Fault Diagnosis Of Nonlinear System Based On Kalman Filter

Kalman filtering is an important research direction in the field of fault diagnosis.It makes the estimated state vector closer to the real state,which has attracted more and more attention of the researchers in the world.Kalman filter has the characteristics of small calculation and low requirements on storage capacity.It provides convenience for online real-time computing and computer implementation,which has broad application prospects.First of all,through the development of fault diagnosis in recent years,the methods of fault diagnosis technology are summarized.Besides,the research significance and the current development status of fault diagnosis are proposed.The research focuses on fault diagnosis based on model analysis.After comparing hardware redundancy and analytical redundancy,select the mainstream redundancy method in the current era.Then make the detailed explanation to the generation and evaluation of the residual and the selection of threshold.Secondly,through the research of traditional Kalman filtering method,the approximate probability distribution is added on the basis of Kalman,and the UKF algorithm which can improve the accuracy of fault detection is formed.The basic principle and development status are briefly described.Therefore,the fault isolation technique based on Kalman filtering for residual smoothing and the multi-filter residual fault diagnosis algorithm are proposed to achieve accurate fault locating.Then,the square root volume method is used in the Kalman filter algorithm.SCKF has the advantages of high precision,excellent stability and low computational complexity for solving the state estimation problem of complex nonlinear systems.For the nonlinear stochastic dynamic system with actuator failure,SCKF is used to estimate the system state.The residual time signal is designed by sliding time window technology to detect the fault occurrence,according to the state estimation result.Finally,using the traditional typical flight control system and adding the Kalman filter algorithm,the dynamic model of the flight control system is established to realize the real-time detection and diagnosis of the aircraft during the flight.In addition,theinfluence of Gaussian and non-Gaussian noise is analyzed,and the estimation performance of Kalman filter is tested,to prove the superiority of proposed Kalman filter based fault diagnosis.