基于随机Petri网的大规模网络服务系统行为适配模型及方法

项目名称： 基于随机Petri网的大规模网络服务系统行为适配模型及方法

项目编号： No.61472004

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 章昭辉

作者单位： 东华大学

项目金额： 83万元

中文摘要： 在资源受限和异常扰动时，如何柔性适配系统行为，提升系统应变力，保持系统稳定的良好服务状态是大规模网络系统可用性技术和理论面临的严峻挑战。本项目围绕群体行为变异原子感知和负载动态均衡的行为自适应重构两个关键科学问题：提出基于随机Petri网行为势能平衡理论与方法，探讨系统受事件扰动下群体行为变异诱发机理；以行为原子性的观点研究群体行为变异的感知方法，构建基于原子行为势能的敏捷感知随机Petri网模型与算法；以群体行为势能与系统动态负载自适应均衡分配为核心，研究系统行为与用户群体行为自适应重构方法，构建系统行为适配的随机Petri网模型及算法；最后在交通信息服务系统中进行验证与应用。项目的研究，将为大规模网络服务系统行为适配提供理论和技术支持，丰富大规模网络服务系统高可用性以及Petri网的理论与方法，为面向民生的大规模网络服务系统提供变异行为敏捷检测、工作流性能适配等应用支撑。

中文关键词： Petri网；随机Petri网；系统建模；系统性能分析；工作流

英文摘要： It is a critical challenge of technologies and theories on usability of the large-scale network service systems that system behavior is adapted softly to keep the good system service state with the fixed resources capacity and disturbance. There are two key problems: one is atomic perception of group behavior variation and another is adaptive behavior reconstruction under load balance. There are four reserch tasks to solve above problems. Firstly, the theories and methods on stochastic Petri Nets of behavior potential equilibrium are presented to discuss the induced mechanism on group behavior variation. Secondly, it is to be built that models and algorithms of agile perception of group behavior variation based on stochastic Petri Nets of atomic behavior potential. Thirdly, considering adaptive equalization distribution of group behavior potential and dynamic service load, adaptive reconstruction of system behavior and user group behavior is to be researched ,and some stochastic Petri Net models and its methods of soft behavior adaption are to be built. Lastly, all new theories and methods are to be verified and applied in the Traffic Information Service System. All research results are valuable to technologies and theories on usability,and Petri Nets.The results can support agile variant behavior detecting and workflow performance adapting for the large-scale network service systems.

英文关键词： Petri Net;Stochastic Petri Net;System Modeling;System Performance Analysis;Workflow