BP神经网络结合正交试验法优化Bacillus velezensis Z-27菌株培养基组分

Culture Medium of Strain—Bacillus velezensis Z-27 Based on BP Artificiai Neural Network Combined With Orthogonal Experiment

将BP人工神经网络(Artificiai neural network)技术与传统的正交试验方法相结合,提出一种新的试验分析和处理方法,利用神经网络特有的自学能力,对主要影响因素进行仿真优化,获得Bacillusvelezensis Z-27菌株培养基组分,即玉米粉2.0%、豆饼粉1.5%、MnSO4.H2O 0.07%,起始pH为7.0、接种量为2.0%,应用优化得到的培养基组分进行验证试验,取得了较好的效果。

In this article, a new method of test analysis and data treatment which combined BP artificial neural network and traditional orthogonal experiment was proposed to obtain optimized medium composition of Bacillus velezensis Z-27 strains. By this method the main factors could be simulated and optimized with the help of specific learning capability of neural net- work. The results showed that the medium components of B. velezensis Z-27 strains was corn flour 2.0%, soybean powder 1.5%, MnSO4＂H20 0.0＂1% at initial pH 7.0, with inoculation quantity 2.0%. The optimized culture medium was then verified, and showed satisfactory effect.