Flight Regime Recognition Analysis for the Army UH-60A IMDS Usage

Abstract

Usage Monitoring requires accurate regime recognition. For each regime, there is a usage assigned for each component. For example, the damage accumulated at a component is higher if the aircraft is undergoing a high G maneuver than in level flight. The objective of this research is to establish regime recognition models using classification algorithms. The data used in the analysis are the parametric data collected by the onboard system and the actual data, consisting of the correct regime collected from the flight cards. This study uses Rpart (with a tree output) and C5.0 (with a ruleset output) to establish two different models. Before model fitting, the data was divided into smaller datasets that represent regime families by subsetting using important flight parameters. Nonnormal tolerance intervals are constructed on the uninteresting values; then these values in the interval are set to zero to be muted (e.g. excluded). These processes help reduce the effect of noise on classification. The final models had correct classification rates over 95%. The number of bad misclassifications were minimized (e.g. the number of bad misclassifications of a level flight regime as a hover regime was minimized), but the models were not as powerful in classifying the low-speed regimes as in classifying high-speed regimes.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 2006
Accession Number
ADA462429

Entities

People

  • Ahmet M. Dere

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Air Platforms
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Accuracy
  • Aircraft Equipment
  • Aircrafts
  • Algorithms
  • Big Data
  • Classification
  • Data Mining
  • Data Sets
  • Flight
  • Information Science
  • Level Flight
  • Neural Networks
  • Operations Research
  • Recognition
  • Rotary Wing Aircraft
  • Test Sets
  • Visual Inspection

Readers

  • Aviation Science / Aeronautics.
  • Regression Analysis.