Research On Multiagent Self-adaptivity Negotiation Strategy Based On Correlate Issues Finding

Production-marketing collaboration is the important point in Multi-agent manufacture supply chain collaboration, however, during the process of productionmarketing, manufacturer and retailer always have production-marketing conflicts because of the difference in information, goal and so on. If these conflicts can’t be solved as soon as possible, they will reduce the efficiency of supply chain running to a great extent. Negotiation is an effective method to solve the conflict of supply chain production-marketing conflicts. In consideration of the correlation between issues is suitable to the realist negotiation environment, the research on how to learn the opponents’ s information and come out self-adaptation negotiation strategy are becoming the important question of resolving multi-agent supply chain productionmarketing collaboration conflicts.This paper mainly research on self-adaptation negotiation strategy based on issues grouping.To solve the correlation of different issues in intelligent negotiation, this paper divide the issues into different issue group based on multi-dimensional stereoscopic volume theory and modified DAI model. Firstly, both agents calculate the correlation degree between issues and then form the issue group based on multi-dimensional stereoscopic volume theory. Manufacture agent and retail agent submit their issue groups to coordinate agent, then coordinate agent finally integrate issue groups based on modified DAI model.To solve the poor self-adaptation and low efficiency of traditional negotiation, in the premise of maxmizing the overall efficiency and shortening the negotiation time, this paper design a two-stage bilateral multi-issues negotiation model. In the first stage, named the whole negotiation pushing stage, in order to reach an agreement on more issues, both angets learn the concession degree of opponents, and make concession degree insistently, Therefore, this paper designed a opponents’ s concession degree learning model based on Adaboost-SVM, and proposed the self-adaptation negotiation strategy based on opponents’ s concession degree. In the second stage, named the whole negotiation success stage, in order to reach an agreement on the issues which are not solved on the first stage, both agents learn the proposal value of opponents. Therefore, this paper designed a opponents’ s proposal value learning model based on Qreinforcement and RBF neural network, and proposed the self-adaptation negotiation strategy based on opponents’ s proposal value.At last, in order to evaluate the effectiveness of issues-grouping algorithm and self-adaptation negotiation strategy, go on a simulation based on Matlab platform. Designthe product order model between manufacturer agent and retailer agent, build the negotiation model, assign the variables and parameter, respectively experiment on issues-grouping and multi-agent self-adaptation negotiation strategy, at last, analyzed the efficiency brought by correlation issues grouping algorithm and the self-adaptation strategy in this paper.