Automatic Crack Locating Algorithm Using a Convolution Kernel Based on Eddy Current Field for the Fractal Planar Eddy Current Probe

The quantitative evaluation of defects is one of the challenges of eddy current (EC) testing. Efficient algorithm that requires a small amount of time and hardware resources is needed. In this article, based on the principle of the interaction between the EC field and the cracks, a ternary convolution kernel convolution algorithm based on an EC field is proposed to process the C-scan signal of the planar Koch EC sensor and locate the 28 cracks. The proposed algorithm comprises three steps: convolution kernel preparation, signal preprocessing, and convolution and positioning. The absolute and relative errors of the position relation between the #1 and the other cracks were less than 1.95 mm and 2.61%, respectively. This algorithm can be used for differential EC sensors to auto-locate the crack.