Dynamic coupled control with closed-loop robustness for reconciling adaptability, safety, and real-time performance under production line condition switches
Year: 2026
Type: article
Abstract: Context Light industry production lines are characterized by frequent dynamic condition switching, heterogeneous task portfolios, and deeply interconnected equipment. Their intelligent upgrading imposes competing demands for algorithmic adaptability to varying conditions, security against cyber threats, and real-time responsiveness for critical tasks. These three objectives are inherently in conflict, and existing approaches largely address them in isolation, lacking a unified mechanism to reconcile their trade-offs under dynamic operation. Objectives This paper aims to develop a coupled control and robustness assurance framework that jointly optimizes adaptability, safety, and real-time performance under dynamic operating conditions, while maintaining end-to-end robustness across both routine condition switches and extreme fault scenarios. Methods We propose a closed-loop "sensing–decision–execution–feedback" architecture driven by real-time operating-condition prediction. Multi-source perception (sensors, PLCs, MES) feeds an LSTM-based condition forecaster, whose outputs guide dynamic weight allocation among the three objectives. The optimal trade-off point is solved via an adapted NSGA-Ⅲ algorithm under production-line constraints. A hierarchical adaptation strategy tailors control parameters for heterogeneous equipment (PLC/DCS, industrial robots, sensors). Lightweight security primitives (SM3, AES) and a priority-deadline hybrid scheduler are integrated to keep overhead within millisecond bounds. A robustness feedback module continuously monitors performance and triggers re-optimization when deviations are detected. Results Experiments on electronic assembly and food processing lines show that our system achieves a 96.3% adaptability success rate, a security response time of 7.2 ms, and a key-task latency of 38.2 ms under dynamic conditions—meeting the ≤50 ms deadline with 99.2% satisfaction. Under simultaneous multi-device failures, the core-task interruption rate drops to 15%, a 75% relative reduction over the traditional scheme, with anomaly recovery within 80 ms. Conclusion The experimental results demonstrate that the proposed framework effectively reconciles the adaptability--security--real-time trade-off through dynamic coupled control and closed-loop robustness feedback, offering a practically deployable solution for intelligent production-line upgrades in complex, time-varying industrial environments.
Source: Array
Institutions: University of Michigan, UNSW Sydney
Language: English
Subfield: Industrial and Manufacturing Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,680
Funders: Natural Science Foundation of Guangdong Province, Guangdong Provincial Applied Science and Technology Research and Development Program, Major Projects of Guangdong Education Department for Foundation Research and Applied Research, Basic and Applied Basic Research Foundation of Guangdong Province
Awards: G9000374514, G9000374890
OpenAlex ID: W7214461898