【作者】 刘兼唐；

【导师】 赵敏；

【作者基本信息】 南京航空航天大学 ， 测试计量技术及仪器， 2020， 博士

【摘要】 在工业设备和复杂器件检测中,对腔体内壁情况的检测和密闭环境中流场的检测,现有的检测方法受到环境因素、器件结构、被测物材料等制约,都有一定的局限性,无法实现无扰、无损的高精度检测,这已成为迫切需要解决的问题。本文基于正电子湮没原理,面向工业检测的特点,对γ光子散射校正、γ光子3D图像重建等关键技术进行了研究,进而面向工业领域复杂器件内腔检测和致密材料密闭空间的流场检测两类检测需求,充分利用γ光子穿透力强、抗干扰、探测精度高等特点,对γ光子3D成像检测方法进行了系统深入的研究。针对工业检测中因被测物材质等原因造成γ光子散射严重的问题,本文以γ光子2D采集模式下的γ光子单次散射模型为基础,提出了优化的3D采集模式下的γ光子单次散射模型,并基于此模型,提出了遍历散射角的γ光子单次散射校正方法,从空间几何角度分析散射的γ光子运行路径,通过能量窗的设置,将能量窗对应散射角,进而确定散射的γ光子数目,可达到对γ光子散射准确校正的目的,通过仿真实验及与其他散射校正方法的比较,验证了该方法的校正效果和技术优势。针对工业检测中γ光子探测采样时间短影响成像质量的问题,本文提出了采用γ光子均分补偿的方法,基于OSEM算法迭代和利用OTSU算法进行目标增强的3D图像重建方法,并借助GATE仿真平台进行图像重建实验验证成像效果。同时,本文将深度学习理论引入到图像重建中,选用卷积神经网络,采用迁移学习实现快速采样的小样本γ光子图像的深度学习,进而将深度学习的识别与图像重建结合起来,采用γ光子成像特征算子进行超分辨率图像重建,实现对γ光子3D图像超分辨率显示,为提高成像质量探索新的技术途径。针对工业复杂器件内腔缺陷检测和定位问题,本文基于正电子湮没和放射性核素衰变原理,提出通过选择合适的放射性核素,采用化学方法制备正电子源试剂,将放射性核素溶液注入器件内腔后,采用γ光子阵列探测器环来记录湮没事件,最后通过数学建模及3D图像重建算法实现器件内壁及缺陷的3D成像检测。本文搭建了激发正电子、探测γ光子进行3D成像的实验系统,通过对不同材料和结构的试验件进行3D成像检测实验,验证了γ光子3D成像检测方法的可行性和检测精度。通过与其他成像检测方法的技术比较及成像比对实验,验证了γ光子3D成像检测方法的技术优势。针对工业领域致密材料密闭空间的流场检测问题,本文提出了运用γ光子3D成像技术,通过建立γ光子成像与流场场态的映射关系进行流场检测的方法。通过将正电子核素标记在工作介质中,在工作场态下进行γ光子探测与3D成像检测。本文设计了小球漂流和喷雾场模拟两个实验,验证了γ光子3D成像在流场检测中的可行性。本文模拟工业燃烧场景,设计了燃烧流场的检测实验,获得了场态分布情况,并进行了CFD仿真比对,验证了检测结果,为工业设备中的密闭管道以及容器、腔室等内部的流场状态进行无扰、无损检测和成像提供了理论依据。本文的研究工作,为γ光子3D成像在工业领域的应用提供了技术支撑,为解决工业领域,特别是高端装备制造领域关键部件检测及设备工艺流程监测提供了新的无损检测方法,并为开发工业用γ光子探测成像仪器提供理论依据。更多还原

