Fault Detection For Networked Control System With Access Constraints And Packed Dropout

Networked control system has been widely used in industrial automation, robot,aerospace and other fields because of its advantage such as flexible installation, low cost,easy to extend and maintain. However, safer and more reliable networked control systemsare demanded for this application. In this paper, we studied the robust fault detectionproblem of network systems with packet losses and communication constraints in order toimprove the safety and reliability of NCS. The main research results are as follows:First, this paper consider the problem of robust fault detection for delayed networkedcontrol system, which has packet losses and communication constraints under actuatorfaults and external interference conditions. Packet losses exist between the controller andthe actuators, communication constraints exist between the sensor and controller andbetween the controller and the actuator. By using linear matrix inequality, new methods offault detection filter and threshold were proposed, and our result proved this method coulddetect fault faster.Secondly, the problem of fault detection was studied for Lipschitz nonlinearnetworked control systems with packet dropouts and communication constraints.Furthermore, these are described as Bernoulli models and exist between the channel. Then,we designed a fault detection filter based on linear matrix inequality which could make themean square exponential stable and error system satisfy the robust H performanceindex. Our results indicated that fault detection filter could quickly detect the fault, whensudden faults happened in the actuator.Finally, the problem of robust fault detection for Lipschitz nonlinear networkedcontrol systems was studied in which the packet dropouts and communication constraintsare described by Markov chain. Based on the linear matrix inequality, the sufficientconditions were obtained for the existence of fault detection filter. The design of faultdetection filter was not only sensitive to the fault, but also robust to the externaldisturbance. We firstly put forward the new threshold, then jungled whether a faultoccurred by compare the error evaluation and the threshold. The further simulation experiment showed the our method was characterization of faster fault detection and lowermissing reported rate.