Iterative learning identification with an application to aerodynamic drag coefficient curve extraction problem

Abstract

The iterative learning identification method is proposed for effectively extracting projectile’s optimal fitting drag coefficient curve C _df from radar measured velocity data. The performance index is in a minimax sense and the initial curve can be arbitrarily chosen. A general convergence analysis is given. A guideline for choosing learning parameters, varying w.r.t. both time t and iteration number i, is suggested. A bi-linear scheme is proposed for an even better convergence speed. Its related identification results are also given. From the results of this chapter, it is clearly demonstrated that the iterative learning idea is very effective in identifying a nonlinear function in a nonlinear dynamic system.