Study Of Artificial Bee Colony Algorithm For Solving Constrained Optimization Problems

Artificial Bee Colony (ABC) algorithm was firstly proposed for unconstrained optimization problems and showed superior performance. This paper studies two nov-el ABC algorithms for constrained optimization problems.The first algorithm of this paper is Coevolutionary Artificial Bee Colony (CABC) algorithm. The constrained optimization are transformed to an unconstrained opti-mization problem by the augmented Lagrange penalty function. Then the variable vector and the Lagrange multiplier vector are optimized alternately by coevolution-ary computation. CABC algorithm is tested on 4 well-known test problems and the results are compared to Coevolutionary Particle Swarm Optimization (CPSO). The result shows that CABC algorithm is robust than CPSO algorithm.Another algorithm is modified Artificial Bee Colony (MABC) based feasibility rules. On the basis of an Artificial Bee Colony (ABC) algorithm solving constrained optimization problem based feasibility rules, four modifications related with boundary constraints, selective probabilities, the onlooker bee and the scout bee strategies are made to enhance its global search ability by making full use of the best and the worst of current generation. We test the new algorithm on 13 well-known benchmark prob-lems and the results are compared to the original algorithm. The experiment shows that the strategies modified can enhance the ability of global search. Three engineer-ing design problems are tested on MABC algorithm. The result shows that MABC algorithm is efficient for solving practical application problems too.