基于ANP-RBF神经网络的化工行业绿色供应商选择

基于绿色供应链理念,提出了化工行业绿色供应商选择的特色指标,构建了化工行业绿色供应商选择的ANP-RBF神经网络模型。通过ANP确定各指标权重,再结合RBF神经网络,从训练数据中提取隐含的知识和规律,能够方便地用于新供应商的选择。该模型求解算法为增量算法,具有很好的可扩展性,从而增加了评价的动态性。算例验证结果表明,将ANP-RBF神经网络模型用于化工行业绿色供应商的选择具有较强的实用性。     

Locality properties of radial basis function expansion coefficients for equispaced interpolation

Natasha Flyer ^{Many types of radial basis functions (RBFs) are global in termsof having large magnitude across the entire domain. Yet, incontrast, e.g. with expansions in orthogonal polynomials, RBFexpansions exhibit a strong property of locality with regardto their coefficients. That is, changing a single data valuemainly affects the coefficients of the RBFs which are centredin the immediate vicinity of that data location. This localityfeature can be advantageous in the development of fast and well-conditionediterative RBF algorithms. With this motivation, we employ hereboth analytical and numerical techniques to derive the decayrates of the expansion coefficients for cardinal data, in both1D and 2D. Furthermore, we explore how these rates vary in theinteresting high-accuracy limit of increasingly flat RBFs.}     

Prediction of membrane spanning segments and topology in β‐barrel membrane proteins at better accuracy

Prediction of membrane spanning segments in β‐barrel outer membrane proteins (OMP) and their topology is an important problem in structural and functional genomics. In this work, we propose a method based on radial basis networks for predicting the number of β‐strands in OMPs and identifying their membrane spanning segments. Our method showed a leave‐one‐out cross validation accuracy of 96% in a set of 28 OMPs, which have the range of 8–22 β‐strand segments. The β‐strand segments in OMPs and the residues in membrane spanning segments are correctly predicted with the accuracy of 96% and 87%, respectively. We have developed a web server, TMBETAPRED‐RBF for predicting the transmembrane β‐strands from amino acid sequence and it is available at http://rbf.bioinfo.tw/～sachen/tmrbf.html . We suggest that our method could be an effective tool for predicting the membrane spanning regions and topology of β‐barrel membrane proteins. © 2009 Wiley Periodicals, Inc. J Comput Chem 2010     

A numerical approach based on the meshless collocation method in elastodynamics

In this paper, a collocation technique with the modified equilibrium on line method （ELM） for imposition of Neumann （natural） boundary conditions is presented for solving the two-dimensional problems of linear elastic body vibrations. In the modified ELM, equilibrium over the lines on the natural boundary is satisfied as Neumann boundary condition equations. In other words, the natural boundary conditions are satisfied naturally by using the weak formulation. The performance of the modified version of the ELM is studied for collocation methods based on two different ways to construct meshless shape functions： moving least squares approximation and radial basis point interpolation. Numerical examples of two-dimensional free and forced vibration analyses show that by using the modified ELM, more stable and accurate results would be obtained in comparison with the direct collocation method.     

FLNG低温软管内衬波纹管多目标优化设计

低温柔性管道是海上浮式液化天然气生产储卸装置系统中输送液化天然气的核心装备之一。U型波纹管作为低温柔性管道最内层结构,直接影响低温管道的安全性。本文选取波径和环板长度作为独立结构设计参数进行结构分析,解决了在传统的U型波纹管灵敏度分析中,波高会随波距的改变而改变的耦合计算问题。构建了径向基函数神经网络代理模型,采用遗传算法对U型波纹管结构开展了多目标优化,获得以拉伸刚度最大和弯曲刚度最小为优化目标时的帕累托最优解集,为U型波纹管在实际工程应用中提供有益的设计参考。

     

A RECONSTRUCTION METHOD FOR FINITE VOLUME FLOW FIELD SOLVING BASED ON INCREMENTAL RADIAL BASIS FUNCTIONS

由于流场参数重构中, 用于重构的基网格单元的物理参数波动量相对于均值较小, 径向基函数（RBF） 直接插值方法重构会产生较大的数值振荡, 论文提出了一种增量RBF 插值方法, 并用于有限体积的流场重构步, 明显改善了插值格式的收敛性和稳定性. 算例首先通过简单的一维模型说明该方法的有效性, 当目标函数波动量相对于均值为小量时, 增量RBF 插值能够抑制数值振荡; 进一步通过二维亚音速、跨音速定常无黏算例、静止圆柱绕流非定常算例以及超音速前台阶算例来说明该方法在典型流场数值求解中的通用性和有效性. 研究表明增量RBF 重构方法可陡峭地捕捉激波间断, 可有效改善流场求解的收敛性和稳定性, 数值耗散小, 计算效率高.     

