A Sequence Based Genetic Algorithm with Local Search for the Travelling Salesman Problem

The standard Genetic Algorithm often suffers from slow convergence for solving combinatorial optimization problems. In this study, we present a sequence based genetic algorithm (SBGA) for the symmetric travelling salesman problem (TSP). In our proposed method, a set of sequences are extracted from the best individuals, which are used to guide the search of SBGA. Additionally, some procedures are applied to maintain the diversity by breaking the selected sequences into sub tours if the best individual of the population does not improve. SBGA is compared with the inver-over operator, a state-of-the-art algorithm for the TSP, on a set of benchmark TSPs. Experimental results show that the convergence speed of SBGA is very promising and much faster than that of the inverover algorithm and that SBGA achieves a similar solution quality on all test TSPs.