Analysis On A Self-organizing Overall-regional Competition Network And Application To The TSPs

As the application of the traveling salesman problem become more and more popular, any the methods which can be used to optimize the TSP, will be received the strong attention. The paper carries out the detailed analysis and a lot of experiments to the improved algorithm of SOM based overall-regional competition for the TSP. We analyze the parameters which can affect the performance of the ORC-SOM algorithm and get the value which suit the algorithm. And we also study the computation complexity and the space complexity of the ORC-SOM algorithm, and find that the algorithm has the advantage of the computation complexity and the space complexity. By comparison with six typical SOM-based TSP algorithms, our experimental results on structure different TSP data and a set of 16 TSP instances indicate that the proposed algorithm is superior to those TSP algorithms, not only on the total tour length of the TSP, but also on the stability of the algorithm.