Dynamic Event-Triggered Synchronization of Markov Jump Neural Networks via Sliding Mode Control.

This article proposes an asynchronous and dynamic event-based sliding mode control strategy to efficiently address the synchronization problem of Markov jump neural networks. By designing an adaptive law, and a triggered threshold in the form of a diagonal matrix, a special dynamic event-triggered scheme is applied to send the control signals only at triggered moments. An asynchronous sliding mode controller with gain uncertainty is designed by constructing a specified sliding manifold. Then, linear matrix inequalities are used to represent sufficient conditions for guaranteeing system synchronization. The error system trajectories are pushed onto the sliding surface by the controller. Eventually, the availability of the presented control strategy is demonstrated by an illustrative example.

State Estimation for Recurrent Neural Networks With Intermittent Transmission.

This work addresses the state estimation problem for recurrent neural networks over capacity-constrained communication channels. The intermittent transmission protocol is used to reduce the communication load, where a stochastic variable with a given distribution is used to describe the transmission interval. A corresponding transmission interval-dependent estimator is designed, and an estimation error system based on it is also derived, whose mean-square stability is proved by constructing an interval-dependent function. By analyzing the performance in each transmission interval, sufficient conditions of the mean-square stability and the strict (Q,S,R) - γ -dissipativity are established for the estimation error system. Finally, the correctness and the superiority of the developed result are illustrated by a numerical example.

Synchronization of Markov Jump Neural Networks With Communication Constraints via Asynchronous Output Feedback Control.

This article is concerned with the synchronization issue of discrete Markov jump neural networks (MJNNs). First, to save communication resources, a universal communication model, including event-triggered transmission, logarithmic quantization, and asynchronous phenomenon, is proposed, which is close to the actual situation. Here, to further reduce conservatism, a more general event-triggered protocol is constructed by developing the threshold parameter as a diagonal matrix. To cope with mode mismatch between the nodes and controllers due to potentially occurring time lag and packet dropouts, a hidden Markov model (HMM) method is adopted. Second, considering that state information of nodes may not be available, the asynchronous output feedback controllers are devised by a novel decoupling strategy. Then, sufficient conditions based on linear matrix inequalities (LMIs) for dissipative synchronization of MJNNs are proposed with the virtue of Lyapunov techniques. Third, by eliminating asynchronous terms, a corollary with less computational cost is devised. Finally, two numerical examples verify the effectiveness of the above results.

Finite-time cluster synchronization for complex dynamical networks under FDI attack: A periodic control approach.

In this paper, the finite-time cluster synchronization problem is addressed for complex dynamical networks (CDNs) with cluster characteristics under false data injection (FDI) attacks. A type of FDI attack is taken into consideration to reflect the data manipulation that controllers in CDNs may suffer. In order to improve the synchronization effect while reducing the control cost, a new periodic secure control (PSC) strategy is proposed in which the set of pinning nodes changes periodically. The aim of this paper is to derive the gains of the periodic secure controller such that the synchronization error of the CDN remains at a certain threshold in finite time with the presence of external disturbances and false control signals simultaneously. Through considering the periodic characteristics of PSC, a sufficient condition is obtained to guarantee the desired cluster synchronization performance, based on which the gains of the periodic cluster synchronization controllers are acquired by resolving an optimization problem proposed in this paper. A numerical case is carried out to validate the cluster synchronization performance of the PSC strategy under cyber attacks.

Event-Triggered Resilient Filtering With the Interval Type Uncertainty for Markov Jump Systems.

The problem of event-triggered resilient filtering for Markov jump systems is investigated in this article. The hidden Markov model is used to characterize asynchronous constraints between the filters and the systems. Gain uncertainties of the resilient filter are the interval type in this article, which is more accurate than the norm-bounded type to model the uncertain phenomenon. The number of linear matrix inequalities constraints can be decreased significantly by separating the vertices of the uncertain interval, so that the difficulty of calculation and calculation time can be reduced. Moreover, the event-triggered scheme is applied to depress the consumption of network resources. In order to find a balance between reducing bandwidth consumed and improving system performance, the threshold parameter is designed as a diagonal matrix in the event-triggered scheme. Utilizing the convex optimization method, the sufficient conditions are derived to guarantee that the filtering error systems are stochastically stable and satisfy the extended dissipation performance. Finally, a single-link robot arm system is delivered to certify the effectiveness and advantages of the proposed method.

