Scholarlink

<p>Experimental studies on frost thickness and surface roughness on inclined cold surfaces were presented and reliable machine-learning prediction models were developed for 2160 sets of data collected separately. The experimental results showed that frost growth was easily influenced by the inclination angle during the early frosting stage, frost thickness on inclined cold surfaces was thick in the upper and lower regions and thin in the middle part, with a high level of frost surface roughness. Out of these machine prediction techniques, the extreme gradient boosting (XGBoost) models produced the best results for these two frost characterization parameter datasets, with average absolute relative error (AARE) values of 1.68 % and 2.02 %, respectively. Eventually, the leverage method was adopted to show the viability of the machine prediction approaches, and sensitivity analyses were carried out to identify the variables that had the most effects on the two frost parameters.</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

Experimental study and machine learning prediction of frost characteristics on inclined cold plates

  • 期刊论文

  • 发表日期：

    2025年1月31日

查看全文
  • Jingjing Guo,
  • Jianying Gong,
  • Zhen Jiang,
  • Xin Wu,
  • Ruining Huang,
  • Yutao Wang

摘要

Experimental studies on frost thickness and surface roughness on inclined cold surfaces were presented and reliable machine-learning prediction models were developed for 2160 sets of data collected separately. The experimental results showed that frost growth was easily influenced by the inclination angle during the early frosting stage, frost thickness on inclined cold surfaces was thick in the upper and lower regions and thin in the middle part, with a high level of frost surface roughness. Out of these machine prediction techniques, the extreme gradient boosting (XGBoost) models produced the best results for these two frost characterization parameter datasets, with average absolute relative error (AARE) values of 1.68 % and 2.02 %, respectively. Eventually, the leverage method was adopted to show the viability of the machine prediction approaches, and sensitivity analyses were carried out to identify the variables that had the most effects on the two frost parameters.