Scholarlink

<p>This study focuses on optimizing the wire arc additive manufacturing (WAAM) process for creating steel (Fe)–aluminum (Al) bimetallic structures by evaluating the impact of process parameters, such as current (I) and travel speed (TS), on intermetallic thickness (IMT), diffusion length (DL), and micro-hardness (HV). It compares two hybrid statistical approaches, where response surface methodology (RSM) is combined with the genetic algorithm (GA) and Siberian tiger optimization (STO), and two machine learning models, random forest (RF) and support vector regression (SVR), each integrated with the non-dominated sorting genetic algorithm II (NSGA II), for predicting and optimizing parameters to enhance the reliability of the bimetallic interface. The RF-NSGA II model exhibited superior prediction and optimization accuracy for Fe-Al bimetallic structures, achieving the lowest MSE of 0.19%, outperforming RSM-STO, RSM-GA, and SVR-NSGA II. It identified an optimal parameter combination of 35.44 A current and 14.26&#xa0;mm/s travel speed, yielding a heat input of 43.55&#xa0;J/mm, which produced ideal interface properties (IMT = 1.606&#xa0;μm, DL = 1.3576&#xa0;μm, HV = 1.2&#xa0;GPa). Although RF-NSGA II reduced validation MSE by 47.37% compared to RSM-STO, the latter offered 92% faster convergence, making it more suitable for time-sensitive applications. Both RSM-based methods (RSM-GA and RSM-STO) achieved approximately 1% MSE, while ML-based methods, SVR-NSGA II and RF-NSGA II, recorded 2.44% and 0.12% MSE, respectively, all within an effective MSE ≤ 5% for precise WAAM parameter control. A balanced low heat input, such as ~ 43.55&#xa0;J/mm, is critical in forming a reliable Fe-Al bimetallic interface during WAAM.</p> Graphical Abstract <p></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

Optimization of Wire Arc Additive Manufacturing Parameters for Steel–Aluminum Bimetallic Interface: A Comparative Study of Metaheuristic and Machine Learning Approaches

  • 期刊论文

  • 发表日期：

    2025年8月21日

查看全文
  • Somnath Nandi,
  • Shatarupa Biswas,
  • Manidipto Mukherjee

摘要

This study focuses on optimizing the wire arc additive manufacturing (WAAM) process for creating steel (Fe)–aluminum (Al) bimetallic structures by evaluating the impact of process parameters, such as current (I) and travel speed (TS), on intermetallic thickness (IMT), diffusion length (DL), and micro-hardness (HV). It compares two hybrid statistical approaches, where response surface methodology (RSM) is combined with the genetic algorithm (GA) and Siberian tiger optimization (STO), and two machine learning models, random forest (RF) and support vector regression (SVR), each integrated with the non-dominated sorting genetic algorithm II (NSGA II), for predicting and optimizing parameters to enhance the reliability of the bimetallic interface. The RF-NSGA II model exhibited superior prediction and optimization accuracy for Fe-Al bimetallic structures, achieving the lowest MSE of 0.19%, outperforming RSM-STO, RSM-GA, and SVR-NSGA II. It identified an optimal parameter combination of 35.44 A current and 14.26 mm/s travel speed, yielding a heat input of 43.55 J/mm, which produced ideal interface properties (IMT = 1.606 μm, DL = 1.3576 μm, HV = 1.2 GPa). Although RF-NSGA II reduced validation MSE by 47.37% compared to RSM-STO, the latter offered 92% faster convergence, making it more suitable for time-sensitive applications. Both RSM-based methods (RSM-GA and RSM-STO) achieved approximately 1% MSE, while ML-based methods, SVR-NSGA II and RF-NSGA II, recorded 2.44% and 0.12% MSE, respectively, all within an effective MSE ≤ 5% for precise WAAM parameter control. A balanced low heat input, such as ~ 43.55 J/mm, is critical in forming a reliable Fe-Al bimetallic interface during WAAM.

Graphical Abstract