Fault Diagnosis Method And Application Based On Observer

Combined with the sliding-mode observer and unknown input observer related theories, the thesis focuses on the existing problems in the field of fault detection and diagnosis technology and systematically studies non-linear fault detection and diagnosis algorithm in the industry from the aspects of improving the robustness of the fault diagnosis ability and the fault detection of induction motor system.The major contents can be summarized as follows:First, briefly described & analyzed the slide-mode theory. Following are three classical design cases from which the general linear system, the linear system with uncertainties and the sliding-mode observer method are introduced. Based on the above, the convergence examine on more common triangle nonlinear system observer used in project is made.Second, the thesis researches the fault diagnosis of the nonlinear system which partially satisfies Lipschitz condition and transfer the sensor failure into pseudo actuator fault by introducing new state variables. Then, sliding-mode observer is designed in accordance with each kind of fault model to realize multiple sensor failure detection and isolation. Effectiveness of the proposed method is proved by experiments of the single-link robotic arm.Finally, the unknown input observer theory and the d-q mathematical model of induction motor two phase synchronization are discussed & analyzed and current sensor fault model has formed. Since mathematical model of induction motor, an induction motor fault current sensor design method of robust observer is proposed based on the thoughts of unknown input observer and Luenberger observer. A simulation experiment results has demonstrated the correctness and effectiveness of the proposed method.Through the simulation examples and a large number of experimental results, it indicates that the fault diagnosis method based on sliding-mode observer and unknown input observer not only robustness the external disturbances but also reduces the rate of report error and improve the sensitivity of the fault diagnosis, which has a practical value as well as realistic meaning.