基于故障树分析法构建专家系统知识库模型摘要:本文在广泛搜集往复式压缩机故障类型的基础上,探析故障机理。
运用故障分析法,建立故障树模型,并用二维表格将其表示出来。
然后并运用access数据库和vb语言构建知识库链表。
最后,给出故障诊断专家系统知识库维护方法。
关键词:往复式压缩机知识库故障树引言:往复式压缩机由于其自身的特点广泛应用于石油石化企业。
但由于机构复杂、零件繁多,现场维修人员在诊断故障问题时困难重重。
在维护和维修往复式压缩机时,故障诊断专家系统可以给现场维修人员提出宝贵建议的。
在往复式压缩机故障诊断专家系统中,知识库的优劣直接影响到诊断的准确性和真实性。
在构建知识库过程中,故障树分析法直接简明、逻辑性强等特点,所以本文采用故障树模型建立往复式压缩机故障诊断系统的知识库,保证诊断的准确性和真实性。
Building a knowledge base of expert system model based on the fault tree analysis 1,故障树分析法基本知识1.1定义:故障树分析法就是把所研究系统的最不希望发生的故障状态作为故障分析的目标,然后寻找直接导致这一故障发生的全部因素,再找出造成下一级事件发生的全部直接因素,一直追查到那些原始的、其故障机理或概率分布都是已知的,毋需再深究的因素为止。
通常,把最不希望发生的事件称为顶事件,毋需在深究的事件称为底事件,介于顶事件和底事件之间的一切事件为中间事件,用相应的符号代表这些事件,再用适当的逻辑门把顶事件、中间事件和底事件联结成树形图。
这样的树形图称为故障树,用以表示系统或各个部件故障事件之间的逻辑结构关系。
以故障树为工具,分析系统发生故障的各种途径,计算各个可靠性特征量,对系统的安全性或可靠性进行评价的方法称为故障树分析法。
1.The failure analysis1.1 Basic knowledge of fault tree analysisFault tree analysis is that the most reluctant fault condition occurred in the studied system will be as a failure analysis of target; then look for all the factors leading to the most reluctant fault condition; next seek for all the direct factors causing the next level faults till original fault factors、well known failure mechanisms or open Probability distribution of fault factors would be fond out; finally, you can obtain all the original fault factors that can’t be divided. Usually, the most reluctant fault case would be considered as the top incindents; the fault factors that couldn’t be searched would be acted as the bottom incindents; the fault case in the middle of the top incindents and the bottom incindents would be though as intermediate incindents. By appropriate symbols of fault tree analysis expressing the three typle of mentioned incindents and combining the top incindents、intermediate incindents and the bottom incindents in logic relationship, we can make out the model of the fault tree analysis-the graph of fault tree analysis that it would indicate the logic structure for each fault incidents or fault tree analysis.Fault tree analysis is the method that it can evaluate security and reliability of the studied systems accuratelly that by the way of the model of fault tree, analyzing all kinds of faults incindent, caculating vavious characteristic quantities of reliability.1.2故障树分析法步骤故障树分析步骤具体如下:1.对所选定的系统作必要分析,了解系统的组成及各项操作的内容。
2.对系统的故障进行定义,对预计可能发生的故障、过去发生的故障事例及故障统计,作广泛调研。
3.仔细分析各种故障的形成原因并收集各故障发生概率数据。
4.选定系统可能发生的最不希望发生的故障状态为顶事件,作出故障树逻辑图。
5.对故障树结构作定量分析。
如故障树各底事件的故障概率数据为已知,就可以根据故障树逻辑,对系统的故障作定量分析;若底事件概率值系未知数,可假设某个合理值,以便对系统进行可靠性方案的比较。
下图是故障树分析法框图:1.2 the procedure of fault tree analysisThe concrete steps of fault tree analysis as follows:1.Do the necessary analysis for selected fault systems, obtain all the compositons andoperation contents for selected systems. 2.Define the type of incidents for the fault systems, take serious research and wild statisics for forcast possible fault case、already happened fault samples. 3.analyaze all becoming acuases for vavious faults carefully, and take together the probability datas for all kinds happened fault samples. 4.Take the most relutant fault cases as the top incindents; take the original cases or causes as the bottom incindents and so on, then figer out the model of fault tree. 5.Make the quantitative analysis for elements of the fault system. For example, if the datas of the probabilities for the happened cases could be obtained, we could make the quantitative analysis for elements of the fault system on the basis of logic conections in the fault systems; if the datas of the probabilities for the happened cases could not be obtained, we can suppose reliable datas, then analyze the fault systems in order that we can compare the results with the reliability method.图1 故障树分析法的顺序Figure 1 the sequence of fault tree analysis2故障机理分析12.故障表示由于往复压缩机故障诊断专家系统的知识通常是描述过程性的知识。
很多情况下,知识是松散杂乱的,因此非常适合采用产生式规则来组织知识库。
但是往复压缩机故障知识数量很多,为了提高推理速度,需要将其分门别类地存储, 因此推荐采用“框架+ 产生式”规则。
以“往复压缩机二级排气温度高”故障为例,下图中每个故障事件结点就是一个诊断对象, 每个诊断对象对应于一个描述框架。
2.fault representationBecause there were a lot of discription procedure information in expert fault diagnosis system for reciprocating compressor usually. As well as, the information abort fault diagnosis were loose and mess in most cases. So it is very suitable to establish knowledge base by production regulation. But the information of the fault cases were much huge. In order to improve the speed of caculation, we should store the datas regulately and apply “frame +produce”rule propably. Take “high temperature in second grade exhaust of reciprocating compressor” for example, in the following figure, every fault case node is a target of fault diagnosis, and each target of fault diagnosis correspond with each discripting frame.二级压缩缸排气温度高故障树模型The fault tree model of high temporatue in second exhaust 在构建诊断系统知识库时,本文使用Mirconsoft access数据库软件和SQL(结构查询语言)。