Fault Detection For Nonlinear Networked Control Systems

With the development of the computer technology and control technology, networked control systems(NCSs) have been finding their applications in broad areas, owing to their advantages in many aspects. It should be noted that faults in actuators, sensors and other components existing in NCSs are similar to other control systems. The introduction of the networks brings communication constraints to the control systems, which increase the difficulty of the study. Thus, it is important that faults can be promptly detected and appropriate remedies can be applied.Firstly, the history and development of fault detection(FD) are surveyed, and the characteristics and shortcomings of the existing results are analyzed. Some important theorems and lemmas are given.Then, the FD problem of NCSs with transfer delays is learned. Two irrelevant Markov chains are employed to characterize the delays between the actuator-sensor and controller-actuator, respectively. The nonlinear term in the system satisfies the sector restriction. An H-infinity robust fault detection filter is designed for the proposed system.Also, the problem of FD for NCSs with transfer time delays and stochastic data drops is studied. Based on the results of the pervious section, stochastic data drops between sensors and controllers modeled by random variables are considered. A weighting matrix function is added into the fault, in which the sensitiveness of residual to the fault is improved. The analysis and design of a new FD observer are proposed.Next, an FD observer is designed for a class of discrete-time nonlinear NCSs via T-S fuzzy model. Bounded state delay and random packet dropouts induced by the limited bandwidth of communication networks are considered, as well as stochastic transfer delays and data drops in the channels of measurements. An FD filter designed in the NCSs with sector-bounded nonlinearity is added into the sub system.Further, FD for a class of nonlinear NCSs via T-S fuzzy model with communicate constrains is concerned. The infinitely distributed time delays are introduced to describe the delays and data drops in the controllers. A sequence of stochastic variables is introduced to govern the random occurrence of the stochastic transfer delays and data drops in measurements, and all of the stochastic variables are mutually independent. A fuzzy FD filter is designed for the proposed system, and it overcomes the difficulties of the FD problem about data drops between the actuator-sensor and controller-actuator.Finally, the proposed methods in the paper are simulated in several examples. The simulation results show that the methods are effectiveness and feasibility.