Going Nonlinear: Towards Automated Storm Chasing
George S. Young
Department of Meteorology
The Pennsylvania State University
Severe storm chasing via automated aeronautical vehicle (UAV) poses a considerable
challenge, not least of which is command and control. Storm interception
becomes more feasible and less labor intensive if the UAV involved is also
autonomous, i.e. able to make its own routing decisions. This interception
problem is nonlinear and can be solved either analytically or by iterative
optimization. In contrast, the storm mapping problem has no analytic solution.
Thus, while a neural network approach can be applied to both problems, the
first can be handled via supervised learning while the second requires unsupervised
learning. Experiments with the storm interception problem demonstrate that
neural network performance nears 100% only when the problem is geometrically
transformed to remove a discontinuity in the function relating UAV course
to intercept success. Training a neural net to handle the post-intercept
storm mapping problem requires that the traditional back propagation approach
be replaced by a method that does not require a priori knowledge of the
optimal solution. One way to achieve this unsupervised training is via a
genetic algorithm.