基于遗传算法的波长选择方法在绿茶近红外光谱分析模型中的应用

利用遗传算法(GA)提取茶叶的近红外吸收特征波长的方法,研究建立了绿茶水分和氨基酸的近红外分析模型,并对波长选择前后两种成分的模型进行了比较分析。结果表明,经遗传算法波长选择后,简化了分析模型,同时模型的稳健性增强。氨基酸预测集的均方根误差(SEP)减小82.1%,水分预测集的均方根误差减小(SEP)76.6%,它们在波长选择前后对应的分析波长数之比分别为995∶7和1990∶33。     

吸收峰混叠的太赫兹光谱区间组合特征提取算法

太赫兹光谱是物质识别的前沿方法之一。由于不同物质的分子组成或结构各异,许多物质的太赫兹吸收谱会在特定频率上出现吸收峰,可以作为混合物成分检测的重要特征。有效准确地提取这些吸收峰的参数,是提高识别率的关键。多峰拟合算法将光谱曲线拟合成若干个标准峰函数之和,能够同时提取到吸收峰的频率、峰高、峰宽等信息。但是该算法以寻峰算法结果为基础确定吸收峰的大致位置和数量,寻峰结果不一定是最优的拟合结果,而且很难准确识别定位混叠状态的吸收峰。为了提高混叠光谱中吸收峰的识别定位精度,提出以大幅度平滑后的曲线波谷为分界点,将预处理后的光谱分成若干个子区间。然后将子区间组合起来进行多峰拟合,通过遗传算法得到最优的拟合子区间组合和吸收峰频率近似值,拟合时每个子区间中通过峰数递增最优化方法确定拟合的吸收峰数,最后微调优化得到最优的吸收峰频率、峰高值。为了实现物质的识别,通过密度聚类算法得到同一类纯净物在多次测量中的共同吸收峰,以此作为标准数据,通过提出的基于吸收峰特征的光谱匹配算法实现了纯净物和不同含量混合物的快速识别。对10类纯净物的实际光谱数据进行拟合聚类,得到其吸收峰参数,结果与太赫兹光谱数据库一致。通过识别算法对纯净物测试集进行识别的识别率为100%,证明了特征提取和物质识别算法的有效性。对于含有混叠峰的混合物光谱,二阶导数法对葡萄糖-乳糖混合物光谱中被掩盖吸收峰(1.280 THz)的识别率仅为70%,提取到的频率平均值为1.316 THz;而该算法提高识别率至95%,频率平均值为1.281 THz,该算法提高了对混叠峰的分辨能力,能够精确定位混叠峰。对10类纯净物构成的6类不同程度混叠的二元混合物前二、三识别率分别达到90.8%和98.3%,提取到的特征能够有效应用于混合物的成分检测。该算法能够以纯净物数据为标准数据实现成分各异的混合物成分检测,对于太赫兹光谱混合物成分检测有重要意义。     

Cr(VI) and Cu(II) removal from aqueous solution in fixed bed column using rice bran; experimental,statistical and GA modelling

Rice bran, a green and low-cost adsorbent, is used for Chromium (VI) and Copper (II) remotion from its aqueous solution. The influence of different process parameters in a fixed-bed on the removing efficiency has been investigated. The results show that the removal efficiency is higher at a minimum flow rate, low metal ion concentration, and higher bed height. The adsorption studies show that the rice bran has a better affinity to Chromium (VI) than Copper (II). Different kinetic models are used for the prediction of the column performance. This study shows that rice bran could be a potential and eco-friendly adsorbent for chromium (VI) and copper (II) removal and is suitable for developing countries like India. Multiple linear regression and ANN-based genetic algorithm modelling have been applied successfully to predict both metal ions' percentage removal separately.     

电化学DNA传感器的研制及其医学应用

对近几年发展起来的电化学DNA传感器的研究工作进行了讨沦,详细论述了该类传感器的工作原理、探针固定化技术、杂交指示剂的类型以及在基因诊断、药物分析等方面的医学应用,并展望了应用前景和发展方向。     

(BN)n(n≤12)团簇的结构及成键性质

利用遗传算法和Gastreich提出的经验势函数研究了(BN)_n(n≤12)团簇的可能稳定结构, 并对能量较低的异构体在HF/6-31G(d)水平进行优化, 得到了(BN)_n(n≤12)团簇的线状、蒲扇形、单环、双环、三环和笼状结构, 讨论了各种结构的特征及相对稳定性. 分析了BN团簇中原子的成键性质, 在单环结构中, N原子以sp2杂化成键, B原子以sp杂化成键, 而在节点处B原子以sp2杂化成键. (BN)6是唯一没有张力的单环结构.     

