| Abstract: | During the last decade various methods have been proposed to handle linear and non‐linear constraints by using genetic algorithms to solve problems of numerical optimization. The key to success lies in focusing the search space towards a feasible region where a global optimum is located. This study investigates an approach that adaptively shifts and shrinks the size of the search space to the feasible region; it uses two strategies for estimating a point of attraction. Several test cases demonstrate the ability of this approach to reach effectively and accurately the global optimum with a low resolution of the binary representation scheme and without additional computational efforts. Copyright © 2004 John Wiley & Sons, Ltd. |
