Robust Fault Detection For Networked Control Systems With Time Delays And Packet Dropouts

With the rapid development of communication technology, Networked Control Systems(NCSs) which arises from the combination with the traditional control systems have been applied widely in recent years, for the sake of insuring NCSs circulating safe and credible, it is more important for the study on the Fault Detection(FD), while the robustness of fault detection system is the key issue for the FD, as well as one of practical evaluation criterions. The information transmission time delay and packet dropout issues caused by the network not only likely to worsen the performance of systems but also may even lead to the instability of systems, which must be considered during the study of the fault detection of NCSs. At the same time, we need to pay attention to the effect of the uncertainty on the system caused by external disturbances and modeled errors. From the normal controlled object to the complex, and from linear NCSs to nonlinear NCSs, which have been studied gradually on their robust fault detection, and the proposed methods have been proved to be effective through the numerical simulations.First of all, the problem of robust fault detection for networked control systems with time-varying delay and random delay is studied respectively in the paper. By designing the fault detection filter based on the filter form and observer form respectively, the residual signal is produced and the problem of robust fault detection is converted into the robust H∞ filter design problem. The designed robust fault detection filters guarantee the fault detection system(mean-square) asymptotically stable, satisfying a prescribed H∞ performance, Threshold logic of a decision function is used to evaluate the generated residual and a numerical example is given to verify.Secondly, the problem of robust fault detection for networked control systems with two types of random packet dropout is studied respectively, the random packet dropout is modeled as the Bernoulli distribution sequence and Markov chain, and combined with the robust fault detection filter to form an augmented system, using the LMI method, the condition of the system mean-square asymptotical stable and stochastic stable, satisfying a prescribed H∞ performance is derived respectively, robust fault detection filter parameter matrices are got by solving a convex optimization problem about H∞ noise attenuation level ?, and the proposed methods have been proved to be accurate and effective through the numerical simulations.Then, the problem of robust fault detection for nonlinear networked control systems with both random packet dropout and time delay is studied, the nonlinear satisfy the Lipschitz nonlinear condition and sector boundary nonlinear condition respectively. By designing the robust fault detection filter, the residual signal is produced and the problem of robust fault detection is converted into the robust H∞ filter design problem. The designed robust fault detection filter guarantees the fault detection system mean-square asymptotically stable, satisfying a certain H∞ noise attenuation level ?, and the proposed methods have been proved to be effective through the numerical simulations.Finally, the problem of robust fault detection for Linear Parameter Varying(LPV) system is studied under network environment. Using the method of directly estimating the fault signal to design filters, using the LMI method, the condition of the system asymptotically stable, satisfying a prescribed H∞ performance is derived. To a slack matrix is introduced to decouple the coupling between the system matrices and the parameter-dependent Lyapunov matrix, then a new theorem is obtained. The infinite dimensional parameterized linear matrix inequality is transform to the finite dimensional parameterized linear matrix inequality by basis function and gridding technique. Finally, threshold logic of a decision function is used to evaluate the generated residual and a numerical example shows that using the proposed approach can detect the fault fast and precisely.