Dynamic Weight Optimization-Based Real-Time Monitoring and Fault Diagnosis System for Automation Equipment Using Edge Computing

Abstract

With the rapid advancement of Industry 4.0, the demand for precise fault diagnosis in automation equipment has become critical to ensure manufacturing continuity. This paper introduces a dynamic weight optimization-based system for real-time monitoring and fault diagnosis of automation equipment, leveraging edge computing technologies. The proposed five-layer architecture (Sensor, Production Line, Edge, Network, Cloud) integrates multi-threading and parallel computing at the edge layer to enhance PLC data processing efficiency by 40%. The core fault diagnosis algorithm, driven by dynamic weight optimization, completes weight updates within 300 milliseconds, achieving a 12% higher diagnostic accuracy than traditional methods (e.g., BP neural network and SVM). Experimental validation using 128,000 normal operation samples and 40,000 fault samples demonstrates a 98.3% fault diagnosis accuracy under complex industrial conditions, statistically supported by paired t-tests (p < 0.05). The system exhibits minimal latency degradation even at high data flows (2000 samples/second) and 4G network environments, outperforming cloud-based architectures in real-time fault detection capabilities. Compatibility tests across six industrial device types further validate its robust performance, making this framework a reliable solution for intelligent fault diagnosis in modern manufacturing systems.