A multi-source information fusion approach for crack detection of 42CrMo steel during fatigue test

摘要

For non-destructive testing (NDT) to be applicable in mechanical equipment, it is mandatory that NDT is highly flexible, accurate and robust. At present there have been several methods based on acoustic, optical and electromagnetic information to detect the cracks of the mechanical equipment. As a new technology, the fusion of testing information harnesses the advantages of various NDT instruments, and thus improve the accuracy of the result. In this paper, a multi-source information fusion approach for the cracks of 42CrMo steel is presented. First, we collect the information from the acoustic emission, machine vision and infrared instruments. Second, we take advantage of the software Matlab to build a fusion model, which can enhance the relationship of the detective information to increase the crack's recognition rate greatly. Finally, we analyze the cracks' location and length based on the result of the model. We evaluate the proposed fusion model, and the experiment shows that the fusion result achieves a better performance than single testing.