基于数据驱动的卷积神经网络电容层析成像图像重建

针对电容层析成像技术的图像重建问题,提出了基于数据驱动的卷积神经网络图像重建方法。根据气固两相流的流型特点,通过数值模拟的方法随机生成了60000组介质分布图像,并利用有限元法计算了与之对应的电容向量,从而建立了一个“电容向量-介质分布”数据集;然后根据电容层析成像图像重建特点建立了卷积神经网络模型,对数据集中的训练集进行学习和训练,并利用测试集对训练结果进行了验证与评价。在此基础上,对获得的ECT图像重建卷积神经网络模型进行了静态实验和流化床测试实验研究。模拟和实验结果表明：所建立的卷积神经网络能较好地实现ECT图像重建,可直接用于流化床内的颗粒浓度分布测量。     

一种新的电容层析成像电极组合激励测量模式

电容层析成像图像重建为非线性不适定反问题,其敏感场分布不均匀,可获得的测量数据有限,中心处物体的成像效果不佳。增加电极数目,可获得更多的电容测量数据,减小其不定性,同时改善敏感场分布。但同时导致电容测量值变小,测量精度下降。在保证电容测量精度的前提下,提出了24电极组合式电容层析成像传感器结构,研究了两种激励测量方案,并与传统12电极电容层析成像传感器进行了对比分析,包括电容测量值的大小及其动态测量范围、灵敏场分布的均匀性以及不同流型下的重建图像。仿真结果表明,与12电极电容层析成像传感器相比,采用24电极组合式电容层析成像传感器,其电容测量值大,可较好地保证测量精度,其灵敏度分布更加均匀,对中心处物体的成像质量明显提高。     

基于改进的LS-SVM测量油气两相流空隙率

利用ECT电容传感器提供的电容测量值,基于改进的最小二乘支持向量机(LS-SVM),提出了一种油气两相流空隙率测量的新方法.运用该方法测量空隙率时,以ECT电容传感器获取的66个独立电容值作为空隙率模型的输入,计算即可得空隙率.建模阶段,针对LS-SVM使用中存在的问题,首先运用训练数据筛选技巧,对LS-SVM进行稀疏化改进,接着引入实数编码的遗传算法(RC-GA)优化LS-SVM参数,然后运用改进后的LS-SVM和基于RC-GA的参数选取方法建立空隙率测量模型.所提出的空隙率测量方法省去了常用ECT方法测量空隙率时的复杂耗时的图像重建过程,提高了空隙率测量的实时性.实验结果表明,提出的LS-SVM改进和参数优化方法是有效的,提出的空隙率测量方法具有实时性佳的优点,测量精度满足实际应用要求.     

基于改进的RBF网络和自适应小波图像增强的ECT两相流测量系统(英文)

Electrical capacitance tomography(ECT) is a non-invasive imaging technique that aims at visualizing the cross-sectional permittivity distribution and phase distribution of solid/gas two-phase flow based on the measured capacitance.To solve the nonlinear and ill-posed inverse problem:image reconstruction of ECT system,this paper proposed a new image reconstruction method based on improved radial basis function(RBF) neural network combined with adaptive wavelet image enhancement.Firstly,an improved RBF network was applied to establish the mapping model between the reconstruction image pixels and the capacitance values measured.Then,for better image quality,adaptive wavelet image enhancement technique was emphatically analyzed and studied,which belongs to a space-frequency analysis method and is suitable for image feature-enhanced.Through multi-level wavelet decomposition,edge points of the image produced from RBF network can be determined based on the neighborhood property of each sub-band;noise distribution in the space-frequency domain can be estimated based on statistical characteristics;after that a self-adaptive edge enhancement gain can be constructed.Finally,the image is reconstructed with adjusting wavelet coefficients.In this paper,a 12-electrode ECT system and a pneumatic conveying platform were built up to verify this image reconstruction algorithm.Experimental results demonstrated that adaptive wavelet image enhancement technique effectively implemented edge detection and image enhancement,and the improved RBF network and adaptive wavelet image enhancement hybrid algorithm greatly improved the quality of reconstructed image of solid/gas two-phase flow [pulverized coal(PC)/air].     

