Nature-inspired computational model for solving bi-objective traveling salesman problems

Abstract

Bi-objective Traveling Salesman Problem (BTSP) is an NP-hard problem in the combinatorial optimization, which is also important question in the field of operations research and theoretical computer science. Genetic Algorithm (GA) is one type of efficient methods for solving NP-hard problems. However, GA-based algorithms suffer high time computational complexity, low stability and the premature convergence for solving BTSP. This paper proposes an improved method of genetic algorithm based on a novel nature-inspired computational model to solve these problems. The initialization of population of proposed algorithm is first optimized by the prior knowledge of Physarum-inspired computational model (PCM) in order to enhance the computational speed and stability. Then the hill climbing method (HC) is used to increase the diversity of the individuals and avoid falling into the local optimum. A series of experiments are conducted and results show that our proposed algorithm can achieve the better performance.