Scholarlink

<p>This study investigates the tribological behavior of wire arc additive-fabricated SS316L stainless steel, concentrating on optimizing and predicting specific wear rates (SWRs) and coefficients of friction (COFs) using an ensemble machine learning model. This model, which integrates artificial neural networks, long short-term memory, and random forest, attained high prediction accuracy (<i>R</i><sup>2</sup> = 0.9946 for SWR, <i>R</i><sup>2</sup> = 0.9884 for COF), effectively minimizing prediction errors and reducing experimental trials. The optimized parameters (sliding velocity: 40&#xa0;cm/s, load: 27.27&#xa0;N, sliding distance: 603.6&#xa0;m) decreased the SWR (6.62159 × 10<sup>−5</sup>&#xa0;mm<sup>3</sup>/Nm) and COF (0.399738). Scanning electron microscopy analysis indicated a transition from mild abrasive wear at lower loads to severe adhesive wear at higher loads, while EDS confirmed that key elements influence wear resistance. This study highlights the combined efficacy of RSM optimization and ensemble machine learning in improving predictive accuracy and optimizing tribological performance.</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

Assessment of the Tribological Behavior of WAAM-Fabricated SS316L through Advanced Ensemble Machine Learning Predictions and RSM-Based Optimization

  • 期刊论文

  • 发表日期：

    2025年8月4日

查看全文
  • S. Saravanan,
  • Saravanakumar Sengottaiyan,
  • Ra. Aravind,
  • S. Krishnakumar

摘要

This study investigates the tribological behavior of wire arc additive-fabricated SS316L stainless steel, concentrating on optimizing and predicting specific wear rates (SWRs) and coefficients of friction (COFs) using an ensemble machine learning model. This model, which integrates artificial neural networks, long short-term memory, and random forest, attained high prediction accuracy (R2 = 0.9946 for SWR, R2 = 0.9884 for COF), effectively minimizing prediction errors and reducing experimental trials. The optimized parameters (sliding velocity: 40 cm/s, load: 27.27 N, sliding distance: 603.6 m) decreased the SWR (6.62159 × 10−5 mm3/Nm) and COF (0.399738). Scanning electron microscopy analysis indicated a transition from mild abrasive wear at lower loads to severe adhesive wear at higher loads, while EDS confirmed that key elements influence wear resistance. This study highlights the combined efficacy of RSM optimization and ensemble machine learning in improving predictive accuracy and optimizing tribological performance.

Graphical Abstract