Effect of Curing Regime on Bonding Strength of EVA Modified Mortar to Tile

Ethylene-Vinyl Acetate (EVA) redispersible powder and latex were used to modify mortar. Three kinds of curing regimes: standard curing, high temperature curing and freeze-thaw circle curing were adopted to cure the bonded samples. Bonding strength of EVA modified mortar was tested at 28 days. The development of bonding strengths under all three curing regimes were discussed and compared. The experimental results show that bonding strength increases with the increase of EVA content in mortar. The curing regi...     

整数规划的量子行为蝙蝠算法

蝙蝠算法是一种新型群体智能算法,传统的蝙蝠算法在解决整数规划问题时容易陷入局部最优并出现早熟收敛现象,为了解决这些弊端,提出了一种基于势阱的具有量子行为的蝙蝠算法。论述了算法的优化原理和实现方式,并通过仿真实验,与粒子群算法和量子行为粒子群算法进行性能对比。实验结果表明,量子行为蝙蝠算法不仅能够有效地解决整数规划问题,而且比其他算法具有更好的性能。     

基于定性与定量分析的联络中心任务量预测法

任务量预测是联络中心人员排班的第一步,如何精确预测联络中心的任务量并相应地安排适当数量的座席人员来处理客户的需求是联络中心管理人员面临的首要问题.提出采用定性分析与定量分析相结合的任务量预测方法.对于给定的任务量数据,首先,采用观察法进行定性的分析区分异常和正常数据,然后,分别选定合适的定量预测方法进行预测.任务量定量预测方法包括趋势拟合、线性回归、移动平均、时间序列分解等预测方法.通过定性分析与定量分析相结合的综合应用方法,给出更准确的联络中心任务量预测方案,帮助管理人员最大程度的减少放弃和被拥堵在队列中的任务数量,设计制定可行性高的座席人员数量及人员排班.     

结合Bi-2DPCA与CNN的美式手语识别

针对现有美式手语（ASL）识别算法准确率低和模型训练时间长的问题,提出一种结合双向二维主成分分析（Bi-2DPCA）与卷积神经网络（CNN）并基于贝叶斯优化的识别算法。利用Bi-2DPCA算法对原始图像做数据降维处理,提取行、列方向的特征图,使用卷积神经网络对特征图进行训练分类,同时采用贝叶斯优化算法对模型超参数进行自动调参。在24分类ASL数据集上的实验结果表明,该算法的识别准确率达到99.15%,训练时间相比传统CNN算法减少90.3%。     

有限产能下多部件设备视情维修策略与多订单批量生产联合优化研究

针对有限产能下,面对多个不同要求订单时的订单选取和多部件设备维护决策问题,提出了基于客户满意度的多部件设备视情维修策略和多订单批量生产的联合优化模型。首先,基于比例风险模型描述出各部件的具体劣化趋势,并以虚拟年龄表示各部件的健康状态;其次,根据单次批量生产结束后的实时监测结果和维修阈值,确定不同的维修方式;再次,以利润最大化为优化目标,结合有限产能批量问题模型和客户满意度模型,建立视情维修策略和多订单批量生产联合优化模型;最后,利用模拟退火算法和粒子群算法就行模型求解,并通过算例验证了模型的有效性,使模型更具有现实意义,并对模型相应参数进行了灵敏度分析。     

基于主观倾向值和EasyEnsemble算法的虚假评论识别方法

为了有效识别在线虚假评论,提出一种基于XGBoost-EasyEnsemble算法的虚假评论识别方法。首先,根据虚假评论的特点和提出的主观倾向值计算方法,建立多维特征模型;其次,针对评论数据中的类别不平衡问题,EasyEnsemble算法借助集成策略弥补欠采样的缺陷,充分利用样本信息;最后,选择“好而不同”的XGBoost模型作为基分类器训练最终分类器。基于Yelp网站上的评论数据,以AUC作为评价指标,与支持向量机、GBDT、神经网络等热门机器学习算法进行对比,验证了该方法的有效性。     

High-Precision Position Tracking Control of Electro-hydraulic Servo Systems Based on an Improved Structure and Desired Compensation

How to improve the position tracking accuracy of electro-hydraulic servo system is a hot issue today. Full state feedback control has received widespread attention for its ability to significantly improve control performance, however, its practical application range is limited in view of the large influence of measurement noise. In terms of this issue, we propose an adaptive robust controller based on improved structure and desired compensation. Firstly, to reduce the impact of measurement noise, the actual state value is substituted by the corresponding desired value in the controller design based on model compensation and the adaptive model compensator. Then, we introduce a new auxiliary variable into the controller to optimize its structure. In addition, nonlinear robust control laws are integrated in the controller to balance unstructured uncertainties. Simulation analysis shows that the proposed control strategy not only achieves the asymptotic tracking when parameter perturbation exists, but also ensures a specified transient response and final tracing precision under the combined influence of structured and unstructured uncertainties. The results indicate that the control strategy has good control accuracy as well as strong robustness.

     

水平管内流型转换区湿气流经多孔板的压降特性

较之于传统的单孔板,多孔板有压损系数小、整流、低噪声等多方面优点,在湿气流量测量中具有重要应用价值。以空气和水为两相工质,在水平管内湿气流型转换区对小孔分别呈圆形和矩形的2个多孔板的压降特性进行了实验。结果表明:圆孔、槽孔多孔板的湿气压降特性无明显差别;多孔板的气相压降倍率与气相、液相弗劳德数均密切相关;与实验结果相比,现有均相流模型的预报结果均偏高,现有分相流模型的预报结果均偏低;通过气相、液相弗鲁德数双参数对孔板湿气压降模型进行修正,可明显提高压降关联式的预报准确度;基于均相流模型双参数修正所得的多孔板压降关联式的预报精度最优,对圆孔多孔板、槽孔多孔板的预报偏差分别在±7%、±10%以内。     

a-C:F:H薄膜的化学键结构

使用CF4和CH4为源气体,利用射频等离子体增强化学气相沉积法,制备了a-C:F:H薄膜样品.采用拉曼光谱仪、傅里叶变换红外光谱仪、X射线光电子能谱仪(XPS)对薄膜的结构进行了测试和分析.研究发现:该膜呈空间网状结构,膜内碳与氟、氢的结合主要以sp3形式存在,而sp2形式的含量相对较少;在薄膜内主要含有C-Fx(x=1,2,3)、C-C、C-H2、C-H3等以及不饱和C=C化学键;同时,薄膜中C-C-F键的含量比C-C-F2键的含量要高.在不同功率下沉积的薄膜,其化学键结构明显不同.     

Effects of ZrO2 on the structure and magnetic properties of CoPt thin films

[CoPt 1.5 ml/ZrO2 xnm]10 multilayer films were deposited on glass substrates by magnetron sputtering and then annealed in vacuum at 600℃ for 30 min. Their structures and magnetic properties were investigated as a function of ZrO2 content. The results show that the grain size and coercivity first increase and then decrease with the increase in ZrO2 content. The maximum coercivity and grain size are obtained at 37 vol.% of ZrO2. The content of ZrO2 in the film plays an important role in the separation of CoPt grains and in the reduction of intergrain exchange interaction. On the basis of the studies of angular dependent coercivity, it is found that the magnetization reversal of CoPt films with （111 ） texture is different from either the domain wall motion or the S-W type of rotation mode.