Fault Diagnosis For Gas Path Components Of Variable Cycle Engines Based On A Nonlinear Model

Variable Cycle Engine(VCE)is a new generation of aeroengine which has the characteristics of both turbofan engine and turbojet engine.It can realize the advantages of turbojet and turbofan engine under different working conditions.Since VCE has many adjustable geometric components and complex nonlinear conditions,the fault diagnosis of VCE’s gas path components has become a focus of the VCE research.There are two kinds of fault diagnosis methods based on model and data-driven as the main application.Because of the complexity of VCE mechanism,the first step of fault diagnosis is to build its nonlinear model.According to the established model,the Unscented Kalman Filter(UKF)can be directly used to estimate the health parameters representing the failure degree of the gas path components.In addition,data-driven methods can also be integrated with model-based methods,to enhance the ability of realtime diagnosis.In this paper,the nonlinear model of VCE is constructed firstly,and the appropriate health parameters are selected to represent the failure degree of the gas path components.In order to estimate the health parameters that can not be measured directly,the sensitivity,correlation and validity of the parameters are analyzed,and the parameters that can be measured are determined.Then UKF is studied to estimate the health parameters,and it is improved according to the stability and computation problems in UKF.In this paper,the SSUKF method with square root filtering and spherical sampling strategy is studied.The method reduces nearly half of the computation burden and improves the stability of the UKF.To improve the real-time performance of fault diagnosis,a fusion algorithm for real-time fault diagnosis of VCE is proposed by fusing SSUKF with BiLSTM.Experimental results show that the fusion algorithm improves the real-time performance and ensures the accuracy of fault diagnosis by SSUKF.