考虑黄土结构变形的Philip入渗模型参数预报

发布时间：2018-01-16 01:25

本文关键词：考虑黄土结构变形的Philip入渗模型参数预报　出处：《人民黄河》2016年09期 　论文类型：期刊论文

【摘要】：基于黄土高原区区域尺度大田耕作土壤的水分入渗试验,考虑黄土备耕头水地土壤结构的变形特性,建立了Philip半理论半经验入渗模型参数的BP神经网络预报模型,实现了以土壤基本理化参数为输入变量、Philip模型参数为输出变量的BP预报。对Philip模型中吸湿率S、稳渗率A以及90 min累计入渗量的预测值与实测值进行比较,结果显示:吸湿率S的平均相对误差为1.41%,稳渗率A的平均相对误差为2.81%,90 min累计入渗量的平均相对误差为1.75%,三者的平均相对误差值均在3%以下,预测精度很高。这表明以土壤基本理化参数为输入变量的BP神经网络预测是可行的,考虑备耕头水地土壤结构变形使得BP预报结果更符合土壤水分入渗实际状况。
[Abstract]:The Loess Plateau region scale field tillage soil water infiltration test based on considering the deformation characteristics of soil water in Loess plowing head structure, established the Philip semi theoretical and semi empirical infiltration prediction model of BP neural network model parameters, in order to realize the basic soil physicochemical parameters as input variables, the parameters of Philip model for forecasting output variables BP the absorption rate of Philip model. The S, A and the steady infiltration rate of 90 min cumulative infiltration into the predicted values were compared with the measured values, the results showed that the average relative error of moisture absorption rate of S was 1.41%, the average relative error of steady infiltration rate of A was 2.81%, the average relative error of 90 min cumulative infiltration into 1.75%, are below 3% the average value of the relative error of the three, the prediction accuracy is very high. This shows that the soil physicochemical parameters as input variables of BP neural network prediction is feasible, considering soil structure deformation of plowing head water making The results of BP forecast are more consistent with the actual conditions of soil water infiltration.

【作者单位】：太原理工大学;
【基金】：国家自然科学基金资助项目(40671081)
【分类号】：S152.7
【正文快照】： 降水或者灌溉水等地面水进入土壤的过程称为土壤水分入渗。土壤水分入渗是农田水循环的主要过程之一。然而,影响入渗过程的因素众多且影响机制复杂多变,使得获取准确的入渗参数难度很大。基于此,国内外专家学者进行了许多试验研究,其中:解文艳等[1-3]分析了土壤质地、结构、初

【相似文献】