玉米性状的计算机神经网络模型研究
|
摘要
利用神经网络模型研究了品种产量与其他性状的关系,发现神经网络模型在不同次训练结果中的平均绝对误差和平均绝对相对误差的变异系数较小,但在不同次训练结果中的各个自变量相对重要性的变异系数较大。因此,神经网络较适合用于产量的预测,在用于研究各个性状对产量的相对重要性时,应采取多次重复求平均值的方法,以减少分析结果的误差。在所有的性状中,穗粗、穗长对产量的相对重要性最大,随后是百粒重、株高和生育期等。同时,还分析了使产量最大时各个性状的最佳组合。该研究结果对黄淮海地区玉米育种目标的制定具有一定的参考价值。
In order to solve the problems of low level of information and lack of depth mining of experimental data in maize breeding, the relation of yield and other characters was conducted in this study by using neural network model. It was showed that the CV(the coefficient of variation) of the mean absolute error and the mean absolute relative error was small, and the CV of relative importance of each independent variable was larger. Therefore, the neural network model was suitable for production forecast. When applying in the researches on relative importance of each trait to yield, we should calculate the average by repetition, in order to to reduce the error of results. Among all the traits, ear diameter and ear length showed the greatest relative importance to yield, followed with 100-grain weight, plant height and growth period. Meanwhile, the optimal combination of the traits for the maximum yield was studied. The research results had certain reference value for the formulation of maize breeding targets in Huanghuaihai Area.
In order to solve the problems of low level of information and lack of depth mining of experimental data in maize breeding, the relation of yield and other characters was conducted in this study by using neural network model. It was showed that the CV(the coefficient of variation) of the mean absolute error and the mean absolute relative error was small, and the CV of relative importance of each independent variable was larger. Therefore, the neural network model was suitable for production forecast. When applying in the researches on relative importance of each trait to yield, we should calculate the average by repetition, in order to to reduce the error of results. Among all the traits, ear diameter and ear length showed the greatest relative importance to yield, followed with 100-grain weight, plant height and growth period. Meanwhile, the optimal combination of the traits for the maximum yield was studied. The research results had certain reference value for the formulation of maize breeding targets in Huanghuaihai Area.
引文
[1] 鲁珊,肖荷霞,毛彩云,等.玉米杂交种主要农艺性状的相关和通径分析[J].安徽农业科学,2017,45(21):26-27,58.
[2] 史新海,李可敬,孙为森,等.山东省不同年代玉米杂交种主要农艺性状演变规律的研究[J].玉米科学,2000,8(2):33-35.
[3] 李晓花,林永明,谢淑芳,等.玉米产量与主要农艺性状的灰色关联度分析[J].云南农业科技,2016(1):13-15.
[4] 焦李成,杨淑媛,刘芳,等.神经网络七十年:回顾与展望[J].计算机学报,2016,39(8):1697-1716.
[5] 佟屏亚.中国玉米生产的发展方向——质疑“超级玉米”“超高产”[J].农业科技通讯,2005(8):11-13.
[6] 崔俊明.新编玉米育种学[M].北京:中国农业科学技术出版社,2007:12-57.
[7] 周得宝,王娟,王五洲,等.夏玉米品系(种)的产量比较及主要农艺性状的关联度分析[J].安徽农业科学,2017,45(26):48-51,78.
[8] 何文昭.玉米产量和主要农艺性状杂种优势遗传分析[D].北京:中国农业科学院,2017.
[9] 史新海,赵格.山东省紧凑型玉米杂交种主要农艺性状对产量的影响及其演变规律的研究[J].玉米科学,2003,11(2):59-61,85.
[10] 张泽民,刘丰明,李雪英.河南省1963～1993年玉米杂交种籽粒产量及其组成性状的遗传增益[J].作物学报,1998,24(2):182-186.
[11] 丁山,郭去,宋军,等.玉米主要性状与产量的回归模型及相关分析[J].西南农业学报,2008,21(5):1226-1230.
[12] 卓德众,刘启华,郭红甫.灰色关联度分析法在玉米育种中的应用[J].玉米科学,1996,4(3):31-34.
[13] 刘帆,石海春,余学杰.玉米果穗主要性状与产量间的相关与通径分析[J].玉米科学,2005,13(3):17-20.
[14] 陈发波,杨克诚,荣廷昭,等.西南及四川区试玉米组合主要性状分析及育种对策探讨[J].玉米科学,2007,15(4):41-45.
[15] 闫海霞,柳家友,吴伟华,等.含非洲血缘玉米杂交种主要穗部性状与单株产量的相关及通径分析[J].杂粮作物,2007,27(6):379-381.
[2] 史新海,李可敬,孙为森,等.山东省不同年代玉米杂交种主要农艺性状演变规律的研究[J].玉米科学,2000,8(2):33-35.
[3] 李晓花,林永明,谢淑芳,等.玉米产量与主要农艺性状的灰色关联度分析[J].云南农业科技,2016(1):13-15.
[4] 焦李成,杨淑媛,刘芳,等.神经网络七十年:回顾与展望[J].计算机学报,2016,39(8):1697-1716.
[5] 佟屏亚.中国玉米生产的发展方向——质疑“超级玉米”“超高产”[J].农业科技通讯,2005(8):11-13.
[6] 崔俊明.新编玉米育种学[M].北京:中国农业科学技术出版社,2007:12-57.
[7] 周得宝,王娟,王五洲,等.夏玉米品系(种)的产量比较及主要农艺性状的关联度分析[J].安徽农业科学,2017,45(26):48-51,78.
[8] 何文昭.玉米产量和主要农艺性状杂种优势遗传分析[D].北京:中国农业科学院,2017.
[9] 史新海,赵格.山东省紧凑型玉米杂交种主要农艺性状对产量的影响及其演变规律的研究[J].玉米科学,2003,11(2):59-61,85.
[10] 张泽民,刘丰明,李雪英.河南省1963～1993年玉米杂交种籽粒产量及其组成性状的遗传增益[J].作物学报,1998,24(2):182-186.
[11] 丁山,郭去,宋军,等.玉米主要性状与产量的回归模型及相关分析[J].西南农业学报,2008,21(5):1226-1230.
[12] 卓德众,刘启华,郭红甫.灰色关联度分析法在玉米育种中的应用[J].玉米科学,1996,4(3):31-34.
[13] 刘帆,石海春,余学杰.玉米果穗主要性状与产量间的相关与通径分析[J].玉米科学,2005,13(3):17-20.
[14] 陈发波,杨克诚,荣廷昭,等.西南及四川区试玉米组合主要性状分析及育种对策探讨[J].玉米科学,2007,15(4):41-45.
[15] 闫海霞,柳家友,吴伟华,等.含非洲血缘玉米杂交种主要穗部性状与单株产量的相关及通径分析[J].杂粮作物,2007,27(6):379-381.