High Accuracy Trajectory Prediction Model

Introduction

Model_Screen_Shot.JPG

In the past VAST has used the Balloon Track and Near Space Ventures prediction software to conduct trajectory predictions to determine the best launch/landing site of the balloon. Recently the online software was not available the day prior to the launch, so the most up to date predictions were not available for planning. As a result of this inconvenience Austin Howard developed a simple trajectory model in Matlab the night before the launch and it turned out that the simple model was more accurate than the other prediction models because it used data from previous launches. This project aims to adapt the working trajectory model by adding additional features and converting it to Perl code for more flexibility in the use of the software.

In the figure the red line is the actual launch profile, the orange and yellow lines are Austin's predictions and the multi-colored line is the Near Space Venture Profile created 2 days before launch. The same input data was used for each prediction, however different ascent and descent rates were used.

People Involved

Goals

The goals of the project is to create software capable of:

  • Modeling the ascent and descent of a balloon and its payload
  • Accepting complex or non-linear ascent and descent rates (past launch profiles)
  • Generating Monte Carlo dispersions to map prediction uncertainty (uncertainty based on historic atmosphere model variation in local launch area)
  • Running real time during the flight and accepting data from the flight to improve prediction accuracy
  • Outputting trajectory results in multiple formats, including csv and kml
  • Interfacing with other ground station software and functions in software library

Timeline

  • Introduction 10/24/07
  • Goals 10/24/07
  • Rough draft model in Matlab 10/14/07
  • Model converted to Perl 11/15/07
  • Customized user interfaces added 11/10/07
  • Complete software and documentation posted 11/20/07

Suggestions

For the ascent rates, have you guys considered calculating the terminal velocity of the balloon with pressure drag equations and buoyancy and comparing that to actual balloon velocity profiles? I'm pretty sure that would model the ascent rate quite well, and we could come up with a good way to determine these rates on the ground before we let the balloon go… -Cliff