用神经网络方法优化鼓泡塔气含率的关联式  

Optimization of Gas Holdup Correlation in Bubble Columns by Using Artificial Neural Network

作　　者：罗湘华[1] 吴元欣[1] 李定或[1] 

机构地区：[1]武汉化工学院化工系,湖北武汉430073

出　　处：《石油化工高等学校学报》2003年第1期56-59,63,共5页Journal of Petrochemical Universities

基　　金：国家自然科学基金资助(20076036);湖北省教育厅重点科研项目(2001)资助。

摘　　要：　鼓泡塔作为一种结构简单的反应器,由于具有众多的优点,正在越来越广泛的应用于各种工业生产中。气含率是这类反应器的关键设计参数,如何准确地预测鼓泡塔的气含率,一直是人们研究的热点问题。在两千多组公开发表的鼓泡塔的实验数据组成的数据库的基础上,采用神经网络回归方法,在考虑所研究体系的物性参数,操作条件的基础上,结合受力分析,得到有8个对鼓泡塔内气含率有影响的无因次准数。将神经网络回归方法与受力分析相结合,提出了一个预测鼓泡塔内气含率的关联式。采用含有不同参数的模拟方案,进行比较,得到一个优化的神经网络关联式。与文献中6个现有的关联式进行比较,得出该关联式预测的结果与实验结果最为接近的结论。Bubble columns are very important gas-liquid reactors that are widely used in industry due to their merits. Gas holdup is one of the most important parameters of hydrodynamics in bubble columns, it influences heat and mass transfer rate and reaction rate, and it is a key parameter in the design of bubble columns. Hence, it is important to develop a general correlation for prediction of gas holdup in bubble columns with wide range of conditions. On the basis of a large databank consisting of over 2 000 vended experimental results for bubble columns, a state-of-the-art correlation for the prediction of gas holdup in bubble columns was proposed by using neural network regression. Eight dimensionless number influencing gas holdup in bubble columns were identified by applying neural network and force analysis. Two different schemes were applied and comparison was made to derive an optimized correlation. Compared with six existing correlations, it is concluded that the correlation obtained by this work is the optimal one.

关 键 词：鼓泡塔 气含率 神经网络 关联式 优化 

分 类 号：TQ021[化学工程]