基于神经网络的叶绿素含量精细测量建模方法研究

活体植物叶片叶绿素含量SPAD值易受叶片厚度、水分等影响,提出了基于多参数神经网络建模的叶绿素含量精细反演方法。通过测量叶片在中心波长分别为650,940和1 450 nm光照射下的透过率,获得叶片的SPAD值和水分指数WI(water index),同时用数字螺旋测微仪测量相应的叶片厚度并用分光光度法测得其叶绿素含量。利用建模集样本分别建立SPAD值与实测叶绿素含量之间的单参数模型和基于BP神经网络的WI、厚度及SPAD值与实测叶绿素含量之间的非线性模型。利用这两种模型分别计算获得验证集样本的叶绿素含量预测值,对预测值和实测值进行了相关分析和相对误差的分析。实验以340个三种不同植物叶片为样本,用以上方法进行了分析。结果表明,利用BP神经网络建模后,每种植物样本的叶绿素含量预测精度都有不同程度的提高,尤其对于叶片厚度值较大的样本,效果更为明显。数据显示所有混合样本平均相对误差绝对值由单参数模型的7.55%降低到5.22%,实测值与预测值的拟合决定系数由0.83提高到0.93。验证了利用多参数BP神经网络模型可以有效地提高活体植物叶绿素含量预测精度的可行性。     

真空场注入4SMES光场广义电场的4m+2次和压缩

根据量子力学中态的线性叠加原理,构造了由多模真空态、多模相干态、多模相干态的相反态和多模虚相干态的线性叠加所组成的真空场注入四态叠加多模纠缠态(即4SMES)光场.利用多模压缩态理论研究了上述光场中广义电场分量的等幂次高次和压缩特性.结果发现:真空场注入4SMES光场是一种典型的四态叠加多模非经典光场;在一定条件下,该光场的广义电场分量可呈现出周期性变化的广义非线性等幂4m+2次和压缩效应.     

波片光轴偏差对偏振分光光束振幅和相位的影响研究

研究了在线偏振光入射、波片和偏振分束器结合使用时波片光轴偏差对偏振分束器分出的两束光的振幅和相位的影响。给出了在两种放置情况下,任意波片光轴方向偏差对偏振分光光束振幅和相位影响的表达式。详细分析了1/4波片和1/8波片的影响性质,给出了具体的振幅和相位影响关系式。     

基于小波域三状态HMT模型的含噪图像增强

针对含噪图像增强问题,提出一种基于小波域三状态隐马尔可夫树模型的方法,采用三状态的高斯混合模型逼近小波系数的分布,不需要设定精确的阈值,依据期望最大算法训练得到的每个系数所属状态的后验概率,将系数区分为噪声系数、弱边缘系数和强边缘系数,然后通过抑制噪声系数,增强细节特征系数来达到对含噪图像增强的目的,并引入循环平移策略避免人工失真.通过对含噪的标准图像和人脑核磁共振图像进行仿真实验,并与几种经典的图像增强方法作视觉上的对比和定量分析.实验结果表明,本文所提出的方法具有很好的鲁棒性,在突出了图像中更多的细节信息的同时,可以有效抑制噪声.     

加压条件下甲烷/空气预混火焰可燃极限的微重力实验研究

本文成功搭建了适用于中国科学院力学研究所国家微重力实验室(NMLC)落塔的高压对冲火焰实验系统,并首次开展了微重力条件下加压对冲火焰实验,测定了一定张力条件下甲烷/空气层流预混火焰的熄灭极限。实验结果表明,随着压力的增高,甲烷/空气混合气体的可燃极限呈先增后降的非单调变化趋势,峰值发生在0.4 MPa左右。浮力对加压下微弱火焰熄灭极限的影响明显,在常重力条件下,相同张力下的熄灭极限较微重力条件下的偏大,峰值出现的压力略低。微重力条件下的实验结果与使用CHEMKIN的数值模拟的结果相当一致。     

A Fractal Model for Effective Thermal Conductivity of Isotropic Porous Silica Low-k Materials

We establish a new model based on fractal theory and cubic spline interpolation to study the effective thermal conductivity of isotropic porous silica low-k materials. A 3D fractal model is introduced to describe the structure of the silica xerogel and silica hybrid materials （such as methylsilsesquioxane, MSQ）. Combined with fractal structure, a more suitable medium approximation is developed to study the isotropic porous silica xerogel and MSQ materials. Cubic spline interpolation for fitting discrete predictions from the fractal model is used to obtain the continuous function of the effective thermal conductivity versus porosity. Compared with other common models, the effective thermal conductivity predicted by our model presents better agreement with the experimental data for all porosity. These results indicate that the proposed model is valid.     

