Multi-Level Temporal Graph Networks with Local-Global Fusion for Industrial Fault Diagnosis
Summary: arXiv:2604.18765v1 Announce Type: cross
Abstract
Fault detection and diagnosis are critical for the optimal and safe operation of industrial processes. The correlations among sensors often display non-Euclidean structures where graph neural networks (GNNs) are widely used therein. However, for large-scale systems, local, global, and dynamic relations extensively exist among sensors, and traditional GNNs often overlook such complex and multi-level structures for various problems including the fault diagnosis. To address this issue, we propose a structure-aware multi-level temporal graph network with local-global feature fusion for industrial fault diagnosis.
Key Features of the Proposed Model
- Dynamic Correlation Graph Construction: A correlation graph is dynamically constructed using Pearson correlation coefficients to capture relationships among process variables.
- Temporal Feature Extraction: Temporal features are extracted through long short-term memory (LSTM)-based encoder, ensuring that time-dependent information is effectively utilized in the diagnosis process.
- Spatial Dependency Learning: Spatial dependencies among sensors are learned through graph convolution layers, allowing for a more comprehensive understanding of sensor interactions.
- Multi-Level Pooling Mechanism: A multi-level pooling mechanism is employed to gradually coarsen and learn meaningful graph structures, capturing higher-level patterns while retaining important fault-related details.
- Local-Global Feature Fusion: A fusion step is applied to integrate both detailed local features and overall global patterns before making the final prediction.
Experimental Evaluations
To validate the effectiveness of our proposed model, we conducted extensive experiments on the Tennessee Eastman Process (TEP). The results demonstrate that the model not only achieves superior fault diagnosis performance but also excels in handling complex fault scenarios. The proposed approach significantly outperforms various baseline methods, highlighting its robustness and efficiency in real-world applications.
Conclusion
The integration of local-global feature fusion in a multi-level temporal graph network represents a significant advancement in industrial fault diagnosis. By capturing complex relationships among sensors and incorporating both temporal and spatial dependencies, our model offers a more nuanced and effective solution for identifying faults in large-scale industrial systems. This research paves the way for more reliable and safe operations in various industrial processes, emphasizing the importance of innovative approaches in tackling modern challenges in fault diagnosis.