Arbitrary Crack Depth Profiling Through ACFM Data Using Type-2 Fuzzy Logic and PSO Algorithm

Estimating the shape and depth of cracks presented in metallic structures is one of the main issues of non-destructive testing (NDT) in order to evaluate effectively the structural integrity of a component. The alternating current field measurement (ACFM) technique is one of the most frequently used electromagnetic methods in this regard. Given the experimental nature of NDT methods, fuzzy logic-based methodologies have been widely used for solving the inverse problem. Due to some experimental restrictions, the obtained ACFM signals do not have high certainty. This problem usually leads to a high uncertainty of classical fuzzy rules extracted from low-accuracy ACFM signals. Therefore, applying classical fuzzy membership functions (MFs) exactly with maximum and fixed certainty does not lead to the best crack depth estimation. In this paper, a type-2 fuzzy logic system has been proposed to model the existing uncertainties of ACFM signals with a higher accuracy. Moreover, for regulating the uncertainty parameters of type-2 fuzzy MFs in the proposed model, the particle swarm optimization (PSO) algorithm has been used. Combining PSO with some special features existed in the ACFM signals allows the proposed model to be able to control the certainty of the extracted rules for estimating the exact depth of cracks. Then, the results of the proposed method are compared with the other state-of-the-art techniques for different levels of noise and different crack shapes obtained through simulated and empirical ACFM data. The results show the superiority of the proposed method even in conditions where the training database volume is not adequate.