摘要
研究了利用神经网络优化复合纳滤膜制备工艺。以聚砜基膜和聚醚砜酮膜(PPESK)为例,首先,在实验数据的基础上建立并优化了神经网络模型,比较两种不同复合纳滤膜制备工艺的模拟结果与实验结果,充分证明神经网络的适用性和可信性;然后,利用神经网络的预测能力,优化聚砜基膜的制备工艺,确定最佳工艺条件:水相浓度0.6%,有机相浓度0.6%,有机相处理时间6 min。该法不仅可以减少实验成本,且能提供较可信的最优工艺条件,具有一定的实用价值。
The purpose of this paper is to optimize the prepare conditions of composite nanfiltration membranes. A mathematical model of the relationship between the process parameters and the membrane performance （ salt rejection and flux） is established by using BP artificial neural networks. Firstly, the architecture of the ANN model is designed and optimized. By comparing the simulating results with experimental values, the applicability and creditability of the method is proved. Then the ANN is used to optimize the prepare conditions of the polysufone membrane. In the result, the optimal prepare condition is obtained： aqueous concentration 0.6%, non-aqueous concentration 0. 6% and the immersion time 6 min. The experimental results obtained from the ANN simulation are salt rejection 89.5%, flux 60.37 Lm^-2 h^-1, which is better than the orthogonal experimental results： salt rejection 88.4%, flux 52.11 Lm^-2h^-1 It provide a new means for chemical technology optimization.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2007年第8期1055-1058,共4页
Computers and Applied Chemistry
基金
国家自然科学基金(20176001)
关键词
神经网络
优化
复合纳滤膜
制备工艺
ANN, optimize, composite nanofiltration membrane, preparation conditions
