摘要
该文针对在线独立成分分析算法学习速率以及收敛性难以把握的问题,提出了一种利用变窗体移动窗附加在实时信号上的快速独立成分分析(Fast independent component analysis,FICA)改进算法,不但满足在线处理要求,而且不用考虑学习速率的问题,节省存储空间并提高运算效率。利用自组织映射(Self-organizing maps,SOM)神经网络算法在动态分类上的优势,采用变移动窗快速独立成分分析与自组织映射相结合的方法对心动异常数据进行了分类。实验表明,该方法能有效地提高速率和实现实时故障分类。
To deal with the problem of learning rate and convergence in online independent component analysis(FICA)algorithm,an improved fast independent component analysis algorithm with variable moving windows attached to real-time signal is presented here.This algorithm which saves storage space and computing time,can not only meet the requirements of online processing,but also do not need to consider learning rate.With the advantage of self-organizing map neural network algorithm on the dynamic classification,the combined approach based on variable moving window FICA and self-organizing maps(SOM)neural network is used to classify the abnormal echocardiography data.The experiments show that this method can effectively improve the rate and realize real-time fault classification.
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2013年第4期530-535,共6页
Journal of Nanjing University of Science and Technology
基金
辽宁省科学技术计划项目(2010222005)
关键词
快速独立成分分析
自组织映射
心动异常
故障分类
fast independent component analysis
self-organizing maps
abnormal echocardiography
fault classification
