摘要
利用图像处理的方法提取气液两相流动图像的灰度纹理特征值的方法,引入主成分分析法(PCA)来对提取的多维灰度纹理特征向量降维。分别结合单变量的非线性拟合与多变量的支持向量机回归对两相流容积含气率进行了求取。对大量测量样本进行统计分析后,分别求出了测量值第一主成分求取容积含气率的经验公式和测量值第一、二和三主成分求取容积含气率的经验公式,同测试样本的真实容积含气率值进行比较,结果表明:流型图像灰度纹理特征与容积含气率之间有着很好的对应关系,应用测量值第一主成分求取的经验公式适用于低含气率的计算,而应用前三个主成分求取的回归经验公式更加适合高含气率的计算,为容积含气率的计算提出了一个新的思路。
Describe the use of image processing to extract gas-liquid two-phase flow characteristics of the image gray-value method of texture,introduction of PCA to make the dimensionality reduction of the multi-dimensional gray-scale.Compute the two-phase volume fraction with single variable nonlinear fitting and multi-variable support vector machine regression.After statistical analysis of a large number of measurement samples,get the empirical formula to strike a volume fraction with the first principal component measurements and the first,second and third principal component of measured values.compare to the true values of volume fraction,The results show that：The gray-scale of flow image and volume fraction have a good correlation,the single variable method is applied to calculate the low rate,and the multi-variable method is applied to calculate the high rate.The article propose a new idea for calculating of the volume fraction.
出处
《东北电力大学学报》
2012年第1期35-40,共6页
Journal of Northeast Electric Power University
