Aero Engine Fault Diagnosis And Vibration Prediction

With the increasing complexity structure of aero-engine and the rising performance, there are more and more types of vibration fault, which can reduce reliability of aero-engine. Aero Engine Prognostics and Health Management System(PHM) is a sign and one of the key technologies of advanced engine. Since the 1950 s the United States began to study the aero-engine health management technology. And the F119 engine equips PHM system, which plays an important role in fault diagnosis. Therefore, to study aero-engine diagnosis and prediction of vibration can improve engine reliability, and have important meaning in promoting of engine integrated design.This paper combines statistics, dynamic analysis and vibration prediction to form an study idea which links the statistical model, the diagnostic model and the prediction model in one. The main work of this paper includes engine state recognition based on statistics, engine fault diagnosis based on dynamic model and vibration prediction method research. State recognition based on statistical studies the relationship between four statistical factors and vibration. And on this basis, proposed the method of how to make sure of the vibration limit as well as the method of how to identify normal of fault conditions. The diagnostic model based on dynamic studies the three typical failures- rotor blade fouling fault, dual-source beat vibration fault and rotor blades block off fault in the way of failure mechanism characteristic frequency and vibration characteristics. Then the paper comes out the vibration prediction methods and fault identification criterions of the faults. All of the statistical models, statistical factors and fault identification criterions have a support of vibration prediction. The paper studies five kind of stable conditions and transition state prediction methods. Finally, the paper uses a set of measured engine data to verify the methods and evaluate algorithms.Through these studies, the paper mainly obtained the following conclusions.(1) The stable state of the same condition's vibration amplitude is normal distribution. While repeatedly experienced the same speed, the vibration amplitude mean are different, and belong to different conditions in the diagnosis and prediction. So it needs to discussion separately. There are no significant differences in vibration amplitude variance between the different run sets, so it can be used to make sure the limit value.(2) Rotor blade fouling fault, dual-source beat vibration fault and rotor blades block off fault are three typical faults of aero-engine. The values of the fault identification criterions are rt=3.06 ? rt=87.01 and rt=4.97, which in the range of the model acceptance area, means the criterions are useful. The average relative error of linear growth forecast cases and dual-source beat cases are less than 10%. The average relative error of amplitude uprush cases are less than 20%. These means vibration trend forecasting model based on dynamic model can correctly predicts.Standard instrument test and measured data test results show that the proposed method is correct, the accuracy of the results within an acceptable range. This paper's work has strong support in engine fault diagnostic and prediction.