基于BP神经网络的脂肪酶酯化菜籽油脱臭馏出物的仿真研究

该文采用BP神经网络对脂肪酶催化酯化菜籽油脱臭馏出物的工艺条件进行了仿真研究。将中心组合试验和神经网络相结合，采用动量梯度下降法对网络进行训练仿真，并利用训练好的网络对催化酯化工艺条件进行预测。结果表明：经过训练的网络可以很好的模拟反应条件，得到了脂肪酶催化反应的最佳工艺参数，此时脂肪酸甲酯的含量为55.2%，维生素E保留率达到90%,为脱臭馏出物的酯化催化效果的预测提供了一条可行的途径。

Simulation of rapeseed oil deodorizer distillate esterification by lipase reaction based on BP neural network

The Emulation of rapeseed oil deodorizer distillate esterification by lipase reaction based on artificial neural network and central composite experiment was carried out. Back-Propagation algorithm with momentous factorwas adopted to train the neural network followed by the prediction of esterification reaction. Results show that lipase reaction is well simulated and the optimal technological conditions of lipase reaction can be obtained by the trained neural network. Under these conditions, the content of fatty acid methyl ester andthe recovery of vitamin E reaches 55.2% and 90%, respectively. It provids asignificant approach for the prediction of esterification of deodorizer distillate.