Application of deep learning procedure to magnetic multi-sensor matrix transducer data for the need of defect characterization in steel elements

In this paper a machine learning algorithm was applied to analyze the integrated data responses of magnetic multi-sensor matrix transducer. In order to proceed the algorithm, a large database was prepared considering measurement as well as simulation results. A finite element method was utilized to conduct a numerical calculation and provide transducers response signals for the various configurations of defects. The deep artificial neural network was used for implementation of machine learning algorithm and then the model was verified.