Intelligent measurement of coating thickness on steel by neural networks

The physical properties of coatings on steel are known to have an effect on the calibration of coating thickness-measuring instrumentation. Many non-destructive test methods have been applied to the thickness measurement of coatings on conductive substrates. Among them, inductive and eddy current probes have played an important role. The data acquired from eddy current probes, however, is affected by a large number of variables, including substrate conductivity, permeability, geometry, and temperature, as well as probe lift-off. The multivariable properties of coatings on substrates add an even greater level of complexity. All these variables have the effect of increasing measurement uncertainties. This paper reviews methods used in non-destructive test and discusses their merits and limitations as sensing systems applied into coating thickness on steel. An intelligent control system to accurately measure thickness of conductive and non-conductive coatings on steel substrates by using artificial neural networks, then, is investigated and described.