利用C_5开发叔戊基甲醚

介绍了利用C_5生产汽油抗爆剂叔戊基甲醚的工艺进程。其中KOCH公司开发的催化蒸馏工艺是目前最先进的工艺。我国当前尚未有工业化生产叔戊基甲醚的装置。利用流化催化裂化C_5生产叔戊基甲醚是急待开发的课题。     

Formation and Properties of Chalcogenide Glasses in the GeSe2–As2Se3–CdSe System

The glass-forming region in a GeSe₂–As₂Se₃–CdSe system and the dependence of properties on glass compositions as a formula of 70GeSe₂–(30− x )As₂Se₃– x CdSe ( x =0–25) were investigated. Measurements included density, micro-hardness, differential scanning calorimetry (DSC), X-ray diffraction, and IR transmission spectra. A large glass-forming region was shown while addition of Cd increases the crystallization trend of the system. Besides, the Cd content accommodated by the system to form glass is found to be lower in the As₂Se₃-rich region than in the GeSe₂-rich region. With the introduction of CdSe, density, micro-hardness, and T _g of glasses increase whereas an exothermal peak in the DSC curve appears and thermal stability (Δ T ) decreases. The activation energy of crystallization and the Avrami exponent were also obtained using the modified Ozawa equation.     

气浮塔处理含油废水的研究

在实验研究的基础上，结合单级气浮技术和多级板式塔理论，开发出两级气浮塔处理含油废水的新工艺，实现了塔釜一次曝气、多级气浮的分离。实验研究了气浮塔板的流体力学性能、布气性能及操作条件对废水处理效率的影响，结果表明二级气浮塔处理效果很好，是一种具有良好应用前景的新型含油废水处理装置。     

Diffusion of ellipsoidal granular particles in shear flow

The diffusion of nonspherical particles has not been well understood due to the complexity of their contact mechanics and self-organization of their orientations. We perform discrete element method simulations of monodisperse ellipsoids in a shear flow with Lees-Edwards boundary conditions to quantify the relation between the diffusion coefficient and the flow parameters. The results indicate that the particle aspect ratio strongly affects the diffusion coefficient by influencing the particle orientation and alignment. We develop a scaling law for the diffusion coefficient perpendicular to the flow direction, D_yy, which combines the influences of the shear rate , the solids fraction f, the effective particle diameter d_eff and the particle aspect ratio Z. We show that , where k_d is a dimensionless pre-factor, and a fit is obtained for the functional form of χ(f, Z). This scaling law will be useful in developing continuum transport models for applications.     

用新型碳分子筛分析永久气体

本文介绍一种新型气相色谱固定相——碳分子筛(KMS),其最大的优点是在常温下,单柱即可同时分离分析H_2、O_2、N_2、CO、CH_4、CO_2六种混合气体,柱子的保留值重现性好,峰形对称,柱效高。本文求得了在H_2和Ar作载气时KMS对上述六种气体的定性定量特性,配气的定量结果总的平均相对误差为3.67％。同时这种碳分子筛原料易得,制作简单,成本便宜。     

Self-reorganization of mixed poly(methyl methacrylate)/polystyrene brushes on planar silica substrates in response to combined selective solvent treatments and thermal annealing

This article reports reversible changes in surface morphology and surface wettability of mixed poly(methyl methacrylate) (PMMA)/polystyrene (PS) brushes with PS M_n slightly lower than that of PMMA in response to combined selective solvent treatments and thermal annealing. The mixed brushes were synthesized from asymmetric difunctional initiator-terminated monolayers by controlled radical polymerizations. Two sets of experiments were performed. The first set involved treating the samples with glacial acetic acid (AA), a selective solvent for PMMA, followed by treatment with cyclohexane, a selective solvent for PS, or thermal annealing. The surfaces were rough and composed of relatively ordered nanodomains after AA treatment. Subsequent cyclohexane treatment or thermal annealing made the surfaces flattened and surface analysis indicated that the outermost layers were occupied by PS. The second set of experiments involved treating the samples with cyclohexane or thermal annealing, followed by AA treatment. The surfaces were smooth after cyclohexane treatment or thermal annealing and became very rough in the subsequent AA treatment. The nanodomains formed in AA treatment were speculated to be micelles with PS forming the core shielded by PMMA chains.     

