用于光学神经网络硬件系统误差纠正的虚拟神经网络

提出了一种用于修正光学神经网络硬件系统误差的虚拟神经网络模型，光学实验结果表明该网络能够有效地消除硬件系统误差对实验结果的影响。     

Numerical analysis of a model of cure process for composites

We consider a composite material composed of fibres included in a resin which becomes solid when it is heated up (reaction of reticulation). The mathematical modelling of the cure process is given by a kinetic equation describing the evolution of the reaction of reticulation coupled with the heat equation. In this paper, we are interested in the computation of approximate solutions. We propose a family of discretized problems depending on two parameters (β₁, β₂) ε [0, 1]² which split the linear and non‐ linear terms in implicit and explicit parts. We prove the stability and convergence of the discretization for any (β₁, β₂) ε [½, 1 ] × [0, 1]. We present also some numerical results. Copyright © 2006 John Wiley & Sons, Ltd.     

The shear modulus in the fiber direction of unidirectional fiber composites based on the concept of mesophase

A new and accurate relationship for the shear modulus in the fiber direction of unidirectional fiber composites is derived, based on the Kerner model concepts but not using its approximate relationships. Furthermore, this model is extended by taking into account the mesophase between the fiber and the matrix accomodating smoothly the mechanical properties between neighbouring phases. The introduction of the mesophase results in an improvement of the theoretical predictions, which now approach close to experimental values for the moduli of different fibrous composites.     

Micromechanics of ferroelectric polymer-based electrostrictive composites

Electrostriction refers to the strain induced in a dielectric by electric polarization, which is usually very small for practical application. In this paper, we present a micromechanical analysis on the effective electrostriction of a ferroelectric polyvinylidene fluoride trifluoroethylene [P(VDF-TrFE)] polymer-based composite, where the exact connections between the effective electrostrictive coefficients and effective elastic moduli are established, and numerical algorithm for the prediction of the effective electrostrictive coefficients of the composite in terms of its microstructural information is developed. From our calculations, enhanced electrostriction in the composite has been demonstrated, and optimal microstructure for electrostriction enhancement has been identified. Our analysis provides a mechanism for the electrostriction enhancement, where the electrostrictive strain several times higher than that of polymer matrix can be obtained, if the microstructure of the composites can be carefully tailored.     

基于遗传禁忌算法的双资源约束下并行生产线调度研究

并行生产线调度问题兼有并行机器和流程车间调度问题的特点,是一类新型的调度问题。在考虑遗传算法早熟收敛特性和禁忌搜索法自适应优点的基础上,将遗传算法和禁忌搜索法结合起来,提出了基于遗传算法和禁忌搜索算法的双资源并行作业车间的调度优化问题算法,即考虑到了产品的调度受到机器、工人等资源制约的影响,对算法中种群的构造,适应度计算,遗传操作等方面进行了研究,最后给出了实例仿真和结论。     

Effect of the pore surface modification of an inorganic substrate on the plasma‐grafting behavior of pore‐filling‐type organic/inorganic composite membranes

We have investigated the effect of the surface state and surface treatment of the pores of an inorganic substrate on the plasma‐grafting behavior of pore‐filling‐type organic/inorganic composite membranes. Shirasu porous glass (SPG) was used as the inorganic substrate, and methyl acrylate was used as the grafting monomer. The grafting rate increased as the density of silanol on the SPG substrate increased. This result suggests that radicals are generated mainly at the silanol groups on the pore surface by plasma irradiation. The SPG substrates were treated with silane coupling agents used to control the mass of organic material bonded to the pore surface. The thickness of the grafted layer became thinner as the mass of organic material bonded to the pore surface of SPG increased. This decrease in the thickness of the grafted layer could be explained by the decrease in the penetration depth of vacuum ultraviolet rays contained in plasma having a wavelength of less than 160 nm that generated radicals in the pores of the substrate. The thickness of the grafted layer inside the SPG substrates could be controlled through the control of the mass of organic material bonded to the pore surface of the SPG substrate. © 2005 Wiley Periodicals, Inc. J Polym Sci Part A: Polym Chem 44: 846–856, 2006     