Mutation scanning-coupled analysis of haplotypic variability in mitochondrial DNA regions reveals low gene flow between human and porcine Ascaris in endemic regions of China

Haplotypic variation within and among the Ascaris populations representing six provinces in China was investigated. Mitochondrial DNA regions in the cytochrome c oxidase subunit 1 (cox1) and NADH dehydrogenase subunit 1 (nad1) genes were amplified by PCR from total genomic DNA samples (n > 720) from Ascaris individuals from humans and pigs, and subjected to mutation scanning and subsequent selective sequencing. For the cox1, ten different electrophoretic profiles were recorded for human Ascaris, and the same number for pig Ascaris, one of them being common to both host species. For the nad1, 11 different profiles were detected for human Ascaris, and 15 for pig Ascaris. Having defined all haplotypes (20 for pcox1 and 26 for pnad1) by sequencing, their frequencies were estimated in each of the two host species and each of the six provinces. For each mitochondrial region, the frequency of the different haplotypes varied considerably, depending on host species and geographical origin. Analysis of the sequence data (representing all haplotypes for each mitochondrial locus) by F-statistics indicated restricted gene flow between human Ascaris and pig Ascaris, and supported the conclusions from previous molecular epidemiological investigations that pigs are not a significant source of Ascaris infection in humans in endemic regions.     

Improving the neural network method for finite element model updating using homogenous distribution of design points

In developing a neural network technique for a finite element model updating, researchers have been shown that the number of training samples and their quality, significantly affect the accuracy of the NN predication. In this study, based on the genetic algorithm (GA) method, we reduce the number of analyses required to develop the training pairs and reduce the amount of time for training the NN. In the other words, a uniform distribution of design points inside the design space will be obtained by means of this approach. To validate the efficiency of GA sample selection, random generation (RG) method is used for comparison. Two comparisons are made based on a numerical and experimental example. One is updated from 10 degrees of freedom lumped parameters system and the other is updated from a free–free beam using test data. The results indicate that the GA sample selection can reduce the number of training samples without affecting the accuracy of the NN predication. In our present study, also, the advantages of using frequency response function (FRF) data as input to the NN are kept and the drawback of having too many frequency points is overcome by the application of principal component analyses (PCA).     

用遗传算法研究Con(n=3-60)团簇的结构特性

我们采用半经验的Gupta多体势结合遗传算法,系统地研究了Con(n=3-60)团簇的几何结构特性.我们发现在钴团簇的生长中存在一个类fcc构型与类Ih构型之间的竞争,从n=39开始,钴团簇呈现出明显的Ih生长模式.Con(n=3-60)团簇的幻数为13,19,23,38,55,结合钴团簇的平均最近邻原子间距和平均配位数,分析了钴团簇幻数序列存在的原因.我们发现团簇内部原子具有增强团簇对称性和加强团簇稳定性的显著作用.     

红黑排序混合算法收敛速度分析

The algorithm of applying the block Gauss elimination to the Red-Black or-dering matrix to reduce the order of the system then solve the reduced system byiterative methods is called Hybrid Red-Black Ordering algorithm.In this paper,we discuss the convergence rate of the hybrid methods combined with JACOBI,CG,GMRES(m).Theoretical analysis shows that without preconditioner thesethree hybrid methods converge about 2 times as fast as the corresponding natural ordering methods.For the case that all the eigenvalues is near the real axis, the GMRES(m) algorithm converges about 3 times faster than the natural ordering GMRES(m).Various numerical experiments are presented.For large scale prob-lem with preconditioners, numerical experiments show that the GMRES(m) hybrid methods converge from about 3 times to even 5 times as fast as the natural order-ing methods and the computing time is reduced to about 1/3 even 1/6 of that of the natural ordering methods.