Presentation Details
  1. Home
  2. Presentation List
  3. Presentation Details
Gas Generator Blade Fracture Early Warning Method Based on BPIP-OT Feature Reconstruction
ID：100 View Protection：ATTENDEE Updated Time：2025-11-10 15:24:05 Hits：80 Poster Presentation

Start Time：Pending（Asia/Shanghai）

Duration：Pending

Session：No Session »

Presentation File

Tips: Only the registered participant can access the file. Please sign in first.

Abstract
Fracture failures in gas generator blades due to factors such as high-cycle fatigue, hot corrosion, rubbing, and foreign object damage occur periodically. Based on the mechanism of blade fracture fault in casing vibration signals, this study proposes a weak fault feature extraction and early warning methodology to support blade condition monitoring in turbine units. Firstly, based on blade fracture vibration response characteristics, the impact of minor rotational speed variations on blade passing frequency (BPF) amplitude extraction is thoroughly investigated. A Blade Passing Frequency Instantaneous Phase-Based Order Tracking (BPIP-OT) method is proposed, enabling accurate extraction of weak blade fault features. Secondly, utilizing optimized order features as input, a blade fracture early warning model is established through BPIP-OT-based variational autoencoder (VAE),where a comprehensive reconstruction error metric integrating mean square error (MSE) and the top M largest MSE values is designed. Finally, validation is performed via a field case study of blade fracture in an industrial gas turbine. Results demonstrate:(1)Spectra processed by the BPIP-OT algorithm effectively mitigate spectral smearing induced by minor speed variation, yielding concentrated energy distribution and enhanced amplitude accuracy at BPF.(2)Compared to traditional models using original spectra as input, the BPIP-OT-based VAE model proposed herein shows higher recognition accuracy.(3)The optimized dimensionality-reduced features reduce model parameterization and training complexity, demonstrating enhanced suitability for engineering applications in monitoring and early warning of gas generator blade fracture faults.
 
Keywords
Blade fracture fault, Blade passing frequency, Instantaneous phase, Variational autoencoder, Early warning, Gas generator
Speaker
hao gaoyan
Senior Engineer 北京博华信智科技股份有限公司

Submission Author
hao gaoyan 北京博华信智科技股份有限公司
Yingli Li China Petroleum Safety and Environmental Protection Technology Research Institute
FENG Kun Beijing University of Chemical Technology
Peng Zhang Beijing University of Chemical Technology
Submit Comment
Verify Code Change Another
All Comments
Important Date

  • Conference Date

    Nov 21

    2025

    to

    Nov 23

    2025

  • Oct 20 2025

    Draft paper submission deadline

  • Dec 08 2025

    Registration deadline

Sponsored By
IEEE Instrumentation and Measurement Society
South China University of Technology
Organized By
South China University of Technology
Contact Information
Login to view full contact details
Previous Conferences