用人工神经网络模型模拟催化精馏塔

Simulation of Catalytic Distillation Using Artificial Neural Network

尝试用人工神经网络模型模拟醋酸甲酯水解催化精馏塔的操作过程，并寻求最佳工艺条件。以进料水酯比、回流进料比和醋酸甲酯的体积流率与催化剂体积比作为输入层的三个节点，醋酸甲酯的转化率和塔釜中的酸水比为输出层的两个节点，采用ＢＰ算法结合模拟退火算法，利用小试实验数据训练网络并检验训练结果；内插或外推一系列假想的工艺条件，让网络预测操作结果，最后确定最佳操作目标，通过网络模型寻找一组最佳工艺条件。结果表明：以充足可靠的实验数据为基础，神经网络模型能精确预测实验结果；利用人工神经网络模型对工艺条件进行优化，可取得令人满意的结果。

In this paper,an Artificial Neural Network model is used to simulate the catalytic distillation process of the hydrolysis of methyl acetate and optimize the operation conditions.The three nodes of input layer are appointed as the ratios of water flow to the flow of methyl acetate,the reflux to the feed flow and the volume flow of methyl acetate to the volume of catalyst.The conversion rate of methyl acetate and the ratio of acetic acid to water in the bottom are the two nodes in output layer.The algorithm of BP combined with annealing simulation.experiment data are used to train the ANN and the results are examined.Then,a series of supposed operation conditions are interpolated or extrapolated and the results are predicted by the ANN.Finally,the optimum operation object is determined and the optimum operation conditions are found by the ANN.The results show that based on plenty of accurate experiment data,the ANN can predict the operation results of catalytic column accurately and optimize the operation conditions satisfactorily.