Typical Fault Diagnosis For The Wheel Rolling Bearing Of Civil Aircraft Based On Particle Filter

Aircraft wheel bearings play a key load-bearing and transfer motion,and are an important component of aircraft landing gear systems.Aircraft wheel bearings usually use rolling bearings,which is the main source of fault for landing gear systems.The traditional fault diagnosis algorithm can only diagnose the bearing fault in specific operation environment,and the diagnosis algorithm also has strict requirements for the fault diagnosis equipment,so it is difficult to diagnose the wheel bearing accurately.Particle filter(PF)is a Bayesian estimation algorithm based on state space model,and is suitable for state estimation of nonlinear systems.Fault diagnosis based on particle filter is a research hotspot in the field of fault diagnosis.In this paper,based on the particle filter algorithm,the typical fault diagnosis of aircraft wheel rolling bearing is studied.The main research contents are as follows:First of all,the evolution process of particle filter is combed,the implementation steps of standard particle filter are given,and the defects of the algorithm are analyzed Aiming at the problem of importance probability density function design,which can most affect the effect of particle filter state estimation,this paper deeply analyzes the causes and consequences of the problem,uses nonlinear filtering algorithm to obtain approximate posterior probability density,and improves the importance probability density function.Secondly,taking the wheel bearings of Airbus A320 aircraft as typical objects,the typical fault types of rolling bearing are analyzed and determined: inner ring damage fault,outer ring damage fault and rolling body damage fault.Based on the working principle of rolling bearings,the calculation methods of characteristic frequency of three typical faults are defined.Then,in order to improve the effect of particle filter on bearing fault diagnosis,an improved particle filtering algorithm is proposed: Quadrature Particle Filter(QPF).Firstly,a new method of recursive Bayesian filtering was designed by Gauss-Hermit quadrature law and adaptive covariance matching technology and it is called a new Gauss-Hermit quadrature filter.After that,this quadrature filter is used to obtain an approximate posterior probability density function,which is used as the importance probability density function in particle filter,so that the particle degradation problem is solved and the fault diagnosis effect is improved.Finally,a fault diagnosis technology of rolling bearing based on QPF is proposed: The acceleration signal of rolling bearing vibration in operation is collected by acceleration sensor,and based on the typical fault frequency of bearing,the characteristic frequency band is extracted from the Hilbert envelope spectrum of acceleration signal as the bearing health index.According to Paris' law,the physical model of bearing damage which can track the degradation of bearing performance is established.Based on the health index of bearing and the physical model of bearing damage,the health state of bearing is estimated by QPF,so as to realize the typical fault diagnosis of rolling bearing.The vibration data collected from the accelerated bearing degradation experiment platform is used to verify that the fault diagnosis method can effectively track the degradation degree of bearing performance and accurately diagnose the fault location of bearing.