Search

This article studies the input-to-state stability (ISS) problem of cyber–physical systems (CPSs) subject to unknown disturbance and malicious denial-of-service (DoS) attacks. A resilient event-triggering mechanism is first proposed to reduce unnecessary network resource consumption. Then, a novel predictive control protocol is put forward to skillfully compensate for lost signals of CPSs during the case of DoS attacks. It is shown that the resulting closed-loop CPSs can achieve ISS under the proposed control update protocol. One of the key advantages of our proposed control protocol is that the missing information can be compensated, and the CPSs thus, can tolerate more unfavorable DoS attacks. Moreover, an intuitive minimum interevent time is provided so as to exclude the Zeno execution of control signals. Finally, a simulation example on unmanned ground vehicle is presented to verify the effectiveness of the proposed control protocols.

This paper is concerned with noisy sampled-data networked control systems with stochastic network-induced delays and noisy sampling intervals using a stochastic discrete integral quadratic constraint (DIQC) approach. Initially, the noisy sampled-data networked control system is discretized into a stochastic discrete-time system. This system is then transformed into a feedback interconnected system with zero initial conditions. Next, a stochastic DIQC method is developed to obtain a novel stability condition for the noisy sampled-data networked control system. Based on this criterion, a two-stage algorithm is presented to design suitable controllers to ensure the closed-loop system under study is uniformly exponentially stable. Finally, a numerical examples and a fourth-order interleaved flyback modular integrated converter circuit are given to validate the efficiency of the proposed results.

In this paper, the imitation reinforcement learning-based control problem is studied for discrete-time Markov jump systems with external disturbances. First, zero-sum game method is introduced to deal with external disturbances, where control input and external disturbances are regarded as two rival players in adversarial environments. Then, the imitation reinforcement learning problem is formulated, where learner Markov jump systems aim to learn the optimal behavior of expert Markov jump systems. Considering that the dynamics information of both learner systems and expert systems is accurately known, an offline parallel imitation learning algorithm is designed for learner systems to mimic expert behaviors, which contains three steps: 1) policy evaluation, 2) search for weight matrix, and 3) policy improvement. On this basis, by observing the optimal behavior of expert systems, an online imitation reinforcement learning algorithm is presented for learner systems with completely unknown system dynamics. Moreover, rigorous proofs of convergence and stability analysis are provided to guarantee the performance of proposed algorithms. Finally, the effectiveness of the proposed method is verified by a single-machine infinite-bus power systems.

This brief seeks to introduce an observed-based sliding mode control (SMC) for fuzzy systems with stochastic communication protocol characterized by a more general switching rule. In contrast to prior research, the innovative stochastic switching rule for the stochastic communication protocol is unique in that it incorporates both sojourn probability and sojourn time concurrently, aiming to enhance communication efficiency and mitigate computational complexity. Additionally, by capitalizing on observed state information, a fuzzy sliding surface is devised, and a suitable fuzzy SMC law is suggested to ensure the reachability of the designated sliding function. Ultimately, the efficacy of this proposed control strategy is confirmed through a tunnel diode circuit model.

This brief presents an adaptive fuzzy dynamic event-triggered tracking control scheme with guaranteed transient performance for flexible-joint (FJ) robot system. Unlike traditional backstepping control method, we approach the control problem for second-order systems based on fully actuated system approach, which greatly reduces the computational complexity. A dynamic event-triggered mechanism (DETM) is proposed that not only saves resources but also adjusts the number of triggers flexibly. In addition, we design a new time-varying smooth segmentation function and normalized function which removes the constraints on the performance function as well as achieving tracking accuracy in the desired time. At the same time, it is proved by Lyapunov stability theory that the proposed control strategy ensures that all signals of the closed-loop system are bounded and the Zeno behavior can be avoided. Finally, the proposed control strategy is applied to the 2-link FJ robot on hardware experiment platform and the feasibility and superiority of the scheme are verified.

In this article, the exponential stability and controller design of large-scale networked control systems with prediction-based controller are addressed. A matrix dimension transformation method is proposed, which solves the problem of matrix dimension mismatch due to the complex structure. First, a new Lyapunov–Krasovskii functional (LKF) is introduced under the condition of the backstepping-based partial differential equation (PDE) method, and stability criteria with less conservatism are obtained with the help of free-weight matrixes and the inequality deflation technique. Second, combining with the reduction-based ordinary differential equation (ODE) method, by constructing the two-sided mode-dependent loop-based LKF, the information of sampled-data intervals can be utilized more sufficiently. Meanwhile, with the help of the proposed matrix dimension transformation method, controller design algorithms based on the two prediction methods are given, respectively. Finally, the effectiveness of theoretical analysis are demonstrated by using two coupled cart-pendulum systems, where the proposed method can obtain larger upper bounds of the allowable time delay and sampled-data intervals.