Efficiency conditions for hill-climbing control of continuous industrial processes

An industrial process whose mathematical model may be reduced to a sequential connection of a linear dynamic and a quadratic static elements is to be controlled by a two-level hill-climbing system. Under certain conditions the hill climbing is found to be less efficient than stabilization of an a priori optimum even with complete prior data on the process and random perturbations. This condition is formulated as a relationship between the process parameters, perturbations, and measurement noise.