International Journal of Robust and Nonlinear Control最新文献

In this article, a double-layer learning control (DLLC) scheme with iterative disturbance observer (IDO) is proposed for unknown nonlinear batch manufacturing systems subject to nonrepetitive disturbance, by only leveraging the historical batch data of system input and output. The proposed scheme has two layers, where the inner layer is a simple proportional-type iterative learning control (ILC) structure to directly update the control input based on the previous batch data for output tracking, and the outer layer is an IDO based high-order set-point learning control (IDO-HOSPLC) structure to further expedite the convergence speed of tracking error and suppress the adverse effect of nonrepetitive disturbances. The convergence and boundedness of the resulting double-layer learning system is clarified by the contraction mapping and mathematical induction principles. An illustrative steam-water heat exchanger system is used to validate the effectiveness and superiority of the proposed DLLC scheme.

A data-based fault-tolerant controller for spacecraft formation flying (SFF) under external disturbances and actuator faults is proposed in this paper. The overall controller is based on off-policy integral reinforcement learning (IRL) and an adaptive control strategy. First, a model-free optimal controller is designed for the nominal healthy SFF system to track a class of reference trajectories. Second, a data-driven fault-tolerant controller is introduced by incorporating an adaptive approach for SFF when faced with external disturbances and actuator faults. Subsequently, the superiority of the proposed method in this paper is demonstrated and compared with the existing model-based control algorithm. Finally, the simulation results of a typical SFF system verify the efficiency of the proposed overall control approach.

This article investigates the parameter identification problem of the stochastic systems described by the multiple-input controlled autoregressive moving average model. By employing the data filtering technique, the input-output data is filtered and the original system with colored noise is changed into the system with white noise. Then, based on the hierarchical identification principle, the filtered system is decomposed into two subsystems, one containing the parameters of the linear dynamic block and the other containing the parameters of the nonlinear static block, leading to a filtering-based hierarchical stochastic gradient algorithm. To enhance the parameter accuracy, a filtering-based multi-innovation stochastic gradient algorithm is presented for the multiple-input nonlinear stochastic systems. Furthermore, the convergence of the filtering-based hierarchical stochastic gradient algorithm is analyzed, which can guarantee the stability of the algorithm. Finally, a simulation example is provided to verify the effectiveness of the proposed algorithms.

Two-stage anaerobic digestion (TSAD) has emerged as a promising technology for both organic pollutant purification and biogas production. In industrial applications, however, TSAD process measurements are subject to multi-rate sampling, different time delays, and sensor noise, while certain critical process states (e.g., concentrations of anaerobic microorganisms) remain unmeasurable in real time. To address these challenges, this study proposes a multi-rate sampled and delayed output signal prediction-based (MSDOSP) observer for TSAD processes, which simultaneously reduces measurement errors and estimates key unmeasurable states. The proposed MSDOSP observer utilizes multi-rate sampled measurements with time delays and noise to simultaneously reconstruct accurate continuous signals and provide real-time state estimation. The vector small-gain theorem is employed to rigorously establish the stability conditions of the observer. Finally, the performance of the MSDOSP observer was verified on the MATLAB/Simulink platform compared with two existing methods.

This paper develops an algebraic disturbance and state estimation method for a class of multi-input multi-output (MIMO) systems with non-zero-mean output measurement noise. The motivation is to design a novel output-based estimation method for state-feedback anti-disturbance control systems. By using the constructed modulating functions, both the disturbance and system state are explicitly expressed through output-based algebraic integral formulas. To enhance the efficiency of online estimation, a sliding integration window is applied to the proposed algebraic integral formulas. For discrete-time applications, the proposed formulas are easy to implement numerically, and the effects of the corrupting output measurement noise are analyzed. Finally, simulation examples on anti-disturbance control systems illustrate the efficiency and the robustness of the proposed estimation method against non-zero-mean output measurement noise.

Recently, bipartite time-varying formation-containment tracking control has become an important topic for the formation control of multi-agent systems. However, these studies combined the dynamics of leaders and followers with a vector-described dynamic formation. The specific time-varying formations are unclear for practical use, the disturbance attenuations are not efficiently tackled, and the unsuitable uses in edge computing occur. To overcome these inappropriate issues, a distributed disturbance observer-based bipartite formation-change finite-time tracking control (DDO-BFC-FTC) of networked multi-UAV is addressed in this work. A virtual leader signal is first transmitted to the leader UAV of the 1st group. The DDO-BFC-FTC for the serial connection of leader UAVs is then designed to tackle the dominant dynamics of networked multi-UAV. The planned poses among multiple leaders can easily result in different formations. Subsequently, the DDO-BFC-FTC for the other follower UAVs in each group is designed to maintain the 3D pose with respect to their leader, such that the convex hull spanned by the leader UAVs is achieved. These convex hulls do not have intersections to prevent collisions among groups. Besides the stability analyses, the comparison of simulations for the planned 3D pose with formation change validates the effectiveness and practicality of our approach. The average 3D pose error of Residual RNN-BFC-FTC is (0.0198 m, 1.129°), which is slightly better than (0.0199 m, 1.180°) of the proposed control at the expense of 4.16 times the computation time.

In this paper, the trajectory tracking control of an autonomous underwater vehicle (AUV) based on discrete time series controllers is explored. Currently, commonly used AUV controllers rely on digital signal processors in embedded systems, and many researchers have not adequately considered the interference to tracking accuracy caused by the discrete nature of hardware control systems during operation. A thrust structure model is developed for a five-thruster, four-degree-of-freedom AUV. Firstly, the stability of the nonlinear control system is discussed and demonstrated. Secondly, an event-triggering mechanism is added to the tracking process. When the sensor unit detects the relevant data signal value, it compares it with the event-trigger setting threshold. If it reaches the trigger value, the communication mechanism is activated, transmitting the data to the AUV's central controller. This initiates the kinematic and dynamic computations to complete motion control, effectively avoiding unnecessary communication, reducing the controller's computational cycle, indirectly saving energy, and improving the system's response speed. Finally, through several simulations and comparative studies, the algorithm is shown to possess strong robustness under complex disturbances compared with MPC and backstepping control. The proposed control scheme reduces the controller's computational cycles while ensuring high tracking accuracy and fast response of the AUV under complex disturbances.

Distributed stochastic model predictive control (DSMPC) of linear systems with coupled chance constraints under disturbances is investigated in this paper. We consider a practical scenario where only the mean and covariance, but not the exact distribution, of the disturbance is available. A frozen technique is utilized to ensure the satisfaction of the coupled constraints, and a deterministic convex tight reformulation is used for handling the chance constraints based on the available information of the disturbance. Recursive feasibility and convergence of the proposed method are proved. Numerical simulations are given to demonstrate the effectiveness of the proposed algorithm.

In this work, we consider the output regulation problem of heterogeneous multi-agent systems whose agent dynamics are described by the switched linear system. First, when each agent is completely controllable, a novel control law and a switching path are constructed to solve the finite-time output regulation problem, and we can adjust this time appropriately according to the requirements. Second, when at least one agent is incompletely controllable, the key condition to solve the output regulation problem is the uncontrollable sub-mode of the incompletely controllable agent is consistently asymptotically stabilizable. Finally, numerical examples are provided to demonstrate the effectiveness of the theoretical results obtained in this paper.