基于改进谱聚类的提升机故障诊断算法  

Mine hoist fault diagnosis based on improved spectral clustering algorithm

作　　者：陈少达 夏士雄[1] 王志晓[1] 

机构地区：[1]中国矿业大学计算机学院,江苏徐州221116

出　　处：《计算机工程与设计》2015年第12期3241-3245,共5页Computer Engineering and Design

基　　金：国家863高技术研究发展计划基金项目(2012AA0622022);教育部博士点基金项目(20100095110003)

摘　　要：针对传统谱聚类存在需要人为确定聚类数目、对初始聚类中心敏感和鲁棒性较差等问题,通过对谱聚类算法的研究,提出一种改进的谱聚类算法。将谱聚类与数据场模型结合,利用数据场模型剔除孤立数据点,借助数据场模型判断谱聚类算法的k值和初始聚类中心点,利用K-means聚类算法进行聚类划分。基于UCI数据集和提升机轴承故障数据的实验结果表明,将谱聚类与数据场结合能够有效提高提升机故障诊断性能。Traditional spectral clustering method needs to determine the cluster number artificially, and it is sensitive to the ini- tial cluster centers and its robustness is weak. Data field was introduced to spectral clustering, and an improved spectral cluste- ring algorithm was put forward. The similarity matrix was adjusted appropriately, with the help of data field model, the k value and the initial cluster centers that spectral clustering algorithm needs were obtained. The adjusted eigenvectors were clustered using the K-means algorithm. The simulation experimental results show that the improved spectral clustering algorithm has good performance on UCI dataset and mine hoist bearing dataset.

关 键 词：矿井提升机 故障诊断 谱聚类 数据场 聚类 

分 类 号：TP306.3[自动化与计算机技术—计算机系统结构]