基于ECT图像重建算法的多相流检测研究

发布时间：2018-06-09 01:17

本文选题：多相流 + 检测　；参考：《农业机械学报》2016年07期

【摘要】：由于电容层析成像(ECT)图像重建的经典算法Landweber迭代法具有半收敛性,从而限制了其在多相流检测中的广泛应用。为了解决这一问题,从数学角度对Landweber法的半收敛性进行分析证明,并找到其物理本质原因。通过构造压缩算子提出一种全收敛的Landweber法,使得该方法具有稳定的收敛特性。同时,对ECT敏感场的"软场"特性进行分析,针对忽略"软场"特性所导致的经典Landweber法对不同流型的最佳迭代次数不一致的问题,利用最小二乘支持向量机对一定数量的图像样本进行训练,从而得到一个先验条件,并基于该先验条件提出一种自适应的迭代停止判据,实现在图像重建结果最接近多相流实际分布时停止迭代。仿真结果表明,改进后的方法不仅具有稳定收敛性,而且与Landweber法相比,图像重建精度可提高16%~50%。
[Abstract]:Due to the semi-convergence of Landweber iterative algorithm for electrical capacitance tomography (ECT) image reconstruction, its wide application in multiphase flow detection is limited. In order to solve this problem, the semi-convergence of Landweber method is analyzed and proved from the point of view of mathematics, and the physical essence of Landweber method is found. By constructing a contraction operator, a fully convergent Landweber method is proposed, which has stable convergence properties. At the same time, the characteristics of "soft field" of ECT sensitive field are analyzed, and the problem that the classical Landweber method caused by neglecting the characteristic of "soft field" is inconsistent with the optimal iteration times of different flow patterns. The least square support vector machine (LS-SVM) is used to train a certain number of image samples, and a priori condition is obtained. Based on the priori condition, an adaptive iterative stopping criterion is proposed. The algorithm stops iteration when the image reconstruction results are closest to the actual distribution of multiphase flow. The simulation results show that the improved method not only has stable convergence, but also can improve the accuracy of image reconstruction by 16.5% compared with Landweber method.
【作者单位】：西安电子科技大学空间科学与技术学院;
【基金】：国家自然科学基金项目(61571346、61305041、61305040) 中央高校基本科研业务费专项资金项目(JB141305)
【分类号】：TP391.41

【参考文献】