Fault Detection And Isolation For Networked Systems Constrained By Communication Protocols

Fault detection and isolation technology for networked systems is an important method to improve the system reliability and safety.The shared network environment,complex system structure and multi-objective control tasks make the fault diagnosis of the networked systems become a challenge,and the problem of fault detection and isolation for this kind of complex dynamic system has also attracted the continuous attention of researchers at home and abroad.In the control system communication network,multiple sensors need to transmit data through a shared network,and it is easy to raise the problems of data conflict and data loss if the authority of sensor access to the network is not managed.In order to avoid the above problems and reduce the network bandwidth load,effective communication protocols are introduced in this paper to manage the sensor nodes to get access to the network.However,the introduction of communication protocols is likely to make many traditional fault diagnosis methods not suitable for networked systems with communication protocols.At the same time,the existence of communication scheduling principle is also likely to destroy the integrity of the original data packets of the system,and reduce the effectiveness of fault diagnosis.Summarizing the above discussions,in this paper,the fault detection and isolation problem of networked control systems under the constraints of communication protocols is studied.The main work and research results are as follows:To address the problem of robust fault detection for networked control systems with Round-Robin protocol in which the sensor saturation is concerned,the full order fault detection filter is designed.The sufficient condition for the existence of the fault detection filter is derived by applying Lyapunov stability theory and linear matrix inequality technique,and the corresponding optimal filter parameters can be obtained by solving a convex optimization problem with inequality constraints.Finally,a numerical simulation and a DTS200 three-tank leakage detection test verify the effectiveness and the practicality of the proposed method.The desired filter not only ensures the asymptotically mean square stability of the filtering error system but possesses much better disturbance-restraint ability.Aiming at the problem of multi-sensor fusion fault detection in discrete networked singular perturbation systems,considering that the system is constrained by sensor saturation,a Weighted-Try-Once-Discard communication protocol that can manage the permissions of each sensor to access the network is designed.The method for designing local fault detection filters and distributed fusion fault detection filter under the constraint of communication protocol is proposed.Firstly,the models of local fault detection filters and distributed fusion fault detection filters are constructed,and then the sufficient conditions for the existence of local fault detection filters and distributed fusion fault detection filter are derived by using Lyapunov stability theory,linear matrix inequality technology and distributed fusion technology.Finally,according to the convex optimization method,the fusion weight parameters and H? performance index are obtained.To address the problem of fault detection and isolation for networked systems under the hybrid communication protocol,the stochastic cyber-attack,sensor saturation and external disturbances are concerned.A kind of hybrid communication protocols that can realize both dynamic and static scheduling are designed,and a fault detection and isolation method based on residual contribution degree function is proposed.Firstly,the local fault detection filters and a distributed fusion fault detection filter are constructed.Then sufficient conditions for the existence of the local fault detection filter and the distributed fusion fault detection filter are obtained by using Lyapunov stability theory,linear matrix inequality technology and distributed fusion technology.The optimal parameters of local fault detection filters and the distributed fusion weight parameters are obtained by solving the optimization problems with convex constraints.Finally,the fault location is isolated by residual contribution degree function.