运用人工神经网络反传算法研究了异喹啉及其衍生物电子结构与缓蚀性能之间的关系。以缓蚀剂分子氮净电荷、自由价、亲核前沿电荷密度等量子化学参数为输入变量,考察了异喹啉及其羟基、羧基衍生物在30℃,1.0mo1·dm~(-3)HCl溶液中对铁的缓蚀效率,建立了相应的预测模型。对于研究这类缓蚀剂分子在电极表面的吸附模型和缓蚀行为及定量预测同类新分子的缓蚀性能有一定价值。
An approach of neural network was made on its application to the correlation between molecular electronic structure and corrosion inhibition properties of isoquinoline and its derivatives. It was indicated in a reference that the less the net charge and π charge of N atom are ,the higher the inhibition efficiency is, and the net charge of six atoms in pyridine ring increases. The inputs of neural networks were the molecular structure parameters, such as net charge and π charge of N atom, which were obtained by means of HMO and CNDO/2 methods. The neural network's outputs included the corrosion-inhibition efficiencies of isoquinoline and its hydroxyl and carboxyl derivatives for iron electrode in HCl solution or electrode parameters which were determined with electrochemical methods in 1.0 mol·dm-3 HCl solution at 30℃. The corresponding relationship and modeling prediction were established by using the neural networks with 7-15-6, 7-5-5, and 9-10-6 topological structures that were trained beforehand by modified backpropagation (MBP) algorithm respectively. The learning precision was high and the predicting performance was excellent. With the trained neural network the corrosion inhibition properties or electrode parameters of some unknown isoquinoline's derivatives could be predicted quantitatively and the results were quite good.
参考文献
| [1] | 张敬畅,曹维良,王作新.中国腐蚀与防护学报,1986,6(3):217 |
| [2] | 罗明道,毕刚,旷富贵,姚禄安,颜肖慈.化学学报,1994,52:620 |
| [3] | 罗明道,姚禄安,吴庆余,颜肖慈,余晓东,邹津耘,欧阳礼.中国腐蚀与防护学报,1996,16(3):195 |
| [4] | 宁世光,石明理,刘奉岭,余立新,李延芳,章荣玲,程在英.中国腐蚀与防护学报,1990,10(3):383 |
| [5] | 唐子龙,宋诗哲.中国腐蚀与防护学报,1995,15(3):229 |
| [6] | 刘振宇,王昭东,王国栋,张强.钢铁研究学报,1995,7(4):61 |
| [7] | 郭稚弧,金名惠,桂修文,孟厦兰,邢政良,罗逸.腐蚀科学与防护技术,1995,7(3):258 |
| [8] | 郭稚弧,邢政良,金名惠,孟厦兰.中国腐蚀与防护学报,1996,6(4):307 |
| [9] | 李志良,曾鸽鸣,李梦龙,王树信,吉田H,宫下Y,佐佐木S.科学通报,1995,40:16320 |
- 下载量()
- 访问量()
- 您的评分：
-
10%
-
20%
-
30%
-
40%
-
50%