DOUBLE HYBRIDIZED EMBRYONIC GENETIC ALGORITHM FOR OPTIMAL DELIVERY ROUTING
ADRIAN VĂLCU Technical University "Gh. Asachi" Iasi, Department of Engineering and Management, D. Mangeron 63, Iasi, Romania, e-mail: avilcu@misp.tuiasi.ro
This paper presents an efficient embryonic hybrid genetic algorithm for finding a minimum cost Hamiltonian circuit for a variable weight transportation problem. Prior work addresses the design of different heuristic methods, pure genetic algorithms or some hybridizations of genetic algorithm with greedy methods. The proposed algorithm includes a combination between a genetic algorithm, a branch and bound method and a dynamic-programming-inspired method. The basic components of the embryonic genetic algorithm are defined, the ways of using dynamic programming as a hypermutation operator and the realization of the fusion between the branch and bound approach and the basic genetic algorithm are described. The experimental results are compared to those given by other existing methods. The originality of this paper results from a combination of exact techniques - branch and bound method and dynamic programming and genetic algorithm. Except a better solving method for the delivery problem, this new solving technique has design implications on the creation of new methods for solving other difficult practical problems.
