Robust iterative learning control for batch processes with input delay subject to time‐varying uncertainties

Abstract

A robust iterative learning control (ILC) method is proposed for industrial batch processes with input delay subject to time-varying uncertainties, based on a two-dimensional (2D) system description of batch process operation. To compensate the input delay, a 2D state predictor is established to predict the augmented system states, such that a 2D ILC design is developed for the ‘delay-free’ 2D system based on using only the measured output errors of current and previous cycles. Delay-dependent stability conditions for the resulting 2D system are established in terms of matrix inequalities by defining a comprehensive 2D Lyapunov–Krasovskii functional candidate along with free-weighting matrices. By solving these matrix inequalities using a cone complementarity linearisation method, the ILC controller is explicitly derived together with an adjustable H infinity performance index. An important merit is that perfect tracking can be realised for a batch process with arbitrarily long input delay if the delay-free part of the 2D system can be stabilised, in no presence of time-varying uncertainties. Moreover, the time integral of tracking error can be added as an extended 2D system state for ILC design to eliminate steady-state output error for all batches. An illustrative example of injection moulding process is given to demonstrate the effectiveness of the proposed method.