Scholarlink

The prediction of Heat Transfer Coefficient (HTC) plays a critical role in optimizing the thermal performance of systems during heat treatment processes. Traditional numerical methods struggle to solve the Inverse Heat Transfer Problem (IHTP) associated with HTC prediction. In this paper, a novel approach has been proposed that leverages the power of machine learning, specifically Bidirectional Long Short-Term Memory (BiLSTM) networks, to estimate the HTC value. The developed model demonstrates remarkable precision, achieving an impressive ≈ 98.75% accuracy. This outperforms conventional feed-forward networks. The proposed machine learning approach offers several advantages over traditional methods. It provides rapid estimations of the key characteristics of the HTC function, offering quick insights into the heat transfer process. Furthermore, machine learning algorithms have the capability to learn from data and capture complex relationships, making them highly suitable for HTC estimation tasks. Compared to heuristic search algorithms and swarm-based approaches, the present approach significantly reduces computational requirements and maintains excellent predictive performance. The results obtained highlight the potential of machine learning in optimizing heat treatment processes and improving overall performance.

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

Predicting Heat Transfer Coefficient Using Bidirectional Long Short-Term Memory

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ankan Basu,
  • Aritra Saha,
  • Sumanta Banerjee

摘要

The prediction of Heat Transfer Coefficient (HTC) plays a critical role in optimizing the thermal performance of systems during heat treatment processes. Traditional numerical methods struggle to solve the Inverse Heat Transfer Problem (IHTP) associated with HTC prediction. In this paper, a novel approach has been proposed that leverages the power of machine learning, specifically Bidirectional Long Short-Term Memory (BiLSTM) networks, to estimate the HTC value. The developed model demonstrates remarkable precision, achieving an impressive ≈ 98.75% accuracy. This outperforms conventional feed-forward networks. The proposed machine learning approach offers several advantages over traditional methods. It provides rapid estimations of the key characteristics of the HTC function, offering quick insights into the heat transfer process. Furthermore, machine learning algorithms have the capability to learn from data and capture complex relationships, making them highly suitable for HTC estimation tasks. Compared to heuristic search algorithms and swarm-based approaches, the present approach significantly reduces computational requirements and maintains excellent predictive performance. The results obtained highlight the potential of machine learning in optimizing heat treatment processes and improving overall performance.