Edge-Cloud Intelligent Monitoring and Maintenance for Smart Production Lines

INTRODUCTION: Smart manufacturing production lines integrate Computer Numerical Control (CNC) machines, industrial robots, motor drives, conveying units, sensors, and Programmable Logic Controller (PLC) workstations. Their operation and maintenance (O&M) requires low-latency recognition of coupled vibration, temperature, current, cycle-time, and alarm changes under production takt and downtime constraints. OBJECTIVES: To improve real-time condition monitoring and maintenance decision-making, an edge computing-enabled collaborative intelligent O&M method is developed for discrete manufacturing production lines. The objective is to connect equipment state perception with health assessment, resource-constrained maintenance priority, and edge-cloud model updating. METHODS: Multi-source manufacturing data are segmented into sliding status windows at edge nodes, where a lightweight Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM) model extracts local impact features and temporal degradation patterns. The resulting anomaly scores and health levels are jointly modeled with workstation importance, production load, spare-part availability, labor-hour constraints, and downtime cost, while reinforcement learning updates maintenance actions according to execution feedback. RESULTS: Experiments on CNC machining units, industrial robots, conveying units, motor drives, and PLC workstations show that the proposed model achieves an F1-score of 0.946, an Area under the Curve (AUC) of 0.973, and an average detection latency of 126 milliseconds (ms). The maintenance task matching rate reaches 94.1%, and the high-risk task priority resolution rate reaches 96.3%. CONCLUSION: The method links edge-side condition recognition with manufacturing-oriented maintenance execution and cloud-side parameter calibration, providing a deployable closed-loop framework for continuous O&M in smart production lines.