Hybrid Gaussian Mixture Splitting Techniques for Uncertainty Propagation in Nonlinear Dynamics

What is it about?

This paper presents an improved methodology for long-term uncertainty propagation in nonlinear dynamics by combining techniques of initial multi-directional splitting and adaptive entropy-based splitting during the propagation within the Gaussian mixture model framework. As a contribution of this paper, accurate uncertainty distribution is captured with a lower computational load and improved stability compared with entropy-based splitting techniques for highly deformed and elongated uncertainty space.

Featured Image