基于谱聚类提取与数据场模型融合的提升机故障分析  

Hoist Fault Analysis Based on Spectrum Clustering Extraction and Data Field Model Fusion

作　　者：仝部雷 TONG Bulei(Shanxi Coal Staff Training Center, Taiyuan 030006, China)

机构地区：[1]山西省煤炭职工培训中心,山西太原030006

出　　处：《煤矿机电》2020年第6期31-34,38,共5页Colliery Mechanical & Electrical Technology

摘　　要：由于有监督谱聚类方法中需要初始化聚类数目和聚类中心节点,因此存在鲁棒性不足的现象,针对此问题研究了谱聚类算法的改进方法。首先通过数据场模型对样本集合中的孤立数据进行清除,然后再根据数据场模型的设计确定聚类节点个数及聚类分类数,最后再通过K-means算法进行无监督方式下的样本点划分。通过UCI公开数据集与提升机发生轴承故障时所产生的数据集进行了方法验证,实验结果显示将谱聚类技术与数据场模型的融合有效提升了提升机的故障诊断能力。In the supervised spectral clustering method,the number of clusters and the cluster center nodes need to be initialized,so there is a lack of robustness.To solve this problem,an improved method of spectral clustering algorithm has been studied.Firstly,the isolated data in the sample set was cleared by the data field model,then the number of clustering nodes and clustering classification were determined according to the design of the data field model,and finally the unsupervised sample points were divided by the K-means algorithm.The experimental results showed that the fusion of spectral clustering technology and data field model can effectively improve the fault diagnosis ability of hoist.

关 键 词：提升机故障 谱聚类技术 数据场模型 无监督聚类 

分 类 号：TD534[矿业工程—矿山机电]