Emission and capture characteristics of a deep hole trap (H1) in n-GaN Schottky barrier diodes (SBDs) have been investigated by optical deep level transient spectroscopy (ODLTS). Activation energy (E_emi) and capture cross-section (σ_p) of H1 are determined to be 0.75 eV and 4.67 × 10^?15 cm², respectively. Distribution of apparent trap concentration in space charge region is demonstrated. Temperature-enhanced emission process is revealed by decrease of emission time constant. Electric-field-boosted trap emission kinetics are analyzed by the Poole?Frenkel emission (PFE) model. In addition, H1 shows point defect capture properties and temperature-enhanced capture kinetics. Taking both hole capture and emission processes into account during laser beam incidence, H1 features a trap concentration of 2.67 × 10¹⁵ cm^?3. The method and obtained results may facilitate understanding of minority carrier trap properties in wide bandgap semiconductor material and can be applied for device reliability assessment.

In this work, a novel one-time-programmable memory unit based on a Schottky-type p-GaN diode is proposed. During the programming process, the junction switches from a high-resistance state to a low-resistance state through Schottky junction breakdown, and the state is permanently preserved. The memory unit features a current ratio of more than 10³, a read voltage window of 6 V, a programming time of less than 10^?4 s, a stability of more than 10⁸ read cycles, and a lifetime of far more than 10 years. Besides, the fabrication of the device is fully compatible with commercial Si-based GaN process platforms, which is of great significance for the realization of low-cost read-only memory in all-GaN integration.

面向核电磁脉冲等强电磁环境下复合材料壳体平台的电磁环境效应分析需求，根据Maxwell-Ampere定理的积分形式，分析得到了时域有限差分方法在处理弱导电薄层介质材料参数时的等效计算方法，即当介质等效波长远大于模型厚度时，可将薄层模型适当增厚，同时等比例减小其电导率，参数等效前后模型的电磁耦合特性基本相同。该方法通过等效增厚薄层材料从而实现增大空间离散步长，减少网格量的目的，不需要改变传统时域有限差分方法的时间步进格式，不会破坏计算的稳定性。无限大有耗介质薄板、薄层球体、含薄层壳体无人机电磁耦合等算例表明，在包含毫米级厚度弱导电介质薄层壳体平台的核电磁脉冲耦合模拟中，该方法具有较好的适用性。

针对光频梳下变频信号接收过程中存在的频率信息丢失问题，提出了一种基于频率-相位映射的信号频率恢复方法，该方法使用可调光纤延迟线在两路光频梳变频链路之间产生一组固定已知的时延，时延在信号原始频率与下变频信号相位差之间建立映射关系，利用该映射关系可以从测得的相位差计算出信号的原始频率。分析了时延值等参数对频率恢复的影响，估计了该方法对相位测量不确定度的限值要求，最后给出了该方法具体实施方案中关键参数的设置策略。所有下变频信号的相位差可以通过快速傅里叶变换等数据处理一次性得出，因此该方法的时间代价和计算成本几乎不随着信号个数增加而增加。在不考虑下变频信号混叠的情况下，本文所提出的方法在理论上对处理信号的数量没有限制，因此相比于已有的光频梳下变频信号频率恢复方法，在多信号频率恢复方面更具有优势。

针对复杂电磁环境和复杂大型系统电磁干扰测量参数、测量点位及测量工况多的特点，采用传统扫频式频域测量方法具有代价高、耗时长等现实问题，提出了一种低频电磁干扰的多通道时域快速测量与信号计算方法，并研制出低频电磁干扰多通道时域快速测量系统，实验验证表明提出的测量、计算方法和研制出的测量系统可以准确得到复杂电磁环境和大型系统的低频电磁干扰特性，且测量速度快、成本低。

针对低信噪比条件下，现有的雷达辐射源信号识别方法存在识别正确率低、时效性差的问题，提出了一种基于压缩残差网络的雷达辐射源信号识别方法。首先，利用Choi-Williams分布的时频分析方法将时域信号转换为二维时频图像；然后，根据应用场景特点，选择卷积神经网络（Convolutional Neural Networks, CNN）“压缩”范围；最后，构建压缩残差网络来自动提取图像特征并完成分类。仿真实验结果表明，在同等体量的设计下，与当前较为常用的标准CNN以及ResNet模型相比，所提模型能够降低信号识别运行时间约88%，在信噪比为−14 dB条件下对14种雷达辐射源信号的平均识别率高约5%。提供了一种高效的雷达辐射源信号智能识别方法，具有潜在的工程应用前景。