基于改进Alopex算法的烷基化生产过程的优化

为了提高Alopex算法的全局搜索能力,本文提出了一种改进的Alopex算法。该算法将混沌算法引入到Alopex的初始化过程中,提出了具有停滞行为的步长操作方法,在个体连续一定代数没有更新时,引入了高斯变异操作。最后利用改进的Alopex算法求解了烷基化生产过程的最优化问题,优化结果表明了本文算法的有效性。     

三维人脸识别算法研究

近年来,基于三维图像的人脸识别技术已经取得了很大进展,在约束环境下也能获得很好的识别性能,但仍受限于姿态、表情等因素,需要从算法上改进才能解决其影响。本文分别从基于空域直接匹配、基于局部特征匹配和基于整体特征匹配3个角度出发,对人脸匹配算法以及融合算法进行了研究,列出了部分改进算法的实验结果,并分析了算法有效性的原因,总结了目前面临的三维人脸识别算法难以突破的一些困难及未来的研究趋势。     

模拟退火-BP神经网络法测定无机铬的形态

提出了用神经网络方法测定无机铬的形态。在Back Propagation(BP)算法的基础上,吸取了模拟退火算法的优点,解决了BP算法容易陷入局部极小点的问题。经实际样品分析,结果满意。     

基于GPU计算的并行PLS算法研究与实现

偏最小二乘算法(PLS)是与红外、近红外光谱分析结合使用最为广泛的化学计量学算法,然而当前PLS算法通常采用单线程方式实现,当校正模型数量多或样本数量大、波长点数和主成分数较多,模型需对光谱预处理和波长选择方法反复优化时,计算十分缓慢。为大幅提高建模速度,该文提出了一种基于图形处理器(GPU)的并行计算策略,利用具有大规模并行计算特性的GPU作为计算设备,结合CUBLAS库函数实现了基于GPU并行的PLS建模算法(CUPLS)。利用近红外光谱数据集进行性能对比实验,结果表明CUPLS建模算法较传统单线程实现的PLS算法,加速比可达近42倍,极大地提升了化学计量学算法的建模效率。该方法亦可用于其它化学计量学算法的加速。     

数字图像相关方法中基于改进IC-GN算法高精度形变测量研究

在数字图像相关方法对物体形变测量中,FA-NR算法实现了高精度测量,IC-GN算法在此基础上提高了测量效率。为进一步提升测量精度,提出了一种基于IC-GN的改进算法(GIC-GN)。在已知整像素初始位置上,通过梯度法求得更准确的亚像素位移,减小Hessian矩阵计算过程中产生的误差,同时加快迭代的收敛速度,有效提高了测量精度和效率。仿真实验验证结果表明,GIC-GN算法误差能够稳定在10^-4～10^-3 pixel之间,对比IC-GN算法精度提升了10%～60%,耗时是FA-NR算法的0.1倍、IC-GN算法的0.8倍,能够实现对物体形变信息的高精度、高效率测量。     

加权最小二乘支持向量机改进算法及其在光谱定量分析中的应用

为克服异常训练样本对校正模型的负面影响,提出了一种加权最小二乘支持向量机(WLS-SVM)的改进算法,解决了原有算法存在的迭代收敛问题,并将其运用于光谱定量分析.实验结果表明:与原有算法相比,WLS-SVM改进算法显著增强了对异常样本的检测能力,并大幅度地提高了校正模型的稳健性.     

基于小波和轮廓提取的色谱基线算法研究

把基于小波和轮廓提取的基线算法引入了色谱基线提取领域;基于轮廓提取的算法为:通过构造一个滑动窗,沿着色谱数据滑动求窗内的最小值。把这些最小值插值平滑就获得了色谱基线。基于小波和轮廓提取的算法为:先用小波初步提取基线,然后把色谱数据减去小波提取的基线后再用轮廓提取算法获得基线,把小波提取的基线和轮廓提取的基线相加即为原数据的基线。对这两个算法进行了比较实验研究,结果表明:基于小波和轮廓提取的算法比轮廓提取的算法效果好,能更准确地提取色谱基线。     

密度矩阵重正化群计算电子激发态的算法比较研究

激发态电子结构计算是理论与计算化学领域中的重要问题.本文测试了密度矩阵重正化群理论的state specific激发态算法,并在Pariser-Parr-Pople(PPP)模型下以并苯体系和聚乙炔体系为例对state specific算法和传统的态平均激发态算法进行了比较.计算结果表明,state specific算法可以正确地跟踪较低的激发态,并能得到比态平均算法更精确的能量本征值;在高激发态的计算过程中,state specific算法可能出现收敛于错误本征态的问题.     

小波包分析用于重叠分析化学信号的处理

对小波包分析的算法进行了改进,并将此算法成功地应用于多组分重叠色谱信号的解析.结果表明,本文提出的算法解决了MRSD算法的不足,更适合处理分析化学信号,用于重叠信号的解析时不需重构(逆变换),简化了数据处理步骤,加快了数据解析速度,具有较强的解析能力.对于重叠色谱信号的解析,小波包分析比小波分析具有更强的解析能力.     

