摘要
针对风机状态监测问题,提出一种基于多元状态估计和相似性测度的方法。首先建立正常工况下各监测参数之间的关联模型,构造正常工况下的记忆矩阵;然后根据系统当前观测向量与记忆矩阵之间的相似性程度,对当前观测向量进行估计。通过对正常工况下的监测参数的聚类中心的提取,计算估计值与聚类中心的相似性测度值,确定风机工作状态。最后,以山西某电厂送风机为监测对象进行应用研究,结果表明该方法可以准确估计风机振动状况,尽早发现设备异常,实现风机状态的在线监测。
A method was proposed to solve fan condition monitoring, based on the multivariate state estimation tech- nique （MSET） and similarity measure. Correlation model of monitoring parameters in normal work condition was built firstly, and the memory matrix was also built reasonably. Then, according to the similarities between the current ob- served vector and memory matrix, the current vector was estimated by MSET. The cluster center was extracted from the normal condition monitoring parameters, the similarity was calculated, and work condition of the fan was pronounced. Finally, the method was applied on the vibration condition monitoring of the air feeder in a power plant in Shanxi, the results show that the method can monitor the vibration of fans, discover the abnormal earlier, and achieve the purpose of online state monitoring.
出处
《华北电力大学学报（自然科学版）》
CAS
北大核心
2013年第3期91-94,112,共5页
Journal of North China Electric Power University：Natural Science Edition
关键词
多元状态估计
相似性测度
状态监测
风机
multivariate state estimation technique （MSET）
similarity measure
condition monitoring
fan
