【摘要】 目的 采用人工神经网络和遗传算法相结合，优化雷帕霉素发酵培养基。方法 首先通过Plackett-Burman设计实验，筛选影响雷帕霉素产量显著因素；再采用Box-Behnken实验建立数据样本训练人工神经网络模型，最后耦合遗传算法对模型全局寻优。结果 黄豆饼粉和赖氨酸对雷帕霉素的合成有显著的正效应，葡萄糖对雷帕霉素的合成具有显著的负效应。遗传算法-人工神经网络的决定系数与相对误差分别为0.998与2.29%。最终获得影响雷帕霉素发酵主要因素配比：葡萄糖6.5 g·L^-1,黄豆饼粉23.2 g·L^-1,赖氨酸7.9 g·L^-1。结论 优化后培养基的发酵水平较原培养基提高了21.1%,达到预期效果。更多还原

【Abstract】 OBJECTIVE Using a combination of artificial neural networks and genetic algorithms to optimize the rapamycin fermentation medium.METHODS Firstly, Plackett Burman designed experiments to screen for significant factors affecting the production of rapamycin; Then, the Box Behnken experiment is used to establish a data sample to train an artificial neural network model, and finally, a genetic algorithm is coupled to a global optimize the model.RESULTS Soybean cake powder and lysine have a significant positive effect on the synthesis of rapamycin, while glucose has a significant negative effect on the synthesis of rapamycin.The determination coefficient and relative error of the genetic algorithm artificial neural network are 0.998 and 2.29%,respectively.Finally, the main factors affecting the fermentation of rapamycin were determined as follows: glucose 6.5 g·L^-1,soybean cake powder 23.2 g·L^-1,and lysine 7.9 g·L^-1.CONCLUSION The fermentation level of the optimized culture medium increased by 21.1% compared to the original culture medium, achieving the expected effect.更多还原