A Neural Network Prototype for Predicting F-14B Strains at the B.L. 10 Longeron

Abstract

A neural network prototype was developed to predict strain from data obtained from an F-14 flight test program. Data from two flights were available: Flight 400 consisted of standard structural maneuvers, and Flight 401 consisted of maneuvers performed during typical fleet operations. Several variables were monitored during flight, including Nz, Mach number, altitude, wingsweep angle, roll rate, angle of attack, a weight-on-wheels indicator and the strain at B.L. 10 of the F-14B. The neural network was trained on Flight 400, and tested on Flight 401. A forward-stepwise-regression was also performed on Flight 400 and the selected model was tested on Flight 401, for comparison. Results were evaluated by comparing the correlation coefficients between the predicted and measured strains. The correlation coefficient obtained by the neural network was 0.93 and by the regression equation was 0.94. Based on these preliminary results, the conclusion is made that the neural network approach offers a viable alternative to standard regression analysis for predicting strains on airframes.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 01, 1992
Accession Number
ADA255272

Entities

People

  • Margery E. Hoffman

Organizations

  • Naval Air Warfare Center Aircraft Divison

Tags

Communities of Interest

  • Air Platforms

DTIC Thesaurus Topics

  • Aircraft Equipment
  • Aircrafts
  • Airframes
  • Altitude
  • Computer Programming
  • Computer Programs
  • Computers
  • Information Science
  • Mach Number
  • Maneuvers
  • Models
  • Neural Networks
  • Prototypes
  • Regression Analysis
  • Standards
  • Structural Components
  • Vehicles

Readers

  • Aviation Science / Aeronautics.
  • Neural Network Machine Learning.
  • Regression Analysis.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Neural Networks