In this paper, convergent property of a saturation-strategy-based iterative learning control (ILC) law is first investigated for a class of two-dimensional linear discrete first Fornasini–Marchesini model (2-D LDFFM) with input saturation. A three-dimensional dynamical process is transformed into a 2-D dynamical process by row scanning approach or column scanning approach. As a result, it is theoretical proved no matter which method is adopted, perfect tracking on the desired reference surface is accomplished by virtue of the 2-D linear inequality theory. Numerical simulation on a practical thermal process is used to illustrate the effectiveness and feasibility of the designed ILC law. In addition, ILC convergence analysis for 2-D LDFFM with input delay and input saturation is discussed.

This paper first investigates robust tracking problem of a high-order iterative learning control (HOILC) algorithm for two classes of two-dimensional linear discrete time-varying Fornasini–Marchesini systems (2-D LDTVFMS) and 2-D LDTVFMS with input delays with iteration-dependent reference trajectory described by a high-order internal model (HOIM) operator, boundary states and disturbances. An extended high-order linear discrete inequality is proposed to guarantee the ILC tracking error of 2-D LDTVFMS converge robustly to a bounded range, the bound of which is relied on iteration-dependent uncertainties from reference trajectory, boundary states and disturbances. A simulation example on a practical thermal process is used to validate the effectiveness and feasibility of the proposed HOILC law. Additionally, it is verified by theory analysis and simulation that no matter how the ILC controller gain is selected, the convergence condition in Theorem 2 of Wan and Li [(2021). High-order internal model based iterative learning control for 2-D linear FMMI systems with iteration-varying trajectory tracking. IEEE Transactions on Systems Man and Cybernetics: Systems, 51, 1462–1472] is not satisfied. Using the proposed HOILC law, perfect tracking result in Theorem 2 of Wan and Li (2021) can be obtained.