基于神经网络的膜法薏苡仁油脱胶过程动态模拟和预测

应用BP神经网络对膜法薏苡仁油脱胶过程的膜通量动态模拟与预测.研究了神经网络的构建、训练、结果测试和模型泛化,并用得到的神经网络对不同运行条件下(操作压力和温度)膜通量进行预测.结果表明:神经网络能够很好地模拟膜法脱胶过程的通量变化,预测结果与实验实测结果能够很好的吻合.与传统模型法相比,神经网络能快速简便地得到模拟和预测膜分离过程.

Dynamic simulation and prediction of degumming of coix seed oil using ultrafiltration based on neural networks

Back-Propagation(BP) network was applied to dynamically simulate and predict the membrane fluxes during degumming of coix seed oil using ultrafiltration in this work.Neural networks' structure,training,and testing were studied.Membrane fluxes under different operating conditions(transmembrane pressure and operating temperature) were predicted by BP network.The results showed that the membrane fluxes can be well simulated using neural networks and the predicted results were in good agreement with experimental data.Compared with traditional model methods,the neural network can simulate and predict membrane processing more rapidly and simply.