Research On Aero-engine Deterioration Evaluation

With the continuous development of aviation industry,people are paying much more attention to aviation safety issues.As the main power source of aircraft,aero-engine is such important as to ensure aviation safety.Aero-engines are usually operated under harsh conditions such as high temperature and high pressure for a long time.As the time on the wing increases,its working status changes constantly,and the performance status of each component also deteriorates.Evaluating the deterioration status of aero-engine is helpful to improve the safety,economy and reliability of aero-engine.Based on the historical data of aero-engine,this paper uses methods such as fuzzy C-means to analyze and excavate the operation rules of components and whole of aero-engine,and then evaluates the aero-engine deterioration status.The work in this paper can provide a good theoretical reference for failure detection and maintenance of aero-engine.The main contents are as follows:The first is data preprocessing.The aero-engine is operated in a volatile environment,where its parameters data are easily affected by environmental factors such as noise.In this paper,by pre-processing in four aspects: outlier extraction,missing value processing,smoothing processing and correlation analysis,the quality of original data is improved.The second is deterioration evaluation.In view of the multi-parameter problem of aero-engine,the principal components analysis algorithm is firstly used to reduce the attributes of aero-engine performance parameters.And the principal components affecting the engine performance are determined according to the contribution rates of different parameters.After extracting the evaluation criterion using the fuzzy C-means,the degree of deterioration is decided by analyzing the distance between each data point and the evaluation criterion.In addition,considering the multi-working conditions of aero-engines and the differences in the evaluation criteria for different working conditions,the k-means algorithm is firstly used to judge the current working condition,and then the single-working condition evaluation method is used for deterioration evaluation.The third is the extraction and prediction of safety boundary of performance parameters.Due to differences among individuals and differences among working environments,there are some discrepancies about parameters of each aero-engine during actual working process.In order to refine the standard theoretical boundary of aero-engine,based on the fuzzy C-means algorithm,the residuals between the predicted values and the actual values are used asquantitative indicators to extract the safety boundary,and the extreme learning machine algorithm is used to make predictions.What's more,the accuracy and generalization ability of safety boundary extraction and prediction are improved by correcting for special cases.When the performance parameters of the engine change with the increase of the wing time,the safety boundary determination method proposed in this paper can provide more accurate theoretical basis for performance parameters evaluation and prediction.Finally,it is the improvement of deterioration prediction model.On the basis of the evaluation result of the deterioration status of aero-engine,the heuristic segmentation algorithm based on nonlinear fitting is used to divide the deterioration data into "slow deterioration" stage and "fast deterioration" stage in this paper.Based on the improvement of the traditional model,the cumulative time effect of the recession process and the effect of the deterioration status cased by the deterioration trend itself are both considered.At the same time,the model is modified by adding prediction data into the training set,and the accuracy of the prediction is improved.