Impulsive Synchronization Control Of Coupled Neural Networks And Its Application

With the continuous innovation and widespread application of neural network in artificial intelligence and other fields,coupled neural networks research has attracted more scholars’ attention due to its diverse dynamic characteristics and complex nonlinear dynamic behaviors.Coupled neural networks is a specific complex network composed of multiple neural networks,its synchronization phenomenon has become one of the popular research directions in neuroscience,computer science,complex network research.For the coupled neural networks cannot achieve synchronization through internal coupling structure and information exchange,the input control signal is necessary for the synchronization with desired trajectory.Impulsive control method,as a practical control strategy capable of saving communication resources,is still faced with unresolved issues regarding its implementation during synchronization,operation under stochastic disturbances,the relationship between impulsive effect and performance indicators,the application in image encryption and other related aspects.In summary,this article has carried out research on impulsive synchronization control of coupled neural networks and its application,the main contributions are as follows:The impulsive synchronization control issue of heterogeneous coupled neural networks with mismatch parameters is studied.Firstly,the mathematical model of coupled neural networks with mixed time-varying delays is constructed,and the synchronization target with mismatched parameters is given.Based on the of pinning impulsive control strategy,all nodes are divided into different clusters.At each impulsive instant,the set of controlled nodes were updated according to the sorted synchronization errors,which could effectively synchronize the heterogeneous coupled neural networks.The synchronization error system under the pinning impulsive control is provided.The criterion conditions and the synchronization error bound are obtained through the Lyapunov stability theorem,the concept of average impulsive interval and the matrix measure method.The synchronization error bounds under different parameters are analyzed.Finally,some numerical simulations are used to verify the conclusions.The impulsive synchronization control issue of coupled neural networks under stochastic disturbances is studied.In order to reflect the disturbance in communication channels accurately,the randomly occurring uncertainties described by Bernoulli random variables is introduced to construct the distributed event-triggered impulsive controller and event-triggered condition.Afterwards,the impulsive synchronization problem of the networks in the noise environment under a class of event-triggered impulsive controllers subject to impulsive interference is considered.For the sake of hybrid impulses,an eventtriggered impulsive controller and its triggering condition are presented based on the average impulsive effect and average impulsive interval.With the utilization of the Lyapunov stability theorem,M-matrix,extended comparison principle and delayed impulsive comparison system,the sufficient criteria for the realization of synchronization and the convergence rate are derived.In addition,the proposed event-triggered conditions eliminate Zeno behavior during the synchronization process.Finally,the effectiveness of the theoretical results and control strategies is verified through some numerical simulations.The application of coupled neural network impulsive control synchronization in image encryption is investigated.The image encryption structure based on the impulsive synchronization control of coupled neural network and dynamic S-box is firstly presented.The cryptographic digest generated by the plaintext image and the hash algorithm are iterated as initial values to generate chaotic sequences.The transmission of chaotic sequence is realized via synchronizing the driver system and response system composed of coupled neural network instead of utilizing the traditional channel.The plaintext image is scrambled by random sequence and sort transformation,and the ciphertext image is generated by image patch,XOR operation and substitution operation according to dynamic S-box and the ciphertext feedback mechanism.Finally,the effectiveness and security performance of the encryption system are proved by simulation experiments and security analysis.