Conductivity Classification of Multi-Shape Nonmagnetic Metal Considering Spatial Position Drift Effect with a Triple-Coil Electromagnetic Sensor

The primary step in metal recovery is metal classification. During eddy current testing (ECT), the shape of the sample can have an impact on the measurement results. To classify nonmagnetic metals in three shapes—planar, cylindrical, and spherical—a triple-coil electromagnetic sensor that operates as two coil pairs is used, and the difference in the phase tangent of the impedance change of the two coil pairs is used as a feature for the classification. The effect of spatial position drift between the sensor and the sample divided into lift-off vertically and horizontal drift horizontally on this feature is considered. Experimental results prove that there is a linear relationship between the feature and lift-off regardless of the metal shape, whereas horizontal drift has no effect on this feature. In addition, the slope of the curve between the feature and the lift-off is different for different shapes. Finally, a classification method eliminating the effect of lift-off variation has been constructed, and the classification accuracy of Cu-Al-Zn-Ti metals reached 96.3%, 96.3%, 92.6%, and 100%, respectively, with an overall correct classification rate of 96.3%.