Purpose The purpose of this study is to develop a resilient model predictive control (RMPC) framework capable of maintaining optimal control performance in networked industrial cyber-physical systems subject to unpredictable disturbances, communication delays and actuator faults. The proposed approach aims to enhance the resilience, stability and reliability of intelligent manufacturing processes by integrating adaptive fault handling, disturbance compensation and real-time constraint adjustment within a unified predictive control architecture. Design/methodology/approach The proposed RMPC framework employs a disturbance observer-based adaptive MPC structure enhanced by a resilience index for real-time constraint regulation. It integrates multi-disturbance estimation, adaptive fault reconfiguration and stability-guaranteed optimization using Lyapunov-based analysis. Simulation studies on a representative industrial heating, ventilation and air conditioning (HVAC) control benchmark validate the method's capability to sustain control performance under faults and network delays, demonstrating the framework's computational efficiency and applicability to intelligent manufacturing systems. Findings Results demonstrate that the proposed RMPC significantly enhances fault tolerance, stability and tracking performance compared with conventional MPC and proportional-integral-derivative controllers. The framework reduces cumulative tracking error by up to 65%, achieves 50–60% faster recovery from actuator faults, and maintains constraint satisfaction under multiple disturbances. These findings confirm the RMPC's suitability for complex, networked manufacturing environments requiring robust, adaptive and computationally efficient control solutions. Originality/value This study introduces a unified resilient predictive control architecture that integrates disturbance estimation, adaptive constraint regulation and fault recovery into a single optimization framework. Unlike traditional MPC approaches, the proposed method explicitly quantifies resilience and guarantees stability under network-induced uncertainties. The originality lies in bridging resilient control theory with intelligent industrial applications, providing a practical pathway for deploying fault-tolerant, adaptive controllers in next-generation smart manufacturing and cyber-physical systems.
Rahmati et al. (Wed,) studied this question.