复杂系统脆性相对熵及其在分布式系统中的应用  

Relative entropy of complex system brittleness and its application in distributed systems

作　　者：冯丽媛[1,2] 姚绪梁[1] 曹然[1] 邹艾利[1] 

机构地区：[1]哈尔滨工程大学自动化学院,哈尔滨150001 [2]黑龙江科技大学电子与信息工程学院,哈尔滨150022

出　　处：《控制与决策》2015年第1期185-188,共4页Control and Decision

基　　金：国防科工局技术基础研究基金项目(Z192011B001)

摘　　要：分布式系统安全性和可靠性检测的难点在于缺乏对系统脆性的动态评估.针对这一难点,提出一种新的概念脆性相对熵来衡量系统的脆性,并给出评估方法.利用脆性相对熵可以动态地衡量当前概率分布与系统崩溃概率分布之间的相对距离,有效地评估系统当前状态,并对系统脆性的概率风险加以定量分析.仿真结果表明,脆性相对熵可以衡量系统的脆性特征,且越接近系统脆性分布,脆性相对熵越小.The difficulty of distributed system security and reliability testing lies in the lack of dynamic assessment in the system brittleness. Therefore, a new concept called brittle relative entropy is proposed to measure the system brittleness, and assessment methods are given. The brittle relative entropy can dynamically measure the relative distance between the current probability distribution and the collapse distribution, which can effectively assess the current state of the system and analyse the probabilistic risk of brittleness. The simulation results show that the brittle relative entropy can be used to measure the brittle characteristics of the system. When the brittle relative entropy is closer to the brittle distribution system, the brittle relative entropy is smaller.

关 键 词：脆性相对熵 分布式系统 脆性条件相对熵 非负性 凸函数性 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]