作　　者： ; ;

机构地区： 吉林化工学院化学工程系

出　　处： 《吉林化工学院学报》 2004年第1期10-15,共6页

摘　　要： 在Matlab/Simulink中对精馏塔进行了动态模拟研究,建立了精馏塔的通用动态数学模型,在Simulink中以S 函数的方式进行了实现,为了提高计算速度以实现实时动态模拟,相平衡计算采用了BP神经网络.文中用均匀设计法构造了训练集用于训练神经网络,在此基础上进行了不同组成条件下混合物泡、露点的预测,并对醋酸分离流程中的低沸点塔按两种开工情况进行了动态模拟.实际应用表明,神经网络方法具有良好的预测精度,计算速度大大高于迭代法,在分离过程动态实时模拟领域有良好的应用前景. Dynamic simulation of distillation tower was performed in Matlab/Simulink. Generalized dynamic mathematical models of distillation tower were presented and used as S-function in Simulink. Artificial neural network (ANN) was used in phase equilibrium calculation for improving calculating speed in order to realize real time dynamic simulation. Training datasets were chosen using the technique of uniform design. ANN was trained and used to predict the bubble point and dew point respectively. Two startup cases of low boiling point tower of acetic acid separation were simulated. In addition to its accuracy, the method shows high computation speed that will be very useful in real time dynamic simulation.

关 键 词： 神经网络方法 精馏塔 动态数学模型 精馏过程

领　　域： [化学工程] [自动化与计算机技术] [自动化与计算机技术]