基于机器学习的CPS系统可靠性在线评估方法  

Method for evaluating reliability of Cyber-Physical Systems online based on machine learning

作　　者：邹青丙 刘羽 何明[1] 余沛毅[1] 王立辉[3] 

机构地区：[1]解放军理工大学指挥信息系统学院,南京210007 [2]总后后勤科学研究所,北京100071 [3]东南大学仪器科学与工程学院,南京210096

出　　处：《计算机工程与应用》2014年第10期128-130,170,共4页Computer Engineering and Applications

基　　金：国家自然科学基金(No.61203192);江苏省自然科学基金资助项目(No.BK2012326)

摘　　要：为了对CPS系统可靠性进行有效分析与量化,提出了一种实时的CPS可靠性自动在线评估方法,该方法采用机器学习思想构建了评估框架,设计了在线排队算法,实现了对CPS可靠性的实时在线分析与评估,并能及时采取预防措施,确保系统正常无间断运行,极大地提高了系统可靠性。仿真实验结果验证了评估方法的有效性及广泛的应用前景。In order to carry out the analysis and quantification for the reliability of CPS, an automatic online assessment method for the reliability of CPS is proposed. It builds an evaluation framework based on knowledge of machine learning, designs an online rank algorithm, and realizes the online analysis and assessment in real time. The preventive measures can be taken timely. And system can operate normally and continuously. Its reliability has been greatly improved. The result of simulation experiments has verified the validity of the assessment method and the broad application prospects.

关 键 词：信息物理融合系统(CPS) 可靠性 机器学习 排队模型 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]