Control And Applications Of Networked Semi-Markovian Jump Systems

Different modes switching,parameters jumping,structure changing and other hybrid characteristics are involved in most of the practical systems,as one of the hybrid systems Markovian Jump Systems(MJS)are normally used to describe this characteristic.However,the transition probabilities matrix in the MJS are time-invariant,which limits the range of applications through the MJS.Compared with the MJS,the transition probabilities matrix of Semi-Markovian Jump Systems(SMJS)is time-varying,which extends the scope of applications.In addition,data transmission through networks in control systems has been reckoned an important communication and control pattern,but we have to deal with the time delays,unknown disturbances,limited resources,and other adverse constrained conditions while taking advantage of its conveniences.Therefore,it has a capital significance to study networked SMJS both in control theory and practical applications.Based on the related research works,the aim of this thesis is to research the SMJS under the networked control framework,and consider the stability of system,tracking control,filtering,synchronized control and so on.Meanwhile,the mode information would be applied to construct the mode-dependent event-triggering scheme to decrease the control conservatism.All the theoretical results have been verified in the simulation examples of the related practical systems.The main research findings are as follows:①The tracking problem of networked SMJS with uncertain parameters has been tackled.The mode-dependent event triggering communication approach and controller are designed separately,the communication frequency has been reduced effectively and the SMJS has been guaranteed the H∞ tracking error problem on the basis of the system stability.At last,the proposed approaches have been simulated verified in unmanned vertical take-off and landing helicopter systems.②The filtering problems of networked SMJS have been investigated.First of all,the mode-dependent quantitative event-triggering filters are made for a class of SMJS with external disturbances.The passivity filtering problem under the limited communicate bandwidth constraints was solved,and the states of the system could be well estimated.Then the theoretical results obtained are applied in a single-link robotic manipulator system to prove.In view of the prior research results,the distributed filtering problem of the sensors networks systems would be further discussed.The mode-dependent event-triggering distributed passivity filters are built,subject to asynchronous information sampling framework.In closing,the sufficient conditions for the stability of systems which is satisfied with passive performance are presented and the effectiveness of the theoretical results are validated in simulations of the flexible robotic manipulators systems.③The projective lag synchronization of semi-Markovian jump coupled networks has been discussed.The mode-dependent reliable controller has been produced with memorized information and potential actuator failure,then the mean-square stability of the master-slave coupled networks synchronization error system has been achieved,and the sufficient conditions for the optimal solution of the controller have been obtained.The advantages of this designed method have been proved in the simulation examples of artificial neural networks and chaotic cellular neural networks lastly.④The synchronization of singular semi-Markovian neural networks has been proposed.A novel mode-dependent event-triggering synchronization controller has been introduced on account of singular characteristics and time delays of the system.The sufficient conditions of the admissibility of synchronization error and stochastic stability under a given performance criterion are established,and the optimization method of the corresponding controller.Finally,the feasibility of the control scheme has been verified in the numerical simulations.