Research on Fault Diagnosis Method for Rotating Machinery Based on Edge Computing

Rotating machinery, as a vital and inevitable component in industrial production and processing, plays a crucial role in ensuring the normal operation of production processes. However, most of the existing fault diagnosis methods for rotating machinery are either offline or cloud-based online approaches, which suffer from long latency and large data volumes, making them unable to meet real-time requirements. To reduce latency and data transmission volume, this research proposes a fault diagnosis method for rotating machinery based on edge computing. This research constructs an edge node that integrates signal acquisition, data preprocessing, feature extraction, and fault diagnosis classification to accurately and in real-time identify the fault status of equipment. To address the issues of low fault diagnosis recognition rate and data redundancy associated with single sensors under complex working conditions, this research proposes a fault diagnosis method based on dual-channel CNN decision-level fusion. To alleviate the computational pressure on edge nodes, the equipment fault status diagnosis model is trained on the upper computer, and the data preprocessing and diagnosis model are embedded into the edge nodes. The correctness and real-time performance of the proposed method were validated through comparisons with other methods and online fault diagnosis experiments.