International Journal of Robust and Nonlinear Control最新文献

Tracking control for nonlinear systems considering unknown disturbances and tracking-error constraint. Firstly, a dynamic linearized model is presented and, to decrease the computational burden, the pseudo-partial derivatives and the unknown disturbance are estimated based on an event-triggered mechanism. Secondly, an output observer is proposed due to the event-triggered scheme and the boundedness of the output estimation error is ensured using Lyapunov theory. The proposed event-triggered condition is designed based on the dead-zone operator to avoid the Zeno phenomenon during the process. The novelty of this work lies in integrating prescribed performance functions with adaptive terminal sliding mode control under a novel event-triggered design, which simultaneously guarantees finite-time bounded tracking, reduces chattering, and avoids unnecessary updates. More precisely, according to the Lyapunov theory and the proposed control law, it is guaranteed that the trajectories of the terminal sliding surface stay in a bounded region in finite time. Compared with existing event-triggered or data-driven SMC approaches, the proposed framework eliminates the need for complex disturbance estimators, achieves robustness without precise model information, and enhances computational efficiency. Finally, two numerical simulations and one experimental implementation using the Speedgoat Baseline device are considered to verify the efficacy and superiority of the proposed control method.

This paper aims at solving the infinite-horizon stochastic linear quadratic (SLQ) optimal control problem online for continuous-time systems with both additive and multiplicative noises. To eliminate the requirement for prior knowledge of system dynamics, a novel policy iteration approach is proposed, which leverages integral reinforcement learning (RL) techniques to iteratively solve the stochastic algebraic Riccati equation (SARE) using real-time state and input data. The proposed approach is an off-policy RL algorithm, where the learning process can be executed by using identical state and input data collected online over fixed time intervals, thereby enabling the optimal control law to be computed. The convergence of the proposed algorithm to the solution of the SARE is verified, and the effectiveness is validated through a numerical example.

Wearable exoskeletons hold promise in aiding patients with lower-limb dysfunction to regain mobility. However, due to movement disorders and nerve function impairments in patients, ensuring the accuracy and stability of exoskeleton robot motion control is crucial. This paper proposes a discrete-time sliding mode control (DSMC) framework with an adaptive reaching law for high-precision gait tracking in rehabilitation exoskeletons. Initially, a discrete-time dynamics model is established based on the Lagrangian discretization criterion for the lower limb exoskeleton rehabilitation robot. Subsequently, a novel controller incorporating a discrete-time fast terminal sliding mode surface and an adaptive reaching law is designed to address system uncertainties and disturbances. The adaptive adjustment of gain parameters enhances error convergence speed while mitigating chattering. Additionally, the stability of the control system is proven using the Lyapunov theory. Finally, the effectiveness of the proposed algorithm is verified through simulation and experimental tests.

In nonlinear filtering, the performance of Kalman filters is highly dependent on the characteristics of the measurements. However, in practical applications, measurements contain different noise components due to complex factors, making the robustness and accuracy of the estimator not guaranteed. In this article, a modified progressive extended Kalman Filter (MPEKF) is presented for nonlinear systems with compound measurement noises. To enhance the robustness of the system, a hypothesis testing method is employed to identify and remove outliers. Subsequently, considering the problem of non-orthogonality in the filtering process, a new set of measurement update equations is proposed to improve state estimation accuracy. Moreover, a termination condition is proposed to improve the computational efficiency and adaptability of the filter. Furthermore, through theoretical analysis of the MPEKF method, it is proven that the state estimation error remains bounded during the progressive process. Finally, the simulation of a nonlinear system example demonstrates that MPEKF has higher robustness and accuracy than EKF, IEKF, and PEKF.

