Synchronization Control Research Based On Associative Memory Neural Network With Its Application In Encryption And Decryption

The associative memory neural network is a nonlinear dynamic system that simulates human cognitive and activity processes,possessing powerful self-learning,self-adaptive,and self-organizing abilities that allow for efficient processing of information related to objects and environments.Due to its large-scale network structure and complex connectivity,as well as the influence of the nonlinear characteristics of the memristor,it can exhibit rich dynamical behaviors.Among them,synchronization is an important phenomenon that refers to the coordinated and coherent behavior among different neurons in the network.This synchronization behavior can not only help the neural network process information more efficiently,but also provide insights for understanding the operating principles of cognitive neural networks in humans.Currently,the synchronization control of neural networks has become an important topic in the field of neural computation.This article investigates the synchronization control problem of the associative memory neural networks and applies it to the field of secure communication.The main work and innovative results are as follows:1.This thesis investigates the problem of predefined-time synchronization control for memristive complex-valued bidirectional associative memory neural networks with leakage time-varying delays,aiming to address the limitations of finite-time synchronization and fixed-time synchronization.Based on theories such as differential inclusion,set-valued mapping,and Lyapunov stability,a feedback controller is designed,and sufficient conditions to ensure the drive-response system achieves predefined-time synchronization are derived.This ensures that the stability of the error system is not affected by the initial conditions,and the stable time can be intervened manually.Moreover,an image encryption and decryption scheme is proposed based on the network’s chaotic characteristics and synchronization control principles,successfully applying the network to the field of image protection and achieving good encryption and decryption effects.2.This thesis addresses the problem of the large computational cost of complex-valued neural networks,and studies the predefined-time synchronization control problem of memristive complex-valued bidirectional associative memory neural networks based on the complex-valued undivided method.A new approach to processing complex-valued neural networks is proposed by introducing complex norm and signum function.Moreover,a new predefined-time stability theorem is presented,which is more generally applicable in its judgment form.Finally,a simulation-based secure communication scheme for analog signals is proposed and its effectiveness is validated through simulation.3.This thesis investigates the predefined-time projective synchronization control problem of memristive multidirectional associative memory neural networks,in response to the diversity of predefined-time synchronization theorem forms.By defining a modified function projection synchronization,the network can exhibit multiple synchronization modes,which to some extent enhances the engineering practicality.Additionally,an integrated predefined-time stability theorem is proposed,which reduces the parameter quantity in the judging condition.By designing a feedback controller,sufficient conditions for achieving predefined-time projective synchronization of the drive-response system are obtained.Based on this,an audio encryption and decryption scheme is proposed,and the effectiveness of this scheme is verified through a simulation example.4.A set of encryption and decryption systems for images,signals,and audio has been developed based on MATLAB App Designer.This system is easy to operate,highly scalable,and can visually display the simulation results of this article.