摘要
凉山半细毛羊初生到断奶的5个生产性状(羔羊初生重、羔羊断奶重、初生-断奶日增重、断奶毛长度、断奶毛细度)预测成年羊毛(2.5年)的3个生产性状(成年剪毛量、成年毛长度、成年毛细度),用Matlab6.5软件构建广义回归神经网络预测模型,并通过预测结果和实测结果的统计分析验证本研究所构建的广义回归神经网络预测模型的有效性。预测结果和实测结果的统计分析结果显示该预测模型具有较高的准确性,基于广义回归神经网络构建的预测模型在凉山半细毛羊成年羊毛3个性状上的预测被证明是有效的。
We predicted three 25-year traits of Liangshan semifine-wool sheep （wool clip, staple length and wool fibre diameter at 30 months） based on five traits （birth weight, weaning weight, daily gain weight from birth to weaning, weaning staple length and weaning staple fibre diametel） ranging from birth to weaning, The generalized regression neural network prediction model was built by using Matlab software. The validity of the generalized regression neural network prediction model was validated through the statistical confidence of the prediction result and practical ease. The prediction resuhs have higher consistency with the practical case. The generalized regression neural network prediction model was available on predicting three 2.5-year wool traits of Liangshan semifine-wool sheep.
出处
《现代畜牧兽医》
2010年第2期60-64,共5页
Modern Journal of Animal Husbandry and Veterinary Medicine
基金
德国联邦科技与教育部(BMBF)资助(CHN 00316)
关键词
广义回归神经
预测模型
预测
凉山半细毛羊
生产性状
Generalized regression neural network
Prediction model
Prediction
Liangshan semi fi ne-wool sheep
Traits
