人工神经网络在米曲霉菌体固定化研究中的应用

在液体培养基中，将米曲霉3042菌体培养成直径为1～2mm、菌壁上含较高氨基酰化酶浓度的菌丝球，以甲醛为交联剂，明胶为酶活性保护剂对米曲霉菌丝球进行固定化研究。在正交实验L₁₆(4⁵)的基础上，选用结构为410-15-1的BP(Back propagation)人工神经网络，对固定化工艺条件进行优化和预测，得到了优化的固定化条件。实验测定在此条件下制备的固定化米曲霉菌体比酶活为1500u，比酶活保留率达到83%，说明人工神经网络可以用于米曲霉菌体固定化工艺条件的优化。

Study of Immobilization of Aspergillus oryzar Mycelium Aided by Artifical Neural Network

Aspergillus oryzae\% mycelium pellets with abundant aminoacylase were cultured by liquid fermentation. Mycelium pellets were immobilized by crosslinking method with reagent of gelatin and formaldehyde. Basing on the orthogonal design table L\-\{16\}(4\+5), artificial neural network(ANN) model with back\|propagation(BP) of error structured 4\|10\|15\|1 was used to optimize the immobilization conditions. And the optimal conditions were obtained. Then the activity of the immobilized cells prepared under the optimal conditions was assayed. Compared with the free mycelium activity of 1800u, the residual activity rate was 83%, which was in good accordance with that predicted by ANN. It illustrated that artificial neural network can be used to find optimal conditions for cell immobilization.