An Improved Angle-Based Crossover Tabu Search for the Larger-Scale Traveling Salesman Problem

Abstract

An angle-based crossover tabu search (CTS) has been presented by author of this paper in. In this paper, an improved CTS is presented using the tabu-recorder, which record the tabu times of tabu-active attributes, to confirm new initial resolutions except the first cycle, and enlarge the global search, which is applied for the larger-scale traveling salesman problem (TSP). The key strategies of the improved CTS are intensification strategy, diversification strategy and tabu-recorder. The CTS implement the crossover operator of the genetic algorithm (GA) as the diversification strategy, and the strategy of selecting elite solutions as the intensification strategy. The tabu- recorder is another way of the diversification strategy, and used to confirm the initial routes with some constrains except in the first cycle. the improved CTS with the angel-based idea of sweep heuristic are used to solve twelve same TSP problems from VLSI TSP. The results showed the improved CTS have better performances than other algorithms.