Adaptive neighbourhood for locally and globally tuned biogeography based optimization algorithm

Abstract

Biogeography Based Optimization (BBO) is a population based metaheuristic algorithm using the idea of migration and mutation operation of species for solving complex optimization problems. BBO has demonstrated good performance on various unconstrained and constrained benchmark functions. It has also been applied to real world optimization problems of type linear or nonlinear, nominal or ordinal as well as mixed variables. But, it is realized that adaptation of the intensification and diversification for solving complex optimization problems are challenging tasks. To cope with these challenges, we develop a novel migration model for BBO which inherits features of the nearest neighbour of the local best individual to be migrated along with a global best individual of the pool. Furthermore to select the local best individual for the habitat to be migrated an adaptive local topological structure has been used. We name it as “Adaptive Neighbourhood for Locally and Globally Tuned Biogeography Based Optimization algorithm (ANLGBBO)”. This maintains the balance between intensification and diversification i.e., improve solution by exploiting the accumulated search space and exploring the large space by identifying regions with high quality solutions. We have carried out an extensive numerical evaluation and comparisons for experimental tests using twenty benchmark functions with different features to measure the efficiency of the algorithm. The experimental study confirms ANLGBBO draws clear line of other variants of BBO algorithms in terms of population diversity and establish the accuracy of global optimal solution.