Comparison of Artificial Neural Network Approach and Data Mining Technique for the Prediction of Surface Roughness in End Milled Components with Texture Images

This article presents a procedure for using machine vision data to predict surface roughness parameter of the end milled components. Stylus based surface roughness measurements were used and compared to vision based prediction of surface roughness. Wavelet decomposition was used to extract features from vision based data. In this article wavelet Energy features of approximations and details were extracted. The proposed method utilized two different classification techniques. M5P Decision tree was used as one of the technique to classify and correlate surface roughness of milled components. Artificial Neural network was another technique. The obtained surface roughness values were compared with the stylus type surface roughness measurement. It is found that artificial neural network classification outperforms the M5P decision tree.