Re-entry filtering, prediction, and smoothing.

The statistical estimation of trajectories of maneuverable lifting re-entry vehicles is studied via the minimum-variance estimation technique applied hitherto to satellites and space vehicles. The re-entry application differs appreciably from the orbit estimation problem since the dominant forces on the vehicle are aerodynamic rather than gravitational. The problems associated with applying linear filter theory to the re-entry problem are discussed, and means of circumventing these problems are recommended. Measurements considered include ground-based tracking of range, azimuth, and elevation; arid onboard measurement of linear accelerations and inertial body rates. Numerical results are presented which demonstrate the characteristics of re-entry filtering. Both firstand second-order statistics are presented. The mathematical model includes the effects of earth oblateriess and rotation on the threedimensional particle motion of the vehicle. The data demonstrate the effects of using a single reference trajectory, the advantages arid disadvantages of updating a nominal reference trajectory, and the timewise convergence of the estimated trajectory during single-pass processing. The iterative convergence of multiple passes through the data is demonstrated also. Finally, the results show the effects of data rate, data noise magnitude, and data type (tracking vs accelerometer) on the resulting estimate and its covariance.