Accelerated biogeography-based optimization with neighborhood search for optimization

作者：Lohokare M R^*; Pattnaik S S; Panigrahi B K; Das Sanjoy

来源：Applied Soft Computing, 2013, 13(5): 2318-2342.

DOI：10.1016/j.asoc.2013.01.020

• Biogeography-based optimization (BBO)
• Memetic algorithm (MA)
• Evolutionary optimization
• Rectangular microstrip patch antennas (MSAs)
• Convergence
• Neighborhood search

Biogeography-based optimization (BBO) inherently lacks exploration capability that leads to slow convergence. To address this limitation, authors present a memetic algorithm (MA) named as aBBOmDE, which is a new variant of BBO. In aBBOmDE, the performance of BBO is accelerated with the help of a modified mutation and clear duplicate operators. Then modified DE (mDE) is embedded as a neighborhood search operator to improve the fitness from a predefined threshold. mDE is used with mutation operator DE/best/1/bin to explore the search near the best solution. The length of local search is a choice that balances between the search capability and the computational cost. In aBBOmDE, migration mechanism is kept same as that of BBO in order to maintain its exploitation ability. Modified operators are utilized to enhance the exploration ability while a neighborhood search operator further enhances the search capability of the algorithm. This combination significantly improves the convergence characteristics of the original algorithm. Extensive experiments have been carried out on forty benchmark functions to show the effectiveness of the proposed algorithm. The results have been compared with original BBO, DE, CMAES, other MA and DE/BBO, a hybrid version of DE and BBO. aBBOmDE is also applied to compute patch dimensions of rectangular microstrip patch antennas (MSAs) with various substrate thicknesses so as to be used a CAD formula for antenna design.

• 出版日期2013-5