<scp>A</scp> novel <scp>proportional‐integral‐derivative</scp> ‐type iterative learning control strategy based on <scp>2D</scp> model predictive iterative learning control optimization for batch processes with partial actuator failures

Abstract For control problems involving partial actuator failures during batch processes, this paper proposes a novel proportional‐integral‐derivative (PID)‐type iterative learning control strategy based on two‐dimensional model predictive iterative learning control optimization. First, a two‐dimensional extended non‐minimal state space model was constructed using actuator outputs, process variable outputs, and tracking error information, providing more freedom for subsequent controller design. At the same time, a setpoint learning strategy was constructed by introducing historical batch error information. Secondly, a novel model was constructed by combining the two‐dimensional extended non‐minimal state space model and the setpoint learning strategy, which significantly improved the iterative learning ability of the corresponding controller. Furthermore, a novel PID‐type iterative learning controller is constructed by combining an incremental PID controller and a traditional PID‐type iterative learning controller. Finally, the designed control law is used to optimize the parameters of the PID‐type iterative learning controller in real time, ensuring that the control system can simultaneously achieve control performance along the time axis and convergence performance along the batch axis. Based on the model mismatch in the injection moulding machine holding pressure control system, the effectiveness of the proposed strategy was verified by considering two types of partial actuator failures.