摘要
在超声波射频(RF)信号处理过程中,分别以标准试块、锻铝、水的超声反射回波信号为研究对象,进行时域波形特征值的提取及选择,并以降维分析的方法对测试样品进行聚类识别。研究表明:应用t假设检验方法,可以有效的对特征值进行评估及分析,进而取得最佳特征子集;聚类分析结果表明,本文提出的特征值对被检对象具有较强的识别能力。
In this paper, the ultrasonic echo signals of standard-specimen, forging-aluminium and water are studied for feature selection in the time-domain extraction. The preliminary experimental results show that the feature can be evaluated and analysed effectively applying t hypothesis-testing method so as to obtain the optimun feature subset; the results of the cluster analysis testify that the feature supplied in the paper possesses effective recognition effect.
出处
《数理统计与管理》
CSSCI
北大核心
2004年第4期24-27,共4页
Journal of Applied Statistics and Management
关键词
特征值提取
聚类分析
降维分析
feature extraction
cluster analysis
decrease-dimension analysis
