Stochastic optimization framework for spacecraft maneuver detection

Abstract

The main focus of the paper is the detection and characterization of unobserved spacecraft maneuvers, by considering a stochastic optimal control approach. Specifically, the propagation of uncertainty associated with state estimation errors for a spacecraft is investigated using a Monte Carlo type nonlinear simulation framework. Gaussian statistics were assumed for the initial spacecraft states and possible maneuvers. The nonlinear simulation framework helps identifying the classes of maneuvers that steers the uncertainties associated with state estimation from an initial probability distribution to a final probability distribution. The initial state sample points are generated based on conjugate unscented transformation, capturing the initial statistics of the state estimates. A linearized optimization framework is also developed for rapid computation of the stochastic optimal control metric that measures the maneuverability of the uncertainties between given boundary probability distributions. Numerical simulation results comparing the linear and nonlinear framework are provided.