摘要
针对复杂时变工业过程实时故障诊断问题,提出了一种基于提升小波(lifting wavelet,LW)与递归增量聚类(recursive incremental clustering,RICLUSTER)相结合的实时故障诊断方法(lifting wavelet-recursiveincremental clustering,LW-RICLUSTER)。该方法首先通过LW变换对数据实时去噪,再通过RICLUSTER实时监控。由于采用LW与RICLUSTER相结合的方法,节省存储空间和运算时间的同时提高了诊断精度。实验结果表明,LW-RICLUSTER集合方法能有效实现时变过程监控,在诊断精度、速度和适应性方面,优于传统单一型CLUSTER方法。
An ensemble real-time fault diagnosis method based on lifting wavelet （LW） and recursive incre- mental clustering（RICLUSTER）, called LW RICLUSTER, is proposed to realize real-time monitoring for com- plex time-varying industrial processes. Firstly, data are denoised by LW transform in real-time, then RICLUS- TER is used for real-time monitoring. With the ensemble approach, storage space is saved and computing time is shortened, while the precision of diagnostic is increased. Experiment results show that the LW-RICLUSTER algorithm can monitor time-varying process. The LW-RICLUSTER is superior to the traditional single CLUS- TER in diagnosis precision, rate and adaptability
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2013年第1期161-167,共7页
Systems Engineering and Electronics
基金
国家自然科学基金(60974070)
辽宁省科学技术计划项目(2010222005)资助课题
关键词
故障诊断
实时监控
提升小波
递归增量聚类
fault diagnosis
real-time monitoring
lifting （RICLUSTER） wavelet （LW）
recursive incremental clustering