汽车涂料使用的颜料（Ⅰ）

汽车涂料是一种高质量涂饰剂，其制造必须要使用一些具有多方面优良性能的高级颜料。本文对汽车涂料使用的各色颜料的选择作以推荐。     

融合语义差别和流型学习的偏标记学习方法

偏标记学习是一种重要的弱监督学习框架。在偏标记学习中,每个实例与一组候选标记相关联,它的真实标记隐藏在候选标记集合中,且在学习过程中不可获知。为了消除候选标记对学习过程的影响,提出了一种融合实例语义差别最大化和流型学习的偏标记学习方法(partial label learning by semantic difference and manifold learning, PL-SDML)。该方法是一个两阶段的方法：在训练阶段,基于实例的语义差别最大化准则和流型学习方法为训练实例生成标记置信度;在预测阶段,使用基于最近邻投票的方法为未知实例预测标记类别。在四组人工改造的UCI数据集中,在平均70%的情况下优于其他对比算法。在四组真实偏标记数据集中,相比其他对比算法,取得了0.3%～13.8%的性能提升。     

Nd(OR)_3催化异戊二烯聚合动力学及聚合物结构

研究了Nd(OR)_3—A1(i—Bu)_3一A1Et_2Cl体系催化异戊二烯聚合的动力学和所得聚异戊二烯的分子结构。结果表明,本体系的有效利用率为6％左右。在Al(i—Bu)_3用量较高时,主要发生活性中心向A1的链转移;在A1/Nd≤20,聚合温度低于室温条件下可获得“活的”聚合物。聚合物分子量高、分布宽,主要受A1(i—Bu)_3用量和聚合温度影响。分子量分布宽与其GPC图呈双峰有关。催化剂的相分离实验表明,体系中有两种不同的活性种存在:高活性生成高分子量聚合物的固相活性种和低活性生成低分子的可溶性活性种。聚异戊二烯的顺1,4—链节含量在95％左右,其余为3,4—链节。     

Re-GAN：残差生成式对抗网络算法

目的 生成式对抗网络（generative adversarial network,GAN）是一种无监督生成模型,通过生成模型和判别模型的博弈学习生成图像。GAN的生成模型是逐级直接生成图像,下级网络无法得知上级网络学习的特征,以至于生成的图像多样性不够丰富。另外,随着网络层数的增加,参数变多,反向传播变得困难,出现训练不稳定和梯度消失等问题。针对上述问题,基于残差网络（residual network,ResNet）和组标准化（group normalization,GN）,提出了一种残差生成式对抗网络（residual generative adversarial networks,Re-GAN）。方法 Re-GAN在生成模型中构建深度残差网络模块,通过跳连接的方式融合上级网络学习的特征,增强生成图像的多样性和质量,改善反向传播过程,增强生成式对抗网络的训练稳定性,缓解梯度消失。随后采用组标准化（GN）来适应不同批次的学习,使训练过程更加稳定。结果 在Cifar10、CelebA和LSUN数据集上对算法的性能进行测试。Re-GAN的IS （inception score）均值在批次为64时,比DCGAN （deep convolutional GAN）和WGAN （Wasserstein-GAN）分别提高了5%和30%,在批次为4时,比DCGAN和WGAN分别提高了0.2%和13%,表明无论批次大小,Re-GAN生成图像具有很好的多样性。Re-GAN的FID （Fréchet inception distance）在批次为64时比DCGAN和WGAN分别降低了18%和11%,在批次为4时比DCGAN和WGAN分别降低了4%和10%,表明Re-GAN生成图像的质量更好。同时,Re-GAN缓解了训练过程中出现的训练不稳定和梯度消失等问题。结论 实验结果表明,在图像生成方面,Re-GAN的生成图像质量高、多样性强;在网络训练方面,Re-GAN在不同批次下的训练具有更好的兼容性,使训练过程更加稳定,梯度消失得到缓解。