The Diagnostics and the Fault Detection of Gas Turbine Engines by Using Different Neural Networks

Abstract

This paper deals with the results of a research carried out for the evaluation of Neural Networks for engine diagnostics. The study continues the works made in the past on the same subject. Now the main aim is to find new methods for improving the effectiveness of Neural Nets. The results presented here concern with two different Neural Nets: the Back Propagation Neural Networks (BPNN) and the Adaptive Resonance Theory Neural Networks (ART1-2). As regarding BPNN particular attention is paid to the improvement of training time and their robustness. The study of ART1-2 considers the use and the insertion of the probability of fault happening in the training patterns. The paper shows in full details all activities carried out as well as all improvements with respect to past results.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 10, 1999
Accession Number
ADA373338

Entities

People

  • Giovanni Torella
  • Roberto Torella

Organizations

  • American Institute of Aeronautics and Astronautics

Tags

Communities of Interest

  • Air Platforms

DTIC Thesaurus Topics

  • Coding
  • Computer Programs
  • Computers
  • Engines
  • Exhaust Gases
  • Flow Rate
  • Gas Turbines
  • Gases
  • High Pressure
  • Mass Flow
  • Neural Networks
  • Personal Computers
  • Simulations
  • Test And Evaluation
  • Training
  • Turbines
  • Turbofan Engines

Fields of Study

  • Engineering

Readers

  • Neural Network Machine Learning.
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks