| With the rapid development of science and technology,inverter has been widely used in all walks of life,small to household appliances,large to aviation industry and national defense,its safety and stability in industrial production and manufacturing has an important impact.However,in the working process,because of the aging of components,overload,short circuit and other problems,the inverter often fails.After a fault occurs,if the fault cannot be removed in time,it will have a serious impact on personnel health and enterprise economy.Therefore,it is urgent to solve the problem of inverter component failure.This paper studies the LCL inverter,including the following three aspects:Firstly,the switching characteristics of LCL inverter are analyzed,the switching system model is established,the fault characteristics in the system are mined,and the system fault model is established.According to this fault model,the main idea of fault diagnosis is to design a Luenberger observer,and determine whether a fault occurs by comparing the estimated output of the observer with the actual output of the system.If a fault is detected,locate and identify the fault.If you do not know which component is faulty,observe and identify each component.Therefore,an identification observer should be designed for each component,and an adaptive law should be designed for parameter identification.Secondly,considering the finite frequency characteristics of faults,the problem of fault diagnosis is studied in the range of finite frequency domain.Since this is a direct treatment of the performance index of finite frequency,the conservative problem of the system is avoided greatly.By using the generalized KYP lemma,the frequency domain problem can be transformed into the form of state matrix inequalities,and the parameters with finite frequency characteristics can be solved.The basic idea is to design an observer,perform robustness analysis and finite-frequency analysis on the error system,and design a fault detection observer combining finite-frequency performance and robustness,so as to achieve the purpose of fault detection.Finally,combining the characteristics of sliding mode observer and proportional integral observer,a fault diagnosis method based on integral sliding mode observer is proposed.This method not only enhances the robustness of the system,but also can detect and identify system faults quickly.The basic idea is to design an integral sliding mode observer for the fault model,and judge whether a fault occurs by comparing the estimated output of the observer with the residual of the actual output of the system.If the residual is zero,it indicates that the actual output is consistent with the estimated output and no fault occurs.If the residual is not zero,the actual output is different from the estimated output,and a fault occurs.Then three components are designed to identify the observer,fault isolation is carried out,and appropriate adaptive law is selected for parameter identification. |
