Deep Learning-Enabled Fault Diagnosis in Rotating Machinery Using Multimodal Sensor Fusion

Fault diagnosis in rotating machinery is critical for preventing catastrophic failures and reducing maintenance costs in industrial applications. This paper presents a novel deep learning framework that leverages multimodal sensor fusion for enhanced fault detection and classification. We propose an attention-based fusion architecture that integrates data from accelerometers, vibration sensors, temperature sensors, acoustic sensors, and current sensors to capture comprehensive machinery health information. The model employs convolutional neural networks (CNN) for spatial feature extraction, long short-term memory (LSTM) networks for temporal dependency modeling, and an attention mechanism for adaptive feature weighting across modalities. Experimental validation on the Case Western Reserve University bearing dataset and real-world industrial data demonstrates that our approach achieves 96.4% classification accuracy, outperforming state-of-the-art unimodal and traditional fusion methods by 8.2% and 3.7%, respectively. The results indicate significant improvements in early fault detection, with a 94.3% precision for incipient fault identification. This work contributes to the advancement of intelligent maintenance systems and Industry 4.0 applications.

Hali tarjima qilinmagan