基于ECT和蚂蚁算法的油气两相流空隙率在线测量

基于电容层析成像系统（ECT）和蚂蚁算法,提出了一种油气两相流空隙率在线测量的新方法。该方法利用电容层析成像系统12电极电容传感器所获取的66个测量电容信息,首先根据电容层析成像系统所获取的流型辨识结果确定对应流型下的实际空隙率测量模型参数f和b,然后利用蚂蚁算法的信息素信息,找到当前测量状态下对空隙率起主要作用的组合电容集合和相应的权重系数,从而实现空隙率测量。与流型相对应的空隙率测量线性模型参数f和b基于先验数据通过最小二乘方法确定。油气两相流的实验结果表明,该方法对空隙率的在线测量是有效的,避免了复杂的图像重建计算,实时性能佳,测量时间小于0.08 s。在几种典型流型下,提出的空隙率测量方法与常用的快关阀方法相比最大测量偏差小于5.5%。     

一种用于在线测量的电容层析成像图像重建算法

提出了一种新的、快速的电容层析成像算法。在分析极小范数解的基础之上针对ECT逆问题的特点对其进行改进,并利用奇异值分解定理证明这种改进的数值稳定作用,从而从极小范数解的角度解决了ECT逆问题数值解的唯一性和稳定性问题;在此基础上从最优化的角度推导出进一步提高重建图像质量的公式;数值实验表明这种改进是有效的,其图像重建时间近似LBP,然而成像质量比LBP、Tikhonov和Landweber迭代法要好。     

用于两相流流型显示和空隙率测量的电容层析成像技术

引　言两相流流型的实时显示和空隙率的在线测量对两相流系统的控制、运行乃至机理研究等均具有重要意义 ,一直是两相流领域中重要的研究方向 .现已提出多种实现流型显示和空隙率测量的方法 ,例如用于流型显示的核辐射线法和光学法 ,用于空隙率 (或浓度 )测量的机理估算法、核辐射线法、电学法、光学法、微波法、热学法和核磁共振法等 ,但是总的来讲还未能满足应用要求 ,实际应用的例子也较少[1] .电容层析成像 (ｅｌｅｃｔｒｉｃａｌｃａｐａｃｉｔａｎｃｅｔｏｍｏｇｒａｐｈｙ ,简记ＥＣＴ)技术由于可在不干扰流场的情况下获取反映两相流…     

基于遗传算法的快速X射线计算机层析成像多相流测试技术(Ⅰ)测试原理和数值仿真

针对局部浓度场具有二值分布特征的特定多相流体系,提出引入多相流物理特征的快速X射线计算机层析成像（XCT）技术,基于围绕被测物有限角度的X射线投影数据,建立了改进的遗传算法（GA）求解不完整投影数据集的二值图像重构算法。通过有限角度的同时投影,实现对两相流浓度场的瞬间图像冻结和快速图像采集,既拥有XCT的高空间分辨率特征,同时满足高时间分辨率的动态测量要求,并由于投影角度数要求的大幅度缩减而降低了XCT的硬件成本。以气液两相流的多泡体系为研究对象,通过详细的数值仿真实验获得了较理想的图像重构效果,验证了GA-XCT的理论可行性。在3～24个有限角度投影数据的情况下,GA-XCT表现出明显优于传统CT图像重构算法（即滤波反投影算法）的二值图像重构能力。同时,GA-XCT具有很好的抗噪声能力,而且不依赖于气泡形状。     

Simulation of the performance of electrical capacitance tomography for measurement of dense phase pneumatic conveying

A method for the estimation of image distortion which can occur in the reconstruction of electrical capacitance tomography (ECT) data for pneumatic conveying process is presented. Image distortion is associated the finite sampling rate of tomographic instrumentation which arises from changes in the distribution of phases occurring during a measurement cycle. The method uses an algorithm which provides the solution of the inverse and forward problems for an idealized 2-D sensor and to visualize the movement of powder slugs. The influence of the geometry of the sensor electrodes and the nature of slug perturbation on the fidelity of the reconstructed image is discussed. The practical implications of the simulation for the application of capacitance tomography for monitoring and control of dense phase conveying are discussed.     