介观生化反应的随机和混合模拟算法

随着细胞等介观体系中研究的深入,介观生化反应的模拟方法日益引起人们的关注。一方面,这类体系中涨落效应显著,化学反应本质上是随机的离散过程,经典的确定性模拟方法不再适用,需要使用随机模拟方法;另一方面,这种体系中子反应的速率和反应物的分子数目可能有显著差别,存在多尺度特性,使一般随机算法的模拟效率极低,对反应的准确高效模拟提出了新的要求。本文简述了Gillespie提出的一系列基本随机算法,综述了近年来发展的针对反应多尺度特性的改进随机算法和混合算法,介绍了各类算法的特点、实现的技术问题及相对其他算法和实际生物信号描述表现出的优缺点。     

Nature of the entropy versus self-diffusivity plot for simple liquids

The empirical relation (D(*))(alpha) = a exp[S] between the self-diffusion coefficient D(*) and the excess entropy S of a liquid is studied here in the context of theoretical model calculation. The coefficient alpha is dependent on the interaction potential and shows a crossover at an intermediate density, where cooperative dynamics become more important. Around this density a departure from the Stokes-Einstein relation is also observed. The above relation between entropy and diffusion is also tested for the scaled total diffusion coefficient in a binary mixture.     

大洋底多金属结核样品中主、次、微量元素的X射线荧光光谱测定

制定了一个用X射线荧光光谱法测定大洋底多金属结核样品中主、次、微量元素的分析方法。样品用石墨坩埚、混合熔剂低倍稀释熔融,用人工标样作为校准标准。计算了COLA方程的理论α系数,用于校正28个分析元素的吸收-增强效应。本法测定结果与化学法结果相吻合。     

理论α系数-X射线荧光光谱法分析水系沉积物中主次量元素

本文将NBSGSC程序移植到长城0520计算机中,求得一组COLA模式的a系数并输入到3080E3型XRF光谱仪计算机中,作联机分析,只需10个标样即可得到满意的结果,可与用50个标样的经验系数法相媲美。     

SP-PLP-EPR study of chain-length-dependent termination in free-radical polymerization of n-dodecyl methacrylate, cyclohexyl methacrylate, and benzyl methacrylate: evidence of "composite" behavior

The chain-length dependence of the termination rate coefficient in n-dodecyl methacrylate (DMA), cyclohexyl methacrylate (CHMA), and benzyl methacrylate (BzMA) bulk free-radical homopolymerizations at ambient pressure and at temperatures from -20 to 0 degrees C is deduced using the recently developed technique of SP-PLP-EPR: pulsed-laser polymerization (PLP) in which time-resolved EPR measurement of radical concentration, cR, is made following each single pulse (SP) of an excimer laser. The decay of cR results from termination of radicals of almost identical size. Their chain length, i, increases linearly with time, t, after applying a SP. The rate coefficient, kt(i,i), for termination of two radicals of size i is determined by fitting the experimental cR vs t data. This process demonstrates that (at least) two power-law exponents are necessary to describe kt(i,i) over the extended chain-length range of i = 1 to 1000. This is consistent with the so-called "composite model" , which uses power-law exponents alpha(S) and alpha(L) to describe termination of radicals either shorter or longer, respectively, than a crossover chain length, ic. The fourth parameter obtained from fitting the SP-PLP-EPR data with this model is kt(1,1), the termination rate coefficient for two radicals of degree of polymerization 1. Previous DMA experiments are reanalyzed while new experimental results are reported and analyzed for CHMA and BzMA. The parameter values for CHMA and BzMA termination at 0 degrees C are almost identical-kt(1,1) approximately 3 x 10(7) L mol(-1) s(-1), alpha(S) approximately 0.50, ic approximately 90, and alpha(L) approximately 0.21-and they are close to those for DMA at 0 degrees C: kt(1,1) approximately 1 x 10(7) L mol(-1) s(-1), alpha(S) approximately 0.64, ic approximately 50, and alpha(L) approximately 0.18. The results fully support the composite model in that the chain-length dependence is more pronounced for shorter than for longer radicals, i.e., alpha(S) > alpha(L). Moreover, the power-law exponent that characterizes termination of long-chain radicals is close to the theoretical value of alpha(L) = 0.16. In fact all parameter values-including the small differences between DMA and CHMA/BzMA-are more-or-less in accord with expectations based on polymer dynamics. Furthermore, our results suggest that termination of methacrylate radicals with large cyclic or long n-alkyl substituents may be affected by steric shielding of the radical functionality.     

