On-line Prognosis Of Fatigue Crack Growth With Guided Wave Based Monitoring And Particle Filtering

Manufacture and maintenance of modern aircrafts become more and more costly due to their increasing level of integration.To deal with this problem,the prognostic and health management(PHM)technique was developed,and has been gaining more and more attention.Whether to deploy this technique becomes one of the symbols of advanced aircrafts.As the foundation of an aircraft,the airframe structure is the main target of the PHM technique.Accurate prognosis of its fatigue crack growth plays a core role in the PHM development.However,traditional deterministic models for crack growth prognosis may encounter large errors when dealing with a new structure,since fatigue crack growth is a stochastic process affected by numerous kinds of uncertainties.There is an urgent need for more effective and accurate prognosis methods,consequently guaranteeing the safety and improving the economy of the aircraft.This dissertation studies the on-line fatigue crack growth prognosis method that combines the guided wave based structural health monitoring(SHM)technique and the particle filter(PF)algorithm.It organically fuses SHM data with a physical crack growth model in the Bayesian framework to deal with uncertainties affecting fatigue crack growth,which is a novel prognosis method.The main innovations and works of this dissertation are given as follows:(1)The prognosis method based on the guided wave based SHM and the Gaussian weight-mixture proposal particle filter is proposed.To deal with the particle impoverishment problem during prognosis,particles are drawn from the mixture of the measurement probability density function and the prior state transition probability density function.This strategy can effectively increase the particle diversity so that significantly improving the prognostic accuracy.Validation on fatigue tests of the aircraft attachment lug structure shows the advantage of the proposed method,where the attachment lug involves complicated interactions between the pin and the lug hole.(2)The performances of different improved particle filter algorithms,which are combined with the guided wave based SHM for fatigue crack growth prognosis,is studied for the first time.The accuracy,stability,particle diversity,computational cost,and the dependency on the SHM accuracy are explored,optimizing the application strategy of the proposed prognosis framework.(3)To deal with the SHM accuracy affected by uncertainties,the prognosis method based on the on-line updating Gaussian process measurement model and the particle filter algorithm is proposed.The SHM accuracy for a new structure is improved via the proposed on-line updating strategies of the Gaussian process measurement model.On this basis,the accuracy of on-line crack growth prognosis is further improved.(4)An on-line multiple crack growth prognosis method is proposed by integrating the guided wave based multiple crack monitoring and the Gaussian weight-mixture proposal particle filter.The state vector and crack growth equations are constructed and on-line updated according to the multiple crack monitoring result,based on which multiple crack growth prognosis is carried out.Besides,fatigue tests of hole-edge cracked structures are performed for validation,showing the effectiveness of the proposed method.