The Modeling And Learning Control For Descriptor Networked Control System

With the maturing of network communication technology, the great progress of computer-controlled technology and the mutual promotion of industrial automation, networked control system (NCS) came into being. It is a set of integrated multidisciplinary system. Since it relates to a country’s national security and economic lifeline, it has attracted the general attention of the world and academia. Therefore, the networked control system has developed rapidly, significant progress has been achieved in its theoretical level and its application field has been expanded rapidly, while its basic theoretical system still has not been formed.Currently, research achievements on networked control system are fruitful. The research covers systems modeling, scheduling methods, control strategy, architecture and other aspects. Study includes the normal system and the generalized system of two categories, and obtains significantly the effectiveness of the study for the normal system. The descriptor system, as a promotion for normal system, exists in many actual system models. But there are few people participate in the study of descriptor networked control system (DNCS) which is the descriptor system for the controlled object in networked control system. Because the descriptor system is different from the normal system, the theory applied to the normal system does not necessarily apply to the descriptor system. The study for descriptor network control system becomes relatively difficult and challenging.The study for the basic theory of descriptor network control system would enrich and develop the theory of networked control system and promote the industrial practical applications at the same time. Thus, it has scientific significance and potential value.There are a lot of traditional control methods such as robust control, predictive control, have been widely applied in the study of networked control system in order to obtain better control effects. But it is difficult to guarantee the good control performance by the re-use of traditional control methods for the complexity of the controlled object. With the development of intelligent control, applying it in networked control system becomes one of the research directions for scholars and achieves good control effect and greater development. Learning control, as the senior branch of intelligent control, can effectively improve the control performance of system with the characteristics of potential law as well as adapting to the environment. Iterative learning control (ILC), as one of research focus in learning control, attracts extensive concern of the academic circles in recent years. As the simple algorithm and outstanding control effect, it is suitable for the system with repetitive motion nature. So it is more and more applied to networked control system and gradually developed into a research hotspot.Taking the descriptor system and descriptor networked control system as the study subject, we explore the control effect of the descriptor system under the iterative learning control and analyze the feasibility and effectiveness of the iterative learning control used in the descriptor networked control system. Specific research contents are as follows:1、 On account of the study subject in linear time-invariant descriptor system, under the clear research premise of descriptor control system, the characteristics, advantages and disadvantages of the open-loop and closed-loop and open-loop iterative learning rate were explored, the control effect and convergence conditions of the system under the open and closed loop iterative learning control were analyzed, and the algorithm effect of the open-closed loop iterative learning control was also verified. At last, the feasibility of its use in a generalized networked control system was put forward.2、For the continuous generalized network control system, the difference of convergence of the system under the iterative learning control was given according to the existence of time-delay and different situation of packet dropout. The iterative learning control for a descriptor networked control system was initially explored.3、Given the discrete descriptor network control system, the iterative learning control problem for the system with time-delay and packet dropout under the case of single-input single-output (SISO) and Multiple Input Multiple Output(MIMO) analysis system was analyzed and the convergence condition of the system stability was also considered.4、The simulation for iterative learning control for descriptor network control system was considered based on the Matlab/Simulink. Both the influence of time-delay and packet dropout on system performance and the feasibility of iterative learning control for networked control system were verified.The study of the networked control system is in full swing, descriptor networked control system is also poised to take off. The preliminary exploration for iterative learning control in descriptor networked control system, which is a useful attempt for the subsequent applications for descriptor networked control system, and the iterative learning control, and even learning control, is the purpose of this research.