Comparative study on camphor enantiomers behavior under the conditions of gas-liquid chromatography and reversed-phase high-performance liquid chromatography systems modified with alpha- and beta-cyclodextrins

The dependence of retention and selectivity parameters of camphor enantiomers on the concentration of alpha- and beta-cyclodextrins were studied under conditions of GLC (matrix solvent: Glycerol, 95 degrees C) and RP-HPLC (matrix solvent: Aqueous methanolic, 20 degrees C). It has been found that beta-cyclodextrin forms complex of 1:1 stoichiometry and does not recognize enantiomers of camphor. In contrast alpha-CD forming complexes of 1:2 stoichiometry appeared to be very efficient chiral selector of (+) and (-)-camphor. Relatively considerable differences have been observed between stability constants determined by GLC and RP-HPLC, what may be explained by the various natures of the matrix solvents and the various temperatures of the measurements. On the contrary, the enantioseparation factor alpha observed at higher concentrations of alpha-cyclodextrin stabilizes on the very similar value alpha+/-(GLC) approximately = alpha-/+(HPLC) approximately = 1.6. Simple theoretical considerations focusing on the differences in the mechanisms of the studied processes have been performed. According to them the enantiomer forming the more stable complex with the cyclodextrin should be eluted from the RP-HPLC column first and GLC column last. This fact has been confirmed experimentally.     

Measurement and theoretical characterization of electronic absorption spectrum of neutral chrysene (C18H12)and its positive ion in H3BO3 matrix

The ultraviolet and visible spectrum of chrysene and its radical cation formed by ultraviolet irradiation were measured in boric acid glass at room temperature. The theoretical electronic absorption spectrum of any polycyclic aromatic hydrocarbon (PAH) in boric acid matrix is calculated for the first time using semi empirical methods. Earlier reported theoretical results of electronic spectrum are calculated in free state and the results are compared with the spectrum of aromatic systems in glassy or other matrices. The interaction between the trapped PAHs (neutral and ions) and its environment induces strong perturbations of the energy levels which results in large shifts of the electronic transitions as compared to the ideal case of a free, isolated PAH molecule. This shifting due to perturbation has largely been ignored in earlier calculations, while comparing the calculations with the experimentally measured spectrum, in other matrices. The spectrum of singlet and doublet state of chrysene are computed in aqueous medium and also in free state to estimate the spectral shift. Several other geometric (bond length and bond angles) and spectroscopic parameters of chrysene like difference of HOMO-LUMO, ionization potential, dipole moment and polarizability are calculated using semi empirical methods, namely Austin Model 1 (AM1) and Parametric Method 3 (PM3). To get an idea about how the symmetry of chrysene molecule varies upon ionization, the mean polarizability (alpha) as well as its tensor components alpha(xx), alpha(yy) and alpha(zz) are calculated within a field of 0.005 a.u. The lasing action in neutral chrysene and in its cationic form is also discussed for the first time.     

NIR to UV absorption spectra and the optical constants of phthalocyanines in glassy medium

Optical absorption studies of phthalocyanines (Pc-s) in borate glass matrix have been reported for the first time. Measurements have been done corresponding to photon energies between 1.1 and 6.2 eV for free base, manganese, iron, nickel, molybdenum, cobalt and copper phthalocyanines. Several new discrete transitions are observed in the UV-vis region of the spectra in addition to a strong continuum component of absorption in the IR region. Values of some of the important optical constants viz. absorption coefficient (alpha), molar extinction coefficient (epsilon), absorption cross-section (sigma(a)), band width (delta lambda), electric dipole strength (q2) and oscillator strength (f) for the relevant electronic transitions are also presented. All the data reported for Pc-s in the new matrix have been compared with those corresponding to solution, vapor and thin film media.     

Expansion and hydrodynamic properties of cellulose-stainless steel powder composite matrix for expanded bed adsorption

For better understanding the influences of solid phase properties on the performance of the expanded bed, the expansion and hydrodynamic properties of cellulose-stainless steel powder composite matrix with a series of densities was investigated and analyzed in an expanded bed. Two kinds of matrix particle diameter fractions, the small one (60-125 microm) and the large (125-300 microm), were used in the present work. In general, the expansion factors decreased obviously with the increase of matrix density. A linear relation between the mean density of matrix and superficial velocity at expansion factor of 2.5 was found for same series of matrices. The Richardson-Zaki equation could correlate the bed expansion and operation fluid velocity for all matrices tested. The theoretical prediction of correlation parameters (the terminal settling velocity U(t) and expansion index n) was improved with the modification of equations in the literature. The residence time distributions were investigated to characterize the hydrodynamic property in expanded bed. Compared with three evaluation factors (the height equivalent of theoretical plate, Bo number and axial distribution coefficient D(ax)), the results indicated that D(ax) is the best parameter to analyze the bed stability of expanded bed under various operation conditions and matrix properties. In addition, it was found that fluid velocity is the most essential factor to influence the hydrodynamic properties in the bed. A linear relation between the D(ax) and superficial fluid velocity for all matrices tested was established.     

一种新的基体校正方法在Mn基合金XRF定量分析中的应用

采用谭秉和等人提出的多元素理论影响系数单标样定标的数学校正法及基本参数法对Ｍｎ基合金进行了分析，并对新方法进行了检验，结果良好。