基于SVM的轴类大锻件热处理工艺参数优化  

Parameter Optimization of Heat Treatment Process for Heavy Shaft Forgings Based on Support Vector Machine

作　　者：张建荣[1] 程秀全[1] 黄正[2] 夏琴香[2] 

机构地区：[1]广州民航职业技术学院飞机维修工程学院,广东广州510470 [2]华南理工大学机械与汽车工程学院,广东广州510640

出　　处：《热加工工艺》2014年第8期195-198,共4页Hot Working Technology

基　　金：广东省重大科技专项资助项目(2009A080304004);广州市工业科技攻关计划项目(08A39080443);韶关市科技计划项目(2011CXY/C15)

摘　　要：提出了支持向量机(SVM)与遗传算法(GA)相结合的长轴类大锻件调质热处理工艺参数的优化方法。以热处理加热温度和保温时间为优化对象,以加热时间和最大残余应力为优化目标,对长轴类大锻件的热处理工艺进行了优化。以正交试验数据为样本,采用灰色关联度分析方法把多目标转换为单目标,通过SVM神经网络建立了优化目标的回归模型;采用遗传算法对模型进行了优化并获得了最优的工艺参数。结果表明:优化工艺相对于传统的调质工艺,加热时间减少了20%,最大残余应力下降了24%。A method combining SVM （Support Vector Machine） with GA （Genetic Algorithm） to optimize the processing parameters during heat treatment of heavy shaft forgings was put forward. Taking heating temperature and preservation time as the optimum parameters, heating time and residual stress as the optimum objective, the heat treatment process of heavy shaft forgings was optimized. Based on the sample data obtained through the orthogonal experiment, the multiple optimum target was converted to a single target by the gray correlation analysis method. The regression model for the optimum objective was established based on SVM neural networks. The regression model was optimized by the GA method and the optimum heat treatment processing parameters were obtained. The results show that compared with the conventional heat treatment process, the heating time is shortened by 20% and the residual stress is reduced by 24% with the optimized orocess.

关 键 词：热处理 参数优化 正交试验 支持向量机 遗传算法 

分 类 号：TG156.6[金属学及工艺—热处理] TP301.6[金属学及工艺—金属学]