一种基于多智能体进化的广义图染色算法

T-Coloring Algorithm Based on Multiagent Evolution

基于对广义图染色问题的研究,提出了一种求解广义图染色问题的多智能体进化算法(multiagent evolutionary algorithm for T-coloring problem,简称MAEA-TCP),并将该算法应用到实际中的频率分配问题上,取得了良好的效果.该方法中每个智能体作为一个候选解被固定在智能体网格上,为了增加自身能量而与邻域当中的智能体展开竞争或者合作,同时智能体也可以利用自身的知识进行自学习来增加能量.根据广义图染色问题的特点,为智能体设计了3种算子:竞争算子、自学习算子和变异算子,以引导其进化,并用进化的方式来控制各算子,以协调智能体之间的相互作用.在实验中,分别使用大规模的随机图实例和费城实例来测试算法性能,同时给出参数测试结果和最佳取值区间.比较结果表明,该算法优于其他方法,具有良好的收敛性和实用价值.

Based on the study of T-coloring problem, multiagent systems and evolutionary algorithms are integrated to form a new algorithm, multiagent evolutionary algorithm for T-coloring problem （MAEA-TCP）. Then, this method is used to deal with the realistic frequency assignment problem, and has achieved encouraging results. In this algorithm, each agent is fixed on a lattice point of agent lattice as a possible solution. In order to increase energies, they compete or cooperate with their neighbors. They can also use knowledge to achieve their aims. Three evolutionary operators are designed for simulating the intelligent behaviors of agent, such as competition, self-learning and so on. The evolutionary operators are controlled through evolution, so that the populations can evolve. Experiments on large random graph instances and Philadelphia instances show that MAEA-TCP is a more encouraging algorithm than other methods.