Effects analysis of processing factors on surface quality in optical grinding using -support vector regression

摘要

The aim of this study was to quantitatively analyze the effects of the processing factors on the surface quality in precision optical grinding. A novel identifying model which incorporates an effect factor is proposed based on -support vector regression (-SVR). Experiments were designed and performed to investigate the effects of the processing factors comprising the technological parameters and processing condition factors on the surface quality, and the experimental data were used to train the -SVR. Subsequently, the values of effect factor were solved to quantify the effects of the respective processing factors on the surface quality. Further experiments were performed to verify the effectiveness of effect factor. -SVRs of which the input vectors were multiplied and not multiplied by effect factor were respectively used to predict the surface quality including the surface roughness and surface shape peak-valley value. The values calculated by -SVR using effect factor were found to be much more accurate than those calculated without using effect factor. The results confirmed the effectiveness of identifying model for precision optical grinding.