Optimization of measurements in accelerated algorithms of extremal control

The measurements of the output of an inertial process to be optimized are distributed with constraints imposed on their number and the time interval between them. Solution of this problem minimizes the performance criterion for a proeoss controlled by a digital hill-climbing system. Examples are given in which the distribution is obtained in control of a first-order process by algorithms in which the discontinuity of the highest derivative is computed and the steady state value is forecast. The results are compared by the probability of the search bias with an optimal and a uniform sampling.