Artificial intelligence implementation in the APS process diagnostic

Thermal spray process is a technique of coating manufacturing implementing a wide variety of materials and processes. This technique is characterized by up to 150 processing parameters influencing the coating properties. The control of the coating quality is needed through the consideration of a robust methodology that takes into account the parameter interdependencies, the process variability and offers the ability to quantify the processing parameter-process response relationships. The aim of this work is to introduce a new approach based on artificial intelligence responding to these requirements. A detailed procedure is presented considering an artificial neural network (ANN) structure which encodes implicitly the physical phenomena governing the process. The implementation of such a structure was coupled to experimental results of an optic sensor controlling the powder particle fusion state before the coating formation. The optimization steps were discussed and the predicted results were compared to the experimental ones allowing the identification of the control factors.