Engine bearing fault prediction method based on multiple stages

Abstract

The invention discloses an engine bearing fault prediction method based on multiple stages, relating to the field of engine health management. According to the engine bearing fault prediction method based on the multiple stages, bearing faults of an engine can be predicted in the multiple stages, and finally the purposes of reducing the maintenance cost and guaranteeing the operation safety of theengine are achieved. Prediction is carried out according to the following steps: 1) processing an original vibration signal of a bearing; 2) counting the rms characteristic curve containing a large number of data points according to a certain statistical value; 3) finding out mutation points by using statistical process analysis, and analyzing curves before and after the mutation points by usingstatistical overview; 4) dividing rms counted in the step 2) into multi-stage signals; and 5) inputting the obtained multi-stage data into a long-short-term memory network for prediction. Multi-stageprediction can be carried out on bearing faults of the engine, and the purposes of reducing maintenance cost and guaranteeing operation safety of the engine are achieved.

Description

technical field [0001] The invention relates to the field of engine health management and the field of civil aviation transportation safety research. Background technique [0002] Aeroengine bearings are an essential part of its load-bearing transmission system, and the performance of bearings directly affects the service life and reliability of aeroengines. Among the various major mechanical failures that have occurred in aero-engines in the past, the failure of rotating parts is as high as 80%, mainly the blades, discs, shafts and bearings in the rotor system. Therefore, the fault prediction of aero-engine bearings is of great significance for improving the safe and reliable operation of aero-engines, reducing aircraft maintenance costs and ensuring flight safety. [0003] Fault prediction is divided into prediction methods based on physical models, prediction methods based on statistical analysis, and prediction methods based on quantitative knowledge data analysis. The...

AI Technical Summary

Problems solved by technology

Although there have been studies combining deep learning and statistical process analysis, it has not been applied to the prediction of multi-stage failure of aeroengine bearings.

Images