旅行家問題是眾所熟知的NP-hard問題，它無法在多項式的時間內被解決。然而，基因演算法是一個常見的啟發式演算法，對於旅行家問題而言，可以在合理的時間內得到近似最佳解；而且，其中的幾何特性屬於問題特徵知識，可以被利用來提高基因演算法搜尋最佳解的能力。譬如某些路徑線段距離太長而不會出現在最短的路徑上，這樣的資訊可以幫助基因演算法著重於較合適的路徑線段上，避免太過頻繁地考慮距離較長的路徑線段。所以，我們提出智慧型OPT混合式基因演算法(Intelligent-OPT Hybrid Genetic Algorithm (IOHGA))，以縮短執行的時間與改善子代的品質，根據上述的概念，賦予區域性最佳的路徑線段擁有較高的優先權，藉此路徑線段分成兩個獨立的候選集合與非候選集合，並依據路徑線段的優先權特性，我們設計了兩個基因演算法的運算子，為Skewed Production (SP) 與 Fine Subtour Crossover (FSC)；此外，我們修改了2-OPT區域搜尋演算法，稱為智慧型OPT演算法Intelligent OPT (IOPT)，並將之與傳統基因演算法結合。我們建構了模擬測試以評估IOHGA演算法的效能，實驗結果顯示IOHGA演算法優於混合式基因演算法及利用基因表示演算法改良的基因演算法，可以在較短的時間內得到較高精確度的近似最佳解。IOHGA演算法是一個相當有效率的演算法，如果在不強調必須找到最佳解的情況下，IOHGA演算法可以使用少量的時間，而提供高精確度的近似最佳解。因此，可將IOHGA演算法與叢聚演算法結合，運用於即時環境下的行動代理者規劃問題。

Traveling salesman problem (TSP) is a well-known NP-hard problem which can not be solved within a polynomial bounded computation time. However, genetic algorithm (GA) is a familiar heuristic algorithm to obtain near-optimal solutions within reasonable time for TSPs. In TSPs, the geometric properties are problem specific knowledge can be used to enhance GAs. Some tour segments (edges) of TSPs are fine while some maybe too long to appear in a short tour. Therefore, this information can help GAs to pay more attention to fine tour segments and without considering long tour segments as often. Consequently, we propose a new algorithm, called intelligent-OPT hybrid genetic algorithm (IOHGA), to exploit local optimal tour segments and enhance the searching process in order to reduce the execution time and improve the quality of the offspring. The local optimal tour segments are assigned higher priorities for the selection of tour segments to be appeared in a short tour. By this way, tour segments of a TSP are divided into two separate sets. One is a candidate set which contains the candidate fine tour segments and the other is a non-candidate set which contains non-candidate fine tour segments. According to the priorities of tour segments, we devise two genetic operators, the skewed production (SP) and the fine subtour crossover (FSC). Besides, we combine the traditional GA with 2-OPT local search algorithm but with some modifications. The modified 2-OPT is named the intelligent OPT (IOPT). Simulation study was conducted to evaluate the performance of the IOHGA. The experimental results indicate that generally the IOHGA could obtain near-optimal solutions with less time and higher accuracy than the hybrid genetic algorithm with simulated annealing algorithm and the genetic algorithm using the gene expression algorithm. Thus, the IOHGA is an effective algorithm for solving TSPs. If the case is not focused on the optimal solution, the IOHGA can provide good near-optimal solutions rapidly. Therefore, the IOHGA could be incorporated with some clustering algorithm and applied to mobile agent planning problems (MAP) in a real-time environment.

CHAPTER 1 INTRODUCTION 1
CHAPTER 2 LITERATURE REVIEW 4
2.1 THE GENE ALGORITHM 7
2.2 MUTATION OPERATORS 9
2.3 OPT LOCAL SEARCH ALGORITHM 13
CHAPTER 3 INTELLIGENT-OPT HYBRID GENETIC ALGORITHM 15
3.1 THE MINIMAL TOUR SEGMENT 24
3.2 THE SKEWED PRODUCTION FOR THE FIRST GENERATION 26
3.3 FINE SUBTOUR CROSSOVER 45
3.4 EXCHANGE MUTATION 55
3.5 GENETIC INASSIMILATION 59
3.6 INTELLIGENT OPT LOCAL SEARCH 63
CHAPTER 4 SIMULATION DESIGN 69
4.1 SMULATION PROCEDURE 70
CHAPTER 5 SIMULATION RESULTS 76
5.1 SIMULATION 1 - COMPARISON OF IOHGA WITH HGA(SA) 76
5.2 SIMULATION 2 - COMPARISON OF IOHGA WITH BURKOWSKI’S ALGORITHM 82
5.3 SIMULATION 3 - PERFORMANCE ANALYSIS OF SP 87
5.4 SIMULATION 4 - CONVERGENCE ANALYSIS ON POPULATION SIZE 92
5.5 SIMULATION 5 - PERFORMANCE ANALYSIS OF IOHGA 94
CHAPTER 6 CONCLUSION 98
REFERENCES 99

