Constrained model predictive fault-tolerant control for nonlinear batch processes with time delay by integrating a LRF method and a switching strategy

To guarantee the control performance of batch processes under time delays and actuator failures, a novel iterative learning model-based predictive fault-tolerant control approach is developed. First, a two-dimensional switched model was developed by linearizing the system at multiple operating points so as to describe the actual process with and without actuator faults. Second, a switching controller with greater robustness and diminished conservativeness was developed based on this switched model. Compared to conventional faulttolerant control methods, the switching controller has stronger fault-tolerant ability, less conservativeness, and higher energy efficiency. Third, by integrating the Lyapunov-Razumikhin function (LRF) method, sufficient conditions for system stability could be provided based on the robust positive invariant and terminal constraint sets. Solving the stability conditions could then allow us to obtain the control law gains under faulty or normal cases. Compared to the existing Lyapunov-Krasovskii function method (LKF) method, the developed stability conditions were simpler, less complex, and less conservative. Finally, the effectiveness of the proposed method is tested on a continuous stirred reactor (CSTR).