Font Size: a A A

Methods On Closed-component Fault Diagnosis For Nonlinear Systems

Posted on:2015-05-15Degree:DoctorType:Dissertation
Country:ChinaCandidate:J J HuangFull Text:PDF
GTID:1222330479995590Subject:Mechanical and electrical engineering
Abstract/Summary:PDF Full Text Request
With the improvement of automation and integration degree,the systems are becoming more and more complex. As the premise foundation of active fault-tolerant control, fault diagnosis methods and technology of modern systems have been the most advanced research hotspot. Modern systems mainly consist of the component(mechanical system) and the control system. The corresponding fault diagnosis has been developed into two different research directions. the former aims monitoring and diagnosing the process, the typical fault diagnosis is the mechanical rotor, which is one parts of the component. The latter is the purpose for control, which is mainly focused on the actuators and sensors fault diagnosis, but the component fault diagnosis is little studied. Therefore, based on the component characteristics and the impact on the system performance, the closed-component is defined. For the control purpose, the fault diagnosis study of closed-component is carried out. And research results are applied to the fault detection of the fixed-wing unmanned aerial vehicles(UAVS). This not only has the important scientific significance for fault diagnosis methods and enriching the fault diagnosis content, but also has the broad prospects in the field of engineering applications.The state-space method is a common and effective method in time domain. To closed-component faults, the sensor faults and the actuator faults of control systems, the state-space model of a class of continuous uncertain nonlinear system is described. This lay the theoretical foundation for the subsequent analysis and the method research of the fault diagnosis.Suppose that the system fault is a single fault. To realize the fault isolation of sensors, actuators and components in nonlinear systems, a scheme based on the two sliding mode observers is put forward. The actuators are regarded as the separate system and sliding mode observers are designed for the whole nonlinear system and actuators respectively. The residual changes of two observers can judge the sensors, actuators and components fault one by one. In the end the simulation result shows that the proposed method is effective.Actuators fault diagnosis has played a decisive role to confirm the closed-component fault in fault isolation. So Aiming at containing multiple actuators in nonlinear systems, the fault diagnosis method based on multi-observers is proposed. The full order observer is independently designed for each actuator and according to the residual change of the observers can diagnose whether there exist in the fault actuators, the number of fault actuators and which actuator fails. The simulation result shows actuators fault detection and isolation is realized. A reduced order observer is designed, and then curve fitting is used. So the actuators faults are reconstructed and the design procedure is given.The actuator fault and the component fault have a similar form of expression in the state space model, which makes it difficult to isolate the faults when fault occurs. So a fault detection method is raised for closed-component itself. Only the component nonlinear model is built. The model input is the output of the actuator. Assuming that there is no fault actuator(actuator fault is detected by sensors or observed by observer), without regard to system disturbances, the state variables abnormality of the nonlinear model can be considered to be caused by only the component fault. The discrete model of the closed-component is deduced. The observability and controllability conditions of linear discrete systems are analyzed. The uniformly stable conditions of the optimal filter are considered and real-time states are estimated by Kalman filter. The fault, especially early fault, is detected through the defined fault feature index. In the end, to the two type’s component faults, which are the step jump fault and slow drift fault, the simulation is completed and verifies the feasibility of the proposed method.The above methods of the fault isolation and detection are applied to the closed-component of the subminiature unmanned aerial vehicle(SUAV) for purpose of control. Firstly, the kinematic and dynamic equations are established and deduced under the condition that the moment of inertia is no reduction. And in certain conditions the linear longitudinal and linear lateral models are obtained. Secondly, to eliminate the disturbances and noises of the models, the optimal disturbance decoupling observer is designed. The closed-component fault diagnosis is realized through simulation. Finally according to the characteristics and structure of the flight control systems for fixed-wing SUAV, the main design idea is developed and analyzed in view of the disturbances and faults for the closed-component. And the module with component fault diagnosis is designed, which is based on the original flight control. Or the new flight control system with closed-component fault diagnosis is independently developed. Their main structures are given here. This provides the new idea for the reconfigurable control system in closed-component fault, and also enriches the content of the reconfigurable control system.
Keywords/Search Tags:Nonlinear Systems, Closed-Component, Multi-actuators, Fault Isolation and Detection, Sliding Mode Observers, Fixed-wing SUAV
PDF Full Text Request
Related items