Abstract:
A search algorithm for accelerated optimization is designed which contains a linear part and a nonlinear responses with variable parameters of the linear part and the effect of random and also of monotone disturbances. For the design, at each search step the unknown parameters are represented as piecewise-linear functions and, in the case of monotone disturbances, as a polynomial function, with the measurements processed by a modified Prony method. The stability of the computing process is considered and simulation results given.