Scholarlink

<p>This study presents a machine learning-based optimization framework for enhancing the thermal–hydraulic performance of plate-fin heat exchangers. Using Gaussian process regression (GPR) and artificial neural network (ANN) models trained on orthogonal design data, we established mappings between fin geometry parameters and performance indicators (<i>j</i> and <i>f</i> factors). The GPR model outperformed ANN, achieving validation <i>R</i><sup>2</sup> values of 0.9840–0.9629 for the <i>j</i> and <i>f</i> factors, respectively. Sensitivity analysis identified fin height and length as the most influential parameters. Through NSGA-II optimization, Pareto-optimal designs achieved a 9.19% higher <i>f</i> factor at fixed <i>j</i> factor or a 20.38% higher <i>j</i> factor at fixed <i>f</i> factor. The optimal configuration improved the comprehensive performance evaluation criterion by 9% under varying flow conditions. This research provides an effective optimization strategy for heat exchanger design, contributing to the enhancement of the comprehensive performance of heat exchangers.</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 fin structure parameters for plate-fin heat exchangers based on the machine learning model

  • 期刊论文

  • 发表日期：

    2026年7月8日

查看全文
  • Qifan Wang,
  • Jiahui Zhang,
  • Pengyang Ye,
  • Kejing Lu,
  • Shengqi Zhang,
  • Zhentao Liu

摘要

This study presents a machine learning-based optimization framework for enhancing the thermal–hydraulic performance of plate-fin heat exchangers. Using Gaussian process regression (GPR) and artificial neural network (ANN) models trained on orthogonal design data, we established mappings between fin geometry parameters and performance indicators (j and f factors). The GPR model outperformed ANN, achieving validation R2 values of 0.9840–0.9629 for the j and f factors, respectively. Sensitivity analysis identified fin height and length as the most influential parameters. Through NSGA-II optimization, Pareto-optimal designs achieved a 9.19% higher f factor at fixed j factor or a 20.38% higher j factor at fixed f factor. The optimal configuration improved the comprehensive performance evaluation criterion by 9% under varying flow conditions. This research provides an effective optimization strategy for heat exchanger design, contributing to the enhancement of the comprehensive performance of heat exchangers.