Research On Reliability Data Analysis And Prediction Method Of Civil Aviation Engines

As the core and foundation of general quality characteristics,reliability is an important factor that determines the safety and economy of engine.Relaibility design plays a decisive role in engine reliability performance,while prediction of reliability at the design phase is an important part of reliability design.Traditional prediction methods are often inaccurate,require too much information or can not describe the prediction results very well,so it is urgent to develop new and more applicable prediciton models.As our country has just started the research and development of civil aviation engines,all kinds of reliability work inevitably need to refer to foreign advanced engines and extract useful information from their operation data.Therefore,based on the current problems,this thesis firstly selects the most appropriate parameters that can reflect the engine reliability performance after comparing and analyzing the reliability parameters of domestic and foreign manufacturers and airlines,and establishes the reliability parameter system of engine system.In order to carry out reliability work,a large number of reliability data are needed as support,so it is mainly based on the data collected in the actual operation of airlines.In order to ensure the completeness and comprehensiveness of data analysis,the following work has been done: outlier testing and distribution parameter estimation for data;filling with multiple imputation methods for missing data;introducing faultless data analysis method.After the data is analyzed and processed,it is used as the expected input to obtain the required reliability parameter values through the prediction model.After in-depth study of various prediction methods,three prediction methods are proposed for different parameters and phases by analyzing the shortcomings and limited scope of traditional prediction methods.The first one is the monte carlo simulation method based on the fuzzy reliability model.Based on the establishment of the reliability model,the uncertainty factors in the actual problems are expressed in a fuzzy way,and simplified by Monte Carlo simulation.On the basis of simulation results,the importance analysis is carried out to determine the weak links.The second one is the method of multiple regression analysis and neural network prediction through the correlation analysis of performance parameters and reliability parameters.The third one is to make full use of expert's engineering experience knowledge in the design phase.Fuzzy comprehensive evaluation is carried out on the basis of analyzing the influencing factors of the engine system,suitable for reasonably estimating the parameters with less data and more difficult to predict.Through the predicition analysis of the engine system,it is found that the Monte Carlo simulation is suitable for the detailed design phase,and the reliability index of the system can be calculated very accurately.Performance parameter analysis and fuzzy comprehensive evaluation are applicable to the preliminary design stage,which can roughly estimate the reliability performance of the system and can provide a certain reference value for reliability work.They complement each other and make up for the deficiency of existing prediction models.