基于支持向量机模型的激光相变硬化性能预测研究

【作者】 朱明华； 张国栋； 冯亿生； 张富巨； 李志远；

【机构】 武汉大学材料工程系； 华中科技大学材料科学与工程学院；

【摘要】 本文基于支持向量机(SVM)的理论,建立了激光相变硬化表面性能预测模型,通过对模型结构的分析,提出了模型学习参数的选取方法,并采用序列最小最优化(SMO)算法实现模型参数的优化。用得到的优化模型预测和检验了多种工艺参数下激光相变强化淬硬区尺寸和淬硬区平均硬度,和实验值进行比较,其预测相对误差很小。其结果表明SVM预测模型的预测精度高,泛化能力强,可应用于激光相变硬化性能预测和指导激光相变强化工艺参数制定。更多还原

【Abstract】 In this paper,a prediction model based on SVM is developed for forecasting the properties of laser trans- formation hardening in roller steel hard facing.A new approach of selecting learning parameters is proposed on analysis of model structure,and an algorithm of sequential minimal optimization(SMO)is used for optimizing the model parameters. The optimized model is applied for forecasting and verifying dimensions and average hardness of hardening zone of steels after laser phase-change hardening with several kinds of processing parameters.Comprise to the actual experimental val- ues,the forecasting relative error is litter.The result shows that SVM forecasting model has high accuracy and good gener- alized performance and it could be used for forecasting the properties of laser transformation hardening and drawing up the process parameters of laser phase-change hardening.更多还原