Nonlinear smoothing for the continuous-discrete problem with application to rocket health monitoring
Eric T. Baumgartner, Bruce K. Walker · Guidance, Navigation and Control Conference · 1990
The nonlinear fixed-point and fixed-lag smoothing algorithms are presented for the problem of a continuous time dynamic system with discrete time measurements. The extension of nonlinear smoothing to this problem is accomplished using a formulation similar to the extended Kalman filter. The resulting nonlinear fixed-point and fixed-lag smoothers are then applied to a parameter estimation problem associated with the health monitoring of a space propulsion system, namely the Space Shuttle Main Engine. Since the nonlinear fixed-lag smoother is an unstable algorithm, an alternate formulation of the fixedpoint smoother is developed which results in a stable form for an on-line smoothing algorithm. The new smoothing algorithm is also applied to the health monitoring problem and the robustness of the new smoothing algorithm to unmodeled dynamics is investigated.