基于VMD-ISSA-LSTM的短时交通流预测研究

针对城市短时交通流随机波动性强、可靠性低、预测精度差等问题,将变分模态分解（VariationalMode Decomposition,VMD）和改进麻雀搜索算法（ImproveSparrowSearchAlgorithm,ISSA）与长短期记忆（LongShort-Term Memory, LSTM）神经网络相结合,建立一种短时交通流预测模型（VMD-ISSA-LSTM）。首先利用VMD对历史原始交通流数据进行分解;然后采用佳点集、正弦函数扰动和Tent混沌映射等策略对标准的SSA算法加以改进,增强ISSA算法的寻优能力;最后,将每个分量送入ISSA-LSTM中进行预测,同时将预测结果线性叠加,得到交通流量预测值。以上海市中山北路-曹杨路口2018年11月1日—30日的历史交通数据对模型进行验证。结果表明,与LSTM、VMD-LSTM、VMD-SSA-LSTM等传统预测模型相比,VMD-ISSA-LSTM模型的预测结果的平均绝对百分比误差为1.278 4%,能够更好地应用于短时交通流预测中。     

New 4H silicon carbide metal semiconductor field-effect transistor with a buffer layer between the gate and the channel layer

A new 4H silicon carbide metal semiconductor field-effect transistor (4H-SiC MESFET) structure with a buffer layer between the gate and the channel layer is proposed in this paper for high power microwave applications. The physics-based analytical models for calculating the performance of the proposed device are obtained by solving one- and two-dimensional Poisson's equations. In the models, we take into account not only two regions under the gate but also a third high field region between the gate and the drain which is usually omitted. The direct-current and the alternating-current performances for the proposed 4H-SiC MESFET with a buffer layer of 0.2 μ m are calculated. The calculated results are in good agreement with the experimental data. The current is larger than that of the conventional structure. The cutoff frequency (f_T) and the maximum oscillation frequency (f_max) are 20.4 GHz and 101.6 GHz, respectively, which are higher than 7.8 GHz and 45.3 GHz of the conventional structure. Therefore, the proposed 4H-SiC MESFET structure has better power and microwave performances than the conventional structure.     

Macro-performance of multilayered thermoelectric medium

The effective properties of thermoelectric composites are well known to depend on boundary conditions, which causes the macro performance of thermoelectric composite to be difficult to assess. The overall macro-performance of multilayered thermoelectric medium is discussed in this paper. The analytical solutions are obtained, including the heat flux, temperature,electric potential, and the overall energy conversion efficiency. The results show that there are unique relationships between the temperature/electric potential and the electric current/energy flux in the material, and whether the material is independent of or embedded in thermoelectric composites. Besides, the Peltier effect at the interface can significantly improve the overall energy conversion efficiency of thermoelectric composites. These results provide a powerful tool to analyze the effective behaviors of thermoelectric composites.     

Two-dimensional numerical computation of the structure-dependent spectral response in a 4H-SiC metal-semiconductor-metal ultraviolet photodetector with consideration of reflection and absorption on contact electrodes

A two-dimensional model of a 4H-SiC metal-semiconductor-metal(MSM) ultraviolet photodetector has been established using a self-consistent numerical calculation method.The structure-dependent spectral response of a 4H-SiC MSM detector is calculated by solving Poisson’s equation,the current continuity equation and the current density equation.The calculated results are verified with experimental data.With consideration of the reflection and absorption on the metal contacts,a detailed study involving various electrode heights(H),spacings (S) and widths(W) reveals conclusive results in device design.The mechanisms responsible for variations of responsivity with those parameters are analyzed.The findings show that responsivity is inversely proportional to electrode height and is enhanced with an increase of electrode spacing and width.In addition,the ultraviolet (UV)-to-visible rejection ratio is > 103.By optimizing the device structure at 10 V bias,a responsivity as high as 180.056 mA/W,a comparable quantum efficiency of 77.93%and a maximum UV-to-visible rejection ratio of 1875 are achieved with a detector size of H = 50 nm,S =9μm and W = 3μm.     

基于激光诱导击穿光谱技术的咖啡豆中咖啡因含量快速检测方法

应用激光诱导击穿光谱(LIBS)技术研究了快速检测咖啡豆中咖啡因含量的可行性。将咖啡豆磨粉压成片状作为采集LIBS光谱数据的样本,应用原子吸收分光光度计测量每个样本中咖啡因的含量。应用基线校正,小波变换和归一化等数据预处理方法;针对基于全部变量的偏最小二乘(PLS)模型会出现过拟合,分别应用回归系数和主成分分析(PCA)选择特征变量,并建立了基于特征变量的PLS和BP神经网络模型。结果表明：基于回归系数所选特征变量的PLS模型中,建模集相关系数R_c=0.96,预测集R_p=0.91;基于PCA提取特征变量的PLS模型中,R_c=0.94,R_p=0.90;基于PCA所选特征变量的BP神经网络模型中,R_c=0.96,R_p=0.96。两种方法所提取特征变量均对应C,H,O,N,Na,Mn,Mg,Ca和Fe,且基于上述两种方法所选特征变量的PLS模型均对预测集样本有较好的预测结果,说明上述元素与咖啡因含量存在联系,应用回归系数和PCA选择的特征变量是有效的,但是咖啡豆内C,H,O,N,Na,Mn,Mg,Ca,Fe与咖啡因含量的确切关系需要进一步研究。基于PCA所选特征变量的BP神经网络模型有更优的预测结果,说明所选特征变量适用于不同的建模方法。研究表明LIBS技术结合化学计量学方法可以实现咖啡豆中咖啡因含量的快速检测。     

