Comparison of Machine Learning Methods for Mission Generated Orbital Debris Estimation in Space Operations Planning
Open AccessThe lack of quantitative federal and international regulations on orbital debris production has caused exponential growth in the debris environment, resulting in an increased number of collisions with spacecraft and other pieces of debris. This praxis presents a single launch debris estimation model by comparing 18 machine learning models to predict the mass of orbital debris that the launch vehicle and associated payloads will contribute to the debris environment. The data set was obtained from the Federal Aviation Administration, the European Space Agency, and the Satellite Catalog maintained by the United States Combined Space Operations Center. 5,457 individual launches were identified with 10 associated features for each data point.This Praxis demonstrates how launch authorities and private industry can use machine learning methods to support prediction of mission generated orbital debris to assist with the development of detailed regulations and sustainability practices. The final model selection predicts, with under 4% margin of error, the mass of orbital debris that a single launch event and its associated payload will contribute to the orbital debris picture over the life cycle of all payloads. This research identifies the 124 independent features that are determined to have the most significant influence on the output variables. The Voting Regression model was selected to be the most effective method in predicting the mass of mission generated orbital debris from a single launch.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.