Research On Fault Diagnosis Methods Based On Iterative Learning Algorithm

With the development of science and technology,control systems become increasingly complicated,and many components of the system might be suffered from faults.These faults will reduce the system performance,or even cause catastrophic accidents for systems.In order to increase the system security and reliability,timely fault diagnosis is particularly important.Fault estimation is an important part of fault diagnosis,which provides the basis for subsequent fault-tolerant control.In this article,the main research on fault estimation methods based on iterative learning algorithm.The research background and research status is introduced in the first chapter.Some lemmas and definitions are presented in the second chapter,which will be used in this article.In the third chapter,considering the fact that many faults usually occur in finite frequency ranges.For the discrete-time singular linear systems,this chapter investigate the case of two different types occur fault and design fault observers in finite frequency ranges.By employing the generalized KYP lemma and two finite frequency H? performance indices,the sufficient conditions of the proposed observer are given in terms of linear matrix inequalities.Furthermore,by using slack variables,improved results on the fault estimation observer design are obtained,which are convenient to calculate fault estimation observer parameters for different frequency ranges.Compared with the traditional design methods in full frequency domain,the method proposed in this chapter has a better estimation effect and less conservative.Simulation results are presented to illustrate the advantages of the theoretic results obtained.Chapter 4 studies the iterative learning strategy to estimate the fault of a nonlinear continuous system with actuator faults and disturbances.The designed proportional differential type iterative learning estimator with robust performance not only can estimate the system fault effectively,also can estimate the system state changes accurately.The introduced virtual fault parameter value is determined by the virtual fault value of the previous moment,the output error value decision and the output error derivative value.If the fault occur in systems,the iterative learning fault estimation algorithm continuously updates the virtual fault to achieve an accurate estimation of the type and time range of system faults.The output feedback is used to deal with the disturbance and reduce the impact on the normal operation system.The convergence of the diagnostic algorithm is proved by the method of contraction mapping.Finally,the electromechanical control system simulation is used to verify the rationality of the method.The fifth chapter is consistent with the thought in the previous chapter.In this chapter,a iterative learning estimator with robust performance is designed to study the fault estimation of discrete nonlinear systems.Different from the fourth chapter,the introduced virtual fault parameter value is determined by the virtual fault value of the previous moment,the output error value decision and the output error difference value.The concrete scheme is to construct the form of iterative learning fault estimation observer in the form of fault estimation term and robust term.Then set the threshold range,if the output residual of the system exceeds the pre-defined threshold,it indicates that the system occur faults.The estimator of the observer constantly updates virtual fault to achieve the magnitude of the system faults.The robust term is used to ensure the asymptotic stability of the estimator.The given simulation show that the proposed method is effective.At last,conclusions and further research directions are given in the sixth chapter.