摘要
经典粗糙集理论只能处理离散数据,不能将其直接应用于实际系统的数据挖掘中。本文引入样本之间的相似性和改进的属性广义区分度的概念,并定义了属性的全局相似性程度,然后利用容差关系对连续数据集直接进行属性约简,避免了数据离散化过程中信息的丢失。最后将其应用于汽轮机组故障诊断系统中,实验结果表明该方法的有效性。
Classical rough set theory is just suitable for discrete data; therefore it can not be directly applied to data mining in practical systems. The notions of similarity between objects and improved general discrimination of attributes are introduced in this paper, which define the general similarity measure between objects. A direct reduction method is applied to continuous attributes using tolerance relation, which avoids losing information in data discretization progress. Finally, the method is applied to fault diagnosis of a steam turbine unit, and experiment result shows that the proposed algorithm is effective.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2008年第7期1522-1525,共4页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(60674092)
江南大学创新团队发展计划资助
关键词
粗糙集
属性约简
容差关系
故障诊断
rough set
attribute reduction
tolerance relation
fault diagnosis
