Process control for plasma electrolytic removal of TiN coatings

The paper discusses two main problems of creating a control system for plasma electrolytic removal of titanium nitride coating from stainless steel hardware. The first feedback problem of unobservability of the surface state has been solved using indirect identification with an informative model of the process. The informative model has been developed on the base of the current signal processing and correlation analysis. DC component of the current and power spectral density at 1 kHz are in correlation with the surface state parameters. The second problem-inverse ill-posed control problem has been solved using inverse neural network, an object with artificial intelligence. For the solution of the inverse problem, the surface state and feedback parameters were used as the neural net inputs and voltage as the output. Because of approximation power of the neural network, the inverse problem has been resolved, and the treatment program for coating removal has been generated. Computational and practical tests show that the control system allows to intentionally change the state of the surface with 10% accuracy.