Intermediate population based differential evolution algorithm

Abstract

In the present paper propose two novel variants of Differential Evolution (DE), named IP-OBL and IP-NSDE, have been proposed. In IP-OBL the initial population is generated by using the intermediate positions between the uniformly generated random numbers and opposition based numbers. While in case of IP-NSDE, the initial population is generated as an intermediate of uniform random numbers and numbers generated by Nelder Mead Method. The proposed algorithms are further modified by selecting best NP/2 individuals to perform in population evolution. The modified variants are termed as MIP-OBL and MIP-NSDE. The numerical results of 10 benchmark problems indicate the competence of the proposed algorithms. © 2011 Springer-Verlag.