Research On Prediction Of Aeroengine Performance Development Based On QAR

Aeroengine is the heart of aircraft,its health status has a significant impact on flight safety and maintenance economy.At present,airlines generally rely on foreign engine manufacturers(OEMs)engine health management system based on ACARS message data,which has problems such as high cost of use,limited function expansion in data analysis,etc.The engine PHM technology based on QAR data can solve the above problems to a certain extent.Therefore,the engine condition monitoring and prediction algorithm based on QAR data is studied in this paper,which can be used to predict the engine health status and remaining life in advance.Engine baseline model is the basis of engine performance degradation state prediction.In order to reduce the dependence on OEM condition monitoring system and fully develop QAR data for engine condition monitoring,QAR data instead of ACARS data is used in this paper,combined with deviation trend chart data,and support vector machine method is adopted to mine engine baseline model.It provides a basis for getting the deviation value of engine gas path monitoring parameters.In order to find out the over limit state and predict the remaining life of the engine in advance,a prediction model of engine performance degradation state is established by using the deviation value of engine gas path monitoring parameters.According to the purpose and time span of the prediction,the model includes the short-term prediction of engine performance monitoring parameters and the long-term prediction of engine remaining life.In view of the existing problems of short-term prediction,an adaptive gasvm online prediction model based on sliding time window strategy is proposed.Compared with the traditional time series prediction method based on SVM,the improved prediction method not only has good dynamic adaptability but also significantly improves the prediction accuracy.Aiming at the problem that the prediction accuracy of engine remaining life based on Fleet EGT decay rate is not high,the paper puts forward the engine life interval prediction based on fuzzy information granulation,and the result shows that the method can reflect the overall trend and range of remaining life change well.In view of the problem of low efficiency of engine performance anomaly detection or difficulty in obtaining fault data,engine fault detection based on QAR data is studied in this paper,and a method of engine anomaly detection based on unsupervised clustering of DBSCAN is proposed,which can successfully detect abnormal engine,thus solving the problem of low efficiency and monitoring of over limit anomaly detection technology The problem of small number of abnormal samples in supervised learning detection method.