Scholarlink

<p>Accurate prediction and optimization of thermal deformation behavior are critical for cleaner production and sustainable manufacturing. This study develops a support vector regression (SVR) model to predict the hot-compression flow stress of a 2D-reinforced magnesium alloy based on temperature, strain rate, and strain. The baseline SVR achieved <i>R</i>2 = 0.9998, AARE = 1.15% and MSE = 0.897 MPa2 on validation. After metaheuristic optimization, the SSA-optimized SVR achieved <i>R</i>2 = 0.9996, AARE = 0.74% and MSE = 0.413 MPa2 on independent test data, outperforming both GA- and SPO-tuned variants. This level of predictive accuracy can minimize experimental iterations thereby reducing energy consumption and material scrap during forming processes. By integrating machine learning with advanced optimization techniques, our work provides a novel, data-driven pathway toward greener manufacture of 2D-reinforced lightweight alloys, directly aligning with the principles of cleaner production and resource efficiency.</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

AI-Optimized Prediction of Hot-Compression Behavior of 2D-Reinforced Magnesium Alloy for Cleaner Production

  • 期刊论文

  • 发表日期：

    2026年2月2日

查看全文
  • Ayoub Elajjani,
  • Chaoyang Sun,
  • Rami Farhat,
  • Amer Baras,
  • Yuyan Zhang,
  • Xianyou Mo,
  • Sun Zhihui

摘要

Accurate prediction and optimization of thermal deformation behavior are critical for cleaner production and sustainable manufacturing. This study develops a support vector regression (SVR) model to predict the hot-compression flow stress of a 2D-reinforced magnesium alloy based on temperature, strain rate, and strain. The baseline SVR achieved R2 = 0.9998, AARE = 1.15% and MSE = 0.897 MPa2 on validation. After metaheuristic optimization, the SSA-optimized SVR achieved R2 = 0.9996, AARE = 0.74% and MSE = 0.413 MPa2 on independent test data, outperforming both GA- and SPO-tuned variants. This level of predictive accuracy can minimize experimental iterations thereby reducing energy consumption and material scrap during forming processes. By integrating machine learning with advanced optimization techniques, our work provides a novel, data-driven pathway toward greener manufacture of 2D-reinforced lightweight alloys, directly aligning with the principles of cleaner production and resource efficiency.