Decision and behavior evolution in MAS based on neural network and swarm intelligence

This paper proposes a method using neural networks and swarm intelligence technology for the decision-making in the multi-agent systems (MAS). In this paper, a neural network is used for behavior decision controller. The inputs of the neural network are decided by the last actions of other agents. Then the outputs determine the next action that the agent will choose. The weight values are updated by particle swarm optimization algorithm, and they imply the behavior evolution of agents. The validity of the decision model is verified through simulation experiment, and the results show that this method has the ability of adaptive learning and can prevent the collision between agents to obtain the Pareto optimal.