基于BO-RF的多道熔覆层表面平整度的优化
Optimization of surface flatness of multi-pass cladding layer based on BO-RF
梁强 1徐永航 2杜彦斌 1何国华 2周志杰3
作者信息
- 1. 重庆工商大学制造装备机构设计与控制重庆市重点实验室,重庆 400067;重庆工商大学机械工程学院,重庆 400067
- 2. 重庆工商大学机械工程学院,重庆 400067
- 3. 广东工业大学机电工程学院,广东 广州 510006
- 折叠
摘要
为了在20Cr13 马氏体不锈钢表面熔覆多道高平整度的15-5PH合金粉末熔覆层,提出一种基于贝叶斯优化、随机森林回归的 15-5PH合金粉末激光熔覆工艺参数优化方法.以激光功率、扫描速度、送粉速率设计三因素五水平的全因子实验,并构建工艺参数与单道熔覆层形貌的BO-RF回归预测模型.通过NSGA-Ⅱ算法对单道激光熔覆工艺参数进行多目标优化,获得Pareto最优解集,并利用TOPSIS结合EWM对Pareto解集进行综合评价.基于单道次熔覆层形貌和平顶重叠模型计算单层多道激光熔覆的搭接率,确定出最佳工艺参数组合进行工艺实验.结果表明:采用最佳工艺参数制备的熔覆层表面平整度达到 83.5%.采用该优化方法进行多道次激光熔覆可以改善熔覆层的表面平整度和减少后续机械加工余量.
Abstract
In order to coat the surface of 20Cr13 martensitic stainless steel with multiple layers of high flatness 15-5PH alloy powder cladding layer,a method for optimizing the process parameters of 15-5PH alloy powder laser cladding based on Bayesian optimization and random forest regression was proposed.A full factorial experiment with three factors and five levels was designed with laser power,scanning speed and powder feeding rate,and a BO-RF regression prediction model for process parameters and single pass cladding layer morphology was constructed.Multi objective optimization of single-pass laser cladding process parameters was carried out using the NSGA-Ⅱ algorithm to obtain the Pareto optimal solution set,and the Pareto solution set was comprehensively evaluated by TOPSIS combined with EWM.Based on the single-pass cladding layer morphology and flat-top overlap model,the overlap rate of single-layer multi-pass laser cladding was calculated,and the optimal combination of process parameters was determined for process experiments.The results show that the surface flatness of the cladding layer prepared with the optimal process parameters reaches 83.5%.Using this optimization method for multi-pass laser cladding can improve the surface smoothness of the cladding layer and reduce subsequent machining allowances.
关键词
激光熔覆/15-5PH合金粉末/随机森林回归/平整度优化Key words
laser cladding/15-5PH alloy powder/random forest regression/flatness optimization引用本文复制引用
基金项目
重庆市高校创新研究群体(CXQT21024)
重庆英才计划"包干制项目"(cstc2022ycjhbgzxm0056)
2021年度校内资助项目计划(2152026)
出版年
2024
NETL
NSTL
万方数据