A simple quasi-distributed fiber sensing interrogation system based on random speckles is proposed for weak fiber Bragg gratings (WFBGs) in this work. Without using tunable lasers or spectrometers, a piece of multimode fiber is applied to interrogate the WFBGs relying on the wavelength sensitivity of speckles. Instead of the CCD sensor, an InGaAs quadrant detector serves as the receiver to capture the fast-changing speckle patterns. A supervised deep learning algorithm of the multilayer perceptron architecture is implemented to process speckle data and to interrogate temperature changes or dynamic strains. The proposed demodulation system is experimentally demonstrated for WFBGs with 0.1% reflectivity. The experimental results demonstrate that the new system is capable of measuring temperature change with an accuracy of 1°C and achieving dynamic frequency of 100 Hz. This speckle-based interrogation system paves a new way for distributed WFBGs sensing with a simple design.

为研究飞秒激光加工硬脆透明材料时存在的“微裂纹”与“诱导条纹”等共性工艺问题，利用飞秒时间分辨泵浦探测阴影成像技术，对飞秒激光多脉冲烧蚀石英玻璃过程中的电子动力学过程进行成像，分析了激光脉冲电离材料初期（700 fs之前）等离子体丝的演化情况。多脉冲诱导微结构的存在使成丝区域分布在微结构的两侧与光脉冲传播的轴线方向，前者主要是由微结构侧壁对光脉冲的折射造成的，而后者则是由微结构底面与侧壁形貌不同导致的光程差引起的。实验结果揭示了多脉冲加工过程中脉冲串诱导微结构对后续光场的重塑效应，该效应影响了等离子体成丝区域与能量沉积的分布，这是共性工艺问题产生的核心机制。

激光诱导损伤阈值（LIDT）是光学元件发展中不可或缺的一项重要指标，提高其测量结果的准确性仍是当前人们致力于研究的方向。基于蒙特卡罗法提出了一种损伤测试点优化分配方法，以提高LIDT拟合结果的准确性。针对测试样品有限的辐照测试面积及辐照光斑大小，模拟了一种非线性简并缺陷损伤模型，对不同通量水平下测试点变化对拟合LIDT结果的影响进行了敏感性分析。根据设定的损伤模型参数建立模型生成相关损伤数据，通过控制变量法对每次指定通量水平处的测试点数进行变更，在其余通量处测试点数不变的情况下，采用蒙特卡罗法对所有损伤数据进行多次模拟计算，绘制拟合结果均方根误差和测试点的关系曲线图。计算其相应测试点数对损伤阈值拟合结果标准差的敏感性。从而以此敏感性为权重对各通量下的测试点进行更合理的分配。结果表明，该敏感性权重法的拟合结果的标准差为0.272 J/cm²，相比于标准平均分配方法的标准差0.395 J/cm²减小了约31%。

精准的大气能见度预报对空气污染治理、保障公共交通安全等方面具有重要意义。基于2019年12月1日至2020年9月23日国家气象信息中心观测的大气能见度站点数据，分别采用ConvLSTM模型和PredRNN模型对中国中东部地区的能见度进行12 h预报，并对这两种模型的预报结果进行评价。试验表明，PredRNN模型相对于经典的ConvLSTM模型在大气能见度预报、图像质量评价指标和预报指标上都有更好的表现。此外，分析还表明，相对于ConvLSTM模型，PredRNN模型对4000 m中等级别雾区预报效果随时间延长有明显提升。

该文针对新工科软件专业图像分类任务的新生项目课程存在的3个问题进行分析设计，包括Python语言不熟悉，图像分类入门时间短以及实操环境学生难以搭建，提出了基于百度的AI Studio平台和以PaddleHub预训练模型的应用为中心的课程教学设计，包括融合线性代数的实际应用，预训练模型fine-tune实践以及组队分工线上线下结合学习等，后续可以改进为由学生自主选择设计识别任务内容。该课程设计内容有助于学生奠定工程基础，提高专业技能，培养团队合作能力，最终达到增强专业兴趣和专业信心的课程设计目标。

针对Pillar编码点云的三维目标检测算法中存在一定细粒度信息的丢失以及对点云特征提取能力不足等问题，基于PointPillars提出一种融合逐点空间注意力机制和跨阶段局部网络的三维目标检测算法。首先在支柱特征网络层中融入逐点空间注意力机制，增强网络对局部几何信息的提取并保留深层次信息，使得到的关键特征更适合检测任务；其次将对点云伪图像进行高维特征提取的降采样模块中的普通卷积替换为跨阶段局部网络，进一步提升网络的学习能力；最后算法在高速公路的应用场景下，以KITTI数据集中car类作为检测目标，与基准网络相比，在简单、中等和困难三种情况下的3D检测精度分别提高了2.23%、2.25%和2.30%。实验结果表明，所提算法在检测性能上有明显提升，同时检测速度达到实时检测水平，对自动驾驶技术的优化和完善具有一定的积极意义。

