基于支持向量机的铝基碳化硅磨削表面质量预测

目的针对传统粗糙度指标评价具有凹坑缺陷的铝基碳化硅磨削表面质量的局限性,提出基于三维形貌的改进表面粗糙度评价指标及其预测模型。方法基于磨削表面三维形貌构建等高线图,获取等高线轮廓间面积占比与轮廓高度的关系曲线,提出表面三维形貌体积相对于采样区域面积的算术平均偏差和凹坑最大偏离高度评价指标,用于表征包含凹坑缺陷的磨削表面质量。基于支持向量机建立和优化三维形貌算术平均偏差和凹坑最大偏离高度的预测模型,并分析磨削工艺参数对评价指标的影响规律。结果三维形貌算术平均偏差和凹坑最大偏离高度评价指标包含凹坑缺陷等更多表面特征,评价指标预测值与实验值误差在5%以内,且随着砂轮转速的增大而减小,随着进给速度与磨削深度的增大而增大。结论采用三维形貌算术平均偏差和凹坑最大偏离高度评价包含凹坑缺陷的磨削表面质量是合理的,评价指标测量和确定方法是可行和有效的。基于支持向量机的评价指标预测方法具有正确性,为铝基碳化硅磨削表面质量评价和使用性能研究打下了基础。

Surface Quality Prediction of SiCp/Al Composite in Grinding Based on Support Vector Machine

The work aims to propose the modified surface roughness evaluation indexes based on topography and the predic-tion model since the traditional surface roughness indexes have limitations on the evaluation of cavity defect in the grinding process of SiCp/Al composite. The contour map was constructed based on surface topography to obtain the relationship curve between the projection area of adjacent contour and contour height. The average deviation of whole topography to the sampling area and the maximum deviation height index of cavity defect were proposed to characterize grinding surface quality including cavity defect. Based on Support Vector Machine (SVM), the two prediction models of average deviation of topography and the maximum devia-tion height of cavity were proposed and optimized and the influence of laws of grinding parameters on evaluation index was ana-lyzed. The evaluation indexes of the average deviation of whole topography and the maximum deviation height of cavity defect in-cluded more surface features. The errors between the predicted and experimental results were both within 5% and decreased with the increase of wheel speed, and increased with the increase of feed speed and grinding depth. It is reasonable to use the arithmetic average deviation of topography and the maximum deviation height of cavity to evaluate the grinding surface quality including pit defects, and the measurement and determination method of evaluation index is feasible and effective. The prediction method of evaluation index based on support vector machine is correct, which lays a foundation for the evaluation of grinding surface quality of aluminum-based silicon carbide and the study of application performance.