Reliability Analysis And Optimization Of Aero-engine Gears Under Hybrid Uncertainties

Aero-engine is one of the core components of an aircraft.Its quality and reliability are the cornerstones of aviation safety.Nowadays,almost every nation pays a great attention to the development of aero-engine technology,especially for aero-engines with large complex structures.As aero-engine design and manufacturing are rapidly advancing today,the vital performance parameters in the design,including thrust-to-weight ratio are facing stricter demands.The gear plays a key role in an aero-engine and its reliability will be the focus of this thesis.Ranging from material forming and characteristics,processing and manufacturing,service,shape and size,to external environment,a variety of factors should be taken into consideration to determine the gear’s reliability.Therefore,the study of its hybrid reliability has a significant prospect in both theory and application.This thesis focuses on the reliability analysis method and reliability optimization design of aero-engine gear under hybrid uncertainties,including the following research:(1)Research about multiple sources of uncertainties in aero-engine gear from manufacturing errors,operating environment,material properties,etc.The uncertain parameters in aero-engine gear are quantitatively characterized based on random,interval and fuzzy theory.Also,a unified quantification method of random variables and fuzzy variables based on entropy equality is introduced,so to consider and deal with uncertainty factors in the following reliability analysis and optimization design problems.(2)A structural reliability analysis model under hybrid uncertainties variables is established.Considering the characteristics of hybrid uncertainties,the learning function of active learning surrogate model method in random uncertainty is improved to solve the reliability problem under the circumstances which random and interval variables both exist.Also,based on the concept of variance,a new convergence criterion is added to solve the problem that the results of the surrogate model are stable but still not convergent.The calculation accuracy and efficiency of the proposed method is proved by a numerical examples and aero-engine gear example.(3)Based on the idea of sampling and re-sampling,the Sobol sensitivity analysis method is improved,which is more suitable for complex calculation models,and it comes with the improvement of calculation efficiency and accuracy.The improved method is applied to the reliability sensitivity analysis of aero-engine gear,and the uncertainty parameters which have important influence on the uncertainty of failure probability are calculated.(4)The reliability optimization model under hybrid uncertainty variables is established.Based on the principle of entropy equality,the fuzzy variables are transformed into normal random variables.Based on conservative idea,a reliability optimization design method for mixed random variables and interval variables is established considering the situation when the failure probability reaches the maximum.Then the optimization method is decoupled,and the improvement of effectiveness and computational efficiency of the method is verified by a numerical example and an aeroengine gear example.