Dynamic control of adaptive parameters in evolutionary programming

Abstract

Evolutionary programming (EP) has been widely used in numerical optimization in recent years. The adaptive parameters, also named step size control, in EP play a significant role which controls the step size of the objective variables in the evolutionary process. However, the step size control may not work in some cases. They are frequently lost and then make the search stagnate early. Applying the lower bound can maintain the step size in a work range, but it also constrains the objective variables from being further explored. In this paper, an adap­tively adjusted lower bound is proposed which supports better fine-tune searches and spreads out exploration as well.