Comparative results for a special class of robust nonlinear trackingalgorithms
Frank D. Gorecki, Michael J. Piehler · Journal of Guidance Control and Dynamics · 1989
This work presents the extension of two robust, adaptive estimators, both based on linear system theory, to the nonlinear case of an accelerating spacecraft in its ascent phase. This scenario is complicated by nonlinear system dynamics (due to the presence of the gravitational field and vehicle staging) and a nonlinear measurement model (consisting of tine-of-sight range, azimuth, and elevation measurements). The first algorithm in our presentation is an adaptive batch estimator (ABE) that utilizes the innovations process to estimate a piecewise constant acceleration approximation to the real-system dynamics. The second algorithm is an iterative least-squares estimator (ILSE) that models the system dynamics with the derivative of acceleration held constant. To account for gross modeling errors, such as discontinuities encountered in the target's acceleration profile due to its staging, a stochastic detector is employed in the later estimator that allows the algorithm to be restarted in the event of divergence. In order to provide a reasonable comparison for these two algorithms, both are implemented in a realistic tracking scenario in which their robustness is demonstrated. Although the simulation results demonstrate the feasibility of using either algorithm, the accuracy of the iterative least-squares algorithm is significantly better than that of the ABE.