Prediction of Hardness in TiA1N Coating Process using Support Vector Machine

Muhammad A. Mohamad, Nor A. Ali, Roselina Sallehuddin, and Habibollah Haron


Support Vector Machine, RSM-Fuzzy, hardness, TiAlN coatings, PVD magnetron sputtering


This paper comes up with new approach on predicting of hardness performances for Titanium Aluminium Nitride (TiA1N) coating process. A new application in predicting the hardness performances of TiA1N coatings using a method called Support Vector Machine (SVM) is implemented. TiAlN coatings are usually used in high-speed machining due to its excellent surface hardness and wear resistance. Physical Vapor Deposition (PVD) magnetron sputtering process has been used to produce the TiA1N coatings. Based on the experimental dataset of previous work, hence the SVM model is used in predicting the hardness of TiA1N coatings. The radial basis function (RBF) was selected as a kernel function with considered the C and Gamma as RBF parameter. For the process, the selected input parameters are the substrate sputtering power, bias voltage and temperature while the output parameter is the coating hardness. The results of proposed SVM model compared against the experimental result and hybrid RSM-Fuzzy model from previous work. The comparisons of SVM against hybrid RSM-Fuzzy was based on predictive performances in order to obtain the most accurate model for prediction of hardness in TiA1N coating. In terms of predictive performance evaluation, four performances matrix were applied that are percentage error, mean square error (MSE), co-efficient determination (R2) and model accuracy. The result has proved that the proposed SVM model shows the better result compared to the actual data (experimental result) and hybrid RSM-fuzzy model. The SVM model result of average percentage error, MSE, R2 and model accuracy were 0.4%, 0.076, 0.911 and 95.69% respectively. The good performances of the results obtained by the SVM method shows that this method can be applied for prediction of hardness performances in TiA1N coating process with better predictive performances compared to hybrid RSM-Fuzzy.

Important Links:

Go Back