Research On Synchronization Control And State Estimation Of Markov Jump Inertial Neural Networks

In recent years,tremendous focuses have been garnered in the aspect of a special kind of neural networks not only because neural networks greatly improve the disordered search performance of neural networks,but also because of increment in the complexity of the dynamic performance of designed system.Although some achievements obtain in the synchronous control and state estimation for inertial neural networks,current challenges still need to be urgently addressed.For instance,the impacts of the external environment on random jumping parameters among neurons is unclear,synchronous controls strategies of inertial neural networks via adopting the information of state estimation under unmeasurable system is also rarely focused,and the limitation on integrating synchronization control and state estimation problems of inertial neural networks with jumping parameters in the case of limited network resources and cyber-attacks during data transmission,should be further highlighted and needed to provide elaborate illustration.To solve the mentioned problems,we adopt Markov jump process to describe the random jumping phenomenon of inertial neural networks in this dissertation.Stochastic analysis methods,robust control theory,and linear matrix inequality techniques are also adopted to provide the stability conditions of the closed-loop system,controller,and estimator design methods under the influence of the limited network bandwidth,and cyber-attacks.Also,the corresponding methods are also applied in the fields of secure communication.The core contributions of this work are presented as follows:(1)The extended dissipative filtering problem of Markov jump inertial neural networks is discussed when multi neuron signals are transmitted in parallel but the network resources are limited.The weighted try-once-discard protocol is used as an important scheduling mechanism to determine which nodes can be accessed between the sensor node and the filter.Then,based on the Lyapunov function and the improved decoupling method,a set of sufficient conditions that ensure the desired properties is derived,and the filter gain is obtained.Finally,an illustrative example is employed to verify the validity of the proposed method.(2)The non-fragile extended dissipative synchronization problem for Markov jump inertia neural networks subject to limited network resources is studied.Firstly,the event-triggered control strategy is introduced to determine whether the current,data needs to be transmitted.Then,a design scheme of non-fragile controller described by Bernoulli distribution is proposed.Based on the Lyapunov stability theory and the improved inequality technology,sufficient conditions are obtained to ensure that the system is stochastically stable and satisfies the extended dissipative performance.Finally,a numerical example is provided to verify the effectiveness of the proposed method,and the results are successfully applied in the field of chaotic secure communication.(3)For the Markov jump inertial neural networks with multi-sensor,the problem of H∞ secure state estimation under replay attacks is studied.On the basis of a tag-based replay attacks detection algorithm,the redundancy of the sensor is used to obtain the effective measurement data,and the state estimator that can resist replay attacks is designed.Then,a set of sufficient conditions for the stability of the estimation error system is derived and the gains of estimator are obtained.Finally,the superiority and effectiveness of the method proposed in this dissertation are verified through comparative experiments with or without detection and defense mechanisms under replay attacks.(4)When the system subjects to random deception attacks and the system states are unavailable,the observer-based synchronization control problem for inertial neural networks with Markov jump parameters is studied.Firstly,the random deception attacks model is described by the Bernoulli sequence.According to the Lyapunov theory,some sufficient criteria are established to ensure that the synchronization error system is stochastically stable and can achieve the expected H∞ performance under deception attacks.Finally,the validity of the obtained results is verified by a numerical example,which can resist the influence of deception attacks on the system to a certain extent.