基于部分可观Petri网的机车故障诊断方法研究  

Research on fault diagnosis method of locomotives dispatching based on partially observed Petri nets

作　　者：方欢[1] 陆阳[2] 方贤文[1] 王丽丽[1] 

机构地区：[1]安徽理工大学理学院,淮南232001 [2]合肥工业大学计算机与信息学院,合肥230009

出　　处：《电子测量与仪器学报》2015年第5期722-729,共8页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金(61070220;61472003;61272153;61340003;61402011);高等学校博士学科点专项基金(20120111110001);国家"863"计划(2011AA060406);安徽省教育厅高等学校自然科学研究重点(KJ2014A067)资助项目

摘　　要：机车调度的故障检测与诊断是保障行车安全的必要条件。为了实现故障的检测和准确诊断,以仅有部分库所可见的部分可观Petri网为建模工具,研究机车调度系统中传感器监控节点数目和位置的确定方法。首先,给出了加权Petri网系统中故障准确诊断方法,提出故障定位表和监控库所集的确定算法(FLT&MPD),并进一步证明所提出的算法是正确的、有解的且具有多项式级时间复杂度。进一步,基于FLT&MPD算法,给出了保证故障准确诊断下的部分可观系统设计方法,并且证明利用FLT&MPD可以实现监控节点数目最少的最优监控。最后,以机车调度系统中的两类典型故障为例,通过构造对应的层次颜色Petri网模型,给出机车调度系统中传感器监控节点的安放位置,为工业应用奠定了理论基础。The fault detection and diagnosis is the necessity of ensuring the safety of locomotives. In order to realizing the fault detection and unambiguous diagnosis, partially observed Petri Nets are used to research the positions deter- mination method of monitoring sensors, in which only partial place can be observed and all transitions can＇ t be ob- served. First of all, the fault unambiguous diagnosis methods of weighted Petri Nets are proposed. Then the fault lo- cation table and monitoring places determination algorithm （FLT＆ MPD） is given. It is proved that the proposed FLT＆ MPD algorithm is correct, solvability and behaves polynomial complexity. Furthermore, based on the FLT＆ MPD algorithm, the constructing method of partially observed system is given. And it is proved that this constructing method realizes the optimum monitoring with the minimum number of monitoring sensors. Finally, two kinds of faults of locomotives dispatching system are considered. Through corresponding hierarchical colored Petri Nets model, the monitoring sensors positions are determined, which provide profound theoretical foundation for industrial application.

关 键 词：部分可观系统 故障诊断 PETRI网 传感器监控 机车调度 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]