Multi-algorithm co-evolution strategy for Dynamic Multi-Objective TSP

Dynamic Multi-Objective TSP (DMOTSP), a new research filed of evolutionary computation, is an NP-hard problem which comes from the applications of mobile computing and mobile communications. Because the characters of DMOTSP change with time, the method of designing a single algorithm can not effectively solve this extremely complicated and diverse optimization problem according to NFLTs for optimization. In this paper, a new approach to designing algorithm, mufti-algorithm co-evolution strategy (MACS), for DMOTSP is proposed Through multi-algorithm co-evolution, MACS can accelerate algorithmpsilas convergence, make Pareto set maintain diversity and make Pareto front distribute evenly with a complementary performance of these algorithms and avoiding the limitations of a single algorithm. In experiment, taking the three-dimensional benchmark problem CHN144+5 with two-objective for example, the results show that MACS can solve DMOTSP effectively with faster convergence, better diversity of Pareto set and more even distribution of Pareto front than single algorithm.

[1]  Keld Helsgaun,et al.  An effective implementation of the Lin-Kernighan traveling salesman heuristic , 2000, Eur. J. Oper. Res..

[2]  Chen Yu-ping,et al.  A gene-pool based genetic algorithm for TSP , 2008, Wuhan University Journal of Natural Sciences.

[3]  Brian W. Kernighan,et al.  An Effective Heuristic Algorithm for the Traveling-Salesman Problem , 1973, Oper. Res..

[4]  Zbigniew Michalewicz,et al.  Inver-over Operator for the TSP , 1998, PPSN.

[5]  Richard Durbin,et al.  An analogue approach to the travelling salesman problem using an elastic net method , 1987, Nature.

[6]  Leen Stougie,et al.  Algorithms for the On-Line Travelling Salesman1 , 2001, Algorithmica.

[7]  Aimin Zhou,et al.  Benchmarking algorithms for dynamic travelling salesman problems , 2004, Proceedings of the 2004 Congress on Evolutionary Computation (IEEE Cat. No.04TH8753).

[8]  B. Freisleben,et al.  Genetic local search for the TSP: new results , 1997, Proceedings of 1997 IEEE International Conference on Evolutionary Computation (ICEC '97).

[9]  Hartmut Schmeck,et al.  An Ant Colony Optimization approach to dynamic TSP , 2001 .

[10]  David H. Wolpert,et al.  No free lunch theorems for optimization , 1997, IEEE Trans. Evol. Comput..

[11]  Martin Middendorf,et al.  Pheromone Modification Strategies for Ant Algorithms Applied to Dynamic TSP , 2001, EvoWorkshops.

[12]  José Ignacio Hidalgo,et al.  A hybrid heuristic for the traveling salesman problem , 2001, IEEE Trans. Evol. Comput..

[13]  R. Jonker,et al.  The symmetric traveling salesman problem and edge exchanges in minimal 1-trees , 1983 .