Observer-based cooperative distributed fault-tolerant model predictive control with imperfect network communication and asynchronous measurements

Intermittent actuator and sensor faults tolerant are simultaneously considered in a distributed control system with imperfect communication network. The asynchronous measurements of different output variables in one sampling period are synchronized through a novel two-stage model-based projection method. Different from centralized control network, in both layer-to-layer and in-layer communication, the packet delay, loss and disordering are corrected by the predicted data from model predictive control. Moreover, a completely distributed state observer is established for both system states and sensor faults problem with bounded noise uncertainties. For the intermittent actuator faults, actuator plug-and-play design methods based on model predictive control has been introduced, making the actuator faults estimation omitted. The distributed stability conditions are derived for the proposed fault-tolerant controller, and the online feasibility is explained in detail. Numerical simulation is given to verify the design procedure.