A fast bee colony optimization for traveling salesman problem

This paper presents a modified bee colony optimization (BCO) by pattern reduction to reduce the computation time, called BCOPR. Although BCO was robustness optimization, but likes the other algorithm for solving optimization problem, BCO has many reduncation computations on its convergence process, as a consequence, it will more computation time. Two operators are developed to BCOPR in this paper. The first one BCOPR1, is used to cut down the computation time by avoiding performing the same process on the preferred edge. On the second operator, BCOPR2, a bee is possible to duplicate the best previous-iteration tour if her first stage tour is the same with part of the best previous-iteration tour. In addition, likes BCO original, BCOPR is also use local search to enhance the quality solution. To evaluate the performance of the proposed algorithm, BCOPR uses some various benchmarks of TSP. Our experimental results show BCOPR reduce computation time as well as achieve good solution.