Quasisynchronization for Neural Networks With Partial Constrained State Information via Intermittent Control Approach.

This work addresses quasisynchronization (QS) of the master-slave (MS) neural networks (NNs) with mismatched parameters. The logarithmic quantizer and the round-robin protocol (RRP) are used to deal with the limited communication channel (CC) capacity, then the intermittent control strategy is employed to improve the efficiency of CC and the controller. A transmission-dependent controller is designed, and the synchronization error system (SES) is established. The QS with a boundary is ensured for the MS NNs by a developed sufficient condition, and the controller design method is given. A numerical simulation is given to show the effectiveness of the obtained method.

Consensus-based distributed receding horizon estimation.

This paper studies the distributed state estimation over sensor networks based on receding horizon estimation (RHE). Firstly, a new scheme of centralized RHE is introduced, which gathers the decomposition terms instead of collecting the measurements of each node. Then, we present a distributed estimate algorithm based on the centralized RHE. To avoid the quadratic programming (QP) problem, the proposed algorithm takes advantage of the analytical solution of the centralized RHE and performs consensus steps to generalize the distributed estimation for each node, which greatly reduces each node's computation. Under the assumption of collective observability over networks, the proposed algorithm can guarantee the stability of estimation error in the case of enough consensus steps. Finally, the simulation results verify the effectiveness of the proposed method.

Reliable impulsive synchronization for fuzzy neural networks with mixed controllers.

This work studies the synchronization of the master-slave (MS) fuzzy neural networks (FNNs) with random actuator failure, where the state information of the master FNNs can not be obtained directly. To reduce the loads of the communication channel and the controller, the simultaneously impulsive driven strategy of the communication channel and the controller is proposed. On the basis of the received measurements of the master FNNs, the mixed controller consisting of observer based controller and the static controller is designed. The randomly occurred actuator failure is also considered. According to the Lyapunov method, the sufficient conditions are achieved to ensure the synchronization of the MS FNNs, and the controller gains are designed by using the obtained results. The validity of the derived results is illustrated by a numerical example.

Nonfragile Finite-Time Synchronization for Coupled Neural Networks With Impulsive Approach.

This article addresses the problem of the average stochastic finite-time synchronization (ASFTS) for a set of coupled neural networks (NNs) with energy-bounded noises. Due to the channel capacity constraint, the impulsive approach is introduced so as to cut down the communication times among the leader NNs and the follower NNs. Then, a nonfragile controller is designed to improve the robustness of the controller with randomly occurred uncertainty. The sufficient conditions that guarantee the ASFTS of the coupled NNs and the leader NNs are achieved. The boundary of the synchronization error is also obtained by constructing the monotonic increasing functions. Finally, the controller gains are given based on the derived conditions, and their effectiveness is illustrated by a numerical example.

Quasi-Synchronization of Time Delay Markovian Jump Neural Networks With Impulsive-Driven Transmission and Fading Channels.

The problem of quasi-synchronization (QS) for the Markovian jump master-slave neural networks with time-varying delay is studied in this article, where the mismatch parameters and unreliable communication channels are considered as well. A set of stochastic variables with different expectations are used to describe the fading phenomena of parallel communication channels. An impulsive-driven transmission strategy is designed to reduce the communication load, and a corresponding impulsive controller is then designed. A synchronization error system (SES) is obtained, and a convex QS condition is established for the SES. A linear matrix inequality-based iterative algorithm is proposed to reduce the bound of the SES, and the corresponding controller gains are calculated. A numerical example is provided to illustrate the effectiveness of the developed result.

Observer-Based Impulsive Synchronization for Neural Networks With Uncertain Exchanging Information.

This article investigates synchronization for a group of discrete-time neural networks (NNs) with the uncertain exchanging information, which is caused by the uncertain connection weights among the NNs nodes, and they are transformed into a norm-bounded uncertain Laplacian matrix. Distributed impulsive observers, which possess the advantage of reducing the communication load among NNs nodes, are designed to observe the NNs state. The impulsive controller is proposed to improve the efficiency of the controller. An impulsive augmented error system (IAES) is obtained based on the matrix Kronecker product. A sufficient condition is established to ensure synchronization of the group of NNs by proving the stability of the IAES. An iterative algorithm is given to obtain a suboptimal allowed interval of the impulsive signal, and the corresponding gains of the observer and the controller are derived. The developed result is illustrated by a numerical example.