The Study Of Synchronization Control For Memristive Neural Networks Based On Event-triggered Mechanism

Memristor,as a type of resistor with memory capability,has the advantages of nanoscale,fast switching and low power consumption.It is considered to be an ideal component for simulating synapses and realizing brain like computing The emergence of memristor provides a new research direction for the contradiction between computation and storage in Von Neumann architecture.Memristor based artificial neural networks(memristive neural networks)has attracted extensive attention in the fields of pattern recognition,artificial intelligence and other fields.As one of the research hotspots of neural networks,synchronization plays an important role in secure communication,pattern recognition,associative memory and information science.Therefore,it is an important work to study the synchronization for memristive neural networks.At present,most of the synchronization control methods are based on the feedback control of time-triggered mechanism,and the amount of information transmission is large.However,the sensors and other devices are connected through networks in the control systems,and the communication environment is usually congested and dispersed in actual systems,in order to further improve the control effect of the system and reduce the utilization of communication bandwidth,this dissertetion studies the synchronization of memristive neural networks based on event-triggered mechanism.The following is the main contents of this dissertetion.The quasi-synchronization control of time-varying delayed memristive neural networks with parameter mismatches is considered.Firstly,the first order differential equation of Lyapunov function is obtained by nonsmooth analysis method and using inequality expansion and contraction technique.Then,by using the comparison principle of impulsive system and the variable parameter formula,the bound of synchronization error and the exponential convergence rate are obtained based on event-triggered impulsive and state feedback control.Furthermore,a collaborative design method is proposed to make the synchronization error converge to a predetermined level.Finally,a self-triggered scheme is proposed to avoid the continuous monitoring the event-triggered condition.Numerical examples show the effectiveness of the obtained results.A novel hybrid impulsive control algorithm,which combines time-triggered and eventtriggered impulsive control,is proposed to realize the quasi-synchronization of delayed memristive neural networks.Firstly,when the initial value of Lyapunov function is larger than the initial value of event-triggered threshold function,the Lyapunov function can be smaller than the event-triggered threshold function in finite time via time-triggered impulsive control of finite times.Furthermore,when the Lyapunov function is smaller than the event-triggered threshold function,only the event-triggered impulsive control can realize the quasi-synchronization of memristive neural networks.This method can effectively reduce the number of impulsive control.In addition,by adjusting the parameters of the event-triggered threshold function,the synchronization error norm can be controlled in any small range,which can reduce the conservatism of the existing quasi-synchronization results.Numerical examples are given to show the effectiveness of the results.The quasi-synchronization of memristive neural networks with communication delays is considered.By designing the event-triggered impulsive control strategy,and by using differential inclusion theory and analysis techniques,several algebraic conditions are given for the quasi-synchronization of memristive neural networks.Moreover,a novel switching event-triggered mechanism based on communication delays is established.More to the point,there is no restriction on the derivation of Lyapunov function.Then,in order to further reduce the time of continuous sampling and monitoring,we further propose a switching event-triggered mechanism depending on the communication delays and the aperiodic sampling,which is more economical and practical and can directly avoid Zeno behavior.Finally,numerical examples show the effectiveness of the obtained results.The synchronization control of memristive neural networks with unknown parameters is considered,where the unbounded discrete and bounded distributed time-varying delays are involved.By combining the event-triggered mechanism with the adaptive control,the adaptive law of the weight of the response system and the adaptive law of the controller gain are designed,the sufficient conditions for the synchronization between the response system and memristive neural networks are obtained.By constructing suitable Lyapunov function and using inequality expansion and contraction technique,the influence of unbounded discrete delays and bounded distributed delays are solved.The event-triggered mechanism is introduced to adaptive control,which reduces the number of controller updates and eliminates the chattering caused by symbolic function.In addition,a relative threshold strategy,which is relative to fixed threshold strategy,is proposed to increase the inter-execution intervals and to improve the control effect.Finally,one simulation is presented to validate the effectiveness of the proposed results.Finally,the research results of this paper are summarized,and the future research directions are prospected.