基于Agent的体系过程A-GERT网络“刺激-反应”学习模型  

A stimulus-response learning model for Agent-based system process A-GERT network

作　　者：方志耕[1,2] 夏悦馨 张靖如 熊仪 陈静邑 FANG Zhigeng;XIA Yuexin;ZHANG Jingru;XIONG Yi;CHEN Jingyi(College of Economics and Management,Nanjing University of Aeronautics and Astronautics,Nanjing 211100,China;Institute of Grey System,Nanjing University of Aeronautics and Astronautics,Nanjing 211100,China)

机构地区：[1]南京航空航天大学经济与管理学院,江苏南京211100 [2]南京航空航天大学灰色系统研究所,江苏南京211100

出　　处：《系统工程与电子技术》2022年第8期2540-2553,共14页Systems Engineering and Electronics

基　　金：国家自然科学基金面上项目(71671091);南京航空航天大学研究生创新基地(实验室)开放基金(kfjj20200908)资助课题。

摘　　要：针对图示评审技术(graphic evaluation and review technique, GERT)网络应用于体系活动决策过程中未考虑网络节点自身的学习能动性问题,基于智能体(Agent)技术,结合刺激-反应模型构建A-GERT网络模型。首先,根据体系活动的逻辑,搭建A-GERT网络框架,并通过提出网络传递效用函数,设计带有反馈机制的A-GERT网络。然后,利用刺激-反应模型的自适应优势构建学习方程,以GERT网络期望概率与期望时间等指标度量刺激强度,进一步拓展刺激-反应模型。最后,给出网络智能决策节点的学习步骤,并以创新技术开发体系活动进行算例研究,结果表明了所提方法的有效性与实用性。In view of the graphic evaluation and review technique(GERT) network applied in the decision-making process of system activities without considering the learning initiative of network nodes themselves, an A-GERT network model is developed based on Agent technology and stimulus-response model. Firstly, the framework of the A-GERT network is developed, and through the utility function of network transmission, the A-GERT network with feedback mechanism is designed. Then, we employ the adaptive advantages of the stimulus-response model to design the learning equation, and measure the stimulus intensity with the expected probability and expected time of the GERT network, so as to further expand the stimulus-response model. Finally, the learning steps of network intelligent decision nodes are given, and an example of innovation technology development system is given. The results show the effectiveness and practicability of the proposed method.

关 键 词：图示评审技术 智能体技术 复杂适应系统 刺激-反应模型 自适应学习 

分 类 号：N945[自然科学总论—系统科学]