Synthesis of ILC-MPC Controller With Data-Driven Approach for Constrained Batch Processes

The iterative learning control (ILC) combining with model predictive control (ILC-MPC) is an effective control method for constrained batch processes. However, in real applications, model uncertainty usually makes it slow for the controlled process to converge to the reference trajectory. To eliminate the restrictions in previous works, a data-driven approach is proposed, which directly describes the relationship between inputs and outputs according to the past data. Based on this method, a novel data-driven ILC-MPC controller is proposed, where the two-mode framework and the invariant updating strategy are employed to guarantee the convergence. Since the outputs caused by model uncertainty are partly known from the past data, better performance can be achieved by the proposed design which is verified by experimental studies on a manipulator.