摘要
针对滚动轴承的运行状态识别问题,利用典型DP混合模型良好的聚类特性,提出了基于DPMM的滚动轴承运行状态识别算法,并推导了算法聚类的详细步骤。利用轴承状态监测数据进行了验证和分析,结果表明:DPMM算法不依赖于训练样本,模型结构能够随着观测数据的变化实现自适应变化和动态调整,自动识别轴承的运行状态数;同时,识别结果不依赖于DPMM算法初始参数的选择,具有较强的稳定性和适应性。
Aiming at recognition problem for operating states of rolling bearings,the good clustering properties DP mixture model is used.The algorithm for recognizing operating state of rolling bearings DPMM,and the detailed steps of algorithm clustering are deduced.The verification and analysis are carried condition monitoring data of bearings.The results show^that the DPMM algorithm does not depend on training samples.The model structure is able to realize adaptive change and dynamic adjustment with variation and the operating states of bearings are recognized automatically.At the same time,the recognition on choice of initial parameters for DPMM algorithm,so it has strong stability and adaptability.
作者
瞿家明
周易文
王恒
黄希
QU Jiaming;ZHOU Yiwen;WANG Heng;WUANG Xi(School of Mechanical Engineering,Nantong University,Nantoog 222019,China)
出处
《轴承》
北大核心
2018年第9期58-62,共5页
Bearing
基金
国家自然科学基金项目(51405246)
江苏省自然科学基金项目(BK20151271)
江苏省研究生科研创新计划项目(KYCX17_1913)
南通市应用基础研究项目(GY12016010)