基于BP神经网络和显微结构的甘蓝型油菜抗倒性评价  

Evaluation of lodging resistance on BP neural network and microstructure in rapeseed(Brassica napus L.)

作　　者：刘唐兴[1,2] 官春云[1] 雷冬阳[2] 付爱斌[2] 陶甲寅[2] 梁勇[2] 

机构地区：[1]湖南农业大学油料作物研究所,湖南长沙410128 [2]湖南生物机电职业技术学院,湖南长沙410127

出　　处：《中国油料作物学报》2009年第3期322-326,共5页Chinese Journal of Oil Crop Sciences

基　　金：国家948项目(2003-Q04);湖南教育厅项目(06D057)

摘　　要：以不同抗倒性的18个油菜品种为材料,分析了农艺性状、物理性状及主茎显微结构,结果表明:抗倒性强的品种中双9号根冠比(干重)为22.04%,显著高于其中的9个品种,其茎秆各茎段和主花序的组成比例适中;在抗倒性强品种中双9号和富油3号的茎秆纵向显微结构中,维管束部分细胞排列比较整齐,细胞排列方式与主茎方向呈平行状态的线形分布;而抗倒性一般的材料如XYY2和XYY6,其维管束部分细胞排列存在许多不规则现象;用BP神经网络模型对农艺性状和物理性状与倒伏指数的关系进行模拟预测,其倒伏指数的拟合率高,两品种倒伏指数的预测值处于置信区间。Rapeseed （Brassica napus L. ） is one of the most important oil Crops in the world. In this study, agronomic and physics characteristics, and the microstructure of stem were analyzed using 18 rapeseed varieties. The results showed that ratio of root/top（dried） of Zhongshuang 9 was 22.04%, which was significantly higher than other 9 rapeseed varieties. Zhongshuang 9 has moderate proportion of stem and main inflorescence. The results indicated that cells in vascular bundle of Zhongshuang 9 and Fuyou 3 were lined linearly along the direction of stem growth and were more regularly than XYY 2 and XYY 6. The relationship between lodging index and characteristics of agronomic and physics was fitted by BP neural network model： the fitting ratio of lodging index was rather high, and the fitting ratio to lodging index forecasted of two varieties was in confident range.

关 键 词：甘蓝型油菜 农艺性状 抗倒性 显微结构 BP神经网络 

分 类 号：S565.401[农业科学—作物学]