In this work, a two-step metal organic chemical vapor deposition (MOCVD) method was applied for growing β-Ga₂O₃ film on c-plane sapphire. Optimized buffer layer growth temperature (T_B) was found at 700 °C and the β-Ga₂O₃ film with full width at half maximum (FWHM) of 0.66° was achieved. A metal?semiconductor?metal (MSM) solar-blind photodetector (PD) was fabricated based on the β-Ga₂O₃ film. Ultrahigh responsivity of 1422 A/W @ 254 nm and photo-to-dark current ratio (PDCR) of 10⁶ at 10 V bias were obtained. The detectivity of 2.5 × 10¹⁵ Jones proved that the photodetector has outstanding performance in detecting weak signals. Moreover, the photodetector exhibited superior wavelength selectivity with rejection ratio (R_{250 nm}/R_{400 nm}) of 10⁵. These results indicate that the two-step method is a promising approach for preparation of high-quality β-Ga₂O₃ films for high-performance solar-blind photodetectors.

Expanding the optical communication band is one of the most effective methods of overcoming the nonlinear Shannon capacity limit of single fiber. In this study, GeSn resonance cavity enhanced (RCE) photodetectors (PDs) with an active layer Sn component of 9%–10.8% were designed and fabricated on an SOI substrate. The GeSn RCE PDs present a responsivity of 0.49 A/W at 2 μm and a 3-dB bandwidth of approximately 40 GHz at 2 μm. Consequently, Si-based 2 μm band optical communication with a transmission rate of 50 Gbps was demonstrated by using a GeSn RCE detector. This work demonstrates the considerable potential of the Si-based 2 μm band photonics in future high-speed and high-capacity optical communication.

为了给航空动平台激光通信的附面层效应校正系统提供验证条件，设计了一种基于液晶空间光调制器来模拟附面层效应的模拟器。首先，从几何光学的角度分析附面层效应，将其等效为负透镜。然后，利用计算机对不同飞行条件下附面层效应与等效负透镜焦距之间的关系进行数值分析仿真。之后在液晶空间光调制器上导入不同焦距透镜的相位灰度图以实现变焦透镜的功能，通过改变变焦透镜的焦距来模拟航空平台不同飞行条件下的附面层效应。最后通过实验验证模拟的准确性，在环境温度下，模拟附面层效应的液晶空间光调制器得到的光斑图像由红外相机拍摄，之后进行图像处理分析实际光斑大小，并与理论计算光斑大小比较，得出误差曲线图。研究结果显示，基于液晶空间光调制器模拟附面层效应引起散斑效应的理论光斑大小与实际光斑大小的均方根误差为0.043 75，验证了所提出的方法的可行性和有效性。

针对当前单次通过型数字束流位置探测器（BPM）有效采样点数少、数据信噪比低、测量分辨率差的问题，提出了一种基于“束流单次通过”的数字BPM测量改进算法。该算法通过模拟信号的功分-延迟-合成，增加了有效采样数据，并通过采样数据的截取与数据拼接、数字滤波等数据处理方式，提高了采样数据的信噪比，并最终实现了BPM测量分辨率的提高。实验结果表明，该算法在不改变模数转换器（ADC）采样率的情况下，将单次通过型BPM的测量分辨率提高了约2倍。该测量方法为提高单次通过型BPM的测量分辨率提供了一种新的解决方案，已成功应用于北京正负电子对撞机（BEPCII）和高能同步辐射光源（HEPS）直线加速器项目中。

针对无人机端目标检测中存在图像尺度变化大、目标尺寸小和无人机机载嵌入式计算资源有限的问题，提出一种应用于无人机平台轻量化的目标检测网络。该网络以YOLOv5作为基准模型，首先增加检测分支以处理尺度变化的问题；然后提出基于归一化Wasserstein距离与传统IOU混合的小目标检测度量方法，用于解决小目标检测精度低的问题；随后提出FasterNet与C3融合的C3_FN轻量化网络结构，降低网络计算量，使其更适合无人机平台使用。最后将算法分别在仿真平台与嵌入式平台上利用无人机目标检测数据集VisDrone进行性能测试。仿真平台上的测试结果表明，本文提出的网络相较于基准网络在mAP_0.5指标上提升了6.6%，mAP_0.5-0.95指标上提升了4.8%，推理时间仅需45.9 ms，对比其他主流的无人机目标检测网络具有更好的检测效果。在嵌入式设备NVIDIA Jetson Nano上的测试结果表明，本文算法能够在有限的硬件资源下获得高精度接近实时的检测性能。