Research On The Performance Of Feedback Control In Cyber Physical System

Cyber-physical system is a new system which is based on the wireless sensor network, the computer control system and the complex network. On the basis of full awareness, CPS is for controlling more completely. It is a novel research field for structuring a controllable, credible, extensible and secure CPS network. CPS has the nature of control system, communicating access network makes the system more complex in modeling and analyzing.For real time, CPS is strict with the response time. In this case, there is a big problem that how to reduce the influence of random time delays.Because of the different nature of time delay, we can divide it into two parts, the sensor-to-controller time delay which can be measured using timestamp and the controller-to-actuator time delay which could not be measured since it has not happened. So, we proposed a dynamic timeout scheme and actuator compensation method to solve this problem respectively. The main contents are as follows:1. we abstract a control system block diagram of CPS. And we explain that different node driven mode makes different system timing model. We derive the system state space equation.2. Compared with the constant delays, the random independent delays and Markov delays, we find that these three models simplify the model of time delay, and they can not describe the model exactly. we model the delays as a Hidden Markov Model. Actually, the time delay is infinite and continuous. To model the time delay, we quantize the time delay, that will make the observable variable be a finite and discrete set.3. We propose a dynamic timeout scheme and an actuator compensation method to solve the two cases in controller design. The dynamic timeout is used to solve the problem that controller nothing to do only waiting the data coming. We determine the appropriate timeout value by predict the network state using hidden markov model. We also take the system criticalness into consideration, that makes the CPS coupling with the physical world more tightly. In actuator compensation method, we predict the network state using hidden markov model. Based on the network state, the controller calculates a set of control lows and package them to the dispatch node, then the dispatch node can calculate the timestamp and choose the appropriate control law. An optimal control law is designed to compensate directly for the effect of the whole time delay.If we can get the whole system state, we can use state feedback control, but if the whole system state could not be got, we can use output feedback control.4. We introduce the truetime box and design a simulation platform based on TrueTime. We give the detail parameter design of the simulation platform. We do two sets of experiments. The first set of experiments, we design the controller ignoring or considering the stochastic time delay, the result shows that we can not ignore the time delay in designing the controller. The second set of experiments we consider the dynamic timeout scheme and actuator compensation method. The average system stability time is analyzed. The results show that the dynamic timeout scheme and actuator compensation method can improve system performance.