In this article, a novel reinforcement learning-based distributed self-organizing swarm optimal control method is presented to address the challenges of real-time collision avoidance and optimal trajectory tracking for multiple unmanned surface vehicles (USVs) operating under uncertain environmental disturbances and in the absence of accurate dynamic models. To overcome the structural limitations of traditional formation methods during collision avoidance, a distributed self-organizing swarm mechanism with an all-round collision avoidance strategy is constructed, enabling the swarm to dynamically adjust its structure and behavioral modes in real time according to environmental changes and tracking tasks. To mitigate the unnecessary wide-ranging evasive maneuvers caused by a fixed obstacle avoidance radius, a velocity-regulation-based dynamic avoidance method is developed to reduce unnecessary path detours. Furthermore, to alleviate the control complexity arising from model uncertainties, high-frequency noise disturbances, and abrupt changes in system dynamics, an innovative identifier–critic–actor (ICA) synchronous learning neural network architecture based on a low-frequency (LF-ICA) learning identifier is proposed to achieve LF reconstruction while learning uncertain system dynamics, thereby ensuring the optimal consistency of the swarm posture during collision avoidance and trajectory tracking. Based on Lyapunov stability theory, it is shown that all signals in the swarm system remain bounded under the proposed control method. Finally, simulation results are presented to demonstrate the effectiveness of the proposed control approach.
This article investigates generalized conditions for the negative and positive definiteness of matrix-valued polynomial inequalities, with applications to linear systems with time-varying delays. To address the negative- and positive-definiteness problems of high-order matrix-valued polynomials, generalized negative- and positive-definiteness conditions for matrix-valued polynomial inequalities are proposed. By incorporating an $m$ -degree matrix-valued polynomial in the time-varying delay into the construction of the Lyapunov–Krasovskii functional (LKF), a novel delay-degree-dependent LKF is developed. To handle the reciprocally convex terms arising in the derivative of the LKF, a parameter-associated reciprocally convex combination lemma is proposed. On the basis of the proposed methods, several stability criteria are established. The effectiveness and advantages of the proposed approach are demonstrated through two benchmark numerical examples. Furthermore, the results demonstrate that different arrangements of the matrices in the generalized negative-definiteness condition can affect the estimation accuracy of high-order matrix-valued polynomial inequalities involving time-varying delays. Finally, to demonstrate the practical applicability of the proposed techniques, they are applied to a practical load frequency control (LFC) problem in a power system, and a hardware-in-the-loop experiment is conducted for further validation.
This article investigates the quasi-consensus problem for a class of heterogeneous multiagent systems (HMASs) subject to parametric uncertainties. To address this issue, a novel intermittent asynchronous impulsive control (IAIC) strategy is proposed, integrating the advantages of asynchronous impulsive control (AIC) and intermittent control. Under this framework, each agent operates with a distinct impulsive sequence confined to active control windows, thereby reducing communication overhead and control burden while preventing instantaneous power overload. Furthermore, an event-triggered IAIC scheme is formulated to dynamically determine the activation instants of these windows, further minimizing unnecessary resource consumption. By employing the concept of the intermittent average impulsive interval (AII), sufficient conditions are derived to guarantee quasi-consensus of the HMASs while strictly excluding Zeno behavior. Finally, numerical simulations involving Chua’s oscillators and single-link manipulators with flexible joints are provided to validate the effectiveness of the derived results.
This article investigates the resilient control problem for nonlinear multiagent systems (MASs) under false data injection (FDI) attacks on both sensors and actuators. To enhance the operational resilience and coordination reliability of networked cyber–physical systems under such attacks, a resilient control framework is developed based on a robust selection mechanism and a cooperative interaction framework. First, to address compromised output measurements, the robust selection mechanism identifies trustworthy information from the attacked outputs. Based on the selected outputs, a state observer is employed to estimate the system states, and an adaptive dynamic programming (ADP)-based nominal consensus controller is subsequently developed. Second, to mitigate actuator attacks, a cooperative framework involving the MASs and two virtual systems is established. These systems embed the effects of actuator attacks into computable virtual dynamics, based on which a resilient compensation input is generated without explicitly estimating the attack signals. The resulting resilient controller combines the nominal consensus input with the virtual-system-based compensation input. Subsequently, the system states are proven to be uniformly ultimately bounded (UUB). The effectiveness of the proposed method is validated through a nonlinear numerical example, a connected vehicle system, and a multiple-manipulator system. Comparative studies further demonstrate its advantages over existing control methods.
This article addresses the problem of adaptive output-feedback prescribed-time tracking control using preset trajectories for uncertain active suspension systems (ASSs) with electromagnetic actuator dynamics, actuator faults, and input voltage saturation. The prescribed time represents a system-level suspension-recovery requirement. A key contribution lies in the introduction of a time-varying zero-initialization function that extends preset-trajectory-based control to an adaptive output-feedback design. Combined with a preset error trajectory, this function defines state-estimation, parameter-estimation, and control errors with zero initial values despite unknown initial values of the unmeasured states and nonzero parameter estimates. A neural-network-based adaptive state estimator reconstructs the unmeasured states from the suspension-deflection measurement, while adaptive laws compensate for unknown nonlinearities and actuator faults. An adaptive fault-tolerant controller is then designed using the preset error trajectory and an input-saturation compensator. A projection-based adaptive law keeps the loss-of-effectiveness estimate positive, and conditions for selecting trajectory coefficients are derived. Stability analysis establishes practical prescribed-time tracking and semiglobal boundedness of all closed-loop signals. To represent practical operating conditions, comparative simulations are conducted under an ISO 8608 Class A road profile with localized depressions, actuator faults, and input voltage limits. The results show that the proposed method achieves earlier suspension recovery with a substantially lower peak actuator voltage than the existing fault-tolerant controller. Furthermore, simulations at different prescribed times quantify the tradeoff among sprung-mass acceleration, body displacement, and actuator demand, and 30 Monte Carlo runs show consistent responses under variations in model parameters, fault occurrence time, and road profile.
