Finite-Dimensional Density Representation for Aerocapture Uncertainty Quantification

View Video Presentation: https://doi.org/10.2514/6.2021-0932.vid

Aerocapture is a problem of high interest for future interplanetary missions. It is dominated by a high sensitivity to a number of uncertainties, especially atmospheric density variability. The common state of practice for uncertainty quantification in this context is the Monte Carlo simulation methodology. A number of methods exist which may be superior to Monte Carlo methods but require a lower finite-dimensional representation of the random inputs. A Karhunen–Loève expansion approximation of perturbed atmospheric density is presented, which by reducing the dimensionality of the atmosphere model could enable the application of more advanced techniques such as polynomial chaos expansion. This representation is then demonstrated by implementation in a Monte Carlo simulation of entry dynamics for aerocapture. Early results of applying compressed sensing polynomial chaos expansion to this problem are also presented, and compared with the Monte Carlo baseline. Compressed sensing-based polynomial chaos expansion is shown to converge faster than Monte Carlo for this problem, and filtering out trajectories that impact the surface is shown to further improve convergence, but further validation is needed.