Machine learning approach for estimating coating layer thickness in an electric discharge coating process with ANN and ANFIS

Machine learning is an efficient method for optimizing parameters and predicting models, surpassing other numerical-statistical prediction tools. It improves system quality, productivity, and reduces manufacturing costs. This study focuses on coating Mg alloy substrates with Cu-Ni green compact electrodes using an electric discharge coating (EDC) process. This experiment is carried out with input parameters of peak current (2, 3, 4A), pulse duration (100, 150, 200µs) and electrode compaction load (50, 100, 150MPa). The main objective of this study was analysis the prediction ability of ANN and ANFIS tools. The results obtained indicate that increasing the peak current leads to an increase in the coating layer thickness. Additionally, achieving a superior coating thickness is feasible by utilizing lower compaction load and pulse duration. In terms of prediction models, the ANFIS model outperforms ANN, offering a more accurate and dependable prediction value.