电容层析成像技术在线测量气固流化床空隙率的研究

基于电容层析成像技术，提出了一种在线测量气固流化床空隙率的新方法。建立了相应的12电极电容层析成像气固流化床空隙率测量系统，可同时实现气固 流化床空隙率分布的在线显示和整体空隙率测量。选择加权反投影算法进行图像重建以保证空隙率分布显示的实时性和有效性。采用Tikhonov正则化原理和ART算法相结合的组合型新图像重建算法来实现整体空隙率的测量。Tikhonov正则化原理用于克服图像重建过程中的不适定问题，ART算法用于提高最终重建图像的质量。研究表明以上提出的空隙率测量新方法是有效的。空隙率分布在线测量的速度可达25幅／秒以上，整体空隙率测量的最大误差可小于5％。     

Reconstruction algorithm of electrical impedance tomography for particle concentration distribution in suspension

An inverse problem is solved to obtain the particle concentration profile in suspension under pressuredriven flow with electrical impedance tomography (EIT). The finite element method (FEM) is employed in the forward problem and the regularized Newton-Raphson iterative method is used in the inverse problem. Different FEM meshes are used in the forward and the inverse problem not to commit inverse crime. To avoid post-calibration of measurement data, the complete electrode model is introduced. For the evaluation of the robustness of the reconstruction algorithm, several testing cases with measurement error are considered. The proposed algorithm can be used to reconstruct the particle concentration in suspension.     

应用电容层析成像对多相流的动态可视化(英文)

Identifying the flow patterns is vital for understanding the complicated physical mechanisms in multiphase flows.For this purpose,electrical capacitance tomography(ECT) technique is considered as a promising visualization method for the flow pattern identification,in which image reconstruction algorithms play an important role.In this paper,a generalized dynamic reconstruction model,which integrates ECT measurement information and physical evolution information of the objects of interest,was presented.A generalized objective functional that simultaneously considers the spatial constraints,temporal constraints and dynamic evolution information of the objects of interest was proposed.Numerical simulations and experiments were implemented to evaluate the feasibility and efficiency of the proposed algorithm.For the cases considered in this paper,the proposed algorithm can well reconstruct the flow patterns,and the quality of the reconstructed images is improved,which indicates that the proposed algorithm is competent to reconstruct the flow patterns in the visualization of multiphase flows.     

融合电容层析成像先验信息的集合卡尔曼滤波统计估计

提出融合先验信息的电容层析成像(ECT)系统信息融合成像法——基于Bayesian重建算法的集合卡尔曼滤波融合算法(EnKF)。其优势在于能充分利用流体动力方面的先验信息,进一步修正原始迭代算法的重建图像质量。Bayesian重建算法在重建图像的迭代过程充分考虑噪声和图像的概率分布,完成初步成像过程。集合卡尔曼滤波算法利用先验信息实现对多相流流体流动的预测,继而改善成像质量。此外,为了便于估算误差的协方差,多相流的相分布作为EnKF融合方法的估计对象,用状态向量空间的灰度差分统计描述,明显改善了通过状态向量空间模型获得的统计估算结果。仿真和试验的结果都充分表明了在ECT系统中集合卡尔曼滤波融合方法的可行性。     

Neural network multi-criteria optimization image reconstruction technique (NN-MOIRT) for linear and non-linear process tomography

In this work, an analog neural network is utilized to develop a new image reconstruction technique for the linear as well as the non-linear process tomography. The ultrasonic computed tomography (CT) and the electrical capacitance tomography (ECT) are chosen to represent the linear and the non-linear tomography. The image reconstruction technique is based on a multi-criteria optimization, namely neural network multi-criteria optimization image reconstruction technique (NN-MOIRT). The optimization technique utilizes multi-objective functions: (a) the negative entropy function, (b) the function of the least weighted square error of projection (integral) values between the measured data and the estimated projection data from the reconstructed image, and (c) a smoothness function that gives a relatively small peakedness in the reconstructed image. The optimization image reconstruction problem is then solved using the Hopfield model with dynamic neural-network computing. The technique has been tested using simulated and measured data; this technique has shown significant improvement in accuracy and consistency compared with other available techniques for both linear and non-linear tomography.     

基于遗传算法的快速X射线计算机层析成像多相流测试技术(Ⅱ)静态实验验证

采用静态实验模型和一套X射线测试系统模拟了二值多相体系多角度同时投影的X射线CT测量过程。静态实验模型模拟空气-水两相体系,截面相浓度分布满足0-1特征。X射线测试系统由一台工作电压为150 kV的闪光X射线机、一台X射线平板检测器和一套数据采集单元组成。通过旋转静态实验模型,分步实现不同角度的投影数据采集。基于模拟的多角度“同时”采集的投影数据,使用本系列研究所构建的基于遗传算法的快速X射线CT多相流测试技术（GA-XCT）对截面图像进行重建,测试结果表明：在3～24个有限角度实测投影数据的情况下,GA-XCT表现出了明显优于传统CT图像重构算法（滤波反投影算法）的图像重构能力,且具有良好的抗噪声能力。     

