Three-dimensional (3D) Lagrangian Particle Tracking (LPT) was performed on a subsonic jet flow at Mach 0.506 and 0.845 generated by a round nozzle with diameter-based Reynolds numbers of 1.7 $\times$ ${10}^5$ and 3.1 $\times$ ${10}^5$, respectively. The Multi-Pulse Shake-The-Box (MP-STB) technique was employed to reconstruct particle tracks along the four-pulse sequences, which were obtained by using orthogonally polarised light to separate the pulses on camera images. The MP-STB method applied here has a number of differences compared to previous publications, in particular, a new adaptive search radii approach, and an iterative strategy with particle track validation criteria customised for high subsonic/transonic flows. A description of this methodology is given followed by presentation of the instantaneous 3D flow velocity and material acceleration particle tracks. By ensemble-averaging the scattered instantaneous measurements extracted from individual particle tracks into small volumetric bins, highly resolved statistical quantities were obtained. The performance of MP-STB was assessed by comparing velocity profiles with published particle image velocimetry (PIV) data-sets. MP-STB was found to be able to better resolve the steep velocity gradients, in particular the thin jet shear layer near the nozzle exit. At this location the MP-STB results also yielded higher turbulence intensities compared with the reported studies for similar flow conditions. The MP-STB acceleration flow statistics were compared for the two Mach numbers, and for the Mach 0.506 case, higher levels of normalised acceleration and fluctuations were found. The position accuracy of the 3D imaging system was quantified and it was found that the use of two different states of polarisation had a direct impact on the accuracy and the amount successfully tracked particles. Further assessment of the particle imaging quality of each camera revealed a significant disparity between cameras. This was attributed to the particle light scattering intensity variations, which were highly dependent on the particle size, camera angles and different states of polarised light. Despite these challenges, an average of 40,000 individual particle tracks could be reconstructed from a typical particle image density of 0.02 particles per px (and an active sensor area of 1800 $\times$ 2200 px²). Furthermore, the accuracy of the measurement was shown to be relatively high, with respect to PIV.

对由基于直径的雷诺数为1.7的圆形喷嘴产生的0.506和0.845马赫的亚音速射流进行了三维（3D）拉格朗日粒子跟踪（LPT） $\times$ ${10}^5$ 和3.1 $\times$ ${10}^5$， 分别。采用多脉冲摇动盒（MP-STB）技术沿四个脉冲序列重建粒子轨道，这些粒子轨道是通过使用正交偏振光分离相机图像上的脉冲而获得的。与以前的出版物相比，此处使用的MP-STB方法有许多差异，特别是新的自适应搜索半径方法，以及针对高音速/跨音速流定制的具有粒子轨迹验证标准的迭代策略。给出了此方法的描述，然后给出了瞬时3D流速和材料加速度粒子轨迹。通过对从单个粒子轨迹提取的分散的瞬时测量值进行集合平均，以得到较小的体积箱，可以获得高度解析的统计量。MP-STB的性能通过将速度曲线与已发布的粒子图像测速（PIV）数据集进行比较来评估。发现MP-STB能够更好地解决陡峭的速度梯度，特别是喷嘴出口附近的薄射流剪切层。与类似流量条件下的报道相比，在该位置，MP-STB结果还产生了更高的湍流强度。比较了两个马赫数的MP-STB加速度流量统计数据，对于0.506马赫数的情况，发现了更高水平的归一化加速度和波动。对3D成像系统的位置精度进行了量化，发现使用两种不同的偏振态对精度和成功跟踪的粒子数量有直接影响。对每个摄像机的粒子成像质量进行进一步评估后，发现摄像机之间存在巨大差异。这归因于颗粒光散射强度的变化，其高度依赖于颗粒大小，相机角度和偏振光的不同状态。尽管存在这些挑战，但仍可以从每px 0.02个粒子的典型粒子图像密度平均重构40,000个单独的粒子轨迹（有效传感器区域为1800个 $\times$ 2200像素²）。此外，相对于PIV，测量精度显示出较高。