关键词: 表面粗糙度, 最小二乘支持向量机, 智能预测, 铣削, 正交实验

Abstract: A novel prediction model based on Least Squares Support Vector Machine (LSSVM) with application in the milling area was proposed. Firstly, according to the basic of the design theory in orthogonal experiments and features with milling, the effect of milling technological parameters on surface roughness was analyzed. Secondly, through analyzing the experiment data, some laws about the effect of different parameters such as spindle speed, desired rate and cutting depth were found. Finally, LSSVM regression algorithm was used to set up milling prediction model, by which the roughness could be predicted from technological parameters on the limited test data. The practical experimental results showed that the speed of presented model was 23 power level higher, while accuracy predicted was about 10 times, that of BackPropagation (BP) network model under the same circumstances.

Key words: surface roughness, least squares support vector machine, intelligent prediction, milling, orthogonal experiment