摘要
[目的]采用人工神经网络结合粒子群算法优化产多杀菌素发酵培养基。[方法]通过Plackett-Burman设计试验,筛选出水解糖、蛋白胨和植酸这3种培养基组分对产多杀菌素具有显著影响;单因素实验确定这3个因子的合适取值范围;以Hybird设计试验数据为样本,建立神经网络模型,并用粒子群算法对模型全局寻优。[结果]得到最佳培养基组分(g/L):葡萄糖55,水解糖8.5,酵母膏10,蛋白胨7.5,棉籽粉30,菜籽油30,MgSO4.7H2O0.4,植酸0.5。[结论]在该发酵条件下,多杀菌素产量达到790.3 mg/L,较优化前提高36.5%。
[Aims] The fermentation conditions of spinosad were optimized by artificial neural network（ANN） coupling particle swarm optimization（PSO） algorithm.[Methods] hydrolysis sugar,peptone and phytic acid,which had significantly positive effect on spinosad production,were screened out from 8 related factors by Plackeet-Burman design.The reasonable ranges of these three factors were determined by single factor experiment.ANN coupling PSO algorithm based on hybrid design was applied to predict these factors＇ mutual interactions for spinosad production.[Results] The optimal medium composition was determined as follows（g/L）： glucose 55,hydrolysis sugar 8.5,yeast extract 10,peptone 7.5,cottonseed power 30,rapeseed oil 30,MgSO4 ·7H2O 0.4,phytic acid 0.5.[Conclusions] Under this optimized conditions,spinosad production come to be 790.3 mg/L,which increases 36.5% after verification test than before optimization.
出处
《农药》
CAS
北大核心
2012年第11期805-808,共4页
Agrochemicals
基金
浙江省农业科技重大专项重点项目(2009C12062)
浙江工业大学自然科学基金(1001105016408)
关键词
刺糖多孢菌
多杀菌素
人工神经网络
粒子群算法
Saccharopolyspora spinosa
spinosad
artificial neural network（ANN）
particle swarm optimization（PSO）