Techniques for the reconstruction of a distribution from a finite number of its moments

The reconstruction of a distribution knowing only a finite number of its moments is an extremely important but in practice still unsolved question for many fields of science (chemical and process engineering, electronic engineering, nuclear physics, image analysis, biotechnology…). Several methods have been proposed and corresponding mathematical formulations have been introduced in the literature during the last decades. Nevertheless, all these are generally limited to particular, often simple cases and require specific assumptions. It is indeed extremely difficult from a theoretical point of view (it is necessary, however, not sufficient, that all moments are available for a correct reconstruction) as well as from a practical point of view (ill-posed inverse problem) to find an accurate and relatively fast method which can be applied to all scientific areas. In the present paper, different possible methods (prescribed functions, discrete method, spline-based reconstruction) allowing such a reconstruction are explained, compared in terms of efficiency and accuracy, and validated for chemical engineering applications using examples with different degrees of difficulty.     

ECT imaging of three-phase fluidized bed based on three-phase capacitance model

In this work, the electrical capacitance tomography (ECT) with neural network multi-criteria optimization image reconstruction technique (NN-MOIRT), early developed by the authors, is applied to imaging bubble column and three-phase fluidized bed systems in the real time manner. Air, norpar (paraffin) and glass-beads are used as the gas, liquid, and solid phases, respectively. A three-phase capacitance model coupled with a two-region model is proposed to attain the gas holdup and the solids fraction from the permittivity maps of the three-phase system. The two-region model assumes that the solids fraction in the emulsion phase in the no bubble region is the same as in the bubble region. The three-phase capacitance model combines series and parallel capacitance connections among gas, liquid and solid components to relate the three-phase permittivity to each phase holdup. A direct image calculation to obtain the gas holdup from the permittivity map of the three-phase system is also performed by determining the permittivity threshold for the gas bubbles. Comparisons of the gas holdup obtained by ECT with that obtained from liquid head measurement showed a good agreement, validating the applicability of the model and its associated image calculation.     

基于电容层析成像和模糊模式识别的油气两相流流型辨识新方法的研究

基于电容层析成像和模糊模式识别技术别提出了一种油气两相流流型辨识的新方法。建立了12电极电容层析成像流型自动识别系统，该系统利用Tikhonov正则化原理并结合SIRT(Simultaneous Reconstruction Techniques)算法进行图像重建。Tikhonov正则化原理用于克服图像重建过程中的不适定问题，SIRT算法用于提高最终重建图像的质量。根据流型的随机和模糊特性，提出了一种根据管截面重建图像进行流型辨识的模糊流型判别方法。研究结果表明，提出的流型辨识新方法是有效的。对于层状流、核心流、环状流、均相流等流型，流型辨识的准确率高于95％，辨识一个流型所用的时间小于0.3秒。对于塞状流，流型辨识的准确率高于90％。     

基于小波多分辨分析和最小交叉熵的快速CT图像重建算法

引言 两相流系统广泛存在于电力、石油、化工等现代工业生产中,两相流流型的在线识别是实现两相流在线监控的前提,也是两相流测量中重要的研究方向[1-3].     

非规则结构电容层析成像填补法测量的敏感场特性及重构算法改进

利用填补的方式将不规则区域转为规则成像区,可拓展电容层析成像的测量范围。首先分析了填补法对应的敏感场特性,然后分别研究了敏感场、归一化电容、填充区介电常数和图像重构算法对填补法成像的影响。并利用特定的填充区边界条件对Landweber算法进行改进,提出两种改进方法:变换矩阵和拆分矩阵的Landweber图像重构方法。结果表明,填充区使电容信号绝对值增强,非线性程度增加。不同背景的敏感场对成像影响较小,但填充区充满时的敏感场可强化环状流成像。去除填充区噪声的归一化电容矩阵有利于图像重构。填充物质的介电常数与测量区高介电常数相近时成像最佳。拆分矩阵法降低了求解维度,缩短了计算时间,对各种流型成像整体效果较好,基于松弛因子改进还可进一步优化重构图像。最后,在一般非规则结构中对改进方法的可靠性进行了测试验证。研究表明,改进方法可以用于一般性不规则结构的相分布测量,未来还可以用于因高温、腐蚀等无法直接进行电极安装的工业场合中,具有改进多相流测量方法的普遍意义。