电子镇流器故障诊断的变精度粗糙集模型  

Fault diagnosis model based on variable precision rough set for electronic ballasts

作　　者：赵荣泳[1] 李翠玲[2] 高晓康[3] 王昭云[4] 

机构地区：[1]同济大学CIMS研究中心,上海200092 [2]上海海事大学电气自动化系,上海200135 [3]上海应用技术学院机械与自动化工程学院,上海200233 [4]环球迈特照明电子有限公司,上海201102

出　　处：《南京大学学报（自然科学版）》2010年第5期494-500,共7页Journal of Nanjing University（Natural Science）

基　　金：上海市教委支出预算项目(2008088);上海市重点学科建设项目(J50602);教育部留学回国人员科研启动基金

摘　　要：目前大批量电子镇流器在线质量检测环节中的故障诊断是典型的信息不确定性问题,连续属性和离散属性并存,故障根源间的关系复杂,也无法建立故障根源与故障之间的精确数学模型.为此,本文引入变精度粗糙集理论(VPRS),构建电子镇流器的智能故障诊断模型(VPRS-ID).针对连续属性的离散化问题,引入了自组织映射神经网络模型(SOM),自动的实现了连续属性的离散化;针对训练集合中论域对象的不一致问题,提出了基于频度的论域对象更新模型;针对部分诊断对象与已发现规则之间的不一致性问题,通过合理控制变精度因子,发现了隐藏在产品故障数据与故障根源之间的规则.在满足电子镇流器诊断精度要求的基础上,发现了有用的诊断知识,改善了经典粗糙集理论规则集合的适应性.针对目前低效的电子镇流器的人工诊断方式,进一步提出电子镇流器的在线智能诊断程序架构,为准确和高效的自动化故障诊断,提供决策支持.最后,通过电子镇流器故障诊断的工程实例,验证了该智能诊断模型的有效性.In the mode of large-scale manufacture,the fault diagnosis of electronic ballasts for online quality testing is a typical uncertainty problem due to the complex relationship between the fault causes and the faults nowadays.A precise mathematic model cannot be set up for the fault diagnosis of electronic ballasts.Thereby the variable precision rough set theory is introduced to discover the rules between ballasts testing data and fault roots in this paper.An intelligent diagnosis model based on variable precision rough set is established as VPRS-ID model.The self-organizing-mapping（SOM） neural network is introduced in the clustering process for the continuous attributes discretization.An update model based on object frequency is presented to solve the inconsistency between the training objects.The variable precision factor is adjusted reasonably to solve the inconsistency between the some testing objects and established rules in classic rough set theory.Some meaningful knowledge is discovered to improve the adaptability of classic rough set rules,based on the diagnosis precision requests of electronic ballasts.A program structure of the online intelligent diagnosis for electronic ballasts is proposed to change traditional,low-efficient and manual diagnosis into automatic and intelligent diagnosis,acting as a decision support tool.Finally an engineering sample illustrates the feasibility of this intelligent model.

关 键 词：变精度粗糙集 属性约简 电子镇流器 故障诊断 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]