应用神经网络和遗传算法优化利福霉素B发酵培养基

对培养基进行优化时，会获得大量的实验数据，但这些实验数据往往不能被进一步利用。如果使用一些不同的方法对历史数据进一步挖掘就可能得出额外的规律。本文使用神经网络对历史数据进行处理，得到了利福霉素B培养基成份和发酵效价之间的关系；再以这个关系作为遗传算法的适应度函数，在一个相对大的空间内快速搜索最优解。优化的培养基使摇瓶发酵效价提高了17．9％，说明所建立的模型能够实现培养基优化的目的并具有数据挖掘的功能。

Optimization of the fermentation medium for rifamycin B production by using artificial neural network and genetic algorithms

During optimization of medium to promote productivity,lots of data was cumulated. If these data could be dealt with some different methods were a new relationship among them will be found. In this paper,artificial neural network(ANN) was used to predict the relationship between the potency of rifamycin B and the fermentation medium minor error. Two optimized fermentation formula for rifamycin B production were obtained by genetic Algorithms(GAs),using t-test when alpha was 0.25,and the errors between the actual value and predicted value by ANN were small. The rifamycin B production was increased by 17.9%using the optimized media. Results indicated that the procedure proposed in this work has a great potential for practical application.