International Journal of Robust and Nonlinear Control最新文献

This paper investigates the quantized guaranteed cost control design (GCCD) for networked control systems (NCSs) with interval parameters and random delays. A comprehensive NCS model is established, incorporating three key elements: Interval parameters represented as convex combinations of endpoints, random delays governed by a Markov chain, and quantization effects described by the sector-bound method. Based on the model, sufficient conditions for the GCCD are derived using the Lyapunov functional approach combined with the cone complementary linearization (CCL) algorithm. Numerical simulations demonstrate the efficacy of the proposed method.

This paper studies the discrete-time stochastic zero-sum games by employing the approximate dynamic programming technique. We present on-policy and off-policy policy iteration algorithms to attain the saddle point without using the information of the system dynamics. A comparative analysis of model-free algorithms and their equivalence relationships is examined. Numerical examples are given to illustrate the efficiency of the proposed algorithms.

This paper investigates the adaptive finite-time asymptotic tracking control problem for a class of stochastic nonlinear systems. For the first time, an exponential decay factor is introduced into the theory of stochastic finite-time stability, establishing a novel stochastic finite-time asymptotic tracking criterion. Building upon this theorem, a new adaptive multi-dimensional Taylor network (MTN) backstepping control method is proposed: During the controller design process, the same exponential decay factor is creatively incorporated. This enables rapid finite-time convergence while guaranteeing that the tracking error asymptotically converges to zero as time approaches infinity. Utilizing Lyapunov stability theory, it is proven that the proposed control method ensures all the signals of the closed-loop system are bounded in finite time and enables the tracking error to asymptotically converge to zero as time approaches infinity. Finally, a numerical simulation and a practical example are presented to validate the effectiveness of the proposed design method.

We carry out a detailed analysis of direct voltage control of a Boost converter feeding a simple resistive load. First, we prove that using a classical PI control to stabilize a desired equilibrium leads to a very complicated dynamic behavior consisting of two equilibrium points, one of them can be stabilized for PI gains within certain negative ranges, while the second equilibrium point may also be rendered stable—but for sufficiently large positive tuning gains. Moreover, if we neglect the resistive effect of the inductor, there is only one equilibrium and it is stable for a certain range of negative PI gains. From a practical point of view, it is important to note that the only useful equilibrium point is that of minimum current and that, in addition, there is always a resistive component in the inductor either by its parasitic resistance or by the resistive component of the output impedance of the previous stage. In opposition to this scenario, we recall three nonlinear voltage-feedback controllers that ensure asymptotic stability of the desired equilibrium with simple gain tuning rules, an easily defined domain of attraction, and smooth transient behavior. Two of them are very simple, nonlinear, static voltage feedback rules, while the third one is a variation of the PID scheme called PID-Passivity-based Control (PBC). In its original formulation, PID-PBC requires full state measurement, but we present a modified version that incorporates a current observer. All three nonlinear controllers are designed following the principles of PBC, which has had enormous success in many engineering applications.

This paper proposes a dynamic event-triggered optimal control strategy for unknown nonlinear continuous-time (CT) systems with input constraints based on a state observer and an identifier-critic neural network (ICNN). Firstly, a discounted cost function is introduced to deal with the input constraints for the systems and establish a relevant Hamilton-Jacobi-Bellman (HJB) equation to solve the optimal control problem. Secondly, an observer is designed to reconstruct the state of the systems that relate to the information at the moment of event triggering. Then, to reduce communication and computation burdens, a dynamic event-triggered control (DETC) strategy is applied to keep the signal transmissions and controller aperiodic update, which is better than the traditional static event-triggered control (ETC) strategy and effectively excludes Zeno behavior. Moreover, to relax the assumptions about the dynamics of the systems and obtain the optimal solution of the HJB equation, an ICNN architecture is developed. The identifier neural network (INN) approximates the dynamics knowledge of unknown systems, and the given update law of the weight matrix ensures the convergence of the state errors. By the gradient descent method, the critic neural network (CNN) is used to approximate the optimal performance index function to approximate the solution of the modified HJB equation, and then the theoretical analysis proves the uniform ultimate boundedness (UUB) of the system states and the weight matrix of ICNN. Finally, the effectiveness of the proposed method is verified by a numerical simulation.

The spacecraft attitude tracking problem with prescribed performance in the presence of environmental disturbance, model uncertainty, actuator faults, and saturation presents a significant challenge. Under these tough conditions, this paper proposes an adaptive model predictive control (MPC) with a novel appointed-time performance function (APF) constraint to achieve high-performance attitude tracking trajectories with appointed-time convergence. Firstly, within the proposed MPC framework, an optimal attitude trajectory is obtained, respecting the performance boundary and actuator limitations. Next, by introducing a contraction constraint via prescribed performance auxiliary sliding mode control (SMC), the recursive feasibility and closed-loop stability of MPC are rigorously demonstrated. Additionally, the function-adaptive law is employed to estimate and compensate the total disturbances. Finally, simulations are conducted to demonstrate the effectiveness and validity of the proposed adaptive appointed-time prescribed performance MPC algorithm.

In this paper, we study the control scheme of robust adaptive neural output feedback stabilization for uncertain input-output-quantization nonlinear systems with a high-order output filter. Neural network technology is employed to approximate nonlinear functions, and disturbance estimators are used to estimate unknown noise and disturbances in the system. An adaptive neural observer is well-designed to estimate unmeasured states while preventing quantization errors from affecting the observer's state equations. To stabilize the system, an adaptive backstepping quantization controller is obtained by using two kinds of filters. The first kind of filters (first-order filters) is employed to address the problem of the virtual control laws of differential explosion. The other kind of filter (high-order output filter) is used to make quantized sensor output signals differentiable and solve the problem of noncontinuity of quantized sensor outputs in the traditional backstepping method. The stability analysis illustrates that all signals in the closed-loop system are ultimately uniformly bounded and adjustable. In the simulation results, a fan speed model is constructed to verify the accuracy and effectiveness of the proposed scheme.

This article focuses on linear multi-agent systems (MAS) subject to Denial-of-Service (DoS) attacks on the input/output (I/O) sides under a fading channel environment. An iterative learning control (ILC) algorithm with the event-triggered scheme is developed. Firstly, the process of DoS attacks is formulated by mutually independent Bernoulli sequences with known mean and variance. Then the fading measurements in the I/O channels are characterized as independent Gaussian distributions with known mean and variance. For conserving network communication resources, an event triggering mechanism is employed to reduce the number of controller updates. Subsequently, the repeating system and the presented ILC algorithm involving both iteration axis and time axis are converted into a stochastic 2D Roesser model. Based on the 2D theory, the stability criteria of the system are further derived, the design scheme for controller gains is proposed, and the gains are solved by the LMI technique. To mitigate the adverse effects induced by stochastic fading, an ILC algorithm with a compensation mechanism is proposed, and rigorous theoretical analyses are conducted. Finally, numerical simulations are established to confirm that the presented control algorithm is effective.

This paper investigates iterative learning control for stochastic differential systems of fractional order in the Hilfer sense. Unlike existing studies that treat either fractional dynamics or stochastic effects separately, we develop an integrated framework that combines Hilfer fractional derivatives, Brownian perturbations, and a proportional–fractional integral learning law. The proposed approach captures both the memory effects and random uncertainties inherent in complex systems. As a case study, we apply the method to a gantry robot equipped with a flexible arm. Numerical simulations show that the Hilfer derivative significantly improves tracking accuracy and convergence speed compared to integer-order models, highlighting the potential of the proposed strategy for robotic applications under uncertainty.