Fault Diagnosis And Reliable Control Based On SVM Pole Classification

In the modern life and production,fault diagnosis and reliable control is still an important problem we are facing and must be solved,reliable control in recent years has been gradually complete,but in some precision instrument operation,our controller is more demanding,we should not only make the system stable,but also the performance,such as instantaneous the characteristics of the system.The instantaneous characteristics of the system are closely related to the pole position of the system,so that the distribution of the poles of the system in the characteristic area is beneficial to the performance we need.In recent years,machine learning is booming,and the methods of fault diagnosis are also innovating.Timely reporting errors can help us avoid risks and recover economic losses.SVM is a kind of machine learning method based on statistical learning theory in recent years,which trains and builds model based on the principle of structural risk minimization.The main contents of this study are as follows:(1)a variety of faults often occur to the machine,especially the fault of the actuator channel.Therefore,it is of great significance to design a controller for the system to maintain stable and normal operation in the case of failure.The concept of reliable control is to design a controller for a system,no matter whether actuator or sensor failure occurs,so that it can remain stable and meet certain performance indicators.This paper studies the uncertain problem of reliable controller design of static output feedback pole assignment system,in the basis of actuator failures,the sufficient conditions for the existence of reliable controller static output strip region pole assignment system of feedback uncertainty is given,and using the LMI design method of the controller is given.The controller is guaranteed that the closed-loop system remains stable when there is no failure and actuator failure,and the poles of the system are allocated in the required stripe area.Finally,the effectiveness of the results is verified by numerical simulation.(2)the following methods are given for fault diagnosis and classification.First,the design method of the pole observer is given.By the real-time monitoring of the state of the system,the change trend of the poles of the system is estimated in real time.Secondly,through the change trend of the system poles,we use SVM to classify the system poles,and achieve real-time diagnosis of faults according to the characteristics of different channel failures and the different poles of the system poles.Finally,according to the fault system diagnosis of different types of fault,the system to implement the corresponding fault reliable control.In the third and fourth chapters,SVM is combined with static output feedback pole assignment and state feedback pole assignment problem.The poles are classified,and the fault diagnosis is carried out.Finally,the effectiveness of the results is verified by data simulation.