Scholarlink

<p>Laser transmission welding (LTW) is a promising method for fusing polymers and hybrid structures, widely used in automotive, aerospace, and healthcare applications due to its localized heating and process efficiency. Acrylonitrile–butadiene–styrene (ABS) polymer is widely utilized due to its favorable mechanical and chemical properties. This study develops a hybrid intelligence approach by coupling an artificial neural network (ANN) with the non-dominated sorting genetic algorithm II (NSGA-II) to simultaneously predict, analyze, and optimize weld quality in LTW of ABS. Experiments are conducted to generate welding parameter–weld quality data for ANN by varying laser power, scanning speed, stand-off distance, and clamp pressure, with weld strength and weld width as performance responses. The ANN with a 4-5-2 architecture, trained using the Levenberg–Marquardt algorithm, demonstrates high predictive accuracy, validated by strong regression (<i>R</i> ≈ 0.99) across training, validation, and test datasets, along with low error metrics (RMSE and MAE) on independent test data. The model is further utilized to analyze parametric interactions and quantify sensitivity, identifying scanning speed as the most influential parameter governing both responses. The proposed ANN–NSGA-II framework generates a Pareto front with 19 optimal trade-off solutions, capturing distinct thermo-physical regimes governed by heat input and energy distribution. The best trade-off solution yields a maximum weld strength of 47.52 N/mm and a minimum weld width of 2.31&#xa0;mm, effectively balancing conflicting objectives. Experimental validation confirms close agreement with predictions, demonstrating the robustness and practical applicability of the proposed approach.</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

Analysis and Enhancement of Performance Characteristics in Laser Transmission Welding of ABS via Machine Learning: Leveraging ANN and NSGA-II Integration

  • 期刊论文

  • 发表日期：

    2026年7月8日

查看全文
  • Ghulam Anwer,
  • Bappa Acherjee

摘要

Laser transmission welding (LTW) is a promising method for fusing polymers and hybrid structures, widely used in automotive, aerospace, and healthcare applications due to its localized heating and process efficiency. Acrylonitrile–butadiene–styrene (ABS) polymer is widely utilized due to its favorable mechanical and chemical properties. This study develops a hybrid intelligence approach by coupling an artificial neural network (ANN) with the non-dominated sorting genetic algorithm II (NSGA-II) to simultaneously predict, analyze, and optimize weld quality in LTW of ABS. Experiments are conducted to generate welding parameter–weld quality data for ANN by varying laser power, scanning speed, stand-off distance, and clamp pressure, with weld strength and weld width as performance responses. The ANN with a 4-5-2 architecture, trained using the Levenberg–Marquardt algorithm, demonstrates high predictive accuracy, validated by strong regression (R ≈ 0.99) across training, validation, and test datasets, along with low error metrics (RMSE and MAE) on independent test data. The model is further utilized to analyze parametric interactions and quantify sensitivity, identifying scanning speed as the most influential parameter governing both responses. The proposed ANN–NSGA-II framework generates a Pareto front with 19 optimal trade-off solutions, capturing distinct thermo-physical regimes governed by heat input and energy distribution. The best trade-off solution yields a maximum weld strength of 47.52 N/mm and a minimum weld width of 2.31 mm, effectively balancing conflicting objectives. Experimental validation confirms close agreement with predictions, demonstrating the robustness and practical applicability of the proposed approach.