基于人工神经网络和遗传算法优化用于栽培皱环球盖菇的农业废弃物培养基  

Formula optimization of culture substrate compounded by agricultural waste based on artificial neural network and genetic algorithm for cultivating Stropharia rugosoannulata

作　　者：马婧嘉 李雅茹 雷萍[1] 张文隽[1] 吴亚召[1] MA Jingjia;LI Yaru;LEI Ping;ZHANG Wenjun;WU Yazhao(Microbiology Institute of Shaanxi,Xi’an 710043,Shaanxi,China)

机构地区：[1]陕西省微生物研究所,陕西西安710043

出　　处：《菌物学报》2024年第11期143-151,共9页Mycosystema

基　　金：西安市科技局西安市创新能力强基计划-农业技术研发项目(21NYYF0040);陕西省科学院“一所一品”项目(2021K-02)。

摘　　要：本文采用正交试验设计对农业废弃物的皱环球盖菇菌丝培养基配方进行优化,并对正交试验结果分别进行方差分析(ANOVA)和人工神经网络(ANN)分析。研究表明:ANOVA优化的培养基配方为木屑5 g/L、茶梗5 g/L、油菜秸秆30 g/L、小麦秸秆30 g/L;ANN优化的配方为木屑5 g/L、茶梗5 g/L、油菜秸秆29.99 g/L、小麦秸秆15.36 g/L。ANOVA和ANN优化条件下获得的实际值和预测值基本一致。ANN拟合模型的相关系数(R)、均方根误差(RMSE)、平均绝对误差(MAE)分别为0.92896、0.041463、0.00418,具有更高拟合能力和准确性,拟合效果更好,更适合应用于农业废弃物的皱环球盖菇菌丝培养基配方优化。The formula of culture substrate compounded by agricultural waste for cultivating Stropharia rugosoannulata was optimized by orthogonal experimental design,and the orthogonal test results were varified by variance analysis(ANOVA)and artificial neural network(ANN).The results showed that the optimal substrate formula optimized by ANOVA was sawdust 5 g/L,tea stalk 5 g/L,rape straw 30 g/L and wheat straw 30 g/L and that optimized by ANN was sawdust 5 g/L,tea stem 5 g/L,rape straw 29.99 g/L and wheat straw 15.36 g/L.ANOVA and ANN results are basically the same in the optimal formulas.The correlation coefficient(R),root mean square error(RMSE)and mean absolute error(MAE)of ANN fitting model are 0.92896,0.041463 and 0.00418 respectively.ANN has higher fitting ability and accuracy,and the fitting effect is also better,so it is more suitable for optimizing the formula of S.rugosoannulata culture substrate compounded by agricultural wastes.

关 键 词：人工神经网络 皱环球盖菇 农业废弃物 培养基 优化 

分 类 号：S646[农业科学—蔬菜学] TP18[农业科学—园艺学]