Random Model And Control Of A Class Of Networked Control Systems

Networked control system is a combination of automation, computer network and communication technology. Under the impact of network circumstance, many issues emerged in networked control systems inevitably, such as network induced transmission delays, data packet dropouts from the failed communication, multi-channel transmission brought by the limited bandwidth, the asynchronous clock among network nodes etc. These problems may not only depress the control performance of normally designed systems, but also deteriorate the system stability.The networked induced delay is the most apparent disadvantage from the network connectivity. The constant induced delay is studied in a large amount of references, yet seldom considered the time varying ones. The data packet dropout event is also one of the most important phenomena arising from network connectivity. Therefore, modeling, stability analysis and controller design in networked control system with random delays and data packet dropouts are studied in this thesis, when the controlled plant is linear system.Firstly, the networked induced delays of the NCS are considered as interval variables governed by a Markov chain, and the NCS with data drops is considered as a switch of irregular shutdown. Using the upper and lower bounds of the delays, a discrete-time Markovian jump system with norm-bounded uncertainties is presented to model the NCS with random delays and data drops. Based on this model, a sufficient condition for the stochastic stability of system is present for the closed- loop system, and then an output feedback controller is constructed via a set of linear matrix inequalities. Matlab simulation experiment verifies the validity of the given results. Then, the data packet dropout states are estimated by using Kalman filtering method; the Kalman recursion equations with data packet dropout are obtained. Matlab simulation shows the results using the state estimated by Kalman filtering to feedback are better than the results using the previous state when data packet dropout exists. The results verify that it is effective to use Kalman filtering to evaluate the status in networked control systems with the data packet dropout.