Analysis Of Time-delay, Modeling And Control For Networked Control Systems

With the fast development of control system and network technologies, networked control system (NCS) is widely applied in the modern industry control and received widely attention, which have advantages of less wires, good flexibility and share of information resources, etc. However, since the communication networks have been inserted into the control systems, some new issues such as network delay, data disorder, multi-packet transmission and packets dropout appear. Therefore, the analysis, modeling and control for NCS have been a hot research topic in the control theory and control engineering area.The main achievement of this dissertation is to analyze the characteristic of network-induced time-delay, to establish the mathematic models of NCS, and to research the methodologies of analysis and designs corresponding to NCS, so as to provide meaningful use for reference. The following outlines the research results:(1) On the basis of the research of background, concept, characteristic of NCS, the basic problems are given and a survey of NCS research work is analyzed. Some useful research aspects are summarized and establish the future study foundations.(2) The network delay's composition, cause of formation and the influences caused by different delays to the stability and performance indexes of NCS are studied. The characteristics and capabilities of three typical control networks including ControlNet,DeviceNet,Ethernet are analyzed. Based on the discussion of the influences induced by different control networks to the performance of NCS, the occasions in point and the selection criteria of network protocols are given. The principle of two driven modes of nodes and corresponding four combinations are primarily researched. The mechanism of delay introduced by driven modes is investigated, and the choice standards of driven modes are also presented.(3) The modeling of NCS with time-delays is presented. Based on the continuous models of single-input-single-output NCS, the general continuous models of multi-input-multi-output (MIMO) NCS with system noise and observation noise are established. The discrete models with short delays of single-packet transmission and multi-packets transmission with no packets dropout are considered. Then, the open-loop mathematic models and unified mathematic model of NCS with long time delays which is more than one sampling period are developed when the system is driven by anyone of the four different driven modes. At last, the discrete models of MIMO NCS with any bounded delays disturbed by system noise and observation noise are constructed.(4) Modeling and stability of NCS with and time-delays and data-packets dropout are studied. When the network is considered as a switch based on asynchronous dynamical system theory, proceeding with the modeling of simple unilateral single-packet transmission of NCS with short time-delays and data-packets dropout, the conclusion of the equivalence of bilateral transmission and unilateral transmission of NCS is proposed. Then, the model of single-transmission with long time-delays and multi-packets transmission with short time-delays is built. According to the stabilization analysis of single-transmission with short time-delays, single-transmission with long time-delays and multi-packets transmission with short time-delays using Lyapunov theory, the terms of exponential stability are deduced. And the validity and rationality of the modeling approach is approved by examples.(5) The robust H∞control problem of a discrete NCS with time-variant but bounded time-delays is presented. On the conditions of state feedback control and output feedback control, the terms of robust stability of closed-loop system are deduced by choosing appropriate Lyapunov function, respectively. With the help of LMI, the method of the optimal controller design which satisfies the performance index is proposed.(6) The characteristic, configuration and principle of fuzzy neural network (FNN) are investigated. The self-learning mechanism and control process are chiefly illuminated. A FNN controller as the demand is achieved by off-line training. The NCS with DC motor as the plant is simulated under the conditions of no fixed network delay and no other disturbances, fixed network delay and other disturbances, packet dropout and different control networks, respectively. The comparison results show the FNN controller has not only a comparatively better impact on disturbances than traditional PD controller at the same network setting, but also represent stronger adaptability and robust, which is absolutely suitable for practical application of NCS.(7) The design and control problems of a nonlinear NCS with a strong typical nonlinear characteristic of two inverted pendulums coupled by a spring as the plant are studied. First of all, by means of modeling the object and the method of decomposing the structure of the system, the model will be translated into a number of associated sub-systems, which will transform nonlinear model for the linear model. Then, a robust controller is designed by the use of LQR. Based on the general characteristics of the signal timing of NCS, a common network module is designed. The nonlinear system simulation platforms under a NCS structure are established, which are connected with the common network module and TrueTime network, respectively. And the performance of the robust controller is also tested. The method which deals with the uncertainties of the plant system and network-induced time-delay by a robust controller synthetically, not only reduces the complexity of nonlinear system modeling, but also provides the basis for the research of a nonlinear system under networks.