The Research And Application Of Fault Dignosis Base On Nonliear Filtering

This paper developed by the French HangYu puma helicopter as the research object,established the flight state model more accurately,some typical faults are diagnosed in this paper.First,by using the model and nonlinear linearized model results show that the linear model has some error.It may lead to the phenomenon of false alarm and missing alarm,so as to reduce the accuracy of fault diagnosis.Therefore,a nonlinear filtering method is proposed to diagnose the nonlinear filter.The advantages of different algorithms are compared,and the main work is as follows:First of all,this paper introduces the classification method,research significance and development status of fault diagnosis technology.The fault diagnosis is introduced in detail based on analytic model.Then the hardware redundancy is replaced by analytic redundancy.The residual generation,residual evaluation and threshold selection are explained in detail.Secondly,it introduces the traditional linear Kalman filtering method based on the proposed approximate probability distribution of UKF and PF.And then it introduces the basic principle and development status of the algorithm.Simulation experiments prove that the method of approximate probability does improve the accuracy rate of fault detection.And then it proposes the fault diagnosis algorithm based on particle filter multi filter,residuals of fault isolation and Kalman filtering and residual smoothing.It realizes the accurate fault location.At last,there are two kinds of typical algorithm for flight control system.And it establishes the flight dynamics model,so it achieves that real-time detection and diagnosis of helicopter flight condition.Through a simulation example to compare the two algorithms of filtering estimation performance,it promotes the development of the helicopter fault diagnostic technology under the influence of Gauss noise and non-Gauss noise.