Synthesis of AlPO4 Molecular Sieves with AFI and AEL Structures by Dry-Gel Conversion Method and Catalytic Application of Their SAPO Counterparts on Isopropylation of Biphenyl

AlPO₄-5 and AlPO₄-11 were synthesized by dry-gel conversion (DGC) method. Steam-assisted conversion (SAC) and vapor-phase transport (VPT) techniques were applied for this purpose. The synthesis was successful in presence of a certain minimum amount of external bulk water, without which the crystallization failed. Crystallization by VPT method was slower than corresponding SAC and HTS method. SAPO analogs of the samples, SAPO-5 and SAPO-11 were also synthesized by DGC method. Samples made by DGC methods had higher yield than the conventional hydrothermal synthesis (HTS); otherwise the samples showed similar characteristics as that made by HTS. XRD, SEM and N₂-adsorption results showed high crystallinity and purity of the samples made by DGC, and ²⁷Al MAS NMR spectra indicated the tetrahedral framework nature of Al. SAPO-5 and SAPO-11 were tested for their catalytic activity in isopropylation of biphenyl, and in terms of conversion and selectivity, SAPO-5 was found to be suitable for this application.     

基于商业智能的小区日用水量预测模型研究

商业智能现在已广泛应用于各个领域,利用商业智能可以对趋势预测做出良好的支持,本文将其中一种常用商业智能的工具——数据仓库技术,应用于小区家庭的用水量预测研究。     

纳米ZnO传感阵列对多组份气体的智能识别

采用物理热蒸发法制备纯ZnO纳米线及Al,Ag,Ni掺杂的ZnO纳米线.以纯ZnO纳米线及Al,Ag,Ni掺杂的ZnO纳米线为基料分别制成气敏传感器;利用气体敏感的差异性,构建纳米ZnO基气敏传感器阵列,结合BP人工神经网络来实现纳米ZnO基气敏元件对目标混合气体组分和相对浓度的识别.结果表明:将传感器阵列输出灵敏度值进行处理后,输入BP神经网络完成训练,能够准确识别混合气体(CO,H2,CH4)的组分与相对浓度。     

BP神经网络训练算法的分析与比较

BP神经网络被广泛应用于模式识别、信号处理和自动控制等领域,其广泛性是由于它能实现任何连续映射,但由于BP网络训练所固有的复杂性,目前尚没有任何一种完全的算法能适用于任何BP网络的训练。本文介绍了MATLAB神经网络工具箱中各种训练算法的特点及其函数的参数形式,并对它们的收敛速度和内存消耗情况进行了比较,说明了其各自适用的网络。     

Effective prediction of DEA model by neural network

In this paper,a fast neural network model for the forecasting of effective points by DEA model is proposed,which is based on the SPDS training algorithm.The SPDS training algorithm overcomes the drawbacks of slow convergent speed and partially minimum result for BP algorithm.Its training speed is much faster and its forecasting precision is much better than those of BP algorithm.By numeric examples,it is showed that adopting the neural network model in the forecasting of effective points by DEA model is valid.     

简支梁多位置损伤识别研究

曲率模态是结构损伤识别的敏感标示量,采用数值仿真方法,把曲率模态用于桥梁损伤识别.利用简支梁损伤前后一阶曲率变化率作为识别参数,便可明显确定其多处损伤位置.对相关样本参数进行归一化处理并引入噪声干扰,利用BP人工神经网络进行数值仿真计算,也能对其损伤程度进行有效模拟,验证了BP神经网络也有很强的泛化能力及容错能力.     

基于BP网络的指纹识别系统

针对指纹数据库规模不断增大的情况,为了提高指纹识别的速度和准确度,采用BP神经网络先对指纹进行分类,把指纹分成左环、右环、拱、涡、棚状拱和不辨型6种类型,然后在相应的子类库中根据特征提取进行识别。设计并实现了一个指纹识别系统,该系统能够自动地进行指纹分类,而且还能实时地将结果与指纹数据库进行比对。实验结果表明,该系统能迅速地获得被识别人员的身份信息。     

基于BP神经网络的集装箱箱号识别

提出一种用BP神经网络实现集装箱箱号识别的方法。提出的图像预处理方法对管柱,底面凹凸,脏污物、光线的强弱、倾斜、缺损或模糊有较强的适应性。在BP网络分类器实现时,考虑集装箱箱号的组成特点,采用并行处理方法。在BP算法中,引入动态因子,动态调整学习步长。实践证明利用这种方法来识别装箱箱号,速度快,识别率高,具有较高的实用价值。     

Kernel PCA与BP神经网络相结合的变压器故障诊断*

为了提高变压器故障诊断的准确率和抗干扰能力,提出一种基于核特征量的BP神经网络故障诊断模型。通过核主成分分析将故障样本从低维的特征空间非线性地映射到高维的核空间,提高了样本的可分性,然后以核特征量作为BP神经网络的输入特征量,建立变压器故障诊断模型。实验对比了结构相似、输入量不同的BP神经网络,结果表明采用核特征量的诊断模型具有更好的诊断效果和抗干扰能力。     

基于BP方法的企业核心竞争力关键因素探析

为了有效提升企业核心竞争力,对影响企业核心竞争力的关键因素进行识别和分析.依据人工神经网络理论,基于一较全面的企业核心竞争力评价指标体系,建立企业核心竞争力的人工神经网络专家模型,在此基础上构建一个识别企业核心竞争力关键影响因素的模型,是企业竞争战略的制定与实施的有益工具.对哈药集团进行实证研究表明,该模型能较好地识别哈药集团企业核心竞争力的关键影响因素,这对企业为提高核心竞争力如何进行有针对性的资源投入有较重要的借鉴价值.