Quantitative analysis of defect profile based on multiple linear regression method

In order to quantify the two-dimensional edge contour area of defects in magnetic flux leakage detection, the rule of magnetic refraction diffusion at the defect was obtained based on the magnetic leakage process. The influence of changes in the two-dimensional edge contour area of defects on magnetic flux leakage signals was studied and five groups of characteristic values for axial and radial magnetic flux components were extracted. Using the least square method and multiple regression method, the characteristic value of MFL signal was fitted and analyzed with respect to the two-dimensional edge contour area of defects, and an MFL detection experimental platform for verification was built. The results showed a corresponding relationship between the two-dimensional edge contour area of defects and eigenvalues of axial and radial flux components. The use of multiple linear regression method resulted in higher quantization accuracy for defect evaluation using magnetic flux leakage signals. Multiple linear regression could be a feasible analysis method for defect parameters inversion using leakage magnetic signals.