苯乙烯-马来酸酐本体共聚合过程的神经网络建模

Modeling of Bulk Copolymerization of Styrene and Maleic Anhydride Based on the Artificial Neural Networks

人工神经网络(ANN)是一种有效的建模方法,尤其适用于机理复杂的化工过程,故应用ANN技术来研究苯乙烯-马来酸酐半连续本体共聚合过程的建模方法,并用原始实验数据训练BP网络,来预测本体共聚合过程的目标变量——反应转化率是合适的.由于标准BP训练算法的训练速度较慢, 提出了一种改进的训练算法 (marquardt算法)来提高网络的训练速度.结果表明,改进的训练算法提高收敛速度10倍以上,在不同的初始条件下,如停留时间5小时、聚合温度110～120℃和马来酸酐进料分量7%～10%,能得到满意的收敛点.在3个输入和1个输出(转化率)的情况下,估计结果的最大相对误差为10%～15%,平均相对误差小于5%.转化率的模型预测结果与原始实验数据具有良好的拟合.此方法可以有效地用于此类聚合过程的模型化.

Since the artificial neural network (ANN) is an effective modeling method, especially for chemical processes with complex mechanism, the modeling of semi-continuous bulk co-polymerization of styrene and maleic anhydride based on it was studied. The experimental data were used to train Back-Propagation (BP) network in order to predict the conversion of co-polymerization. To overcome the shortcoming of standard BP network's study algorithm, an improved method (Marquardt algorithm) for BP network was developed to increase the training velocity of the networks. The result showed that the training velocity of the improved method is increased in more than 10 times. Under different initial conditions, such as residence time 3-7 h, polymerization temperature 110°C-120°C, and maleic anhydride content in feed 7%-10% (wt), the satisfied convergence was obtained. In the case of 3 inputs and l output (conversion), the predicted maximum relative error was about 10%-15% and average relative error was less than 5%. The good agreement of the predicted conversion with the experimental data showed that the network model trained by the improved method can be effectively used to model the polymerization process.