Scholarlink

<p>Magnetorheological (MR) dampers have recently been widely applied in various engineering domains owing to their semi-active nature and adaptability. However, their performance can be significantly affected by two major physical defects: leakage and cavitation. This study proposes a novel convolutional long short-term memory (CLSTM) model, which integrates one-dimensional convolutional neural networks (1D-CNN) with LSTM to accurately classify the damper’s operational status into three distinct states: normal, leakage, and cavitation. Experimental validation demonstrates that the proposed model achieves a high classification accuracy of 97.3 % and an F1-score of 0.973 on the test data. Unlike traditional fault detection methods that rely on handcrafted features and classical classifiers such as support vector machines (SVMs), the proposed CLSTM model enables direct classification from raw multichannel time-series sensor data. This work presents a scalable and automated framework for MR damper defect prediction, providing a foundation for intelligent health monitoring in manufacturing and mechanical systems.</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

Defect prediction for magnetorheological damper using a deep learning-based CLSTM classifier

  • 期刊论文

  • 发表日期：

    2026年5月11日

查看全文
  • Yoon-Mok Kim,
  • Jong-Seok Oh,
  • Bo-Gyu Kim

摘要

Magnetorheological (MR) dampers have recently been widely applied in various engineering domains owing to their semi-active nature and adaptability. However, their performance can be significantly affected by two major physical defects: leakage and cavitation. This study proposes a novel convolutional long short-term memory (CLSTM) model, which integrates one-dimensional convolutional neural networks (1D-CNN) with LSTM to accurately classify the damper’s operational status into three distinct states: normal, leakage, and cavitation. Experimental validation demonstrates that the proposed model achieves a high classification accuracy of 97.3 % and an F1-score of 0.973 on the test data. Unlike traditional fault detection methods that rely on handcrafted features and classical classifiers such as support vector machines (SVMs), the proposed CLSTM model enables direct classification from raw multichannel time-series sensor data. This work presents a scalable and automated framework for MR damper defect prediction, providing a foundation for intelligent health monitoring in manufacturing and mechanical systems.