• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2016, Vol. 52 ›› Issue (11): 207-212.doi: 10.3901/JME.2016.11.207

• 制造工艺与装备 • 上一篇    

电火花加工8418钢的工艺预测模型

余剑武1, 2, 何利华1, 段文1, 沈湘2, 易成1   

  1. 1. 湖南大学机械与运载工程学院 长沙 410082;
    2. 湖南大学国家高效磨削工程技术研究中心 长沙 410082
  • 出版日期:2016-06-05 发布日期:2016-06-05
  • 作者简介:余剑武,男,1968年出生,博士,教授,博士研究生导师。主要研究方向为特种加工技术,数字化设计与制造技术,先进材料复杂曲面精密加工技术及数控装备。E-mail:yokenbu@yahoo.com;何利华(通信作者),男,1987年出生,博士研究生。主要研究方向为特种加工技术,超硬磨料砂轮数控精密修整技术与装备。E-mail:helihua0617@yahoo.com
  • 基金资助:
    国家科技重大专项资助项目(2012ZX04003-101)

Prediction Model of Technological Indexes in EDM of 8418 Steel

YU Jianwu1, 2, HE Lihua1, DUAN Wen1, SHEN Xiang2, YI Cheng1   

  1. 1. College of Mechanical &
    Vehicle Engineering, Hunan University, Changsha 410082;
    2. National Engineering Research Center for High Efficiency Grinding, Hunan University, Changsha 410082
  • Online:2016-06-05 Published:2016-06-05

摘要: 在电火花加工中,加工工艺指标的结果与工艺参数的设置密切相关。一般情况下,操作者在进行实际执行之前,只能根据以往的加工规律以及经验手段对其结果进行预判,达到预先评估加工结果的目的。针对这一情况,提出一种适用于电火花加工工艺指标结果预测的模型,该模型的建立是基于支持向量回归理论的数学方法,并利用遗传算法优化该方法中的各参数。以电火花加工8418模具钢为例,结合正交试验方法和经验加工方法选取加工工艺参数,并记录工艺指标结果。为保证EDM工艺指标预测模型的准确性,将试验数据随机分成训练集和测试集,利用训练集训练EDM工艺指标预测模型,可得加工时间模型均方误差TMSE=0.95×10-4,平方相关系数TR2=0.99 1;工件去除率模型均方误差MRRMSE=1.02×10-4,平方相关系数MRRR2 =0.999 3;电极损耗率模型均方误差EWRMSE =1.11×10-4,平方相关系数EWR R2=0.998 9。再利用测试集验证该模型,可见预测结果与试验结果之间的误差在5%以内,从而证明电火花加工8418钢工艺预测模型的准确性和有效性。

关键词: 电火花加工, 工艺模型, 过程参数, 遗传算法, 支持向量回归

Abstract: The technological indexes are closely related to processing parameters in electrical discharge machining (EDM). In general, the output objectives only can be predicted with the help of the past processing laws and empirical methods before actually executing. In view of this situation, a prediction model is presented and established on the base of mathematical method of support vector regression to determine the technological indexes in EDM. The genetic algorithm (GA) is adopted as an optimization solver to seek the best variables in this model. In the case of processing 8418 steel, the proper process parameters are selected by orthogonal experimental method and empirical method, and the results are recorded. To ensure accuracy of prediction models of technological indexes in EDM, the experimental data are divided into a training set and a test set. The prediction model is trained through the training data set. The results show that the model of machining time has the mean squared error TMSE of 0.95×10-4 and the correlation coefficient TR2of 0.999 1; the model of material removal rate has the mean squared error MRRMSEof 1.02×10-4 and the correlation coefficient MRRR2 of 0.999 3; the model of electrode wear rate has the mean squared error EWRMSE of 1.11×10-4 and the correlation coefficient EWR R2of 0.998 9. Then those models are proved by the test samples, and the relative error is less than 5%. Thus the accuracy and effectiveness for the present models are verified.

Key words: EDM, genetic algorithm, prediction model, process variables, support vector regression

中图分类号: