SYNTHESIS OF DBN AND ITS CATALYTIC PERFORMANCE FOR METHANOL CARBONYLATION TO METHYL FORMATE

1 , 5-diazabicyclo['1, 3, 0]non-5-ene(DBN) has been synthesized in our laboratory, and its catalytic activity and selectivity for methanol carbonylation to methyl formate (MF) have been studied. The experimental results have demonstrated that the synthetic processes of DBN used for the present work are reasonable and feasible. The total yield of three steps of DBN synthsis is approximate to 80%. The activity evaluation has shown that DBN can effectively catalyze the carbonylation reaction of methanol in the presence of propylene oxide (PO) promoter. PO-promoted DBN is a novel catalytic system superior to sodium methoxide.     

二溴硝基偶氮胂光度法测定矿物岩石中微量镁

研究了新显色剂二溴硝基偶氮胂与Mg~( )的显色反应及其影响因素,其表观摩尔吸光系数为2.3×10~4Mg~(2 )量在0～20μg/25ml范围内符合比耳定律,可用于矿物岩石中镁的测定,操作简便,结果满意.     

Unique cascade ring-opening/cyclization reaction of azlactones and DBU or DBN: Synthesis of new pyrrolam A analogues

1-Azabicyclo[3.3.0]oct-3-en-2-one derivatives were synthesized efficiently through the facile cascade reaction of 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) or 1,5-diazabicyclo[4.3.0]non-5-ene (DBN) with arylidene azlactones. This research provides a straightforward procedure for the synthesis of a novel class of multicyclic semi-alkaloids which shows good antimicrobial activity. Under catalyst-free, mild, and solventless condition, a wide range of fused tricyclic derivatives was obtained in high yields.     

1,5-二氮杂二环[4.3.0]-壬-5-烯与全氟烷基炔酸甲酯的Michael加成反应

1,5-二氮杂二环[4．3．0]-壬-5-烯(DBN)与全氟烷基炔酸甲酯发生Michael加成反应,生成的中间体发生分子内缩合产生了三环化合物。其结构经元素分析,IR,MS和核磁共振氢谱、碳谱,二维核磁共振^13C-^1H cosy谱,二维核磁共振^1H-^1H cosy谱确认。对反应机理也进行了探讨。     

基于无监督迁移成分分析和深度信念网络的轴承故障诊断方法

针对轴承故障样本少导致识别精度低的问题,提出一种基于无监督迁移成分分析(unsupervised transfer component analysis,UTCA)和深度信念网络(deep belief network,DBN)的故障诊断方法。首先利用UTCA的核函数将不同工况样本特征映射到一个共享再生核Hilbert空间中,使得源域和目标域样本集更加相似,并通过最大均值偏差嵌入法(maximum mean discrepancy embedding,MMDE)判断能够迁移的源域数据,将源域样本迁移到目标域中,为深度学习提供充足的训练样本,解决了实际故障样本较少的问题;然后采用DBN模型对源域样本进行训练,再对映射后无标记的目标域样本进行故障诊断分析。利用不同工况下的滚动轴承实验数据进行算法验证,结果表明,与普通DBN、SVM、BPNN以及传统机器学习-UTCA融合方法相比,本文方法对滚动轴承故障的诊断精度更高。     

TRANSESTERIFICATION OF METHANOL WITH ETHYL ACETATE AND ETHANOL WITH METHYL FORMATE CATALYZED BY DBN

采用三步合成了１，５二氮杂环双环［４，３，０］壬烯５（ＤＢＮ）并考察了其在酯交换反应中的催化活性。结果表明，在没有环氧丙烷（ＰＯ）助剂存在时，ＤＢＮ的催化活性很低；但在有ＰＯ助催化剂存在时，ＤＢＮ对酯交换反应的催化活性极高，其催化活性随ＤＢＮ和ＰＯ浓度的增加而增大。考察了反应温度对ＤＢＮ催化剂的酯交换活性的影响，提出了酯交换反应的催化机理。     

N-Acyl 1,5-diazabicyclo[4.3.0]non-5-ene (DBN) tetraphenylborate salts as O-acylating agents

Air-stable and crystalline N-acyl DBN tetraphenylborate salts have been shown to be effective O-acylating agents, reacting with both primary and secondary alcohols to give the corresponding esters in good yields. In the case of diols, the N-acyl DBN·BPh₄ salts have been shown to acylate regioselectively the primary alcohol functionality in the presence of a secondary alcohol. The DBN hydrotetraphenylborate side product can be readily removed by filtration, providing the ester products without the need for further purification.     

深度学习:多层神经网络的复兴与变革

 人工智能（AI）已经进入一个新的蓬勃发展期。推动这一轮AI 狂澜的是三大引擎,即深度学习（DL）、大数据和大规模并行计算,其中又以DL 为核心。本文回顾本轮“深度神经网络复兴”的基本情况,概要介绍常用的4 种深度模型,即：深度信念网络（DBN）、深度自编码网络（DAN）、深度卷积神经网络（DCNN）及长短期记忆递归神经网络（LSTM-RNN）。简要介绍深度学习在语音识别和计算机视觉领域几个重要任务上的应用效果情况。为便于应用DL,介绍了几种常用的深度学习开源平台。对深度学习带来的启示和变革做了一些开放式的评述,讨论了该领域的开放问题和发展趋势。