区域尺度上土壤入渗模型特征参数传输函数的研究

发布时间：2018-05-03 07:58

本文选题：土壤传输函数 + BP神经网络　；参考：《太原理工大学》2017年博士论文

【摘要】：土壤水分运动的入渗参数决定着地面径流、灌溉水和降水转换为土壤水的速度和分布,进而也影响着灌水质量和灌溉效果(灌水均匀度、储水效果和灌水效率),它是合理确定不同灌水条件下技术参数的主要依据。因此,土壤水分入渗参数的研究便成为改进地面灌水效果、提高灌溉水利用率等技术中迫切需要解决的关键问题。文章以区域尺度上规模化耕作土壤大田入渗试验为依据,利用土壤传输函数理论,在系统地研究土壤入渗模型参数与容易获得的土壤常规理化性状参数间的定量关系基础上,建立了土壤入渗参数与土壤常规理化性状参数间的多元线性、非线性、BP神经网络模型传输函数,实现了通过土壤基本理化参数预测土壤入渗参数的目的。其研究成果可以为实施优化地面灌溉灌水技术参数提供强有力的技术支持,同时,在一定程度上丰富了土壤传输函数理论的发展。文章的主要研究结果表明:(1)土壤水分入渗参数受多种因素的复杂影响,包括土壤结构、质地、含水量、有机质含量、含盐量等。土壤结构、质地、含水量、有机质含量等常规理化参数与土壤入渗参数间存在线性或非线性的定量关系,其中土壤有机质含量与入渗能力成正比,土壤干容重(结构)、初始含水量、黏粒含量(质地)、含盐量等与入渗能力成反比。(2)基于土壤常规理化参数,采用传输函数预测土壤入渗参数是可行的。从预测结果来看,Kostiakov两参数入渗模型、Kostiakov-Lewis三参数入渗模型和Philip入渗模型的显著性检验都是显著的。引入每个自变量的显著性检验和回归方程的显著性检验F值均大于相应的F0.95,说明传输函数的回归系数是显著的,回归方程也是显著的。随着引入变量个数的增加,模型的复相关系数逐渐向1靠近,相关性越来越好,最终各个函数的多元线性回归模型计算值与实测值之间的全相关系数R为0.9~0.95,R2均大于0.81,表明由实验数据所得的多元线性传输函数相关性较好,用土壤常规理化特征参数预测土壤入渗能力及其入渗参数是完全可行的。(3)Kostiakov两参数、Kostiakov-Lewis三参数和Philip两参数入渗模型的参数预报模型中合理的输入变量为:土壤砂粒含量、黏粒含量、不均匀系数、曲率系数、体积含水量、重量含水量、干容重、有机质含量等。(4)Kostiakov模型入渗参数的线性、非线性、BP预测模型相对误差均较低,计算精度较高,相关性较好,拟合度高,均能实现对土壤入渗参数的预测。(5)Kostiakov-Lewis模型入渗参数的线性、非线性、BP预测模型的相对误差都在合理范围之内,相对误差低,显著性强,相关性良好,拟合度高,也能够实现对土壤入渗参数的预测。(6)Philip模型入渗参数的线性、非线性、BP预测模型的相对误差虽在合理范围内,具有一定的可行性,但相对误差值偏大,精度较低。(7)在Kostiakov、Kostiakov-Lewis和Philip模型参数的预测模型中,Kostiakov-Lewis模型参数的预测效果最好,能够很好的反映土壤水分入渗的过程,特别是对于长历时水分入渗,较其它模型具有更高和更稳定的预测精度;BP预测模型的精度要高于多元非线性预测模型,但模型的稳定性不如多元非线性模型。
[Abstract]:The infiltration parameters of soil moisture movement determine ground runoff, irrigation water and precipitation conversion to the velocity and distribution of soil water, and also affect irrigation quality and irrigation effect (irrigation uniformity, water storage effect and irrigation efficiency). It is the main basis for the rational determination of technical parameters under different irrigation conditions. Therefore, soil water infiltration parameters The research has become the key problem to be solved urgently in improving the effect of ground irrigation and improving the utilization rate of irrigation water. The paper systematically studies the soil infiltration model parameters and the regular physical and chemical properties of soil obtained by soil infiltration function theory on the basis of regional scale tillage soil infiltration test on regional scale. On the basis of quantitative relationship between parameters of the soil, the multiple linear, nonlinear, nonlinear, BP neural network model transmission function between soil infiltration parameters and soil conventional physical and chemical parameters is established, and the purpose of predicting soil infiltration parameters through soil basic physical and chemical parameters is realized. The research results can be used to optimize the technical parameters of irrigation and irrigation. It provides strong technical support and, at the same time, enriches the development of soil transport function theory. The main research results of this article show that: (1) soil water infiltration parameters are complex influenced by various factors, including soil structure, texture, water content, organic matter content, salt content, soil structure, texture, water content, organic matter content. There is a linear or nonlinear quantitative relationship between the conventional physical and chemical parameters and the infiltration parameters of soil. The content of soil organic matter is proportional to the infiltration capacity. The dry bulk density (structure), initial water content, clay content (texture) and salt content are inversely proportional to the infiltration capacity. (2) based on the conventional physical and chemical parameters of soil, the transfer function is used to predict soil The soil infiltration parameters are feasible. From the prediction results, the Kostiakov two parameter infiltration model, the Kostiakov-Lewis three parameter infiltration model and the Philip infiltration model are all significant tests. The significance test of each independent variable and the significance test of the regression equation are greater than the corresponding F0.95, indicating the regression of the transfer function. The coefficient is significant, and the regression equation is also significant. With the increase of the number of variables introduced, the complex correlation coefficient of the model is gradually closer to 1, the correlation is getting better and better. Finally, the total correlation coefficient R between the calculated value and the measured value of the multiple linear regression model of each function is 0.9~0.95, and R2 is more than 0.81, indicating the diversity obtained from the experimental data. The correlation of linear transmission function is good. It is completely feasible to predict soil infiltration capacity and infiltration parameters with soil conventional physicochemical parameters. (3) Kostiakov two parameters, Kostiakov-Lewis three parameters and Philip two parameter infiltration model parameter prediction model, the reasonable input variation is: soil sand content, clay content, inhomogeneous system Number, curvature coefficient, volumetric water content, water content, dry bulk density, organic matter content and so on. (4) the linear, nonlinear, nonlinear, nonlinear, relative error of the Kostiakov model is low, the calculation precision is higher, the correlation is better, and the fitting degree is high. (5) the line of infiltration parameters of the Kostiakov-Lewis model. The relative error of the model is within a reasonable range, the relative error is low, the relative error is low, the correlation is strong, the correlation is good, the fitting degree is high, and the prediction of soil infiltration parameters can be realized. (6) the linear, nonlinear and the relative error of the BP prediction model of the Philip model is feasible, although the relative error of the BP model is in a reasonable range, but it is feasible, but it has certain feasibility, but the relative error of the model is reasonable. The relative error value is large and the precision is low. (7) in the prediction model of Kostiakov, Kostiakov-Lewis and Philip model parameters, the prediction effect of Kostiakov-Lewis model parameters is the best. It can reflect the infiltration process of soil moisture well, especially for long diachronic water infiltration, which has higher and more stable prediction accuracy than other models; BP The accuracy of the prediction model is higher than that of the multivariate nonlinear prediction model, but the stability of the model is not as good as that of the multivariate nonlinear model.

【学位授予单位】：太原理工大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：S152.7

【相似文献】