氨具有低成本、易液化和高体积能量密度等特点, 是一种有吸引力的无碳燃料, 用其制成的直接氨燃料电池也备受科研人员青睐, 却受限于阳极氨氧化缓慢的动力学过程。本工作采用无表面活性剂的简单方法通过纳米颗粒(NPs)自组装制备了三维多孔网络结构的PtIr合金气凝胶高效氨氧化催化剂。该结构提供了丰富开放的互联质子传输通道和额外的催化活性位点, 有助于氨电催化氧化中NH₃分子的去质子化过程。当Pt和Ir物质的量比为80/20时, PtIr合金气凝胶展现出最优的氨氧化(AOR)催化活性。实验通过研究NH₃浓度和工作温度对催化剂氨氧化性能的影响发现, Pt₈₀Ir₂₀合金气凝胶催化剂的AOR性能随着氨水浓度或温度的上升而增强, 如当氨水浓度为0.5 mol/L时, 合金催化剂在0.5 V电位下的质量比活性为44.03 A·g^-1, 约是0.05 mol/L 氨水中的4倍。当温度上升至80 ℃ 时, 合金催化剂在0.5 V电位下的质量比活性为148.73 A·g^-1, 约为25 ℃下的12倍。在此温度变化区间, 其AOR起始电位下降了40 mV。利用Arrhenius方程计算发现, Pt₈₀Ir₂₀合金气凝胶催化剂的AOR反应活化能比商业Pt/C催化剂降低了~9.43 kJ·mol^-1。此外, 催化材料的稳定性测试结果表明, Pt₈₀Ir₂₀合金气凝胶催化剂经2000次循环伏安测试后的峰值质量比活性损失为~50.6%, 优于商业Pt/C催化剂(~74.9%)。

We propose an ultra-broadband near- to mid-infrared (NMIR) tunable absorber based on VO₂ hybrid multi-layer nanostructure by hybrid integration of the upper and the lower parts. The upper part is composed of VO₂ nanocylinder arrays prepared on the front illuminated surface of quartz substrate, and VO₂ square films and VO₂/SiO₂/VO₂ square nanopillar arrays prepared on the back surface. The lower part is an array of SiO₂/Ti/VO₂ nanopillars on Ti substrate. The effects of different structural parameters and temperature on the absorption spectra were analyzed by the finite-difference time-domain method. An average absorption rate of up to 94.7% and an ultra-wide bandwidth of 6.5 μm were achieved in NMIR 1.5–8 μm. Neither vertical incident light with different polarization angles nor large inclination incident light has a significant effect on the absorption performance of the absorber. The ultra-broadband high absorption performance of this absorber will be widely used in NMIR photodetectors and other new optoelectronic devices.

Scintillators are the vital component in X-ray perspective image technology that is applied in medical imaging, industrial nondestructive testing, and safety testing. But the high cost and small size of single-crystal commercialized scintillators limit their practical application. Here, a series of Tb³⁺-doped borosilicate glass (BSG) scintillators with big production size, low cost, and high spatial resolution are designed and fabricated. The structural, photoluminescent, and scintillant properties are systematically investigated. Benefiting from excellent transmittance (87% at 600 nm), high interquantum efficiency (60.7%), and high X-ray excited luminescence (217% of Bi₄Ge₃O₁₂), the optimal sample shows superhigh spatial resolution (exceeding 20 lp/mm). This research suggests that Tb³⁺-doped BSG scintillators have potential applications in the static X-ray imaging field.

With rapid advancement and deep integration of artificial intelligence and the internet-of-things, artificial intelligence of things has emerged as a promising technology changing people’s daily life. Massive growth of data generated from the devices challenges the AIoT systems from information collection, storage, processing and communication. In the review, we introduce volatile threshold switching memristors, which can be roughly classified into three types: metallic conductive filament-based TS devices, amorphous chalcogenide-based ovonic threshold switching devices, and metal-insulator transition based TS devices. They play important roles in high-density storage, energy efficient computing and hardware security for AIoT systems. Firstly, a brief introduction is exhibited to describe the categories (materials and characteristics) of volatile TS devices. And then, switching mechanisms of the three types of TS devices are discussed and systematically summarized. After that, attention is focused on the applications in 3D cross-point memory technology with high storage-density, efficient neuromorphic computing, hardware security (true random number generators and physical unclonable functions), and others (steep subthreshold slope transistor, logic devices,etc.). Finally, the major challenges and future outlook of volatile threshold switching memristors are presented.