摘要
针对火电厂锅炉运行过程复杂,而且异常工况难以分析的状况,提出了一种新的分析方法。该方法融合属性模糊聚类和关联规则算法两种数据挖掘技术,利用属性模糊聚类方法建立锅炉各运行参数对类别的不确定性描述,选择出代表性分析参数;利用经典的Apriori算法和改进的关联规则找出这些参数的关系。通过2个超温的实例对该方法进行分析考核。结果表明:它不仅充分利用了电厂DCS系统存储的海量数据,而且可实现故障诊断、运行指导等,并可满足诊断的实时性要求,从而可保证锅炉乃至整个机组的安全运行。
Since operational processes of boilers, in fossil fired power plants, are complex and anomalous behaviors are hard to analyze, a new way of analysis is being suggested which amalgamates the two data mining techniques, i.e. attributive fuzzy clustering and association rule calculation. An uncertain description of all the boiler＇s operational parameter is obtained by the attributive fuzzy clustering method and representative parameters are selected; cognition of the selected parameters then proceeds according to association rules, improved by mutative confidence. The suggested method is being analyzed and verified by actually occassioned instances of overheating. Results indicate, that this method not only can make full use of the enormous amount of data, porform fault diagnosis, as well as serve operational guiding stored by the power plant＇s DCS system, but can also purposes. Moreover, because requirements of real time diagnosis are met, the method is a capable of warranting safe operation, not only of the boiler, but even of the whole set.
出处
《动力工程》
EI
CSCD
北大核心
2007年第3期362-366,共5页
Power Engineering
关键词
自动控制技术
锅炉
超温
数据挖掘
属性模糊聚类
关联规则
automatic control technique
boiler
overheating
data mining
attributive fuzzy clustering
association rules
作者简介
万绪财（1980-），男，山东青岛人，硕士。主要从事电站锅炉设计、开发与应用的研究。
