基于流化床气固两相流型图像的多参数检测方法

Because of the complexity of the distribution of the interface, gas-solid two-phase flow are known as the "capricious fluid",and currently on its relevant parameters testing is still in the stage of exploration. This subject will adopt high speed photography method combined with monocular stereo vision camera to obtain the 2D/3D images of fluidized bed of gas-solid two-phase flow. This subject will use digital image processing techniques , wavelet theory, fractal theory and nonlinear system theory to extract the characteristics of the images of flow patterns, combine with pattern recognition theory will realize the intelligent identification of flow patterns;The theories were talked above will be combined with the optical flow analysis and computer vision theories to test the parameters such as the flow field, velocity field and air void volume of rare phase in conveying flow parameters of the gas-solid two-phase flow.Finally, rely on two phase fluid flow image can implement multiple flow parameters'testing.This subject is of important theoretical significance on the development of multiphase flow parameters testing technology and the study on the nonlinear dynamics of gas-solid two-phase flow. It also be of important practical significance on guiding many related equipments' (power plant of coal powder conveying, pulverized coal burning, all kinds of gas-solid two-phase mixer, the gas-solid separator, non-ferrous industry alumina conveying, food industry's flour transportation, construction industry's cement transportation, chemical industry's gas-solid reactor)design and safety, stable and efficiently operation.

气固两相流动由于其界面分布的复杂性，被称为"难测流体"，目前对其相关参数的检测尚处于探索阶段。本项目拟采用高速摄影法结合单目立体视觉镜头，可获得流化床气固两相流动的二维/三维图像，利用数字图像处理技术和小波、分形等非线性系统理论提取流型图像的特征，结合模式识别理论实现流型的智能识别；再结合光流分析法等计算机视觉理论，对气固两相流动的流场、速度场以及稀相输送中体积空隙率等流动参数进行检测。最后，仅依靠两相流体的流动图像就可实现多个流动参数的同时检测。本项目对发展多相流参数检测技术、气固两相非线性动力学等学科有重要的理论意义，对指导许多相关设备（电厂的煤粉输送、煤粉燃烧，各式气固两相混合器、气固分离器、有色行业的氧化铝输送，食品行业的面粉输送，建筑行业的水泥输送，化工行业的气固反应器）的设计和安全、稳定、高效运行具有重要的现实意义。

气固两相流动由于其界面分布的复杂性，被称为"难测流体"，目前对其相关参数的检测尚处于探索阶段。本项目采用高速摄影法结合单目立体视觉镜头，获得了流化床气固两相流动的二维/三维图像，运用数字图像处理技术分别提取出流化床气固两相流动五种典型流型图像的直方图统计特征、图像傅里叶变换纹理特征和图像小波分形特征，然后再分别使用BP 神经网络、遗传神经网络和概率神经网络进行训练，结果显示三种特征的整体识别率均高于90%，从而实现了流化床气固两相流的智能流型识别；在流场参数检测方面，将光流分析法引入到流化床气固两相流动的流场、速度场和涡量场的检测，克服了PIV法实验步骤繁琐以及PTV法跟踪目标形状要求高的弊端，在与传统的 MQD 互相关法进行相应对比后发现，应用光流分析法检测两相流的流场，具有计算速度快、检测精度高等优点，可进一步推广到参数在线检测中；运用数字图像处理技术，在实现颗粒边缘检测、颗粒标号的基础上，扩展了在稀相输送过程所能检测到的参数内容，实现了颗粒粒径、体积以及空隙率的检测，最终实现了基于流化床气固两相流型图像的多参数检测。. 本项目对发展多相流参数检测技术、气固两相非线性动力学等学科有重要的理论意义，对指导许多相关设备（电厂的煤粉输送、煤粉燃烧，各式气固两相混合器、气固分离器、有色行业的氧化铝输送，食品行业的面粉输送，建筑行业的水泥输送，化工行业的气固反应器）的设计和安全、稳定、高效运行具有重要的现实意义。