Quantitative evaluation of surface crack in ferromagnetic materials based on Bayesian network in eddy current testing

Quantitative evaluation of surface cracks using detection signal is of great significance for accurate prediction of cracks in eddy current testing. It is very difficult to evaluate both the width and depth of small cracks. A quantitative evaluation method based on Bayesian network is proposed for estimating the width and depth of surface cracks in the ferromagnetic materials. First, the simulation model of eddy current testing (ECT) is established and verified by the experimental results. Then, the variation of induced voltage with crack size is studied. Four feature points of real and imaginary part of induced voltage of the receiver coil are selected to characterize the crack size. Finally, a Bayesian network is applied to evaluate the crack size based on numerical simulation results. The evaluation results show that Bayesian network can accurately estimate the width and depth of small cracks.