郭继通,郑方志,徐成宇,朱永伟.基于遗传算法和神经网络的软脆工件研磨加工工艺智能决策系统[J].表面技术,2020,49(4):23-29.
GUO Ji-tong,ZHENG Fang-zhi,XU Cheng-yu,ZHU Yong-wei.Intelligent Decision System for Lapping Process of Soft and Brittle Workpiece Based on Genetic Algorithm and Neural Network[J].Surface Technology,2020,49(4):23-29
基于遗传算法和神经网络的软脆工件研磨加工工艺智能决策系统
Intelligent Decision System for Lapping Process of Soft and Brittle Workpiece Based on Genetic Algorithm and Neural Network
投稿时间:2019-10-29  修订日期:2020-04-20
DOI:10.16490/j.cnki.issn.1001-3660.2020.04.003
中文关键词:  智能决策  神经网络  遗传算法  研磨加工  抛光  工艺决策
英文关键词:intelligent decision  neural network  genetic algorithm  lapping processing  polishing  process planning
基金项目:国家自然科学基金项目(51675276)
作者单位
郭继通 1.南京航空航天大学,南京 210001 
郑方志 2.上海航天精密机械研究所,上海 201600 
徐成宇 1.南京航空航天大学,南京 210001 
朱永伟 1.南京航空航天大学,南京 210001 
AuthorInstitution
GUO Ji-tong 1.Nanjing University of Aeronautics and Astronautics, Nanjing 210001, China 
ZHENG Fang-zhi 2.Shanghai Spaceflight Precision Machinery Institute, Shanghai 201600, China 
XU Cheng-yu 1.Nanjing University of Aeronautics and Astronautics, Nanjing 210001, China 
ZHU Yong-wei 1.Nanjing University of Aeronautics and Astronautics, Nanjing 210001, China 
摘要点击次数:
全文下载次数:
中文摘要:
      目的 解决研磨抛光工艺决策中工艺试验耗时耗力的问题,实现在研磨抛光加工中根据加工工艺参数对加工质量进行预估。方法 采用遗传算法优化的BP神经网络为主要算法,构建智能预测模型,建立研磨加工中输入参数和输出参数之间的映射关系。然后收集有效的输入参数和输出参数作为网络训练和测试的样本数据集,通过遗传算法对神经网络的初始化权值和偏置进行优化,用样本数据集训练神经网络。同时,在决策系统的理论基础上,将神经网络与决策系统进行结合,利用神经网络的学习能力建立智能决策的数据库和规则库,最终建立智能决策系统。结果 与无改进的BP神经网络的决策方法相比,无论是在预测精度,还是学习速度上,遗传算法优化的神经网络性能更加优异,决策系统的决策效果更好。结论 研磨加工工艺智能决策系统是可行的,为研磨加工的工艺决策提供了一种新的思路。
英文摘要:
      In order to solve the problem of time-consuming and labor-intensive process testing in the decision-making process of lapping/polishing, and estimate the process quality according to the process parameters in the lapping/polishing process. The BP neural network optimized by genetic algorithm was used as the main algorithm to construct the intelligent prediction model, and establish the mapping relationship between input parameters and output parameters in the lapping process. Then the effective input and output parameters were collected as sample data sets for network training and testing. The initialization weights and offsets of the neural network were optimized by genetic algorithm, and the neural network was trained with the sample data sets. Meanwhile, based on the theory of decision-making system, the neural network was combined with the decision-making system, and the learning ability of the neural network was used to build the database and rule base of intelligent decision-making, and finally the intelligent decision-making system was established. Compared with the decision-making method without improved BP neural network, the neural network performance optimized by genetic algorithm is better in both prediction accuracy and learning speed, but the decision-making system has better decision-making effect. It verifies the feasibility of the intelligent decision-making system of the lapping process and provides a new idea for the process decision of the lapping process.
查看全文  查看/发表评论  下载PDF阅读器
关闭

关于我们 | 联系我们 | 投诉建议 | 隐私保护 | 用户协议

您是第19516356位访问者    渝ICP备15012534号-3

版权所有:《表面技术》编辑部 2014 surface-techj.com, All Rights Reserved

邮编:400039 电话:023-68792193传真:023-68792396 Email: bmjs@surface-techj.com

渝公网安备 50010702501715号