Fault Diagnosis And Reliable Control Based On Regional Pole Classification

With the increasing importance of control system in production and life,how to ensure the stability,safety and reliability of system components has been widely favored by researchers.As a common intelligent learning machine in the field of fault detection,support vector machine has become a common method to classify data and realize fault diagnosis by virtue of its strong generalization and simple structure.However,the parameter selection of SVM needs to agree on the parameter range in advance,which restricts its applicability largely.Therefore,it has important practical significance for optimizing the technology of parameter selection.Based on multi parameter seeking technology of SVM,considering various regional and the disturbance of external uncertain factors,this paper makes an in-depth study on fault detection and reliable control,and the main results are as follows:Combined with the characteristics of SVM dynamic system and reliable control,an effective method for fault detection of single channel actuator in disk area is given.Through the learning and separation of the pole information by the classifier of SVM,the accurate location of the actuator fault is realized,so as to exchange the reliable controller of the corresponding channel.Finally,the effectiveness of the controller is verified by several examples.The fault detection method based on the modified particle swarm optimization algorithm is studied.At the same time,aiming at the problem that the inertia weight wis easy to be limited,an optimization model of MPSO-SVM is established and the algorithm is improved from the perspective of w adaptability.At the same time,according to the algorithm of pole observer in Chapter 2,the estimated value of pole is given.Finally,a basic example is given,and the reliability of the model is obtained by comparing four methods:SVM?Grid search-SVM?PSO-SVM?MPSO-SVM.The fault detection method of genetic algorithm based on optimized SVMparameters in trapezoidal region is studied.At the same time,combined with the algorithm of pole observer in Chapter 2,the estimated value of pole is given.Finally,the feasibility and effectiveness of GA algorithm for dynamic fault-tolerant control of the system are verified by a simulation example.