Abstract
The multiobjective multiple traveling salesman problem (MmTSP), in which multiple salesmen and objectives are involved in a route, is known to be NP-hard. The MmTSP is more appropriate for real-life applications than the classical traveling salesman problem (TSP), however it has not received the same amount of attention. Due to the high complexity of the MmTSP, a promising algorithm for solving it must be based on a global search procedure. This paper proposes a hybrid global search algorithm, that belongs to the membrane computing framework. The search behavior of the algorithm is enhanced by a communication mechanism. The multichromosome representation is considered to reduce the excess runtime. The effectiveness of the proposed algorithm is tested on the MmTSP with disparately-scaled objective functions, salesmen and problem sizes. The experimental results are compared with the NSGA-II and several evolutionary strategies with decomposition, including simulated annealing algorithm, hill climbing method, pure evolutionary algorithm, and genetic algorithm.