基于改进离散粒子群禁忌算法的并行测试任务调度

针对并行测试任务调度问题特点,提出了一种解决并行测试任务调度的离散粒子群禁忌算法,设计了符合并行测试任务调度特点的粒子初始化编码方式、粒子位置转移公式以及禁忌搜索邻域。通过与禁忌搜索相结合,避免了离散粒子群算法早熟问题。通过实验以及算法对比,表明提出的算法是快速有效的,能够很好地解决并行测试任务调度问题,求得测试总时间最短的任务调度序列。     

NH3-NO/NO2 SCR for diesel exhausts after treatment: mechanism and modelling of a catalytic converter

We review herein the key mechanistic and kinetic features of the reactions involved in the NH₃-NO/NO₂ SCR system investigated by dynamic reactive experiments over a V-based commercial powdered catalyst, eventually leading to the proposal of an original redox scheme which accounts for stoichiometry, selectivity and intrinsic kinetics of the global SCR process.     

降维方法监督分类的比较研究

对ISOMap、LDA、LLE、PCA这4种典型降维算法的主要思想和算法步骤进行了详细分析,并将它们用于有监督分类。从实验结果分析得到结论 ,其可为有监督分类提供有益的借鉴。     

电泵井振动监测及传输系统设计

潜油电泵机组是具有超细长转轴的复杂旋转机械,有复杂的生产及安装工艺,且采用了多级离心泵,在生产过程中会产生强大的振动源,测量并分析机组振动状况,是评估潜油电泵机组的运行状况、提高可靠性的重要手段。本文采用MMA3201振动加速度计监测振动状况,并采用4 mA²0 mA电流环将信号传输到井上MCU进行分析处理,根据监测和处理结果,适时提泵检修,采取合理的减振措施,延长潜油电泵的使用寿命。     

基于F2812高精度抗震动电子天平系统设计

采用F2812作为主控制器,对高精度、低成本、抗震动的电子天平系统进行设计。利用F2812高速运算能力,对AD转换后的称重数据进行数字滤波,通过程序判断有无加卸载称重样品,对滑动平均滤波算法进行改进,不仅提高了电子天平的称重精度,又加快了称重速度。为使其能在地表震动、温度变化频繁等恶劣环境中稳定工作,对电子天平增加了震动和温度补偿,改善了称重系统的动态稳定性。基于VC++开发了电子天平上位机监控软件,方便工作人员在主控室中对电子天平实时称重信息进行监测、补偿模式进行远程控制和查询历史数据。     

基于差分算法的反渗透海水淡化优化调度

以系统运行费用为目标的反渗透海水淡化优化调度是一类带有约束的非线性优化问题。针对这一问题,提出一种改进的差分进化算法。该算法对基本差分进化算法中的变异因子和交叉因子进行改进;定义约束违反度函数,将约束优化问题转化为无约束的优化问题。以24小时为一个周期,通过改进的差分进化算法对系统模型进行优化调度。仿真结果表明,改进的算法可以对机组进行优化操作,有效的降低了系统的生产成本。     

Alkali reduction and reactive dye dyeing of T/N nonwoven fabrics dipped into silicon‐containing,water‐borne polyurethane

Two series of anionic water‐borne polyurethanes with alkali resistance and covalent bonds of a reactive dye were synthesized with different molar ratios of poly(tetramethylene glycol) (PTMG). They were classified with respect to PTMG 1000 and PTMG 2000. The fiber blends of polyester/nylon nonwoven fabrics were dipped into silicon‐containing, water‐borne polyurethane and squeezed to an 80% pickup ratio. Finally, the manmade leather was treated with alkali reduction and dyed with a reactive dye. The alkali reduction and the thermal, mechanical, and dyeing properties of the manmade leather were studied. For alkali reduction, different ratios of NaOH and Na₂CO₃ concentrations were used. Na₂CO₃ was used because of its better spreading and buffering properties. The softness and breaking load were measured and related to the weight reduction. For the dyeing properties, a reactive dye with vinyl sulfone groups was found to bond with the OH group of water‐borne polyurethane by covalent bonding. On the basis of alkali reduction, a mixture of NaOH and Na₂CO₃ with a concentration ratio of 0.1N/0.2N could lead to better softness and alkali reduction of leather. For the mechanical properties, leather of the PTMG 1000 series showed a higher breaking load than leather of the PTMG 2000 series. However, less elongation in the PTMG 1000 series resulted. Differential scanning calorimetry showed an endothermic peak at 50–100°C. This indicated that the glass‐transition temperature of the hard segment decreased with an increasing amount of the soft segment in leather; meanwhile, both the glass‐transition temperature of the soft segment and the melting temperature of the hard segment also decreased as the content of the soft segment increased. For the dyeing properties, the reactive blue dye could reach up to 96.1% dye uptake in the polyurethane part of the leather. Moreover, the washing fastness could be graded as high as 4–5, and the light resistance was also graded to 4–5, in the dyed leather. © 2005 Wiley Periodicals, Inc. J Appl Polym Sci 96: 2324–2335, 2005     

高温变换催化剂制备条件的神经网络优化

将影响低汽气比条件下LB型节能高温变换催化剂活性的主要因素作为人工神经网络的特征输入向量,将全部实验数据分为训练集和预测集,运用Matlab神经网络工具箱,按改进的Bayes自动归一化算法建立反向传播神经网络模型,不仅可防止网络陷入局部最小,而且提高了网络训练精度和泛化能力。适当拓宽正交实验各因素的水平范围,经过不同因素、不同水平间的组合模拟,预测出LB型节能高温变换催化剂的最佳制备条件为氧化铈质量分数0.76%、氧化铜质量分数5.8%、氧化铬质量分数8.6%、氧化镧质量分数1.0%、铁液浓度92 g/L、中和过程最终pH值9.5。在最佳条件下试制催化剂在低汽气比下的平均活性达77.6%。     

图的Steiner最小树问题的降阶回溯算法

图的Steiner最小树问题是经典的组合优化问题,是一个NP难题,在不同的领域有着广泛的应用。研究该问题的部分数学性质,在此基础上给出了该问题的初步降阶方法和下界子方法,形成一个新的回溯算法。该算法具有较低的时间复杂度,还给出了应用实例及其分析。     

自适应混合多目标分布估计进化算法

针对多目标分布估计算法全局收敛性较弱的缺陷,提出了一种自适应混合多目标分布估计进化算法。其基本思想是:在多目标分布估计算法中引入全局收敛性较强的差分进化算法,当函数变化率较大时,用分布估计算法产生新种群;当函数变化率较小即算法可能陷入局部收敛时,用差分进化算法产生新种群。理论分析和数值实验结果表明,这种混合算法不仅具有良好的全局收敛性,而且解的分布性和均匀性较没有考虑目标函数变化率的混合多目标分布估计算法也有了一定程度的提高。