论文部分内容阅读
雄先型核桃雄花疏除(去雄)是提高产量的重要管理措施,为提高核桃去雄的效率,建立二次回归与BP神经网络模型。分别以乙烯利、赤霉素和甲哌鎓为自变量和核桃雄花脱落率为响应指标,进行田间建模试验,建立了二次多项式回归方程和BP神经网络模型,并于翌年进行BP模型田间确认试验。试验数据分为训练集、确认集和试验集,中心组合(二次旋转回归试验设计)田间建模试验得到的20组数据随机划为训练集(17)和确认集(3)数据,试验集为翌年田间确认试验得到的数据,BP神经网络的拓扑结构为3-5-1。1BP神经网络对确认集样本的预测值误差分别为1.3550%、0.4291%、0.3538%;2BP神经网络的预测值与田间确认试验结果相差为2.04%,回归预测值与田间确认试验结果相差为3.12%;3BP神经网络预测比回归预测提高预测精度1.0%以上。将二次多项式逐步回归分析和BP神经网络方法有效的结合使用,既可明确各因子的作用效应亦可得到相对准确的预测结果。
Male first type walnut male flower sparse (to male) is to improve yield an important management measure, in order to improve the efficiency of walnut to male, establish quadratic regression and BP neural network model. Taking ethephon, gibberellin and mepiquat as the independent variables and the rate of male flower off of walnut as the response index, field modeling experiments were carried out to establish the quadratic polynomial regression equation and BP neural network model, and the BP model field Confirm the test. The experimental data are divided into training set, validation set and test set, and the center set (quadratic rotation regression test design) field data set is randomly divided into training set (17) and validation set (3) The topological structure of BP neural network for the data obtained in the field validation test for the following year is 3-5-1. The prediction errors of BP neural networks for the validation set samples are 1.3550%, 0.4291% and 0.3538%, respectively. The predictive value of 2BP neural network And field validation test results were 2.04%, the regression prediction and field validation test results were 3.12% difference; 3BP neural network prediction than the regression prediction to improve the prediction accuracy of 1.0% or more. The stepwise regression analysis of quadratic polynomials and the BP neural network method can effectively combine the effect of each factor and the relatively accurate prediction results.