神经网络模型模拟旋转填充床的有效传质比表面积  

Simulation of Effective Interfacial Area in Rotating Packed Beds by Artificial Neural Network Models

作　　者：刘丽[1] LIU Li(Lvliang University,Lvliang 033001,China)

机构地区：[1]吕梁学院,山西吕梁033001

出　　处：《化学反应工程与工艺》2020年第6期531-537,共7页Chemical Reaction Engineering and Technology

基　　金：山西省高等学校科技创新项目(2020L0708);吕梁市引进高层次科技人才计划专项(Rc2020-115)。

摘　　要：有效传质比表面积(a)是设计旋转填充床的重要参数,其高精度预测尤为重要。采用了5种神经网络模型模拟旋转填充床的有效传质比表面积,包括前馈反向传播神经网络(FFBP)模型、广义回归神经网络(GR)模型、串并联叠层神经网络(CFBP)模型、径向基神经网络(RB)模型和埃尔曼正向反向神经网络(EFBP)模型。选择气体雷诺数(Re_(G))、液体雷诺数(Re_(L))、液体弗洛德数(Fr_(L))、液体韦伯数(We_(L))和填料特征参数(ψ)5个无量纲准数作为模型的输入参数,有效传质比表面积作为模型的输出参数,采用均方误差(E^(2))和相关系数(R^(2))评价模型的准确性。模拟结果与文献的实验值比较表明,5种神经网络模型可以很好地描述旋转填充床的有效传质比表面积,其中GR模型最佳。采用GR模型模拟预测了超重力因子、气体流量和液体流量对有效传质比表面积的影响,预测结果表明:有效传质比表面积随超重力因子和气体流量的增加而增加;在液体流量较低时,有效传质比表面积随液体流量的增加而增加,而超过一定流量后随其增加而减小。The effective interfacial area was a vital important parameter for designing the rotating packed bed(RPB)and then the accurate prediction was essential.In this study,five types of artificial neural network(ANN)models were used to simulate the effective interfacial area in the RPB,including feed-forward back propagation neural network(FFBP)model,generalized regression neural network(GR)model,cascade-forward back propagation neural network(CFBP)model,radial basis neural network(RB)model and elman-forward back propagation neural network(EFBP)model.The dimensionless number groups were set as input data,including gas Reynolds number(Re_(G)),liquid Reynolds number(Re_(L)),liquid Froude number(Fr_(L)),liquid Weber number(We_(L))and packing character parameter(ψ).And the variable of effective interfacial area was set as output data.The mean square error(E^(2))and coefficient of determination(R^(2))were used to test the accuracy of the model.The simulation results were compared with the experimental data from references,which showed that the five ANN models well predicted the effective interfacial area and GR model exhibited the best performance.Then GR model was further used to predict the effects of high gravity factor,gas flow rate and liquid flow rate on the effective interfacial area.The predicated results indicated that the effective interfacial area increased with the increase of high gravity factor and gas flow rate.And the effective interfacial area increased with the increase of liquid flow rate at low liquid flow rate,but decreased when the liquid flow rate was above a certain value.

关 键 词：旋转填充床 有效传质比表面积 神经网络模型 

分 类 号：TQ015.9[化学工程]