飞秒激光加工技术主要是利用激光焦点对材料进行微区去除，结合加工路径的精准规划和激光参数的精确调控，实现各种功能微结构的精密加工。然而，在实际加工过程中，材料表面并非都是理想平面，这引起激光焦点与材料表面的相对距离发生变化，导致材料表面接收的激光焦斑尺寸不一致，进而造成飞秒激光加工的微结构不均匀，最终不满足某些应用场景的实际需求。针对该问题，提出了基于分区域平面拟合和二维插值的两种校正方法，即在待加工区域内以少量采样点近似描述材料表面起伏形貌，并以此为依据校正加工路径的高度坐标，使飞秒激光加工过程中激光焦点和材料表面的相对距离控制在不影响加工效果的范围内。试验结果表明，这两种校正方法都能保证飞秒激光加工大面积微结构的均匀性和一致性，是解决非平整表面不易实现高品质微结构加工的有效方法。

为探索可靠的激光雷达回波处理方式及研究大气对回波相位的影响，进行了激光雷达回波相干性分析。首先通过建立激光雷达时域模型及搭建实验系统获取仿真和实验大气回波。为排除系统误差的干扰对实验回波提取相位并脉间对齐，再基于归一化协方差函数计算仿真回波和对齐后实验回波的相关系数，发现回波脉间不相干。在证明出射脉冲的高相干性的基础上，硬目标回波的脉间积累可以实现本征频率的增强，而对齐后的风场回波脉间积累的效果远差于非相干平均，表明硬目标的漫反射能保留本征信号的相干性，而风场这类软目标严重破坏回波的相干性。

Covalent adaptable network (CAN) polymers doped with conductive nanoparticles are an ideal candidate to create reshapeable, rehealable, and fully recyclable electronics. On the other hand, 3D printing as a deterministic manufacturing method has a significant potential to fabricate electronics with low cost and high design freedom. In this paper, we incorporate a conductive composite consisting of polyimine CAN and multi-wall carbon nanotubes into direct-ink-writing 3D printing to create polymeric sensors with outstanding reshaping, repairing, and recycling capabilities. The developed printable ink exhibits good printability, conductivity, and recyclability. The conductivity of printed polyimine composites is investigated at different temperatures and deformation strain levels. Their shape-reforming and Joule heating-induced interfacial welding effects are demonstrated and characterized. Finally, a temperature sensor is 3D printed with defined patterns of conductive pathways, which can be easily mounted onto 3D surfaces, repaired after damage, and recycled using solvents. The sensing capability of printed sensors is maintained after the repairing and recycling. Overall, the 3D printed reshapeable, rehealable, and recyclable sensors possess complex geometry and extend service life, which assist in the development of polymer-based electronics toward broad and sustainable applications.

针对遥感图像中目标尺度差异较大和方向分布随机等导致检测精度较低的问题，提出一种基于稀疏掩模Transformer的遥感目标检测方法。该方法以Transformer网络为基础，首先引入角度参量，使其适应遥感目标的旋转特性；其次在特征提取部分以多层级特征金字塔为输入，以应对遥感图像目标尺寸变化大的特点，提高对不同尺度目标的检测效果，尤其对小目标的检测效果提升明显；最后以稀疏-插值注意力模块代替自注意力模块，有效缓解了Transformer网络检测高分辨遥感图像时计算量大的缺陷，并且加快了网络的收敛速度。在大型遥感数据集DOTA上的实验结果表明，所提方法的平均检测精度为78.43%，检测速度为12.5 frame/s，与基准方法相比，平均精度均值（mAP）提高了3.07个百分点，证明了所提方法的有效性。

In order to minimize the initial energy storage of tens of MA-class Z-pinch accelerators, an intelligent optimization method was developed based on the transmission line code circuit model and PSOGSA algorithm. Using several input parameters, the four overall parameters of the Z-pinch accelerator could be fast determined, including the connection and parallel combination of LTD cavities, the outer radius of the stack-MITL system, and electrical length of monolithic radial transmission lines. The optimization method has been verified by comparing the results with the Z-300 and Z-800 conceptual designs. By means of this intelligent optimization, some factors that affect the initial energy storage on high-current Z-pinch accelerators have been investigated, such as the operating electrical fields, the diameter of the stack-MITL system, and the inner diameter of the LTD cavity. The suggestions for designing relatively low-cost, efficient LTD-based accelerators have been proposed.

针对遥感图像目标广邻域稀疏、多邻域聚集、方向多样等特性导致检测难度大的问题，提出了一种基于稀疏Transformer的遥感旋转目标检测方法。首先，所提方法在典型端到端Transformer网络的基础上，根据遥感图像的特性，利用K-means算法实现多域聚集，从而更好提取稀疏域下的目标特征；其次，为适配旋转目标的基本属性，在边框生成阶段，利用目标包围框的中心点及边框特征学习的策略高效获取目标回归斜边框；最后，为提升网络对遥感目标的检测率，对网络的损失函数进行了优化。在DOTA和UCAS-AOD遥感数据集上的实验结果表明，所提方法的平均精度分别为72.87%和90.4%，能很好地适应遥感图像中各类旋转目标的形状与分布特性。

利用Nd3+与Yb3+之间的能量传递, 设计了808 nm激发下具有近红外光致发光的GdF3∶Nd3+,Yb3+@NaGdF4磁性纳米探针。通过合成参数调控, 研究了反应时间和前驱物浓度对GdF3∶Nd3+,Yb3+材料形貌的影响。GdF3∶Nd3+,Yb3+在808 nm激发下在近红外(NIR)第一/第二窗口具有较宽的发射峰(970~1 070 nm), 共掺杂体系实现了Yb3+和Nd3+双发光中心发光, 有效提高了近红外区间的发光性能, 同时克服了用980 nm激发的荧光成像引起生物组织过热的缺点。表面包覆NaGdF4惰性壳可以进一步提高GdF3∶Nd3+,Yb3+的近红外发光强度, 研究了不同厚度NaGdF4惰性壳对GdF3∶Nd3+,Yb3+发光增强的影响。由于Gd3+具有的顺磁特性, GdF3∶Nd3+,Yb3+@NaGdF4显示出顺磁性能, 磁化值为0.45 A·m2/kg, 适合作为磁共振成像(MRI)T1探针。GdF3∶Nd3+,Yb3+@NaGdF4纳米颗粒在NIR荧光成像和MRI领域具有广阔的应用前景。