基于Kostiakov二参数入渗模型参数的BP预测

BP Prediction Method Based on Two Parameters Kostiakov Infiltration Model

基于黄土高原区大田耕作土壤的水分入渗试验,建立了Kostiakov二参数入渗模型参数的BP神经网络预测,实现了以土壤基本理化参数为输入变量,Kostiakov二参数模型参数为输出变量的BP预测方法,并分别对二参数模型中的入渗系数k、入渗指数α以及90min累积入渗量H进行了预测值与实测值的精度比较,结果显示对入渗系数k实现BP预测的平均相对误差为6.082 3%,入渗指数α的平均相对误差为1.045 9%,90min累积入渗量H的平均相对误差为4.973 5%,三者的平均相对误差值均在7%以下,预测精度较高,预测效果较好,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。研究结果为获取准确的入渗参数提供技术手段,进而为提高农业灌溉水管理水平和灌水效率提供支撑。

Based on water infiltration test in field farming soil of Loess Plateau Region, this research established a BP neural network prediction model based on two parameters Kostiakov infiltration model, in which the soil physicochemical parameters are selected as input variables and the two parameters of Kostiakov infiltration model are selected as output variables. At the same time, the predic- ted values of infiltration coefficient k, infiltration index a and 90 min cumulative infiltration H are compared with measured values. The results show that the average relative error of k, a, and H is 6. 082 3%, 1. 045 9% and 4. 973 5%, respectively, and all are less than 7 %. So the BP neural network prediction model is proved to be feasible and has a relatively high prediction accuracy and effect. These findings will provide technical methods for getting accurate infiltration parameters to provide support for improving agricultur al irrigation water management level and irrigation efficiency.