一种基于神经模糊系统的协商策略  

Negotiation Strategy Based on Neuro-fuzzy System

作　　者：彭志平[1] 李绍平[1] 

机构地区：[1]茂名学院计算机科学与技术系,茂名525000

出　　处：《系统仿真学报》2008年第3期623-626,共4页Journal of System Simulation

基　　金：广东省自然科学基金(06029281;05011905)

摘　　要：协商策略是自治协商的关键环节之一。针对现有协商策略受限于Nash理论的不足,提出一种由模糊逻辑与神经网络深度融合的神经模糊系统(NFS)的协商策略。分析了这两种计算智能技术的协同机制,引入了广义概率积与广义概率和模糊算子,给出了网络有关参数的学习式子。实验结果表明,NFS是有效的,具有较好的性能;基于NFS的协商策略符合人类协商的思维和常规,能获得更高的协商收益和协商效率。Negotiation strategy is the key technology to implement autonomous negotiation in multi-agent systems. Aiming at the drawback in most of the existing negotiation strategies due to being confined to Nash theory, a new negotiation strategy based on Neuro-Fuzzy System （NFS）, a cooperative entity of fuzzy logic and neural network, was put forward. At first, the cooperative mechanism of these two computational intelligence techniques was analyzed. Then the powerful fuzzy operator of generalized probability product and the one of generalized probability sum were introduced into NFS, and the formulas for training the parameters in NFS were given. The experimented results show NFS is effective and possesses the preferable performance. In addition, the negotiation strategy based on neuro-fuzzy system is consistent with the conventionality of human negotiation, and more incomes and higher efficiency can be achieved with the strategy.

关 键 词：模糊逻辑 神经网络 协同 协商策略 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]