博弈模型在复杂系统连锁故障预防的应用  

Application of game model in cascading failure prevention of complex systems

作　　者：曹然[1,2] 金鸿章[1] 冯丽媛 

机构地区：[1]哈尔滨工程大学自动化学院,黑龙江哈尔滨150001 [2]东北农业大学电气与信息学院,黑龙江哈尔滨150030

出　　处：《电机与控制学报》2013年第5期93-97,共5页Electric Machines and Control

基　　金：国防科工局技术基础研究基金(Z192011B001);黑龙江省青年科学基金(QC2009C87)

摘　　要：针对故障树法和贝叶斯网络方法存在的复杂系统连锁故障非动态性分析的问题,结合博弈论和协同学中序参量的概念提出一种连锁故障预测方法。在系统众多状态参量中,序参量主宰着系统演化的方向。每个序参量都企图主宰系统的格局,这一过程恰恰体现了参与人之间的博弈行为。将子系统中的多个序参量映射为博弈中的参与人,在得益函数的定义中,还考虑到序参量之间除了竞争外也有合作,建立了一种基于协同学的非合作博弈模型。以火灾报警控制网络为应用实例,仿真结果表明接近临界阈值的序参量会导致故障子系统与其他子系统的关联作用增强,为复杂系统连锁性故障预防提供了新的途径。Fauh tree analysis and Bayesian networks cannot adequately capture dynamics of cascading failure in complex systems. Combining game theory and the concept of order parameters in synergetics, a novel cascading failure prediction method was presented. The evolution of the system was dominated by order parameters. The domination of order parameters was involved in a game behavior. Order parameters of subsystems were mapped to the players of the game. Cooperation between order parameters was also taken into account in the definition of payoff function except for competition. A non-cooperative game model based on synergetics was constructed. Example of fire alarm system was given. The results of simulation illustrate the order parameters closed to critical threshold make correlation between failure subsys- tems and other subsystems increase. The proposed method provides a new way to prevent cascading failures of complex systems.

关 键 词：复杂系统 连锁性故障 博弈论 协同学 序参量 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]