In this paper the capabilities of two global search techniques to solve the Traveling Salesman Problem (TSP) are investigated. A modified genetic algorithm (GA) and simulated annealing neural network algorithm (NN) for the TSP are proposed. For the GA, a new type of crossover operator which is a combination of the nearest point insertion heuristic and order crossover developed by Oliver, is incorporated into the solution procedure. A mathematical procedure is developed to adjust the coefficient of the energy function in the proposed NN algorithm to improve its solution quality. The performance of these two algorithms are compared with those from traditional TSP heuristics. The results of the test problems indicate that the GA can provide the best solutions among the three algorithms considered in this study. Although the performance of the simulated annealing neural network can be improved by the incorporation of the coefficient adjustment procedure, its solution ability still have its limit. (A)
Samenvatting