[1]Banzhaf, W., “The “Molecular” Traveling Salesman,” Biological Cybernetics, 64: 7–14, 1990.
[2]Back, T., “The interaction of mutation rate, selection, and self-adaptation within a genetic algorithm,” In R. Miinner and B. Manderick, editors, Parallel Problem Solving 2, Amsterdam, Elsevier, 1992.
[3]Burkoski, F. J., “Proximity and priority: applying a gene expression algorithm to the Traveling Salesperson Problem,” Parallel Computing, Vol. 30, pp. 803-816, May 2004.
[4]Croes, G. A. “A Method for Solving Traveling Salesman Problems,” Operations Research, Vol. 6, No. 6, pp. 791-812, 1958.
[5]David, E. G., “Genetic Algorithms in Search, Optimization, and Machine Learning,” Addison Wesley, pp. 1-88, 1989.
[6]Englert, M., Röglin, H., and Vöcking, B., “Worst Case and Probabilistic Analysis of the 2-Opt Algorithm for the TSP,” In Proceedings of the 18th Annual ACM-SIAM Symposium on Discrete Algorithms. SIAM, Philadelphia, PA, pp. 1295-1304, B. 2007.
[7]Fogel, D. B.,“An Evolutionary Approach to the Traveling Salesman Problem,” Biological Cybernetics, 60: 139–144, 1988.
[8]Fogel, D. B., “A Parallel Processing Approach to a Multiple Traveling Salesman Problem Using Evolutionary Programming,” In Canter, L. (ed.) Proceedings on the Fourth AnnualParallel Processing Symposium, pp. 318–326, Fullterton, CA, in 1990.
[9]Fogel, D. B., “Applying Evolutionary Programming to Selected Traveling Salesman Problems,” Cybernetics and Systems, 24: 27–36, 1993.
[10]Garey, M. R., Johnson, D. D., and Tarjan, R. E., “The Planar Hamiltonian Circuit Problem is NP-complete,” SIAM Journal of Computing, 5:704-714, 1976.
[11]Grefenstette, J. J., “Incorporating Problem Specific Knowledge into Genetic Algorithms,” in Genetic Algorithms and Simulated Annealing, (L. Davi, ed), pp. 42-60, Morgan Kaufmann Publishers, 1987.
[12]Holland, J. H., “Adaptation in Natural and Artificial Systems,” Ann Arbor, MI: Univ. Michigan Press, 1975.
[13]Julstrom, B. A., “Coding TSP Tours as Permutations via an Insertion Heuristic,” SAC ’99, Proc. 1999 ACM Sym. Appl. Comp., ACM Press, pp. 297-301, 1999.
[14]Jyh-Da Wei and Lee, D. T., “A New Approach to the Traveling Salesman Problem Using Genetic Algorithm with Priority Encoding,” In proc. 2004 IEEE Congress on Evolutionary Computation, pp. 1457-1464, 2004
[15]Kido T. “Hybrid Searches Using Genetic Algorithms,” In: Kitano H, editor, Genetic algorithm, Sangyo Tosho, pp. 61-88, 1993.
[16]Katayama, K., Sakamoto, H., and Narihisa, H., “An Efficiency of Hybrid Mutation Genetic Algorithm for Traveling Salesman Problem,” Proc 2nd Australia-Japan Workshop on Stochastic Models in Engineering, Technique & Management, Gold Coast, Australia, pp. 294-301, 1996.
[17]Katayama, K., Hirabayashi, H., and Narihisa, H., “Performance Analysis for Crossover Operators of Genetic Algorithm,” Systems and Computers in Japan, Vol. 30, No. 2, pp. 20-30, 1999.
[18]Katayama, K. and Narihisa, H., ”An Efficient Hybrid Genetic Algorithm for the Traveling Salesman Problem,” Electronics and Communications in Japan, Part 3,Vol. 84, No. 2, pp. 76-83, 2001.
[19]Lin, S. and Kernighan, B. W., “An Effective Heuristic Algorithm for the Traveling Salesman Problem”, Operations Research, Vol. 21, No. 2, pp. 498-516, 1973.
[20]Larrañaga, P., Kuijpers, C.M.H., Murga, R.H., Inza, I., and Dizdarevic, S., “Genetic Algorithms for the Travelling Salesman Problem: A Review of Representations and Operators”, Artificial Intelligence Review 13: 129–170, 1999.
[21]Michalewicz, Z., “Genetic Algorithms + Data Structures = Evolution Programs,” Berlin Heidelberg: Springer Verlag, 1992.
[22]Reinelt, G., “The Traveling Salesman: Computational Solutions for TSP Applications,” , Vol. 840 of Lecture Notes in Computer Science, pp. 64-99, Springer-Verlag, 1994.
[23]Reinelt G. http://www.iwr.uni-heidelberg.de/groups/comopt/software/TSPLIB9-
5/index.html
[24]Syswerda, G., “Schedule Optimization Using Genetic Algorithms,” In Davis, L. (ed.) Handbook of Genetic Algorithms, pp. 332–349, 1992, New York: Van Nostrand Reinhold.