Industrial components are often case hardened to improve their strength and wear characteristics. Traditionally, component samples are collected from the production line at specific intervals and destructively tested for case depth profile assessment. This process is time-consuming, laborious, and can potentially allow an improperly treated component to escape detection. Non-destructive evaluation techniques can be employed for reliable case hardening characterization.
This project presents a novel nonlinear eddy current technique for assessing the case hardening profile based on the premise that the magnetic characteristic of the case hardened region is different from that of the host material.
A custom electromagnetic excitation-sensor array is used to both apply sinusoidal excitations to the component and measure the nonlinear response at multiple excitation frequencies and spatial locations, taking advantage of the different penetration regions due to the skin depth phenomenon. Each response signal obtained from the component under test is compared with that from a reference component subjected to the same excitation. Two pattern recognition algorithms (an artificial neural network and the Iterative Dichotomiser 3 (ID3) algorithm) are then used to process selected characteristics of the difference signal to determine the case depth profile of the component. The nonlinear eddy current technique has been applied to evaluate the case hardening profile of automotive bearing assemblies. This problem is challenging due to the variations in geometry across assemblies as well as the limited accessibility to the case hardened surface.