【Abstract】 In the detection of industrial equipment and complex devices,the detection method of the inner wall of the chamber and the detection method of the flow field in a closed environment are subject to factors including but not limited to environment,device structure and materials,thus undisturbed and nondestructive high-precision detection results are hard to achieve.This has become an urgent problem to be solved.Based on the principle of positron annihilation and the characteristics of industrial testing,this paper investigated the key techniques regarding scattering correction and 3D image reconstruction in detection using γ-photon,in order to meet the testing requirements for two types of detection,i.e.,intra-cavity detection of complex devices in the industrial field and flow-field detection of dense materials in confined space.The 3D imaging detection method using γ-photon was systematically investigated by taking full advantages of characteristics of γ-photon,such as strong penetration,antiinterference,high detection precision,etc.In view of the problem that the materials of the measured object cause serious γ-photon scattering issues in industrial testing,in this paper,based on the γ-photon single scattering model in γ-photon 2D acquisition mode,an optimized γ-photon single scattering model in 3D acquisition mode was proposed,and based on this model,a γ-photon single scattering correction method that traverses the scattering angle was proposed.The running path of the scattered γ-photon was reverted to the real running path of the γ-photon from a spatial geometric angle.Through setting the energy window,the scattering angle of the γ-photon can be corresponding to the energy window.Then,the number of the scattered γ-photon can be determined and the γ-photon scattering can thus be accurately corrected.This γ-photon correction method can effectively improve the 3D imaging quality.Through simulation experiments and comparison with other scattering correction methods,the correction effect and technical advantages of this method are verified.To solve the problem that the short detection sampling time of γ-photon affects imaging quality,this paper introduced the 3D image reconstruction method that enhances the target by equally dividing and compensating γ-photon based on the OSEM iterative algorithm and the OTSU algorithm.The image effect was verified by image reconstruction experiment through GATE simulation platform.Meanwhile,deep learning theory was introduced into image reconstruction.The convolutional neural network was selected and transfer learning was used to achieve the few-shot learning of γ-photon image.Then,combining deep learning recognition with image reconstruction,the super-resolution image reconstruction was performed to realize the super-resolution display of γ-photon 3D images,which explores a new technical approach to improve the image quality.To detect and localize the defects of inner cavity of industrial complex devices,the proposed method in this paper followed the principle of positron annihilation and radionuclide decay.Through selecting appropriate radionuclide,the positron source reagent was prepared by chemical technique.After injecting the radionuclide solution into the inner cavity,the γ-photon array detector ring was used to record the annihilation events.Finally,the 3D image detection of the inner wall and defects of the device were achieved by mathematical model and 3D image reconstruction algorithm.In this paper,an experimental system was established to excite positrons and detect γ-photons for 3D imaging.Through conducting 3D image detection experiments on test pieces with different materials and structures,the feasibility,technical advantages and detection precision of 3D image detection method using γ-photon were verified.Also,the technical advantages of 3D image detection method using γ-photon were verified by comparing the 3D imaging detection method with other imaging detection methods.To detect the flow field in confined space with dense materials in industrial field,a γ-photon 3D imaging technology was proposed in this paper to detect flow field through establishing the mapping relationship between γ-photon imaging and flow field state.Through marking positron nuclides in the working medium,γ-photon was detected in the working field state,and then 3D imaging detection was conducted by using the volume rendering image.The ball rafting experiment and the spray experiment were designed to verify the feasibility of 3D imaging in flow field detection using γ-photon.The industrial combustion was simulated in this research.The experiment on combustion flow field detection was designed and conducted,and the distribution of field state was obtained.Moreover,CFD simulation results were compared.The detection results were verified,which provides a theoretical basis for undisturbed,non-destructive testing and imaging for closed pipelines,the inner flow field of containers,chambers,etc.in industrial equipment.The research work in this paper provides a technical support for the application of 3D imaging using γ-photon in the industrial field and presents a novel non-destructive detection method for key components and process flow monitoring in industry,especially in the high-end equipment manufacturing field,thus providing a theoretical basis for developing industrial image detection device using γ-photon.更多还原