Space-time fusion depth representation fault detection method for persistent test run blade damage of aero-engine

By using a spatiotemporal fusion attention autoencoder model, the shortcomings of multivariate modeling in the detection of blade damage in aero-engines are addressed, enabling highly sensitive unsupervised detection and improving the accuracy and reliability of aero-engine health monitoring.

CN121659074APending Publication Date: 2026-03-13DALIAN UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511851902.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-12-10
Publication Date
2026-03-13

AI Technical Summary

Technical Problem

Existing technologies for detecting damage to aero-engine blades have shortcomings such as poor adaptability to operating conditions, weak multivariate modeling capabilities, low sensitivity, reliance on manual threshold settings, and susceptibility to false alarms and missed alarms. In particular, they are difficult to achieve high accuracy and early warning when fault samples are scarce.

Method used

The Spatiotemporal Fusion Attention Autoencoder (STFAE) method is adopted. By fusing the temporal and spatial characteristics of multi-source sensor data, a spatiotemporal fusion attention autoencoder model is constructed and unsupervised training is performed. The self-attention mechanism is used to capture the temporal dynamic pattern and spatial cooperative structure, so as to achieve high-sensitivity identification of blade damage faults.

Benefits of technology

It achieves high-precision, early identification and warning of blade damage faults, improves the stability and robustness of detection, and can effectively detect faults without faulty samples, making it suitable for health monitoring of aero engines.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121659074A_ABST
    Figure CN121659074A_ABST
Patent Text Reader

Abstract

The invention discloses a space-time fusion depth representation fault detection method for aero-engine persistent test run blade damage, and belongs to the technical field of aero-engine health monitoring and intelligent fault diagnosis. The detection method specifically comprises the following steps: S1, data acquisition and feature selection; s2, preprocessing the multi-source sensor data, constructing a time sequence sample window with a fixed length, and dividing a training set and a test set; s3, designing and constructing a space-time fusion attention auto-encoder model composed of a position code, an encoder with a multi-head self-attention mechanism and a decoder; s4, performing unsupervised training on the auto-encoder model to obtain a model with the minimum reconstruction error on the training set as a reference model; and S5, testing the model and an abnormal judgment strategy to realize fault judgment. According to the method, the blade damage fault can be more comprehensively and accurately identified, the problem of fault sample scarcity is solved, and the method has wider applicability and higher reliability and is particularly suitable for health monitoring of aero-engines.
Need to check novelty before this filing date? Find Prior Art