抄録
An asymmetric traveling salesman problem is one in which the costs for travel between one city and another are not symmetric. In this paper, we propose a method for solving such problems based on chaotic search method. It uses block-city exchange method, block insertion method, and 2-opt exchange as exchanging methods. In the block-city exchange method, we consider several cities as a block, and the whole block is exchanged with other city. In the block insertion method, we consider several cities as a block, and the block is inserted at the specified location. In the proposed method, a switching between two exchange methods is determined by chaotic neurodynamics. In the proposed method, the block-city exchange method uses one of exchange method for any instance. The other exchange method is selected from the 2-opt exchange and block insertion methods according to the standard deviations of the costs city-wise and that of all branches in the instance. The proposed method obtains better or equivalent solutions with the conventional heuristic methods for asymmetric TSPs.
