气液两相流双视角气泡重建算法研究

发布时间：2018-10-16 18:39

【摘要】：气液两相反应体系广泛存在于过程工程的诸多领域,是一典型的复杂系统,其中气泡聚并、破碎动态演化行为,对反应器的传质传热效率有着重要影响,进而影响整个反应器的性能,因此,通过无干扰测量阐明气泡群这一重要介尺度结构的动态演化机理是深入理解气液两相复杂体系的关键。基于X射线的双视角成像对于实时测量气泡的动态行为有着具大的潜力。然而,由于只有两个投影角度,CT成像传统的FBP(Filtered Back Projection,滤波反投影)算法的重建精度过低。基于CS(Compressed Sensing,压缩感知)的重建算法已经被广泛应用扇形束、锥形束及其他扫描形式的CT系统,展示了其从稀疏视角下的投影数据高精度重建图像的能力。用基于CS的CT算法有望提高双视角下的气泡CT图像的重建精度。本文以TV(Total Variation,总变差)最小这一经典的基于CS的重建方法为最优化模型,以SD-POCS(Steepest Desent-Projection onto Convex Sets,最速下降-投影到凸集上)算法来解TV模型,从而实现双视角下的气泡图像的高精度重建。具体来说,做了如下几个工作:(1)用计算机仿真方法仿真出了气液两相流系统和基于面阵探测器的双视角下的CT系统,得到了投影数据。(2)基于D2D(Discrete to Discrete,离散到离散)模型,将CT成像的正过程建模成了一个线性方程组。为解此欠定线性方程组,设计了基于TV最小的最优化模型。为求解TV最小化问题,设计了改进的SD-POCS算法。基于此算法,设计并实现了图像重建程序,并使用仿真出的投影数据重建出了仿真系统中的气泡图像。(3)设计并实现了气泡CT图像的后处理及识别,达到了气泡检测的指标要求。仿真实验表明,本文提出的最优化模型、设计和改进的成像算法能有效用于双视角下的气泡图像重建,提高了重建精度,为下一步研发基于X射线双视角成像的气泡无干扰测量系统提供了算法准备。
[Abstract]:The gas-liquid two-phase reaction system widely exists in many fields of process engineering and is a typical complex system in which the dynamic evolution behavior of bubble accumulation and fragmentation has an important effect on the efficiency of mass transfer and heat transfer of the reactor. Therefore, the dynamic evolution mechanism of bubble group, an important mesoscale structure, is the key to further understanding the gas-liquid two-phase complex system. Dual-view imaging based on X-ray has great potential for real-time measurement of bubble dynamic behavior. However, because of only two projection angles, the reconstruction accuracy of the traditional FBP (Filtered Back Projection, filtering backprojection algorithm for CT imaging is too low. The reconstruction algorithm based on CS (Compressed Sensing, compression sensing) has been widely used in sector beam, conical beam and other scanning CT systems, showing its ability to reconstruct images with high accuracy from the projection data in sparse view. The CT algorithm based on CS is expected to improve the reconstruction accuracy of bubble CT images in two angles. In this paper, the classical TV (Total Variation, reconstruction method based on CS is taken as the optimization model, and the TV model is solved by the SD-POCS (Steepest Desent-Projection onto Convex Sets, steepest descent projection on convex set) algorithm, so as to realize the high accuracy reconstruction of the bubble image with two angles of view. The main works are as follows: (1) the gas-liquid two-phase flow system and the CT system based on plane array detector are simulated by computer simulation, and the projection data are obtained. (2) based on D2D (Discrete to Discrete, discrete to discrete model, The positive process of CT imaging is modeled as a linear system of equations. In order to solve this system of underdetermined linear equations, an optimization model based on TV minimization is designed. In order to solve the TV minimization problem, an improved SD-POCS algorithm is designed. Based on this algorithm, the image reconstruction program is designed and implemented, and the bubble image in the simulation system is reconstructed using the simulated projection data. (3) the post processing and recognition of bubble CT image is designed and realized, which meets the requirements of bubble detection. The simulation results show that the proposed optimization model, design and improved imaging algorithm can be effectively used in bubble image reconstruction with two angles of view, and the reconstruction accuracy is improved. It provides the algorithm preparation for the next research and development of bubble non-interference measurement system based on X-ray dual-angle imaging.
【学位授予单位】：中北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;O359

【相似文献】