为了研究油菜茎秆抗风性能，以提供油菜抗倒栽培和倒伏预测的理论依据和技术支撑。选用华杂62和金油杂158两个品种，设置种植密度（30×104、45×104和60×104株·hm-2）、施氮量（120、240和360 kg·hm-2）和播期（9月20日、10月1日和10月10日）各3个水平的单因素试验，测定花期（初花期、盛花期和终花期）茎秆自振、风振、弯曲特性等力学指标和终花后20 d的倒伏指数。结果表明：增大种植密度，花期油菜茎秆自振频率、阻尼比、风振频率、弹性模量和弯曲应力均先增大后减小，最大振幅和倒伏指数先减小后增大，抗倒性先增后降；增大施氮量，花期油菜茎秆自振频率、阻尼比、风振频率、弹性模量和弯曲应力均逐渐减小，最大振幅和倒伏指数逐渐增大，抗倒性逐渐降低；推迟播期，花期油菜茎秆自振频率、阻尼比和风振频率均逐渐增大，弹性模量、弯曲应力最大振幅和倒伏指数逐渐减小，抗倒性逐渐增强。建立了华杂62和金油杂158的BP神经网络倒伏指数预测模型，经检验，预测值与实测值的均方根误差分别为0.157和0.177，平均绝对百分误差分别为7.68%和8.30%，精确度较高，可实现于花期预测油菜终花后的抗倒性。
The wind resistance of rapeseed stem was studied to provide a theoretical basis and technical support for lodging resistance cultivation and lodging prediction. Two varieties of HZ 62 and JZ 158 were selected, and the single factor experiments at three levels of planting density (30 × 104, 45 × 104 and 60 × 104 plants·hm-2), nitrogen application rate ( 120, 240 and 360 kg·hm-2 ) and sowing time ( Sep. 20, Oct. 1 and Oct. 10 ) were set up. The mechanical indexes of stem vibration, wind vibration and bending characteristics at flowering stage (early flowering, full flowering and end flowering) and lodging index 20 days after end flowering were measured. The results showed that by increasing the planting density, the natural vibration frequency, damping ratio, wind vibration frequency, elastic modulus and bending stress of rapeseed stem at flowering stage increased first and then decreased, the maximum amplitude and lodging index decreased first and then increased, and the lodging resistance increased first and then decreased. The natural vibration frequency, damping ratio, wind vibration frequency, elastic modulus and bending stress of rapeseed stem decreased, the maximum amplitude and lodging index increased, and the lodging resistance decreased by increase of nitrogen application rate. The natural vibration frequency, damping ratio and wind vibration frequency of rapeseed stem at flowering stage increased gradually, the elastic modulus, maximum amplitude of bending stress and lodging index decreased gradually, and the lodging resistance increased gradually by delay of sowing date, t. The BP neural network model for predicting lodging index of HZ 62 and JZ 158 was established.