【作者】 孙振；

【导师】 叶阳东；

【作者基本信息】 郑州大学 ， 计算机软件与理论， 2013， 硕士

【摘要】 故障诊断是一种利用故障信息之间的逻辑关系和故障机理联合分析而进行故障的辨识与定位的技术。建立合理的诊断模型以及消除诊断过程中的不确定性问题带来的故障扰动,一直是诊断技术中亟需解决的关键问题。其中,时间不确定性问题对故障信息的归属以及推理都产生了不利的影响,国内外学者在不同背景下从不同的角度提出了许多分析和处理方法。针对大量故障信息在时序上混杂在一起不能明确其隶属关系的影响,本文基于模糊理论、贝叶斯概率推理和Petri网相关理论,提出了一种基于时间不确定性问题的时间贝叶斯Petri网模型TBPN (Time Bayes Petri net)及其相应的故障诊断方法。该模型定义了4种库所、4种变迁、时间因素相关的支持度函数以及贝叶斯概率推理函数等,可以将原本时间不确定性的问题转化为故障信息的定量描述,从而得出故障可能发生的概率,与其他Petri网故障诊断模型相比,在时间因素的考虑上更为明确,并将时间作为一种参数,参与故障信息对规则的支持程度的计算。该故障诊断方法在信息的归属阶段给出信息的归属策略,不仅可以使用后续没有使用过的信息,而且还对已经使用过的信息进行重新利用；在信息推理阶段,利用已经匹配好规则的信息和贝叶斯概率推理函数进行概率推理。文中给出一个变电站故障诊断应用案例,结果表明：该模型及其故障诊断方法可以对变电站故障发生过程进行建模与分析,可以定量的给出故障的发生概率,尤其,在信息不完备的情况下,能够快速给出故障的发生概率。该模型有自身的数学的形式化描述,有很好的语义描述能力,容错性和适应性很强,可以适用于较为复杂的诊断过程。更多还原

【Abstract】 The fault diagnosis is a technology about logical relationship between the fault information and analysis of failure mechanism. Establishing the reasonable diagnosis model and eliminate the uncertainty problem in the process of diagnosis of fault disturbance are need to solve the key problems in the diagnostic techniques. Among them, the study is a difficulty for the uncertainty. Different scholars both at home and abroad from different perspectives put forward a lot of analysis and processing method.In view of the many uncertainties existing in the fault diagnosis problem,this paper consider the time uncertainty impact of the distribution and reasoning of the temporal information, Based on the fuzzy theory, bayes probability reasoning and Petri net theory, the paper put forward a kind of time bayes Petri net model (TBPN). About the model originally time uncertainty problem, the model can be transformed to quantitative description of fault information, and concluded the probability of failure. TBPN model is divided into the classified information processing stages and information inference phase two parts on the network struct:The information belonging to stage the belonging of the fault information is used to put forward a kind of ownership strategy, and gives the fault information and the corresponding matching rules; According to have good information, the information processing stage calculates failure probability, and finds the failure probability of the failure source. The paper proposes token’s formal representation, the corresponding time ownership strategy, information support function and information, support function.The result of the fault diagnosis example of substation indicates that this model and method can construct the reasonable model and solve the problem. Espically, when the informations are incomplete, the method can complete the fuzzy reasoning smoothly. It can be used in more complicate fault diagnosis process.更多还原