In this study, the problem of simultaneous estimation of system states and disturbances is addressed for twodimensional (2D) linear systems. The considered 2D systems can be described as Fornasini-Marchesini second local state-space model coupled with unknown disturbances in the measurement equation. Asymptotically stable and uniformly ultimately bounded integrated state/disturbance observers are proposed as solutions to the simultaneous estimation problem under different system conditions. For the asymptotically stable integrated state/disturbance observer, a necessary and sufficient condition for its existence is presented and proved. For the uniformly ultimately bounded integrated state/disturbance observer, a sufficient condition for its existence is given and proved, and it can be designed such that its estimation error has arbitrarily small upper bound. Moreover, some constructive design methods are given for the proposed integrated state/disturbance observers. Finally, a thermal process plant and another numerical example are provided to illustrate the effectiveness of the proposed methods.
IntroductionObserver synthesis and estimation problems in the context of two-dimensional (2D) systems have received substantial attention during the past several decades. In the early 1980s, Bisiacco conducted fundamental studies on 2D observers [1,2]. He presented both necessary and sufficient conditions for state reconstruction and the structure of observers for 2D systems on the basis of polynomial matrix theory. In the same period, other basic results concerning 2D observers have been reported, such as the minimal order state observer given by Kaczorek (see [3] and the references therein). Since then, a series of studies on 2D observers have been reported, particularly during the past ten years. Bisiacco and Valcher discussed the unknown input observer (UIO) problem for 2D systems in the frequency domain, where the design of a 2D linear system dead-beat UIO was considered on the basis of polynomial matrix method [4]. This kind of UIO can be used for fault detection and isolation [5,6]. By extending the concept of the Luenberger observer to 2D state-space models and replacing the original observer by two interconnected linear 2D systems, Bisiacco and Valcher introduced a generalised Luenberger observer [7] with less conservative existence condition. Xu et al. [8] presented an asymptotic UIO in the time domain with a sufficient existence condition in terms of rank conditions, and a constructive design method was also proposed based on linear matrix inequalities (LMIs). With a similar design and analysis method as that for asymptotic UIOs, Xu et al. [9] successfully solved the functional observer problem for 2D systems. On the basis of an early study on 2D non-linear digital filters [10] and by utilising a method similar to a study reported for functional observers [9], Wang et al. [11] studied the asymptotic UIO problem for a class of 2D nonlinear systems. In addition, Zou et al. [12], Wang and Zou [13] and Kaczorek [...