Model Reference Control Methodfor Networked Control Systems Based On Jumping System Theory

In recent years, people combined the communication technology, network technology and control technology and then created a new control mode, namely, network control. A system using this kind of technology is called networked controlled system. Compared with the traditional control system, although the network control has a lot of significant advantages, howe ver network control also has its inherent shortcomings due to the existence of some communication constraints, such as data packet loss and network time delay. In this paper, the problem of model reference control for two kinds of discrete-time network control systems with data packet dropout was studied based on markov jump theory.First, an iterative algorithm is given to find an exact solution to the coupled Lyapuno v matrix equations of the discrete-time stochastic liner systems with Markovian jumps. It has been proved that the algorithm can obtain the solution within finite steps in absence of round-off errors. The algorithm has fast convergence speed and good numerical stability. a numerical example is given to verify the effectiveness of the algorithm.Second, we concerned the problem of model reference control for the single individual network control system. An assumption was made that the process of the data packet dropout is a one order markov chain. Then, the signal individual networked control systems is modeled as a markov jumping system by introducing feedback and using the last data packet instead of the lost data packet. The stochastic stability theory is applied to analyze the jumping system and an algorithm is designed to get the feedback controller and the feedforward controller which could make the output of the networked control system track the output of the given linear reference model system. A numerical simulation was given to show the effectiveness of the proposed model reference control approaches.Finally, the phenomenon of data packet dropout was expanded to the multi-agent network system. Through the knowledge of graph theory, different constraint mode was modeled to different topology structure. Thus, we could use a topology set to describe the communication between individuals. The switching process of the communication was supposed to a markov chain. Individuals were divided into the leader and follower. The behavior of the followers is controlled indirectly through the direct effect to the leadership. A controller is proposed to make the output of the multi-agent linear network system track the output of the periodic reference model system. A significant characteristic which could make the tracking property of the stabilization controller better is that it is designed based on the definition of stochastic stability. A analysis is made to analyze the solution procedure under the different background and the influence of the sampling time to the performance of tracking. The same as the abo ve, a numerical example is given to verify the effectiveness of the controller.