Stochastic Adaptive Particle Beam Tracker Using Meer Filter Feedback

摘要

This research develops a realizable proportional-plus-integral (PI) feedback tracker to control a neutral particle beam. The design is based on detecting the photo electron events that are emitted from a laser excited particle beam and the observed events are used by a Meer filter to locate the beam's centerline. The observed events are modeled by a Poisson space time process and are composed of both signal and noise induced events. The Meer filter is a stochastic multiple model adaptive estimator which is composed of a bank of Snyder Fishman filters and is designed to distinguish the signal-induced events from the noise induced events. A target model is developed from a Gauss-Markov acceleration process, and the target states are estimated by a Kalman filter. The objectives of the research were to (1) select the best cost weighting matrices that minimize the RMS tracker error and enhance robustness, (2) simplify the Meer filter for easier on line usage, (3) complete full scale sensitivity and robustness analyses over all the Kalman and Meer filter parameters, and (4) develop on line adaptive estimation of those parameters that greatly affect stability robustness and tracker performance. A fifth objective is to identify the source of instability, and to propose a solution that will insure stability during parameter variations.