Hybrid RSM-fuzzy modeling for hardness prediction of TiAlN coatings

Abstract

In this paper, a new approach in predicting the hardness of Titanium Aluminum Nitrite (TiAlN) coatings using hybrid RSM-fuzzy model is implemented. TiAlN coatings are usually used in high-speed machining due to its excellent surface hardness and wear resistance. The TiAlN coatings were produced using Physical Vapor Deposition (PVD) magnetron sputtering process. A statistical design of experiment called Response Surface Methodology (RSM) was used in collecting optimized data. The fuzzy rules were constructed using actual experimental data. Meanwhile, the hardness values were generated using the RSM hardness model. Triangular shape of membership functions were used for inputs as well as output. The substrate sputtering power, bias voltage and temperature were selected as the input parameters and the coating hardness as an output of the process. The results of hybrid RSM-fuzzy model were compared against the experimental result and fuzzy single model based on the percentage error, mean square error (MSE), co-efficient determination (R2) and model accuracy. The result indicated that the hybrid RSM-fuzzy model obtained the better result compared to the fuzzy single model. The hybrid model with seven triangular membership functions gave an excellent result with respective average percentage error, MSE, R2 and model accuracy were 11.5%, 1.09, 0.989 and 88.49%. The good performance of the hybrid model showed that the RSM hardness model could be embedded in fuzzy rule-based model to assist in generating more fuzzy rules in order to obtain better prediction result.