Nonlinear Recursive Filter for Boost Trajectories

A nonlinear recursive algorithm is formulated for state vector and covariance estimation of boost trajectories. The thrust acceleration vector of the booster is modeled by a vector-differential equation that includes effects of propellant depletion and attitude motions resulting from gravity-turn maneuvers and other steering maneuvers. This new dynamics model is incorporated in an extended Kalman filter with nine state variables that describe the inertial components of position, velocity, and thrust acceleration. Additional algorithms are described for filter initialization using angle-only measurements from a geostationary sensor and for detection and estimation of the final staging event using measurement residuals. Tracking accuracy and covariance fidelity are assessed by Monte Carlo simulation.