High-Speed-Camera-System for 3D-Measurements of Flow Fields

将购买两台高速摄像机和配件。它们构成了可压缩流场的完整三维重建和湍流中的声源识别的关键部件。为此，数值方法是数据同化，它将直接数值计算的合成图像与高速摄像机的真实纹影图像之间的差异降至最低。有条不紊地，从两个不同的角度拍摄流动的时间分辨率图像是用相机拍摄的。例如，这些图像可以基于密度梯度或引入的颗粒或烟雾。所获得的数据得到了现有测量技术的补充，例如借助麦克风和热像仪。在进一步的步骤中，将执行基于N-S方程的相同过程的流动计算。为此，必须指定计算的初始条件和边界条件。这种流动取决于这些条件。特别是对于湍流进口，实际流型是非常独特的。然而，在计划的研究项目中，没有指定初始和边界条件，而是通过优化的方式来确定，以使基于流场和建模的光学装置的高速图像和合成图像之间的差异尽可能小。这个过程也被称为数据同化，并使用了类似的方法，最近在物理信息机器学习中获得了发展势头。工作组中只有一台摄像机和一个平面流动就可以进行相应的前期工作。借助这里建议采购的两台高速摄像机，可以调查复杂的三维流动。相机的速度可以调查跨音速流动和空气声学现象。优化的结果是完整的三维流动。资料同化后，所有流动变量均可用于进一步研究。因此，通常用于数值计算的所有分析都可以在实际实验中进行研究。具体地说，要执行流的模式分解。这将补充小组内部确定连贯湍流结构的工作。到目前为止，这只能在数字的基础上完成。将获得的摄像机将允许在实验中进行这项工作。

Two high-speed cameras and accessories are to be purchased. These form the key component for the complete three-dimensional reconstruction of compressible flow fields and for the identification of sound sources in turbulent flows. The numerical method for this is a data assimilation, which minimizes the difference between synthetic images from a direct numerical calculation and the real Schlieren images from the high-speed cameras. Methodically, time-resolved images of a flow from two different perspectives are to be taken with the cameras. These images can be based, for example, on density gradients or on introduced particles or smoke. The data obtained are complemented by existing measurement technology, for example with the help of microphones and a thermal camera. In a further step, flow calculations of the same process based on Navier-Stokes equations will be performed. For this purpose, initial and boundary conditions for the calculation have to be specified. The flow depends on these conditions. Especially for a turbulent inlet, the actual flow pattern is very unique. In the planned research projects, however, initial and boundary conditions are not specified, but are determined by optimization in such a way that the difference between the high-speed images and synthetic images based on the flow field and the modeled optical setup is as small as possible. This process is also called data assimilation and uses similar methods which recently acquire momentum in Physics-informed Machine Learning. Corresponding preliminary work with only one camera and a plane flow is available in the working group. With the help of the two high-speed cameras proposed for procurement here, complex, three-dimensional flows can be investigated. The speed of the cameras allows the investigation of transsonic flows and aeroacoustic phenomena. The result of the optimization is the complete, three-dimensional flow. All flow variables are available for further investigation after data assimilation. All analyses, which are usually reserved for numerical computation, can thus be investigated on the real experiment. In particular, a modal decomposition of the flow is to be performed. This will complement work within the group to determine coherent turbulent structures. So far, this can only be done on a numerical basis. The cameras to be acquired will allow this work to be carried out on the experiment instead.