,
Professor, Department of Computer Science and Engineering, Al-Ameen Engineering College (Autonomous) , Erode , India
,
Assistant Professor, Department of Information Technology, Al-Ameen Engineering College (Autonomous) , Erode , India
,
Assistant Professor, Department of Information Technology, Al-Ameen Engineering College (Autonomous) , Erode , India
,
Assistant Professor, Department of Computer Science and Engineering, Al-Ameen Engineering College (Autonomous) , Erode , India
,
Assistant Professor, Department of Computer Science and Engineering, Al-Ameen Engineering College (Autonomous) , Erode , India
,
Assistant Professor, Department of Information Technology, Al-Ameen Engineering College (Autonomous) , Erode , India
,
Assistant Professor, Department of Computer Science and Engineering, Al-Ameen Engineering College (Autonomous) , Erode , India
Assistant Professor, Department of Computer Science and Engineering, Al-Ameen Engineering College (Autonomous) , Erode , India
The concept of monitoring conditions with the help of AI has become a significant aspect of Industry 4.0 that enhances machine reliability and provides predictive maintenance. However, the models of anomaly detection based on deep learning are not readily implemented because of their lack of interpretability. The article introduces a novel anomaly detection model of vibration signals using a Transformer and augmented with Shapley Additive exPlanations (SHAP) to provide the accountability of the model. To improve the power of the model in diverse circumstances, the hybrid approach of Wavelet Transform and Variational Mode Decomposition (WT-VMD) preprocessing technique is used to get meaningful time-frequency features. The proposed model was tested on an industrial vibration dataset, and the accuracy of anomaly detection is 99.2%, and the fidelity of SHAP elucidation is 88%. An experiment that used 50 industrial maintenance experts as the subjects showed that the level of trust grew by 45 % and the decision-making process became 30 times faster using explainable exploratory models than using non-explainable models. The results illustrate that the Transformer-based method is more effective in increasing the detection performance and interpretability, which is required in industrial predictive maintenance. This model allows implementing AI in industrial systems by defining fault detection in a clear way that facilitates the realization of the maintenance plans and makes it more reliable. The paper has demonstrated the potential of the application of deep learning, along with an interpretable model, in solving the issue of fault diagnosis and condition monitoring in the complicated industrial environment.
This is an open access article distributed under the Creative Commons Attribution Non-Commercial License (CC BY-NC) License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
0
The statements, opinions and data contained in the journal are solely those of the individual authors and contributors and not of the publisher and the editor(s). We stay neutral with regard to jurisdictional claims in published maps and institutional affiliations.