Investigation of sulfonated poly(ether ether ketone sulfone)/heteropolyacid composite membranes for high temperature fuel cell applications

The sulfonated poly(ether ether ketone sulfone) (SPEEKS)/heteropolyacid (HPA) composite membranes with different HPA content in SPEEKS copolymers matrix with different degree of sulfonation (DS) were investigated for high temperature proton exchange membrane fuel cells. Composite membranes were characterized by Fourier transfer infrared spectroscopy (FTIR) and scanning electron microscopy (SEM). FTIR band shifts suggested that the sulfonic acid groups on the copolymer backbone strongly interact with HPA particles. SEM pictures showed that the HPA particles were uniformly distributed throughout the SPEEKS membranes matrix and particle sizes decreased with the increment of copolymers' DS. The holes were not found in SPEEKS‐4/HPA30 (consisting of 70% SPEEKS copolymers with DS = 0.8 and 30% HPA) composite membrane after composite membranes were treated with boiling water for 24 h. Thermal stabilities of the composite membranes were better than those of pure sulfonated copolymers membranes. Although the composite membranes possessed lower water uptake, it exhibited higher proton conductivity for SPEEKS‐4/HPA30 especially at high temperature (above 100 °C). Its proton conductivity linearly increased from 0.068 S/cm at 25 °C to 0.095 S/cm at 120 °C, which was higher than 0.06 S/cm of Nafion 117. In contrast, proton conductivity of pure SPEEKS‐4 membrane only increased from 0.062 S/cm at 25 °C to 0.078 S/cm at 80 °C. At 120 °C, proton conductivity decreased to poor 0.073 S/cm. The result indicated that composite membranes exhibited high proton conductivity at high temperature. © 2006 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys 44: 1967–1978, 2006     

各向异性接合材料界面端部场数值分析

为了求解各向异性接合材料界面端部奇异性应力场,建立了一种新型杂交元模型.该模型的独特之处在于:基于有限元特征法得到的奇异性场数值特征解建立了一种新型界面端奇异单元.通过算例证明,新型杂交元模型能够利用较少的单元数获得较为精确的数值结果.当前模型应用范围广泛,能够用于复杂结构的界面端部场求解.     

模糊神经网络在复合地基沉降量预测中的应用

复合地基后期沉降变形对于建筑物设计及安全具有重要意义,针对通过长期沉降观测以得到复合地基的最终沉降需要耗费较多资源的问题,提出了一种基于模糊神经网络的预测方法.该方法考虑沉降变化过程有较大的随机性和模糊性,直接将样本数据进行模糊化,所得的模糊数代表了样本点集与控制点集中各分量之间的相关度,并依此建立模糊BP神经网络进行学习和估算.实验结果表明了该方法对沉降进行预测是可行与有效的,且在相对误差的有效控制方面优于BP网络方法与灰色方法.     

掺铬铝酸钇红色颜料的合成、结构和性能表征

研究了高温固相合成制备掺铬铝酸钇红色颜料的方法，着重考察了灼烧温度（1000-1400℃）、灼烧时间（1-2h）、主要原料配比（Al／Y：1：0．5—1：1．5，Cr2O3％：0．3％-5％）等实验条件对产物结构和色泽的影响。XRD等方法对样品进行的相态分析表明，主相为铝酸钇石榴石相；运用反射光谱对红色颜料的色泽进行了反射峰位和相关强度的表征。当Y2O3：Al2O3摩尔比为1：1，掺杂3％的Cr2O3作显色剂，在少量矿化剂存在时，在1300℃下灼烧1 h，得到了色坐标为x=0．577，Y=0．401的红色陶瓷颜料。