摘要
介绍了建立神经网络模型一般要涉及三大要素:训练样本、学习算法和网络结构。论述了如何利用人工神经网络建立了工艺参数(空塔气速,浆液固体颗粒浓度和系统压力)对浆态床反应器的气含率及体积气液传质系数影响的模型。实验中进行了网络模型最优化,并由神经网络模型给出了预测样本三维立体图,图形连续光滑,较好的反映了工艺参数与气含率及体积气液传质系数的关系。
Three elements(training samples,learning algorithm,and network structure)in the establishment of the neural network model were introduced.The model to predict the effects of process parameters(superficial tower velocity,concentration of slurry solids,and system pressure) on gas holdup and volumetric gas-liquid mass transfer coefficient in the slurry bubble column reactor was established by means of artificial neural network.The model was optimized during the experiment.The prediction sample three-dimensional graph given by the model presented as continuous and smooth graph,which demonstrated the relationship between the process parameters and the gas holdup as well as the volumetric gas-liquid mass transfer coefficient comparatively.
出处
《石油规划设计》
2009年第6期17-20,共4页
Petroleum Planning & Engineering
