Research On State Estimation For Two Kinds Of Discrete-Time Memristive Neural Networks

Today,with the rapid development of artificial intelligence,artificial neural networks,as one of the cornerstones of artificial intelligence development,plays an important role in promoting the development of artificial intelligence,However,the traditional artificial neural network is increasingly unable to meet the needs of the intelligent technology industry in the era of big data.The emergence of memristor-based neural networks enable researchers to better simulate the biological characteristics of synapses.Compared with the neural network circuit built by traditional resistors,the memristor-based neural network has a relatively higher integration degree,lower power consumption,higher storage capacity and smaller volume,which makes the memristor-based neural network become the research object of researchers nowadays.At present,thanks to the vigorous development of computer technology,in modern industrial control systems,new digital technologies are also widely used in practical operations.In practical engineering system design,in order to obtain better control performance,people tend to discretize the continuous system through mathematical processing.As for mathematical model of memristive neural networks(MNNs)for continuous or discrete cases,it can be described,respectively,by dynamical differential or difference equations whose parameters are state-dependent.These two models for the MNN system are not only concise and clear,but also conform to the actual situation to a considerable extent.This series of excellent modeling makes the dynamic analysis and application of MNNs get a large number of researchers from the academic and industrial circles.This thesis mainly discusses the state estimation problem for two kinds of discrete memristor neural networks.In the first step,the memory characteristics of the MNNs and the uncertain factors such as time delay and environmental noise in the neural network system are considered to establish a kind of discrete-time memristive neural networks(DMNNs)model to describe the phenomenon.The problem of outlier-resistant l2—l∞ state estimation for a class of DMNNs is discussed,and solve the problem that may cause the performance of the state estimator to deteriorate when outlier occur.Based on the previous research,then we consider the problem of energy-to-peak state estimation for a class of DMNNs under Round-Robin communication protocol scheduling.By introducing the communication protocol,we avoid the channel congestion and packet dropouts when the networked DMNNs system is used for data transmission,and obtain a good performance of energy-to-peak state estimation.In the end,the thesis is summarized and the future work is prospected.In general,the research framework of this paper is as follows:The first chapter elaborates.the research background and significance of the subject,and briefly introduces the research status of the subject.Finally,the research problems in each chapter are explained.In the second chapter,the outlier-resistant l2—l∞ state estimation issue is investigated for a class of DMNNs with time-delays.Measurement outputs could occur unpredictable abnormal data due possibly to outliers from abnormal interferences,cyber-attacks as well as vibration of equipment.Obviously,the estimation performance could be degraded if these abnormal measurements were directly taken into the innovation to drive the estimation dynamics.As such,a novel outlier-resistant estimator for DMNNs with time-delays is developed to diminish the adverse effects from predictable abnormal data.By resorting to the robust analysis theory and the Lyapunov stability theory,some sufficient conditions are established to ensure a prescribed l2—l∞ performance index while achieving the stochastic stability of the estimation error dynamics.Furthermore,the desired estimator gains are derived by solving a convex optimization problem.Finally,a simulation example is provided to demonstrate the feasibility of the proposed design algorithm of outlierresistant state estimators.In the third chapter,an energy-to-peak state estimator has been designed for a series of DMNNs under the round-robin protocol and with time varying delays.By introduce a periodic sequence to account for the scheduling effect of the round-robin protocol in the DMNNs.Sufficient conditions have been developed that ensure the asymptotically stability of the estimation error dynamics and the prescribed energy-to-peak performance is satisfied.Then,by using Matlab to solve a certain LMI to get the estimator gain matrix.Finally,a simulation example has been provided to show the effectiveness of the main results.It is still worth to research about the energy-to-peak state estimation,and it can be using in more complicated systems including sensor networks,genetic regulatory networks and social networksThe fourth chapter summarizes the main results of this article and looks forward to the future work.