Abstract
The inversion crossover in the Genetic Algorithm (GA) is useful for solving the twist of the gene and Hybrid GA combined with the inversion is effective for the TSP. When the inversion is used in a simple GA (sGA), it cannot show a high performance except the special case such as having a long size chromosome. Thus the inversion has been hardly used. Based on such a background, this paper tries to improve the convergence characteristics of an inversion. At first, we propose the inversion which the gene are exchanged according to the result of the gene evaluation. The next, we propose the multi-individuals inversion and then show that this method is effective to improve both of the rate and time of convergence. In the last, through the automatic fuzzy graph drawing, we show that the methods of the inversion proposed in this paper indicate higher performances in convergence characteristics than the usual inversion, Maximal Preservative Crossover (MPX) and Improved Edge Recombination Crossover (IERX).
