Research On Measurement Modeling And Estimation Problem For Networked Control System With Random Delay

Networked control system has attracted much attention in the current control field. However, as a direct consequence of finite bandwidth as well as the colli-sion in the process of information transmission, random delays, packet losses, out-of-sequence measurements and multiple-packet transmission are inevitable in net-worked systems. These problems destroy the performance of the system, even lead to system instability. The analysis and design of networked control system will become more complex. This paper proposes the random uncertain measurement model when there are random time delay and out-of-order measurements in net-worked transmission. Furthermore, the problems of state estimation are studied in some various senses for networked control system. The main contents of this paper are listed as follows:1. Under the assumption of no packet loss, two measurement models are pre-sented for random delay and out-of-order measurement, respectively. By a set of permutations for all measurements at each time, the model is given without using time stamps; while the second model improves the pre-existing model with time stamp, it guarantees each packet must be received and only once, and it also allows to receive multiple packets once time. It is more suitable for network communication protocol.2. The state estimation problem is studied for the unstable networked con-trol systems subject to one-step random delay by using measurements without time stamps. Firstly, we consider the case that two measurements are received at the same time with out-of-sequence, and design the optimal estimation under two different es-timate conditions. The primary difference between these two conditions is whether to guarantee estimation error covariance is uniformly bounded or not. The same es-timator form is obtained based on the average of all received measurements at each time. Finally, the estimator gains can be derived for one-step random delay. 3. The state estimation problem is further studied for the unstable and stable networked control systems subject to bounded random delay N>1, respectively. The problem of out-of-order measurement becomes more complex with bounded random delay N>1. With the assumption of FIFO principle, we then study the so-called unbiased, uniformly bounded linear state estimators and show that the es-timator structure is given based on the average of all received measurements at each time. Without the FIFO assumption, the optimal estimation is obtained for unstable system by analyzing the relation of the number of the received packets and delay packets. Finally, we generalize the results to stable system with FIFO, and get the optimal estimation only in the sense of unbiased.4. With the given probability distribution of the random delays, the optimal estimation and H∞filtering are considered by the measurement model with time stamp, respectively. The measurement model with delays is written as an equiv-alent the measurement with delay free. Then the optimal estimator is derived by adopting a re-organized innovation analysis approach. For the measurements can be divided into certain and uncertain, we firstly design the estimator based on the certain measurement, then give the estimator by the uncertain part. Finally, H∞filtering is obtained and can be guaranteed the exponentially mean-square stability of the closed-loop system.5. The observer-based H∞control problem is considered for networked control system subject to bounded random delay N=1and N>1, respectively. By the measurement model with time stamp, the state estimator is designed by certain mea-surement and uncertain measurement. Then the observer-based controller is obtained by solving linear matrix inequality, which guarantees the exponentially mean-square stability of the closed-loop system and satisfies the prescribed H∞performance.