摘要
为提高公差分配精度,通过分析确定以制造成本和质量损失作为目标函数,建立基于成本和质量的公差分配多目标优化函数模型。使用违背度函数法处理模型约束,对初始种群进行分组并计算组内最优解Xteacher;在“教”阶段,采用DFP变尺度法对Xteacher加强局部搜索,保证学生解向教师学习;在“学”阶段,采用有向学习策略,引导学生向组内最优学生学习,使学习具有一定方向性。最后以减速箱装配体公差设计实例验证所提算法和模型的有效性。
In order to improve the accuracy of tolerance allocation,the multi-objective optimization function model of tolerance allocation based on cost and quality was established by analyzing and determining the manufacturing cost and quality loss in the process as the objective function.The initial population is grouped and the optimal solution within the group X teacher is calculated.In the"teaching"stage,the DFP variable scaling method is used to strengthen the local search of X teacher to ensure that students learn from teachers;in the"learning"stage,the directed learning strategy is used to guide students to learn from the optimal students in the group.So that learning has a certain direction.Finally,an example of gearbox assembly tolerance design is used to verify the effectiveness of the proposed algorithm and model.
作者
王宸
向长峰
王生怀
WANG Chen;XIANG Changfeng;WANG Shenghuai(Shanghai University,Shanghai Key Laboratory of Intelligent Manufacturing and Robotics,Shanghai 201900,CHN)
出处
《制造技术与机床》
北大核心
2019年第10期95-98,共4页
Manufacturing Technology & Machine Tool
基金
国家自然科学基金项目(51475150)
湖北省教育厅青年基金项目(Q20181801)
汽车动力传动与电子控制湖北省重点实验室(湖北汽车工业学院,ZDK1201703)
关键词
公差分配
教与学算法
变尺度
PARETO最优解
tolerance allocation
tethering and learning algorithm
scaleable
pareto optimization
作者简介
第一作者:王宸,男,1983年生,博士,副教授,主要研究方向为智能制造、计算智能。
