Aero-engine residual life prediction method, system, device and medium

By using an improved Transformer network model, combined with position encoding, vector encoding, and a multi-head sparse self-attention module, the problem of accurate prediction of aero-engine sensor data was solved, and efficient and accurate prediction of the remaining life of aero-engines was achieved.

CN115952724BActive Publication Date: 2026-05-29XI AN JIAOTONG UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
XI AN JIAOTONG UNIV
Filing Date
2022-10-21
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

Existing methods for predicting the remaining life of aero-engines cannot accurately predict based on multi-dimensional sensor monitoring data. Traditional physical models rely on complex mathematical modeling, and RNN and CNN methods are insufficient in capturing long-range dependencies under long sequence inputs.

Method used

An improved Transformer network model is adopted. By adding positional and vector encoding of sensor data to the input sequence, and combining a multi-head probabilistic sparse self-attention module and a hybrid projection mechanism, an embedding layer, an encoding layer and a projection layer are constructed to improve the model's ability to model temporal information and extract features.

Benefits of technology

It enables accurate prediction of remaining life of multi-dimensional aero-engine sensor monitoring data, improving the model's generalization ability and prediction accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115952724B_ABST
    Figure CN115952724B_ABST
Patent Text Reader

Abstract

The application discloses an aero-engine residual life prediction method, system, device and medium, comprising: obtaining and screening sensor data of each component in the aero-engine to be predicted to obtain main influence index data; taking the main influence index data as the input of a pre-trained aero-engine residual life prediction model, and outputting the residual life prediction result of the aero-engine to be predicted; wherein the pre-trained aero-engine residual life prediction model is an improved Transform network model; the application adopts position coding information vector coding information of sensor data in the input sequence of the Transform network model, so that the model has time information modeling capability and generalization capability; a multi-head probability sparse self-attention module is introduced to improve the feature extraction capability of the model; and the residual service life of the aero-engine is accurately predicted according to the multi-dimensional aero-engine sensor data.
Need to check novelty before this filing date? Find Prior Art