonstrate the effectiveness and superiority of the proposed method in AE-based SHM of complex metallic structures.

Heterogeneous Cluster Application Communication Optimization and Computer Big Data Management

“…Such problems do not ostensibly benefit from many-core acceleration, but the implementation in manycore clustered FFT shows that many-core cluster architecture is beneficial to increase the effective total bandwidth of single-node memory and processing units. Additionally, the card can store more than the CPU cache, which means large operations can be sent to memory at once and run faster, reducing data to the same level as the average [14]. How to make a processor for high-bandwidth applications, optimization of non-homogeneous parallel computing algorithms usually focuses on reducing the number of data transfers from memory to memory to the processor.…”

Acoustic emission sources localization and identification of complex metallic structures based on nearfield frequency space sparse decomposition

“…Furthermore, although SD‐based AE source localization generally requires as few sensors as possible to reduce the calculation time, a large number of sensors need to be arranged to improve the localization accuracy when the specimen size is large 27,28 . Therefore, the correlation analysis between the signals collected by different sensors has become a hot research direction recently 29,30 . In this process, for n sensors, assuming the time of receiving the signal is, respectively, , thus, the actual time difference between the i th and j th sensors can be calculated by the following Equation (1).…”

“…27,28 Therefore, the correlation analysis between the signals collected by different sensors has become a hot research direction recently. 29,30 In this process, for n sensors, assuming the time of receiving the signal is, respectively, t 1 , t 2 , ÁÁÁ, t n n ≥ 4 ð Þ, thus, the actual time difference Δt i,j between the ith and jth sensors can be calculated by the following Equation (1).…”

Crack damage monitoring for compressor blades based on acoustic emission with novel feature and hybridized feature selection

“…28 By selecting features with hybrid ways and fusing multi-information from multi-sensors, the reliability of results can greatly improve. 29 Therefore, great importance should be attached to multi-sensor information fusion and hybridized FS methods for compressor blades.…”