Quantitative evaluation of residual stress and surface hardness in deep drawn parts based on magnetic Barkhausen noise technology

Abstract

In this study, the method of magnetic Barkhausenn noise (MBN) and tangential magnetic field (TMF) was employed for achieving simultaneous, quantitative prediction of residual stress and surface hardness in box-shape deep drawn parts of SPCC steel. Bi-directional MBN and TMF signals, profiles of surface hardness and residual stress were experimentally measured. Pearson correlation analysis results show that the averaged amplitude of MBN in one magnetization cycle and Root mean square of MBN signal demonstrates approximate linear dependency on residual stress (or surface hardness) in both the measured directions. Three-layer BP models are established after input nodes screening and optimization of neurons’ number in the middle hidden layer. BP models are marked corresponding to scanning or magnetization directions. For each direction, two independent models are established and they are combined to use for realizing high accuracy and simultaneous evaluation of the residual stress and surface hardness.