Enhanced Artificial Bee Colony Algorithm with SPV for Travelling Salesman Problem

Artificial Bee Colony (ABC) Algorithm is an optimization algorithm used to find out the global optima. In ABC, each bee stores the information of feasible solution or candidate solution and stochastically modifies this over time, based on the information provided by neighboring bees, it speculative modifies over time and based on the best solution found by the bee itself.. In this proposed work, enhanced ABC algorithm with SPV for travelling salesman problem is used. In this modified bee colony algorithm, additional phase in the form of mutation is used and SPV rule is used in this work for improving local search. After modification, proposed algorithm is implemented on standard travelling salesman problem for checking the efficiency of proposed work. The experimental results are compared with ABC algorithm and ABC with SPV algorithm.

I. Introduction

The problem of optimization is one of the most crucial problems now a day and lots of work have been done previously to solve these problems. There are many types of optimization algorithm like GA, ABC, PSO, ACO[4], etc. and lots of works in these algorithms are available in the literature. In paper [1], Dervis Karaboga defined an algorithm, called Artificial Bee Colony (ABC) for solving numerical optimization problem. This algorithm is based on the foraging behavior of honey bees. Different versions of ABC algorithm have been proposed like Artificial Bee Colony Algorithm with Mutation [9], ABC with crossover, ABC with SPV [10], ABC with crossover and mutation [11] and many different versions have been proposed. In paper [2], dynamically division of bees into subgroup and the searching process is performed on these generated subgroups. In paper [3], Karaboga defines the modified algorithm for solving real parameter optimization problems and comparative study on different evolutionary algorithms like particle swarm optimization (PSO), ABC algorithm, Differential evolution (DE), Ant colony algorithm (ACO) etc and performance derived on different benchmark functions, are discussed on paper [5], [7]. For improving the exploitation, karaboga proposed a modified ABC algorithm based on global best solution in paper [6]. D. Haijun and F. Qingxian proposed a enhanced artificial bee colony algorithm for function optimization in paper [8].