Dual Control and Identification Methods for Avionic Systems. Part 1. Dual Control

Abstract : The report constitutes Part I of a one-year study on the dual control method for nonlinear systems. The difficulties associated with the general stochastic control problem are discussed, and a new approach to the problem is developed using the concepts of dual control and the principle of optimality from dynamic programming. An adaptive dual control strategy is derived which has the characteristic of actively regulating learning which achieves the control objective. The theoretical development is specialized to a class of problems of controlling a time-varying linear system with random parameters. A specific algorithm is derived for this class of problems, and example problems are presented to demonstrate the computational feasibility of the new algorithm, the performance level of the new algorithm, and to give insight into the present theory. Finally, several areas of applications for the present results are outlined, and abstracts of technical papers supported by this contract are given.