基于ICA的油库阀门内漏声发射定量检测  

Quantitative Detection for Acoustic Emission of Oil Depot Valve Internal Leakage Based on ICA

作　　者：宗福兴[1] 税爱社[1] 黄海波 刘翔宇[1] 

机构地区：[1]后勤工程学院后勤信息与军事物流工程系 [2]78460部队

出　　处：《后勤工程学院学报》2015年第4期90-96,共7页Journal of Logistical Engineering University

摘　　要：基于声发射原理的阀门内漏检测作为一种动态无损检测方法得到了广泛应用,但是定量检测一直是研究难点。现有的内漏率定量检测模型计算方法因参数不易测定,导致计算困难,精度不高。采用独立分量分析(ICA)方法提取声发射信号特征向量,并与参考样本集的特征向量相比较,相似程度最高的即可作为待测阀门的内漏率。内漏模拟实验分别将常规参数、频谱成分和ICA提取特征作为特征向量进行比较,结果表明ICA提取的特征向量维数为15时,均方根误差可以达到0.01 L/min,能够满足工程检测需要。The valve internal leakage inspection based on acoustic emission, which is a new dynamic non-destructive testing technology, is increasingly applied. What＇ s more, a quantitative detection has a main difficulty. It is difficult to calculate with the quantitative model methods and the results are inaccurate, for the parameters are hard to he determined. The one of the feature vec- tors extracted by ICA, which matches samples of the reference set best, can be seen as the valve internal leakage rate. Internal leak- age simulation experiments compare the feature vectors of general parameters and frequency spectrum components with ICA. The re- sults show that the root mean square error is 0.01 L/min when the dimension of ICA vector is 15, which can satisfy the requirement of engineering detection. The method of this paper provides reference for in-depth researches on valve internal leakage.

关 键 词：独立分量分析 阀门内漏 声发射 定量检测 

分 类 号：TH137.86[机械工程—机械制造及自动化] TP206.3[自动化与计算机技术—检测技术与自动化装置]