Multiple Model Parameter Adaptive Control for In-Flight Simulation

摘要

Adaptive control of aircraft model-following systems has shown promising results for in-flight simulation, but the computational expense and slow convergence of conventional parameter estimation techniques for higher order models inhibits their direct use for inflight simulation. Computer simulations of adaptive systems usually assume some knowledge of model parameters in order to maintain tracking fidelity at a reasonable computational cost as parameters change. This thesis incorporates a-priori information into a multiple-model estimation algorithm which assigns a probability weighting of each estimator within a bank of estimators. Final parameter estimates used in adaptive control are formed as a probabalistic weighted sum of individual estimates. Simulations of the system show excellent tracking performance throughout the flight envelope. A moving bank scheme for use over a wide range of flight conditions is recommended as a further area of study.