Modeling of the drain-induced barrier lowering effect and optimization for a dual-channel 4H silicon carbide metal semiconductor field effect transistor

A new analytical model to describe the drain-induced barrier lowering(DIBL) effect has been obtained by solving the two-dimensional(2D) Poisson’s equation for the dual-channel 4H-SiC MESFET(DCFET).Using this analytical model,we calculate the threshold voltage shift and the sub-threshold slope factor of the DCFET,which characterize the DIBL effect.The results show that they are significantly dependent on the drain bias,gate length as well as the thickness and doping concentration of the two channel layers.Based on this analytical model,the structure parameters of the DCFET have been optimized in order to suppress the DIBL effect and improve the performance.     

Microwave damage susceptibility trend of a bipolar transistor as a function of frequency

We conduct a theoretical study of the damage susceptibility trend of a typical bipolar transistor induced by a high-power microwave (HPM) as a function of frequency. The dependences of the burnout time and the damage power on the signal frequency are obtained. Studies of the internal damage process and the mechanism of the device are carried out from the variation analysis of the distribution of the electric field, current density, and temperature. The investigation shows that the burnout time linearly depends on the signal frequency. The current density and the electric field at the damage position decrease with increasing frequency. Meanwhile, the temperature elevation occurs in the area between the p-n junction and the n-n + interface due to the increase of the electric field. Adopting the data analysis software, the relationship between the damage power and frequency is obtained. Moreover, the thickness of the substrate has a significant effect on the burnout time.     

基于改进陷阱模型的新型p型隔离层结合场板结构的4H-SiC MESFET的特性研究与优化

本文提出了一种致力于抑制表面陷阱影响的新型结构。基于能准确表征4H-SiC材料特性的物理模型和经过实验证明能较好表征表面陷阱作用机理的模型,对器件的特性进行了研究。通过与实际器件制作中主流采用的埋栅-场板结构的4H-SiC MESFET以及实验测试的器件特性的对比,本文提出的结构在整体上对器件的特性有所提高 。新结构引入的 p型隔离层能有效地抑制表面陷阱的影响并且能减小器件在大电压微波应用条件下的栅漏电容;P型隔离层结合场板结构改善了栅极边缘的电场分布,同时能减小单一使用场板结构时场板对沟道引入的附加栅漏电容; 作为微波晶体管,由于更好的抑制了表面陷阱,基于本文提出的结构的4H-SiC MESFET比埋栅-场板结构的器件具有更高的栅延迟抑制比;在实现大功率应用方面,新型结构同样能提供更高的耐压。新结构的4H-SiC MESFET的最大饱和漏电流密度为460mA/mm,在漏电压20V的栅延迟抑制比接近90%。交流特性的分析结果表明,本文提出的结构比埋栅-场板结构的器件的特征频率和最高振荡频率分别高出5%和17.8%。此外,新结构的器件能承受较高的击穿电压,进而保证了器件的大功率密度。针对本文提出的结构进行了优化,以使器件能发挥最好的微波特性并对器件的设计提供一定参考。     

pH值调节方式对LAB辅助合成纳米Cu2O微球的影响与组装机理

以Cu(Ac)₂为原料,两性表面活性剂月桂酰胺丙基甜菜碱(LAB)为模板,采用两种不同的调节pH值方式制备了Cu₂O纳米材料.表征结果表明两种调节pH值方式均可获得Cu₂O纳米微球,并都呈立方晶相,而且样品的红外吸收峰、固体紫外吸收峰都不同程度的发生了蓝移;第一种Cu₂O纳米微球由针状纳米粒子积聚而成,针状纳米粒子间空隙孔径主要分布在25~50 nm之间,比表面积为22 m²·g^-1,禁带宽度为2.15 eV;第二种Cu₂O纳米微球由小的纳米球状体堆积而成,球状体间孔道直径集中在25~50 nm和50~125 nm两个区域,比表面积为9 m²·g^-1,禁带宽度为2.46 eV.两种不同的调节pH值方式获得的Cu₂O纳米微球,其反应历程和自组装机理存在不同.     

高校“强基计划”拔尖人才培养的探索与实践——以应用物理学专业为例

“强基计划”以服务国家重大战略需求为出发点,旨在促进基础学科拔尖创新人才培养,是高校招生自主化改革的重要探索。本论文研究了以培养拔尖人才为核心任务的高校,结合学校特色和专业特色,从培养定位、培养机制、培养环节等方面建立的“强基计划”拔尖人才培养模式。以应用物理学专业“强基计划”为期两年的培养实践为例,从学习学风情况、科研参与情况和竞赛获奖情况等方面分析应用物理学专业“强基计划”的育人初步成效。最后指出目前“强基计划”在推进过程中在政策定位、培养体系和拓展育人渠道等方面还应强化的方向不断推进基础学科的拔尖创新人才培养改革,为我国立足世界科技前沿领域提供强大的拔尖创新人才支撑。