Feasibility Analysis of Moving Bank Multiple Model Adaptive Estimation and Control Algorithms
Abstract
This investigation examines the feasibility of a moving bank multiple model adaptive estimation/control algorithm. Sliding bank multiple model adaptive estimation differs from conventional multiple adaptive estimation in that a substantially reduced number of elemental filters is required for the sliding bank estimator (9 elemental filters vs. 100 for the system modeled in this thesis). The positions in parameter space that the reduced number of elemental filters occupy are dynamically re-declared: i.e., the sliding bank of filters is moved about the parameter space in search of the true parameter point. Critical to the performance of the sliding bank estimator is the decision method that governs movement of the bank of elemental filters. Because of this, a number of different decision algorithms are discussed and their respective performance compared. Three controller designs are also examined: a single changeable-gain, a single fixed-gain, and a sliding bank multiple model adaptive controller. States of a damped second order system, with uncertain parameters (damping ratio and undamped natural frequency) are estimated by the sliding bank estimator and then regulated to the quiescent state by the controller. Performance of the sliding bank estimator/controller is compared to a benchmark of a single Kalman filter/LQ controller that has (artificial) knowledge of the true parameter values. Comparisons are based upon Monte Carlo analysis results. Originator-supplied keywords include: Adaptive control systems, Adaptive filters, Multiple model adaptive estimation, and Multiple model adaptive control.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1984
- Accession Number
- ADA152015
Entities
People
- K. P. Hentz
Organizations
- Air Force Institute of Technology