Solution of time-domain acoustic wave propagation problems using a RBF interpolation model with “a priori” estimation of the free parameter

In the last decade, Meshless Methods have found widespread application in different fields of engineering and science. Beyond novelty, their mathematical simplicity and numerical accuracy have been the key of their rapid dissemination. Among meshless techniques, RBF (Radial Basis Functions) based methods can be simple and general to solve the problems related to multiple areas of applied physics and engineering. In the specific field of acoustics, there are usually two possible approaches for solving a problem: time- and frequency-domain. In this paper, the authors propose a local time-domain approach to establish an efficient methodology for the solution of large-scale acoustic wave propagation problems. For this purpose, a local interpolation scheme, based on the reproduction of the local wave field using RBFs (MultiQuadric and Gaussian), is implemented and its accuracy is verified against known closed-form solutions. An explicit time-domain marching procedure is adopted, and the quality of the numerical results is also compared with that obtained using standard space-time Finite-Difference schemes. Additionally, the RBF interpolation model is used to simulate the propagation of a Ricker pulse in two simple test cases, and applied to simulate a more complex configuration, corresponding to an underwater sound propagation problem. In this frame, the results are also compared with those computed using a fourth-order in space and second-order in time Finite-Difference scheme.     

基于ANN的定向井有杆抽油系统悬点示功图特征参数的计算

有杆抽油系统悬点示功图的特征参数是合理选择地面机电设备的主要依据.由于井下工况的复杂性和部分参数难以确定,使得基于求解高维时变非线性方程的传统方法的计算结果存在偏差.本文将传统方法和神经网络相结合,给出了一种能比较精确地确定定向井有杆抽油系统悬点示功图特征参数的方法,避免了建立和求解复杂的非线性动力学方程.首先依据传统方法,计算出简化悬点示功图的特征参数.然后考虑抽油杆柱弹性振动、抽油杆与油管之问的库仑和粘性摩擦、气体和供液能力等因素的影响,利用BP神经网络和RBF神经网络建立了不同工况下悬点示功图特征参数的计算模型.利用现场实测数据对建立的神经网络进行了训练和测试.测试结果表明了本文方法的正确性和有效性.     

稀土萃取分离过程软测量方法的研究

分析了稀土组分含量在线检测技术进展缓慢的原因,分析了稀土萃取分离过程机制,找出影响稀土组分含量检测的主要因素,确定辅助变量。利用从包钢稀土厂采集的数据通过newrb函数创建径向基网络来实现稀土组分含量的软测量。文中通过流程图详细阐述了建模方法,采用50组样本数据对所建立的模型进行计算机仿真验证,仿真结果证明了模型的可行性。     

运用RBF径向基神经网络识别基于小波变换的淫羊藿苷红外光谱指纹特征

运用RBF径向基神经网络对基于傅里叶变换红外光谱（FTIR）法、相关系数比对法及多层次小波变换所提取的淫羊藿药材样本有效成分指纹特征进行识别。样品包括药典选用的５个物种（淫羊藿Epimedium brevicornu Maxim.、箭叶淫羊藿E.sagittatum（Sieb. et Zucc.） Maxim.、柔毛淫羊藿E.pubescens Maxim.、朝鲜淫羊藿E.koreanum Nakai和巫山淫羊藿E.wushanense T.S.Ying）和其他物种共计250个样本。淫羊藿指标成分淫羊藿苷在原药材和甲醇提取物的红外光谱图上,具有较明显的1 259 cm-1特征峰,经典的HPLC法的定量分析结果也佐证了淫羊藿苷的含量与1 259 cm-1峰的位置具有很好的一致性。为此该峰可以用来作为判别各物种淫羊藿药材是否含有淫羊藿苷的重要依据。在此基础上,采用相关系数比对法和多层次的小波变换法消除了因淫羊藿苷含量低、吸收峰弱,信号和噪声大的问题,增强了RBF径向基神经网络的识别效果。初步建立了基于小波变换和RBF径向基神经网络淫羊藿原药材红外光谱快速识别淫羊藿苷指标成分的一种新方法。