Multi-Agent Cooperative Study Based On Fuzzy Theory

Multi-Agent cooperative studies the technology which a group of autonomous Agent accomplishes the complicated controlling or task solution by mutual and interactive intelligent behavior under open and dynamic environment. The multi-agent cooperative model based on contact net improves the efficiency of the whole net through communication between agents. Therefore, it is very important in the multi-agent cooperation research. But this model still has some limitation, such as the node agent is lack of the ability of processing fuzzy information. In fact, the node agent itself always needs the fuzzy description of heart-and-brain status. Only in this way it meets the thought of human. The cooperative relation and processing between agents are always calculated with experiential formula which is very subjective.For this limitation, the dissertation introduces fuzzy theory to study multi-agent cooperation deeply, and then analyses the contact net protocol, extends and improves the traditional contact net model from negotiation protocol, negotiation strategy and negotiation process, at last puts forwards a contact net model based on fuzzy theory. The new model can be applied into the systems for distributed medical treatment and malfunction diagnosis.The contact net improved model based on task acquaintance solves the problem that communication is blocked and negotiation efficiency is reduced owing to the increased node number in the original contract net model. But, this model cause new problems, such as system responsive time oscillating and "hungry die" of partial node agent. For this case the local optimal solution can not be found. The new model introduces the fuzzy and combined calculation to find potential cooperation relationship between agents, improves the negotiation process to let the information cooperation among the nodes communicate effectively. The local optimal solution can be achieved in some range, further more got the global optimal solution. The new model can solve the system oscillating problem effectively, choose suitable negotiation strategy dynamically. The contact net model can be more intelligent and adaptable for this case. The last part of paper test responsive time of two systems, one is heating malfunction diagnosis system based on new model, the other is heating malfunction diagnosis system based on contact net model of task acquaintance. The results of the experiment prove that the new model can solve the system oscillating problem effectively.