Moving-Bank Multiple Model Adaptive Algorithms Applied to Flexible Spacecraft Control.
Abstract
Critical to the performace of the moving-bank multiple model adaptive estimator is the decision logic used to determine which elemental filters are implemented in the bank, and when to change this decision. The decision logics discussed focus on three situations: initial acquisition of the unknown parameter values through reducing bank discretization; tracking the unknown parameter values through bank movement; and reacquisition of the unknown parameters following a large jump change in their values through expanding bank discretization. Ambiguity function analysis is used to predict performance in these situations. The system to be controlled is a simplified model of a large scale space structure. Its equations of motion are developed and placed in state space form, the states being the positions and velocities of the rigid body mode and the second and fourth bending modes. The state space matrices describing the system are computed based on nominal values for all physical parameters with the exception of the mass density of the structure arms and their modulus of elasticity. These two parameters are allowed to vary in discrete steps establishing the parameter space. It is then attempted to control the states to the quiescent state, using moving-bank multiple model adaptive algorithms. The results indicate that, although the system under study did not have a great need for adaptive estimation and control the multiple model adaptive estimator performs essentially identically to a single filter artificially knowledgeable of the uncertain parameter values.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1985
- Accession Number
- ADA164016
Entities
People
- Paul G. Filios
Organizations
- Air Force Institute of Technology