A New Approach to Adapting Control Parameters in Differential Evolution Algorithm
In Differential Evolution, control parameters play important roles in balancing the exploration and exploitation capability, and different control parameters are required for different types of problems. However, finding optimal control parameters for each problem is difficult and not realistic. Hence, we propose a method to adjust them adaptively in this paper. In our proposed method, whether or not the current control parameters will be adjusted is based on a probability that is adaptively calculated according to their previous performance. Besides, normal distribution with variable mean value and standard deviation is employed to generate new control parameters. Performance on a set of benchmark functions indicates that our proposed method converges fast and achieves competitive results.
KeywordsDifferential Evolution Adaptive Parameter Control Normal Distribution
Unable to display preview. Download preview PDF.
- 1.Storn, R., Price, K.: Differential evolution - a simple and efficient adaptive scheme for global optimization over continuous spaces. Internation Computer Science Institute, Berkley, Tech. Rep. (1995)Google Scholar
- 4.Liu, J., Lampinen, J.: A fuzzy adaptive differential evolution algorithm. In: TENCON 2002. Proceedings. 2002 IEEE Region 10 Conference on Computers, Communications, Control and Power Engineering, vol. 1, pp. 606–611 (October 2002)Google Scholar
- 5.Das, S., Konar, A., Chakraborty, U.K.: Two improved differential evolution schemes for faster global search. In: GECCO 2005: Proceedings of the 2005 conference on Genetic and evolutionary computation, pp. 991–998 (2005)Google Scholar
- 6.Storn, R.: On the usage of differential evolution for function optimization. In: Fuzzy Information Processing Society, 1996. NAFIPS. Biennial Conference of the North American, June 1996, pp. 519–523 (1996)Google Scholar
- 9.Liu, J., Lampinen, J.: On setting the control parameters of differential evolution method. In: Proc. 8th Int. Conf. Soft Computing, pp. 11–18 (2002)Google Scholar