Spin Filter of Graphene Nanoribbon Based Structure

Spin-dependent transport properties of the zigzag graphene nano-ribbon （zGNR） based structure Al-zGNR-Al are investigated by ab initio technique where density functional calculation is carried out within the Keldysh non-equilibrium Green＇s function formalism. The energy band structure of the infinite zigzag ribbon is sensitive to the dangling bonds of carbon atoms on both edge sides. For the three-circle-width zigzag ribbon with one edge monohydrogenated and the other edge dihydrogenated （zGNR（H-H2））, strongly spin-polarized energy bands are found. A spin-down branch is obtained just below the Fermi level while a spin up band appears above it. For the structure Al-zGNR（H-H2）-Al, where three-circle-width and seven-circle-length （3 × 7） zGNR（H-H2） is coupled by two （100） aluminium electrodes, an obvious spin filter property is found as the bias voltage changes. When the length of the sandwiched zGNR（H-H2） ribbon increaes, the spin-up current is strongly restrained especially under higher bias voltage.     

Discrete Charge Storage Nonvolatile Memory Based on Si Nanocrystals with Nitridation Treatment

A nonvolatile memory device with nitrided Si nanocrystals embedded in a floating gate was fabricated. The uniform Si nanocrystals with high density （3× 10^11 cm^-2 ） were deposited on ultra-thin tunnel oxide layer （- 3 nm） and followed by a nitridation treatment in ammonia to form a thin silicon nitride layer on the surface of nanocrystals. A memory window of 2.4 V was obtained and it would be larger than 1.3 V after ten years from the extrapolated retention data. The results can be explained by the nitrogen passivation of the surface traps of Si nanoerystals, which slows the charge loss rate.     

High pt and photon physics with ALICE at LHC

ALICE,A Large Ion Collider Experiment,is dedicated to study the QCD matter at extreme high temperature and density to understand the Quark Gluon Plasma (QGP) and phase transition.High-transversemomentum photons and neutral mesons from the initial hard scattering of partons can be measured with ALICE calorimeters,PHOS (PHOton Spectrometer) and EMCAL (ElectroMagnetic CALorimeter).Combing the additional central tracking detectors,the γ-jet and π 0-jet measurements thus can be accessed.These measurements offer us a sensitive tomography probe of the hot-dense medium generated in the heavy ion collisions.In this paper,high p T and photon physics is discussed and the ALICE calorimeters capabilities of high-transverse-momentum neutral mesons and γ-jet measurements are presented.     

氧化镧表面反应性对甲烷和氧分子活化的影响

以氧化镧催化剂在甲烷氧化偶联(OCM)反应中的结构敏感性实验研究为基础, 采用周期性密度泛函理论(DFT)计算研究氧化镧(001), (110)和(100)3个晶面及OCM反应物分子甲烷和氧在其上的吸附、 活化和解离. 结果表明, 氧化镧(001), (110)和(100)3个晶面的表面能大小顺序为(110)>(100)>(001), 3个晶面的价带和导带间隙大小顺序为(110)<(100)<(001), 即(001)是3个晶面中最稳定的晶面, 而(110)则是最活泼的晶面. 甲烷分子在氧化镧(001), (110)和(100)晶面上的吸附很弱(0.03 eV), H—CH₃解离吸附能分别为2.16, 0.68和0.90 eV, 解离反应的难易性与晶面的活性顺序一致; 而氧分子在氧化镧(001), (110)和(100)晶面上的分子吸附能分别为-0.04, -0.31和-0.12 eV, 解离吸附能分别为1.22, 0.53和1.52 eV, 即氧化镧晶面结构对氧分子吸附具有明显的影响, 其中, (001)晶面上吸附最弱, (110)晶面上吸附最强, 以致O—O在(110)晶面上可以较低能垒(0.53 eV)解离, 形成亲电的过氧物种. 由于氧分子在氧化镧表面的吸附较甲烷分子强, 因此, 氧化镧在OCM反应中结构敏感性应与氧分子的吸附和活化密切相关. 甲烷和氧分子在氧化镧表面上活化的本质源自于电子自表面流向甲烷和氧分子的反键轨道, 且表面结构的改变会导致不同强度的电子流动驱动.