| Networked systems are closed-loop control systems where the controlled plant,sensors,controllers,actuators and other system components are connected through digital communication networks.The insertion of communication network not only brings a lot of convenience,but also brings a series of challenges,such as delay,packet loss,limited communication bandwidth and so on.The existence of these problems will affect the system stability and even cause faults.Due to the importance of the system security,if faults can not be detected in time,it will cause irreversible losses.Therefore,it is of great application value to investigate the fault detection problem for networked systems.This thesis focuses on the fault detection methods in finite-frequency domain based on Luenberger observer,interval observer and set-membership estimation observer for networked systems with limited bandwidth.The main contents are as follows:The fault detection problem in finite-frequency domain is investigated for nonlinear networked systems under stochastic cyber-attacks.A novel dynamic event-triggered scheme is introduced to mitigate the transmission burden of the network.Under the consideration of fault sensitivity and disturbance robustness,the addressed fault detection problem is converted into an auxiliary H-/H?filtering problem by augmenting the states of the original system and the fault detection observer.Taking sector bounded nonlinearity and stochastic cyber-attacks into consideration,the frequency characteristics of fault signals are considered in the design of H-performance index.Combined with finite-frequency input characteristics,the joint design algorithm is proposed for fault detection observer and dynamic event-triggered scheme under the finite-frequency fault input.Finally,a simulation example of stirred tank reactor system is presented to verify the effectiveness of the proposed Luenberger observer design method.The interval observer-based fault detection problem in finite-frequency domain is investigated for networked linear parameter-varying systems with the communication bandwidth constraint.A novel dynamic event-triggered scheme is introduced to save limited network resource.By using the interval observer as the residual generator and combining the state augmentation method,the addressed fault detection problem is converted into an auxiliary H-/H?filtering problem.By applying the Projection Lemma and the property of finite frequency input assembly,the joint design approach is proposed for fault detection interval observer and dynamic event-triggered scheme under finite frequency fault input.In addition,the residual intervals generated by the interval observer can be directly utilized for fault detection decision-making instead of designing the residual evaluation function and threshold.Finally,a simulation example of vehicle lateral motion system is presented to verify the effectiveness of the proposed interval observer design method.The set-membership estimation-based fault detection problem in finite-frequency domain is investigated for networked linear parameter-varying systems with the communication bandwidth constraint.In order to save limited network resource,a novel dynamic event-triggered scheme is adopted,and the boundaries of the disturbance and measurement noise are approximated by zonotopes to estimate the residual range in real time.By judging the relationship between the zero value and the coverage of zonotopes generated by the residual for fault detection decision-making,the set-membership estimation-based method can achieve the fault detection purpose without designing residual evaluation function and threshold.To improve the robustness of residual to disturbance and sensitivity to finite-frequency fault,the sufficient conditions for the design of set-membership observer are derived and converted as solvable linear matrix inequalities based on the P-radius criterion and the generalized Kalman-Yakubovich-Popov lemma.Finally,a simulation example of vehicle lateral motion system is presented to verify the effectiveness of the proposed set-membership observer design method. |
