Flight Data Processor

This is a python library to process and analyze flight data (e.g. from decoded ADS-B messages). Following functions and algorithms are implemented:

• Extract continuous full or partial flight path data
  • Unsupervised Machine Learning, Clustering using DBSCAN
• Smoothing, filtering, and interpolating flight data
  • Spline filtering
  • Weighted average filtering
  • Time-based weighted average filtering
• Segmenting flight into different phases:
  • using Fuzzy Logic with data interpolation methods
  • supporting phases: ground, climb, descend, cruise, and level flight

Paper and citation

The source code of this repository complements the following publication:

https://arc.aiaa.org/doi/10.2514/1.I010520

If you use the code for your research, please cite:

@article{sun2017flight,
  title={Flight Extraction and Phase Identification for Large Automatic Dependent Surveillance--Broadcast Datasets},
  author={Sun, Junzi and Ellerbroek, Joost and Hoekstra, Jacco},
  journal={Journal of Aerospace Information Systems},
  pages={1--6},
  year={2017},
  publisher={American Institute of Aeronautics and Astronautics}
}

Required software

• Python 3.x
• MongoDB 3
• Dependent Python libraries
  • scipy
  • scikit-learn
  • skfuzzy
  • pymongo

Code examples

1. Flight clustering

1. install MongoDB

2. extract flight from ADS-B positions

   $ python flightextract.py --csv data/sample_adsb_decoded.csv --db test_db --coll flights
   

2. Fuzzy segmentation

You can use previously created collection in MongoDB. Or, using provided pickled data, run:

$ python test_phases.py

The essential code to identify the flight phases is:

import flightphase
flightphase.fuzzylabels(times, alts, spds, rocs)

3. View flights

Use the same previously created MongoDB collection:

$ python flightview.py --db test_db --coll flights

Screen shots

example flight phase identification

example fuzzy logic membership functions

example flight viewer