Multi-agent Cooperative Control And Simulation

With the further research of multi-agent system, multi-agent cooperative control has become a hot topic in the field of control, It has aroused great concern of researchers in the field of disaster search and rescue, urban intelligent transportation, unmanned aerial vehicle cluster battlefield reconnaissance and attack?Because of the complex internal structure of multi-agent system, simulation has become an important method to study its internal characteristics and variation law, which is not only easy to predict but also reduce the cost and risk of physical simulation.This paper designs a multi-agent simulation system based on JADE and World Wind Java. The simulation system,with the background of multi unmanned vehicle cooperative, realizes the multi-node cooperation with the interconnection of local area networks. In this paper, the model of Agent is set up for the object, environment, task, and so on. By using the multi-index task allocation function, the Agent is evaluated and the optimal unmanned vehicle is selected to obtain more accurate target detection information. System has designed the communication and cooperation between Agent and roaming, zoom, and the perspective of the 3D platform. Through simulation experiment, the whole process simulation of unmanned vehicle cooperative task is realized. The main work done is as follows:(1) The simulation system is designed into three levels, including the control center agent, the management agent and the execute agent. Control center agent is responsible for the operation and management of the whole system, with the highest authority of the system; The management agent is responsible for the implementation of the task of the region node; The execute-agent is responsible for the implementation of specific detection tasks.(2) An improved contract net protocol assignment algorithm is proposed. In view of the problems existing in the classical contract net protocol, this paper introduces the distance, speed, revenue and other factors, and evaluate the contract with the particle swarm algorithm. So as to improve the efficiency of task allocation in multi-agent system.(3) Taking into account in the simulation process, the need to assign a number of tasks to some management Agent nodes, may cause some nodes load is too heavy. In this paper, we design and implement a multi-agent load balancing method. The method is based on genetic algorithm, which can detect the load status of each node in real time and adjust it.(4) The system is implemented based on JADE and World Wind Java, which verifies the effectiveness of the proposed method.Finally, the paper summarizes the full text, and points out that is worth further research.