Research On Fault Detection For Networked Control Systems With Time-varying Sampling Periods

Networked control systems(NCSs) are combination of computer technology,control technology and network communication technology. Compared with the traditional control systems, the NCSs have the advantages of high reliability, low cost price and easy maintenance, etc. With the increasing of size and complexity of NCSs,the occurrence of faults can result to unpredictable loss. So the study of fault detection for networked control systems has very importance to preserve the ability and security of the systems.However, the network loads and irregular faults may lead the sampling period change in NCSs, which will reduce the performance of the systems. This thesis studies the fault detection of networked control systems with time-varying sampling periods under different network constraints.First, the thesis studies the problem of fault detection for NCSs with active-varying sampling periods and network delays. By using parameter-based method, the residual systems can be model as a parameter-uncertain systems. A new filter for the parameters-uncertain systems is developed in view of the robust fault detection approach. The sufficient conditions of asymptotic stability for the residual systems with H∞ performance are given with Lyapunov function technique and linear matrix inequalities(LMIs). The effectiveness of the proposed method will be illustrated by a numerical example.Second, the thesis investigates the NCSs with time-varying sampling periods under the condition of communication constraints between controller and actor. The jitter of time-varying sampling periods is modeled as a Markov chain that switches in finite values. Then constructing stochastic jump systems and transforming the problem of design fault detection filter into a LMIs problem with the optimal solution.Third, the thesis researches the problem of fault detection for networked control system with control delays, data dropout and communication constraints. Supposethat the permissible continuous packet loss has upper bound, the data dropout can be modeled as long time-delay, which is integer-periods. By using Lyapunov stability theory and LMIs technology to design the fault detection filter, which meets asymptotic stability and has robustness to the external disturbs of the residual systems.