摘要
提出了一种基于最佳分类数和粗糙集理论的汽轮机轴系振动故障诊断方法。该方法利用模糊C均值聚类算法(FCM)把数据的连续属性离散化,以形成隶属度矩阵及属性分类数,根据隶属度矩阵及属性分类数进行划分系数和划分熵的有效性评判,最终找到连续属性的最佳分类数。然后根据最佳分类数对数据的连续属性进行实际的离散化,将离散化后形成的离散数据根据粗糙集理论,进行数据挖掘,得到诊断规则,有效提高了汽轮机轴系振动故障的诊断水平。
The diagnostic method on stream crankshaft based on optimal classification number and rough set theory is proposed in this paper,which discretizes the continuous attributes of data with fuzzy C-means clustering to obtain the subjected matrix,then it has a validated judgement according to the computed partition coefficient and partition entropy by the subjected matrix and the cluster number,finally finds the optimal classified number of continuous attributes.It actually discretizes the continuous attributes of data according to optimal classified number,and the diagnosis rules are obtained by applying rough set theory on discrete data,effectively enhance the level of vibration fault diagnosis on stream crankshaft.
出处
《计算机与数字工程》
2010年第5期31-34,95,共5页
Computer & Digital Engineering
关键词
模糊C均值聚类
最佳分类数
划分系数
划分熵
粗糙集
fuzzy C-means clustering
optimal classified number
partition coefficient
partition entropy
rough set
