Research On Health Assessment And Prediction Method Of Variable Cycle Engine

Variable Cycle Engine(VCE)is currently one of the most promising nextgeneration flight powers,but ordinary aero engines such as turbofans and turbojet engines are already extremely sophisticated and complex power devices.Compared with turbofan and turbojet engines,the VCE have a more complex structure,a wider flight envelope range,and a more demanding working environment.While the VCE has obtained excellent performance indicators,it has also obtained a higher failure rate and more difficult predictive maintenance decisions than conventional aero engines.Therefore,it is very necessary to research the prognostic treatment for VCE.Aiming at the long-term and short-term prediction of VCE gas path performance,the research work of this thesis is as follows:First of all,to solve the problem of difficult data acquisition of variable cycle engine,based on the principle of variable cycle engine thermodynamics,and correspondingly simplified the thermal process of variable cycle engine,a component model that can reflect the thermal process of each component is built,through mass conservation and energy conservation equations,connect each component model,obtain the variable cycle engine model,and verify the steady state and dynamic simulation of the VCE model.After that,on the basis of the VCE model,the simulation of the propagation process of engine performance degradation is studied,and the fault propagation characteristics of the variable-cycle engine and the fault performance under different operating conditions are analyzed,and the health of the VCE under multiple operating conditions is studied.Then it analyzes the characteristics and requirements of the short-term prediction problem of the VCE,adopts the data-driven prediction methodGRU,and uses the fault simulation model and the engine health evaluation method to realize the high-precision short-term prediction of the variable-cycle engine.Finally,the problem of long-term engine life prediction is analyzed.On the basis of the similarity-based remaining useful life(RUL)prediction,through the trend error correction and time-domain error correction in the similarity calculation process,the adaptive calculation of similarity is realized.The XGBoost algorithm in ensemble learning is introduced,and the similarity adaptive XGBoost model of RUL prediction is proposed,which reduces the point prediction error of RUL and improves the generalization ability of the prediction model.