Title: Research on improved convergence co-evolution genetic algorithm for TSPs

Authors: Ming Shao

Addresses: School of Management, Shanghai University of Engineering Science, 333 Longteng Road, Shanghai 201600, China

Abstract: When the genetic algorithms are applied to solve travelling salesman problems, they are often faced with the problems of early convergence and being trapped in a local optimum. On the basis of co-evolution strategy, this paper proposes an improved genetic algorithm, which dynamically adjusts its fitness function and the probabilities of evolution operator in order to reduce the possible individual inbreeding degradation, and then effectively controls the evolutionary process. Compared with typical genetic algorithms, this improved algorithm can improve the speed of convergence and obtains better performance. Finally, this algorithm is verified on the TSP problem in 144 cities in China. The simulation results show that the improved genetic algorithm has better global searching performance and uses less convergence time.

Keywords: travelling salesman problem; TSP; genetic algorithms; convergence co-evolution; global search performance; China; convergence time.

DOI: 10.1504/IJCONVC.2014.063742

International Journal of Convergence Computing, 2014 Vol.1 No.2, pp.118 - 126

Received: 02 Jul 2013
Accepted: 24 Aug 2013

Published online: 30 Aug 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article