Neural Network Synchronization Based On Quantized Intermittent Control And Its Application In Image Encryption

Chaos synchronization is a nonlinear dynamical phenomenon widely used in communication,cryptography,control systems,and other fields.In recent years,researchers have applied chaos synchronization to digital image encryption to improve the performance of digital image encryption technology.This method has improved the security and reliability of digital image encryption.Therefore,this thesis investigates the chaotic synchronization of several types of neural networks in-depth,mainly including finite-time synchronization and fixed-time synchronization of time-lag neural networks,fixed-time synchronization of complex-valued neural networks,and finite-time synchronization of proportional time-lag neural networks.Based on the finite-time synchronization of proportional time-lag neural networks,a class of digital image encryption and decryption algorithms are proposed in this thesis,and various performance analysis methods verify the effectiveness of the algorithms.The main contributions and contents of this thesis are as follows:(1)Finite-time synchronization and fixed-time synchronization of time lag neural networks are investigated.Based on the quantized intermittent theory,the synchronization controller of the time-lag neural network is designed,the finite-time synchronization and fixed-time synchronization criteria are established respectively,and the convergence times of finite-time synchronization and fixed-time synchronization of the time-lag neural network are obtained.The results show that the convergence time of finite-time synchronization depends on the initial state of the time-lag neural network,while the convergence time of fixed-time synchronization is independent of the initial state.In addition,the feasibility and validity of the time lag neural network theory are verified by numerical simulation.(2)The fixed-time synchronization of complex-valued neural networks is investigated.Based on Lyapunov stability theory and quantized intermittent controller,two sufficient conditions for fixed-time synchronization of complex-valued neural networks are established,and the convergence time of fixed-time synchronization of complex-valued neural networks is obtained.In addition,based on the established two sufficient conditions,two corollaries of fixed-time synchronization are obtained.Finally,the feasibility of the inferences is verified by numerical simulation.(3)Finite-time synchronization of proportional time-lag neural networks and its application in image encryption are researched.Firstly,two synchronization criteria for finite-time synchronization of proportional time-lag neural networks are given by quantized intermittent controller and Lyapunov stability theory,and the synchronization time of proportional time-lag neural networks is estimated.Secondly,a class of digital image encryption and decryption algorithms is proposed based on the finite-time synchronization theory of proportional time-lag neural networks.Finally,the algorithm’s effectiveness is verified by various performance analysis methods,and the algorithm in this thesis is compared with other image encryption algorithms.