用BP神经网络与遗传算法优化γ-氨基丁酸的发酵培养基

Culture medium optimization forγ-aminobutyric acid production by BP neural network and genetic algorithms

为了提高γ-氨基丁酸的产量,建立一个反映因素与指标之间的非线性的模型,保证得到最优组合。本研究运用BP (back propagation)神经网络给γ-氨基丁酸发酵过程中培养基组分与GABA产量的关系建模并预测其产量。利用遗传算法(genetic algorithm,GA)为γ-氨基丁酸发酵培养基中三种组分的配比全局寻优,得出三组分最佳配比为:豆饼粉26.88 g/L,玉米浆11.32g/L,葡萄糖8.97g/L和γ-氨基丁酸最大产量3.84 g/L。

For improving the yield of γ-aminobutyric acid （CABA） ,building a model which can reflected the nonlinear relationship between factors and indexes to ensure it was the optimum combination. BP neural network was used for modeling the relationship of the three components of the medium and the yield of GABA, and predicting the yield of production in batch culture. Based on the model, genetic algorithms were applied to optimize medium. The optimal nutrient concentrations were 26. 88 g/L soybean flour, 11.32 g/L corn plup, 9. 87 g/L glucose. The maximum yield of GABA was 3. 84 g/L.