Scholarlink

<p>This paper presents a comprehensive comparison of deep learning models for predictive maintenance (PdM) in industrial manufacturing systems using sensor data. We propose a framework that encompasses data acquisition, preprocessing, and model construction using various deep learning architectures, including Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, and their hybrid variants. Experiments conducted on three industrial datasets demonstrate the effectiveness of these models in predicting equipment failures and estimating remaining useful life. The CNN-LSTM hybrid model achieves the best performance with 96.1% accuracy and 95.2% F1-score, outperforming standalone CNN and LSTM architectures. Through ablation studies and feature importance analysis, we identify critical components and parameters that influence model performance. The results highlight the potential of deep learning approaches in revolutionizing predictive maintenance practices by enabling more accurate and reliable fault prediction in industrial manufacturing systems. Our findings provide valuable insights for implementing data-driven predictive maintenance strategies in real-world industrial applications.</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

Comparison of deep learning models for predictive maintenance in industrial manufacturing systems using sensor data

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月2日

查看全文
  • Wenjun Li,
  • Ting Li

摘要

This paper presents a comprehensive comparison of deep learning models for predictive maintenance (PdM) in industrial manufacturing systems using sensor data. We propose a framework that encompasses data acquisition, preprocessing, and model construction using various deep learning architectures, including Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, and their hybrid variants. Experiments conducted on three industrial datasets demonstrate the effectiveness of these models in predicting equipment failures and estimating remaining useful life. The CNN-LSTM hybrid model achieves the best performance with 96.1% accuracy and 95.2% F1-score, outperforming standalone CNN and LSTM architectures. Through ablation studies and feature importance analysis, we identify critical components and parameters that influence model performance. The results highlight the potential of deep learning approaches in revolutionizing predictive maintenance practices by enabling more accurate and reliable fault prediction in industrial manufacturing systems. Our findings provide valuable insights for implementing data-driven predictive maintenance strategies in real-world industrial applications.