基于灰色理论-支持向量机的Gardner模型参数的预报模型

Prediction Model of Gardner Model Parameters Based on Gray Theory-Support Vector Machine

为探索获取土壤水分特征曲线模型参数新途径,以山西省黄土高原区农耕土壤为试验材料,进行了系列土壤水分特征曲线试验,拟合得到了Gardner模型参数a和b,同时测定了土壤基本理化参数,建立了由Gardner模型参数a、b和土壤基本理化参数组成的数据样本。采用灰色关联理论,分析了土壤各基本理化参数与Gardner模型参数之间的关联度,基于支持向量机理论建立了以关联度较大的土壤容重、黏粒含量、粉粒含量、有机质含量、全盐量为输入因子,以Gardner模型参数a和b为输出因子的预报模型。研究表明:以土壤容重、黏粒含量、粉粒含量、有机质含量、全盐量为输入因子,对Gardner模型参数a与b进行预报是可行的,参数a、b的实测值与预测值之间相对误差的平均值分别为3.96%、4.68%,吻合度高,预测效果好。该研究结果可为获取土壤水分特征曲线模型参数提供技术手段,同时可促进土壤传输函数理论的发展。

In order to explore the new method of obtaining soil moisture characteristic curve model,the soil moisture characteristic curve test was carried out with the soil of the Loess Plateau of Shanxi Province as the experimental material.The Gardner model parameters a and b and the soil basic physical and chemical parameters were obtained,a data sample consisting of Gardner model parameters a,b and soil basic physical and chemical parameters was established.The relationship between soil physical and chemical parameters and Gardner model parameters was analyzed by Grey Relational Analysis.Based on the theory of support vector machine（SVM）,a prediction model with soil bulk density,clay content,silt content,organic matter content and total salt content as input factors and Gardner model parameter a and b as output factors was established.The results show that it is feasible to predict the parameters of the Gardner model a and b with the soil bulk density,clay content,powder content,organic matter content and total salt content as the input factors;the average relative error between the measured values and the predicted values of parameters a and b is 3.96% and 4.68%,respectively,and the coincidence is high.The results of this study can provide technical means for obtaining parameters of soil moisture characteristic curve model,and promote the development of soil transfer function theory.