Scholarlink

<p>Laser cleaning parameters are numerous and significantly affect both cleaning quality and efficiency. This study aims to optimize laser cleaning parameters to simultaneously ensure effective paint removal and high surface quality for 2024 aluminum alloy. A Box-Behnken design (BBD) was used with laser power, pulse frequency, and scanning speed as input variables, and titanium content, surface roughness, and static contact angle as response variables. Additionally, a response surface methodology (RSM) model and a backpropagation (BP) neural network optimized by an improved atom search optimization (ASO) algorithm were established, and their predictive accuracy and generalization capabilities were comparatively analyzed. Moreover, an enhanced multi-objective particle swarm optimization (MOPSO) algorithm was developed by integrating Logistic chaotic mapping, adaptive inertia weights, and K-means–assisted boundary constraints. While the improved MOPSO was combined with the superior prediction model and benchmarked against several other algorithms based on multiple Pareto-optimal solution sets to evaluate optimization performance. Furthermore, the entropy weight–TOPSIS method was employed to determine the optimal process parameters from the Pareto front, resulting in a laser power of 119.48 W, pulse frequency of 19.81&#xa0;kHz, and scanning speed of 2038.76&#xa0;mm/s. Compared with the optimal specimen from the BBD, while achieving a marginal 5.1° increase in static contact angle, the proposed method reduced titanium content by 28.43% and surface roughness by 54.09%, demonstrating an improvement in cleaning performance. In conclusion, this study provides practical guidance for the application of laser cleaning technology in removing paint layers from aluminum alloy surfaces.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multi-objective optimization of laser paint removal on 2024 aluminum alloy using hybrid neural networks and improved MOPSO

  • 期刊论文

  • 发表日期：

    2025年6月5日

查看全文
  • Wei Wang,
  • Jiaxing Zhong,
  • Wei Wang,
  • Weijun Liu,
  • Hongyou Bian,
  • Hongbo Zhao,
  • Yu Zhou

摘要

Laser cleaning parameters are numerous and significantly affect both cleaning quality and efficiency. This study aims to optimize laser cleaning parameters to simultaneously ensure effective paint removal and high surface quality for 2024 aluminum alloy. A Box-Behnken design (BBD) was used with laser power, pulse frequency, and scanning speed as input variables, and titanium content, surface roughness, and static contact angle as response variables. Additionally, a response surface methodology (RSM) model and a backpropagation (BP) neural network optimized by an improved atom search optimization (ASO) algorithm were established, and their predictive accuracy and generalization capabilities were comparatively analyzed. Moreover, an enhanced multi-objective particle swarm optimization (MOPSO) algorithm was developed by integrating Logistic chaotic mapping, adaptive inertia weights, and K-means–assisted boundary constraints. While the improved MOPSO was combined with the superior prediction model and benchmarked against several other algorithms based on multiple Pareto-optimal solution sets to evaluate optimization performance. Furthermore, the entropy weight–TOPSIS method was employed to determine the optimal process parameters from the Pareto front, resulting in a laser power of 119.48 W, pulse frequency of 19.81 kHz, and scanning speed of 2038.76 mm/s. Compared with the optimal specimen from the BBD, while achieving a marginal 5.1° increase in static contact angle, the proposed method reduced titanium content by 28.43% and surface roughness by 54.09%, demonstrating an improvement in cleaning performance. In conclusion, this study provides practical guidance for the application of laser cleaning technology in removing paint layers from aluminum alloy surfaces.