• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2026, Vol. 62 ›› Issue (18): 348-369.doi: 10.3901/JME.260531

• 材料科学与工程 • 上一篇    

扫码分享

7075铝合金激光除漆性能研究及多目标参数优化分析

王蔚1,2, 李晓鹏1,2, 寇俊源1,2, 刘伟军1,2, 钟家兴1,2, 王梦奎1,2   

  1. 1. 沈阳工业大学机械工程学院 沈阳 110870;
    2. 辽宁省激光表面工程技术重点实验室 沈阳 110870
  • 收稿日期:2025-12-03 修回日期:2026-07-05 发布日期:2026-11-20
  • 作者简介:王蔚(通信作者),女,1984年出生,博士,副教授,博士研究生导师。主要研究方向为先进激光加工技术,包括激光清洗、激光3D打印技术和激光再制造等激光加工技术。E-mail:wangwei_me@sut.edu.cn
  • 基金资助:
    国家自然科学基金(52375456)和辽宁省教育厅科研平台建设(LJ232510142004)资助项目。

Research on Laser Paint Removal Performance of 7075 Aluminum Alloy and Multi-objective Parameter Optimization Analysis

WANG Wei1,2, LI Xiaopeng1,2, KOU Junyuan1,2, LIU Weijun1,2, ZHONG Jiaxing1,2, WANG Mengkui1,2   

  1. 1. School of Mechanical Engineering, Shenyang University of Technology, Shenyang 110870;
    2. Liaoning Key Laboratory of Laser Surface Engineering Technology, Shenyang 110870
  • Received:2025-12-03 Revised:2026-07-05 Published:2026-11-20

摘要: 在激光清洗实验中,工艺参数众多并且对清洗质量的影响呈现出复杂的矛盾关系,因此本研究旨在探索一种激光除漆7075铝合金包铝多目标参数优化方法。实验采用纳秒脉冲光纤激光器去除7075铝合金包铝表面漆层,研究激光平均功率、激光行进速度及激光扫描速度对除漆后表面二维形貌、粗糙度、润湿性、附着力及显微硬度的影响规律,因此本研究针对去除深度、表面粗糙度和显微硬度三个目标,构建回归预测响应面模型和PSO-BP神经网络模型,通过分析发现PSO-BP神经网络模型具有更高的预测精度。结合实验结果提出一种基于粒子驱动和自适应阈值的鲸鱼优化算法。进一步地,与其他主流算法(如NSGA-Ⅱ、PSO、DE等7种算法)进行对比,得到的解集在均匀性、收敛性和多样性表现更好。经实验验证去除深度低至40.8 μm,面粗糙度达到0.913 μm,显微硬度提高到59.98 HV。

关键词: 激光除漆, 预测模型, 多目标优化, 润湿性, 附着力

Abstract: In laser cleaning experiments, there are numerous process parameters and their impact on cleaning quality presents a complex contradictory relationship. Therefore, a multi-objective parameter optimization method for laser paint removal on alclad 7075 aluminum alloy is studied to explore in this study. A nanosecond pulse fiber laser is used to remove the paint layer on the surface of alclad 7075 aluminum alloy, and the influence laws of laser average power, laser travel speed, and laser scanning speed on the post-paint-removal surface two-dimensional morphology, surface roughness, wettability, adhesion, and microhardness are investigated. Consequently, targeting three objectives removal depth, surface roughness, and microhardness, regression prediction models including the response surface model and the PSO-BP neural network model are also constructed in this study. Moreover, it is found that the PSO-BP neural network model has higher prediction accuracy through analysis. Combined with the experimental results, a whale optimization algorithm based on particle drive and adaptive threshold is proposed. Furthermore, compared with other mainstream algorithms (seven algorithms including NSGA-II, PSO, DE, etc), the obtained solution set performs better in uniformity, convergence, and diversity. Experimental verification shows that the removal depth is as low as 40.8μm, the surface roughness reaches 0.913 μm, and the microhardness increases to 59.98 HV.

Key words: laser paint removal, predictive models, multi-objective optimization, wettability, adhesion

中图分类号: