Research on Multi Sensor Fault Diagnosis Method Based on Time-Frequency Graph Neural Network

Jinghe Lin

doi:10.54097/cx36j893

Authors

Jinghe Lin

DOI:

https://doi.org/10.54097/cx36j893

Keywords:

Time-Frequency graph neural network; Graph convolutional neural network; Attention mechanism.

Abstract

With the continuous development of modern industry, numerous sensors have been ubiquitously deployed in industrial systems. Effectively integrating multi-sensor data is of great significance for improving the accuracy and reliability of fault diagnosis. However, most existing methods are limited to a single-level structure and fail to fully exploit the deep-level features within the data, exhibiting notable limitations. To address this issue, this study proposes a fault diagnosis method that integrates time-domain and frequency-domain graph information to more comprehensively extract and utilize fault features from sensor data. Specifically, graph structures in the time and frequency domains are first constructed by analyzing the correlations among different sensors. Then, graph convolutional neural networks are employed to extract features from both types of graphs, while the mutual information between them is minimized to enhance feature independence. Finally, an attention-based fusion strategy is introduced, and a classifier is constructed to improve the reliability of fault diagnosis. Experimental results on the nickel flash smelting furnace system demonstrate that the proposed method achieves a fault diagnosis accuracy exceeding 95.43%, significantly outperforming methods relying solely on time-domain data.

References

[1] Lei L, Li W, Zhang S, et al. Research Progress on Data-Driven Industrial Fault Diagnosis Methods. SENSORS,Vol 25, pp. 2952-2952, 2025.

[2] Jia T, Tian Y, Yin Z, et al. Data-free stealing attack and defense strategy for industrial fault diagnosis system. Chemical Engineering Research and Design, Vol 216, pp. 200-215, 2025.

[3] Lin J, Ren G, Huang K, et al. A novel multi-sensor information fusion method for fault diagnosis of rotating machinery with missing signals. Advanced Engineering Informatics, Vol 68, pp. 103595, 2025.

[4] Carlos F, David D, Jose L, et al. Fuzzy processing applied to improve multimodal sensor data fusion to discover frequent behavioral patterns for smart healthcare. Information Fusion, Vol 123, pp. 103307, 2025.

[5] Deng X, Wang Y, Yang W. A sensitive efficient multiple predictable feature extraction and fusion method for complicated industrial fault detection. CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS,Vol 263, 2025.

[6] Zhu D, Liu G, He W, et al. Fault feature extraction of rolling element bearing based on EVMD. JOURNAL OF THE BRAZILIAN SOCIETY OF MECHANICAL SCIENCES AND ENGINEERING,Vol 43, pp. 1-14, 2021.

[7] Lee C, Le T. Intelligence Bearing Fault Diagnosis Model Using Multiple Feature Extraction and Binary Particle Swarm Optimization With Extended Memory. IEEE ACCESS, Vol 8, pp. 198343-198356, 2020.

[8] Ribeiro J, Ronny F, Santos A, et al. Fault Detection and Diagnosis in Electric Motors Using Convolution Neural Network and Short-Time Fourier Transform. JOURNAL OF VIBRATION ENGINEERING & TECHNOLOGIES, Vol 10, pp. 2531-2542, 2022.

[9] Xiang G, Miao G, Cui L, et al. Intelligent Fault Diagnosis for Inertial Measurement Unit through Deep Residual Convolutional Neural Network and Short-Time Fourier Transform. MACHINES, Vol 10, pp. 851, 2022.

[10] Feng T, Wang Z, Qiu L, et al. Rolling Based on Multi-Source Time-Frequency Feature Fusion with a Wavelet-Convolution, Channel-Attention-Residual Network-Bearing Fault Diagnosis Method. SENSORS, Vol 25, pp. 4091, 2025.

[11] Sun C, Wang Y, Tian Z, et al. Optimal symplectic geometric bi-spectrum mutual information contrastive learning framework for fault diagnosis of helicopter spiral bevel gear. MEASUREMENT SCIENCE AND TECHNOLOGY, Vol 36, pp. 066004-066004, 2025.

[12] Lei L, Li W, Zhang S, et al. Research Progress on Data-Driven Industrial Fault Diagnosis Methods. SENSORS, Vol 25, pp. 2952-2952, 2025.

[13] Ma L, Peng K, Yang Q. A unified representation and fusion framework of multi-source heterogeneous data for fault diagnosis in industrial processes. ADVANCED ENGINEERING INFORMATICS, Vol 67, 2025.

[14] Ma L, Yang Q, Long Z, et al. A multi-source heterogeneous data fusion framework for fault diagnosis in industrial processes with missing image data. MEASUREMENT, Vol 256, 2025.