System Identification Algorithm Based On Network Control System Environment

Various control strategies have achieved results in different aspects of both the industry field and academic literatures in recent years due to flexible applications and many other advantages. Although various algorithms have been proposed for networked control systems(NCS), most control schemes assume that system model is known’a priori’. It is necessary that the control plant in the network system is identified when the model of the actual networked control system is not known. In addition, owing to existing network uncertainty, including the network-induced delay, data packet out-of-order, and data packet loss, the sampled data points of the system are illustrated as a non-synchronized non-uniformly(NSNU) situation. One of the main identification tasks is how to deal with the uncertainty of the system due to network-induced delays and data packet loss. In presented paper, a modified Gram-Schmidt algorithm is applied to confirm the parameters and the model structure. In addition, a discard-packet strategy is developed which takes into account network-induced delays, data packet out-of-order and data packet loss within one framework.This study also presents a deterministic approach to the robust design Hx-estimation theory and least squares estimation of networked control model in the presence of unknown but finite uncertainties in the network identification data. The aim is to solve the difficulties associated with the robust identification method due to lack of a priori knowledge on the uncertainties of the networked identification system. The authors explore the use of Hx-estimation theory and least squares estimation for networked control model without making any assumption and requiring a priori knowledge of upper bounds, statistics and distribution.Finally, the effectiveness of the proposed scheme is shown by using a numerical example. Obviously, the performance of two robust identification algorithms is superior to that of the ordinary recursive least squares identification.