Scholarlink

<p>This study presents an innovative approach for intelligent fault diagnosis in center-opening elevator door systems, designed to significantly enhance the accuracy and reliability of fault detection compared to traditional expert systems and single-scale deep learning models. The proposed method harnesses the synergistic capabilities of adaptive signal decomposition and multi-scale deep learning. To validate the effectiveness of this integrated model, a custom experimental platform was developed that replicates six prevalent elevator door faults under diverse operational conditions. The collected multi-channel vibration data, processed through adaptive signal decomposition, was utilized to train and assess the deep learning model. Experimental outcomes demonstrate the model’s exceptional performance, achieving an average diagnostic accuracy of 97.72%, which significantly outperforms conventional fault diagnosis methods, such as support vector machines and single-scale convolutional neural networks. Additional real-time fault detection tests demonstrated 96.8% accuracy, while cross-dataset validation on bearing fault data confirmed the model’s generalizability with 94.3% accuracy, highlighting its robustness across related electromechanical systems. Noise robustness tests further validated the model’s reliability in noisy environments, maintaining high accuracy under varying signal-to-noise ratios. The high-resolution visualization of decomposed signals and model performance metrics enhanced the interpretability of results. These findings underscore the potential of combining adaptive signal decomposition with multi-scale deep learning for intelligent fault diagnosis in complex electromechanical systems. The model’s ability to detect early-stage faults enhances elevator door system safety, reduces maintenance costs, and improves operational reliability. The approach’s adaptability to non-stationary, noisy signals positions it as a promising solution for broader applications in electromechanical fault diagnosis, contributing to the development of safer and more efficient 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

Intelligent fault diagnosis for elevator door systems using variational mode decomposition and multi-scale convolutional networks

  • 期刊论文

  • 发表日期：

    2025年8月8日

查看全文
  • Anping Wan,
  • Xin Tong,
  • Khalil AL-Bukhaiti,
  • Zhenchao Zhou,
  • Yipo Su,
  • Xiaomin Cheng

摘要

This study presents an innovative approach for intelligent fault diagnosis in center-opening elevator door systems, designed to significantly enhance the accuracy and reliability of fault detection compared to traditional expert systems and single-scale deep learning models. The proposed method harnesses the synergistic capabilities of adaptive signal decomposition and multi-scale deep learning. To validate the effectiveness of this integrated model, a custom experimental platform was developed that replicates six prevalent elevator door faults under diverse operational conditions. The collected multi-channel vibration data, processed through adaptive signal decomposition, was utilized to train and assess the deep learning model. Experimental outcomes demonstrate the model’s exceptional performance, achieving an average diagnostic accuracy of 97.72%, which significantly outperforms conventional fault diagnosis methods, such as support vector machines and single-scale convolutional neural networks. Additional real-time fault detection tests demonstrated 96.8% accuracy, while cross-dataset validation on bearing fault data confirmed the model’s generalizability with 94.3% accuracy, highlighting its robustness across related electromechanical systems. Noise robustness tests further validated the model’s reliability in noisy environments, maintaining high accuracy under varying signal-to-noise ratios. The high-resolution visualization of decomposed signals and model performance metrics enhanced the interpretability of results. These findings underscore the potential of combining adaptive signal decomposition with multi-scale deep learning for intelligent fault diagnosis in complex electromechanical systems. The model’s ability to detect early-stage faults enhances elevator door system safety, reduces maintenance costs, and improves operational reliability. The approach’s adaptability to non-stationary, noisy signals positions it as a promising solution for broader applications in electromechanical fault diagnosis, contributing to the development of safer and more efficient systems.