Distributed predictive control of interconnected systems based on disturbance observation

Abstract

In this study, a distributed model predictive control (MPC) method based on extended state observer is proposed for a class of large-scale interconnected systems by using a reduced state-space decomposition strategy. Firstly, the full state of the system is divided into internal state and external state to decrease computation and communication burden, where the former one is used for local optimisation, and the latter one is used to exchange with its neighbours. Based on the receding optimisation strategy, a reduced state-space decomposition iterative algorithm is proposed, and the convergence of the proposed algorithm is given. To improve the disturbance rejection capability of the system, an extended state observer is designed to estimate the disturbance by using a pole assignment method. By applying a feedforward compensation strategy, a composite controller is derived by combining with local MPC controller and compensation controller, then the input-to-state stability is analysed for the closed-loop system. Finally, an example of a ship engine room power system is given to demonstrate the effectiveness of the proposed method.

Distributed extended state observer design and output feedback anti‐disturbance control for nonlinear interconnected systems: A dynamic memory event‐triggered mechanism

“…In this subsection, under the same sampling period h = 0.0002 s, we compare the dynamic ETC strategy [23] (Scheme II) with the DMET control strategy (Scheme I). The simulation results without and with noise are depicted in Figures [14][15][16] than with the dynamic ETM in Shi et al [23]. Additionally, a comparison of triggering times is presented in Table 4.…”

“…Consider system (1) with Assumptions 1 and 2. For given 𝜃 i𝑗 and 𝛾 i , if there are positive definite matrices Q i ∈ R n i ×n i , P i ∈ R (n i +m i )×(n i +m i ) satisfying the following matrix inequalities: (16) i𝑗 0 Ēi1 * Π( 22) i𝑗 0 Π (24) i𝑗 0 0 Π (27) i𝑗…”

“…In the aforementioned references [13][14][15][16][17][18], the control schemes were implemented in a time-triggered manner. However, unnecessary data transmission may result in a communication channel overload and reduce communication efficiency [19].…”

Distributed Model Predictive Control for Linear Constrained Systems Based on Time-Varying Terminal Sets

“…However, the controllers related to the subsystem will not exchange information with each other to improve the system performance. In this case, distributed predictive control is proposed by some researchers [18], [19], which allows communication information sharing between subsystem controllers, so that the coupling between subsystems can be fully considered to make the system performance better. To the best of our knowledge, the online calculation load is still large.…”

“…Therefore, to reduce the load of online calculations and the conservative brought by the fixed terminal constraint set in DMPC, a novel distributed model predictive control strategy is studied in this paper. Unlike the conventional approaches [19], each subsystem regards the coupling effect of neighboring subsystems as its own disturbance, to directly introduce the nominal system of the corresponding system, thereby reducing the complexity of the DMPC calculation process caused by the disturbance. Subsequently, due to the time-varying terminal constraint set is more practical than constant terminal constraint set, measures to design such terminal constraint set are provided, together with feasibility and stability of the closed-loop system.…”