Research On Life Prediction And Maintenance Decision Of Aeroengine Based On Condition Monitoring Data

With the rapid development of the aviation industry,the safety of aircraft,Reliability gradually become the most concern of the aeroengine in the heart of a plane,is also the only power plant,its operation reliability is important index to measure the safety of an aircraft,therefore,in order to avoid safety accident due to aircraft engine fault,often need to evaluate its reliability,and predicting residual life based on the evaluation results,the reasonable maintenance optimization solution at present,the domestic and foreign research for this class of problems,mainly adopts the method based on failure physics model,the method is widespread low prediction precision,poor efficiency problem In order to improve the prediction accuracy,this paper starts with the analysis of the status monitoring data of aero-engine,and establishes the life prediction of aero-engine based on the status monitoring data.Firstly,before the life prediction and maintenance decision,this paper in order to better understand the aircraft engine system state of related data,the judgment of the predicted and the proper maintenance decisions in this paper,first of all to the structure of the aircraft engine working principle were introduced simply and on this basis,respectively from the failure of performance index is a typical failure mode important performance parameters such as Angle to study the failure mechanism and mode of transmission.Secondly,In maintenance decision making based on state variables,this paper adopts the covariate proportional hazards models as the maintenance decision model by linear regression analysis of K-M method to determine the life of the aircraft engine distribution obeys Weibull distribution,here to set the model of the basis function as the two parameter Weibull distribution function for covariate species should be considered in the model,using stepwise regression method for all significant covariate analysis condition,introduced if meet,will introduce model;Otherwise,it is eliminated;Through constantly introducing the weed out certain model of variable and maximum likelihood estimation method is used to estimate the model parameters to be determined,the Weibull proportional hazards model is build and according to the experience of the airlines,maintenance,this article selects the maximum availability and minimum maintenance cost as the optimization goal of maintenance,repairs as decision-making.Thirdly,Aiming at the advantages of memorylessness and homogeneity of Markov model,this paper improves on Markov model and establishes a multi-variable life prediction method based on fuzzy C-clustering weighted Markov model This article selects the aircraft engine reliability index for prediction of performance indexes,performance index increment sequence as input sequence forecasting model,fuzzy Cof the state of the input sequence interval clustering method,calculation of sync long transition probability matrix and state prediction of state probability,introduction weighting method based on size weight to determine the state probability prediction time in state,the level characteristics determine the prediction of incremental value concept,according to the airlines often USES the failure criterion of judgement to predict whether the performance index of the failure,according to the case of residual life calculation Finally,the simulation results of experimental data are compared with the actual results.Fourth,To improve the prediction accuracy,this paper introduces uncertainty quantification life prediction in the study,in the full analysis of uncertain factors in the process of life prediction,to estimate the reliability function of Weibull parameters are uncertain quantitative,generalized polynomial chaos method was used to construct polynomial agent model,the agent model to predict the failure threshold of boundary value back into fuzzy C-clustering weighted Markov model to predict the remaining life,get life prediction interval,and the prediction results and the actual results of the two methods of comparison,evaluate the accuracy of the prediction results of this method can avoid not only because of the uncertainty caused by failure after repair It can also provide new methods and new ideas for health supervision technology in aviation field.