Authors: Harish Sharma; Jagdish Chand Bansal; K.V. Arya
Addresses: ABV-Indian Institute of Information Technology and Management Gwalior, Morena Link Road, Gwalior, Madhya Pradesh-474015, India ' South Asian University, Akbar Bhawan, Chanakyapuri, New Delhi-110021, India ' ABV-Indian Institute of Information Technology and Management Gwalior, Morena Link Road, Gwalior, Madhya Pradesh-474015, India
Abstract: Differential evolution (DE), like other probabilistic optimisation algorithms, sometimes exhibits premature convergence and stagnation. It is analysed by researchers that the DE is better in exploration of the search space compared to exploitation. In the solution search process of DE, there is enough chance to skip the true solution due to large step size. In order to balance the exploration and exploitation capability of the DE, a power law-based local search strategy is proposed and integrated with DE. In the proposed strategy, new solutions are generated around the best solution and it helps to enhance the exploitation capability of DE. The experiments on 14 un-biased test problems of different complexities show that the proposed strategy outperforms the basic DE and recent variants of DE namely, self-adaptive DE (SaDE) and scale factor local search DE (SFLSDE) in most of the experiments.
Keywords: numerical optimisation; evolutionary computation; memetic algorithms; power law; local search; PLLS; differential evolution; exploration; exploitation.
International Journal of Computational Intelligence Studies, 2013 Vol.2 No.2, pp.90 - 112
Available online: 23 Jul 2013 *Full-text access for editors Access for subscribers Purchase this article Comment on this article