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 be 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.
展开▼