This study proposes a fixed-time adaptive control strategy for a class of strict-feedback nonlinear systems subject to unmodeled dynamics, input saturation, and dead-zone nonlinearities. The key innovation lies in the introduction of a fixed-time dynamic compensator that effectively handles unknown unmodeled dynamics within a fixed convergence time, independent of initial conditions. To overcome the non-smooth characteristics of input saturation and dead-zone, a smooth non-affine approximation is employed and transformed into an affine form using the mean-value theorem. A systematic control design is developed by integrating adaptive backstepping with fixed-time Lyapunov theory, ensuring that all closed-loop signals remain bounded and the tracking error converges to a small neighborhood of zero within a fixed time. The proposed method not only enhances robustness but also guarantees fast and predictable transient performance, which is critical for safety-critical applications. The effectiveness and superiority of the approach are validated through two illustrative simulation examples.

In this paper, the control problem for a class of systems with nonlinear uncertain dynamics under quantized sampled-data measurements is considered. Firstly, the explicit tuning laws for ADRC design with a discrete ESO for promptly estimating the total disturbance are presented to tackle the control problem of nonlinear uncertain systems under quantized sampled-data measurements. Moreover, a feasible set of the ESO's bandwidth is given to ensure the stability of the closed-loop nonlinear system despite various disturbances. Secondly, the performance of ADRC under quantized sampled-data measurements is rigorously analyzed. It is discovered that the tracking error between the system state vectors and the desired trajectory can converge to a desired bound by tuning the bandwidth of ESO, the sampling step, and the quantizing step together. Finally, the necessary and sufficient condition for the tracking error to be infinitesimally equivalent to the sampling step and quantizing step is discussed, and the proposed method is tested on a numerical simulation concerning aircraft longitudinal overload control. The simulation results thoroughly demonstrate the competence of our tuning laws.

This paper investigates the adaptive gain design for discrete-time stochastic systems with fading measurements, aiming to achieve fast transient convergence and high final tracking accuracy. The fading measurements introduce output-dependent noise comprising both multiplicative and additive randomness, where the magnitude of the noise varies from iteration to iteration. The output-dependent noise leads to challenges in gain design and learning-tracking control. We propose a noise-adaptive learning control (NALC) algorithm with adaptive decreasing gain. Relying on observed output error information, the adaptive gain dynamically adjusts its rate of decrease in response to output-dependent noise. When the output-dependent noise is significant, the adaptive decreasing gain decreases rapidly to mitigate the impact of the noise; otherwise, the adaptive decreasing gain remains constant or decreases slowly to accelerate the reduction of tracking errors. The input error is proved to converge to zero in the almost sure sense. The example of a permanent magnet synchronous motor is provided to verify the proposed algorithm.

This paper is dedicated to iterative learning control for nonlinear systems with completely unknown states and time-iteration-varying parameter uncertainties. The unknown states cover unknown iteration-varying initial states and system operational states. The uncertainties are converted to a scalar without requiring an iterative sequence. A reference signal-based adaptive gain observer is developed to estimate the operational states. An iteration factor-based contraction mapping composite energy function is exploited to treat the iteration-varying initial states. The resultant controller is built on reference input and uncertainty estimation. The validity of the proposed method is verified through a circuit model.

This paper addresses the problem of controlling time-varying formation (TVF) in nonlinear multi-agent systems (MASs) that are subject to false data injection (FDI) attacks. A malicious attacker can inject false data into an actuator, which leads to a deviation in the follower's perception of its own state or the state of its neighbors, and thus disrupts the formation control of multi-agent systems (MASs). Meanwhile, the nonlinear nature of the system itself brings problems such as modeling uncertainty, control complexity, and difficulty in ensuring stability. To address the above challenges, this paper establishes a dynamic model of nonlinear multi-agent systems (MASs) under false data injection (FDI) attack, and designs an observation and estimation mechanism with robustness for detecting and compensating the disturbances caused by the attack. On this basis, a distributed formation control strategy is proposed. Specifically, this paper designs a distributed control protocol based on neural networks, utilizes neural networks to approximate and compensate for unknown nonlinear terms, designs an adaptive compensator to defend against false data injection (FDI) attacks, and demonstrates the feasibility of the proposed control scheme with the help of Lyapunov stability theory. Finally, the feasibility of the proposed method is verified by simulation examples.