基于神经网络的CO_2-MDEA-PZ-H_2O体系相平衡计算  

Calculation of Phase Equilibrium for CO_2-MDEA-PZ-H_2O System Based On BP Artificial Neural Network

作　　者：郑洪川[1] 李红艳[1] 贾林祥[1] 

机构地区：[1]哈尔滨工业大学低温与超导技术研究所,黑龙江哈尔滨150080

出　　处：《节能技术》2008年第3期226-227,246,共3页Energy Conservation Technology

摘　　要：虽然已有热力学模型能够对CO2-MDEA-PZ-H2O体系相平衡进行计算,但是误差较大。本文在文献数据的基础上采用两隐含层BP神经网络及VLBP算法对CO2平衡分压进行了预测。训练后的神经网络预测值与实验值的平均偏差仅为1.92%,与热力学模型相比有了很大的改进。因此采用BP神经网络对CO2-MDEA-PZ-H2O体系相平衡进行预测是非常有效的。In order to calculate the phase equilibrium for CO2-MDEA-PZ-H2O system,we have got some thermodynamic models,but their errors are relatively large.Based on literature data,this paper introduces BP artificial neural network(BP-ANN) with two hidden layers to predict the CO2 partial pressure.The mean deviation of predicted value got from BP-ANN trained properly and experiment data is 1.92%,this result is much better than thermodynamic models′.So introducing BP-ANN to predict phase equilibrium of CO2-MDEA-PZ-H2...

关 键 词：神经网络 MDEA CO2 相平衡 净化 

分 类 号：O642[理学—物理化学]