论文部分内容阅读
差异演化法是一种在连续空间全局最优化问题上简单有效的新型智能进化算法.利用非饱和土壤数据库UNSODA的不同数据集和差异演化算法预测土壤特征曲线参数,获得对不同环境(田间、实验室)和土壤湿度(湿润、干燥)的参数估计,以及土壤质地、样本的选取对参数估计和拟合精度的影响.结果表明:基于实验室中测得的干土壤数据的参数估计和特征值具有最高精度;在四种不同质地的土壤中,黏土的均方根误差最小,土壤特征曲线的拟合精度较好,饱和含水量和残余含水量都较高,砂土的相应参数估计精度则较低;样本容量对于参数的拟合均值影响较小.
Differential evolution method is a kind of new intelligent evolutionary algorithm which is simple and effective in global optimization of continuous space.Using different data sets of unsaturated soil database UNSODA and differential evolution algorithm to predict soil characteristic curve parameters, (Wet and dry), and soil texture and sample selection on the parameter estimation and fitting accuracy.The results show that based on the parameter estimation and eigenvalue of dry soil data measured in the laboratory, With the highest accuracy. Among the four soils, the root mean square error of clay was the smallest, the fitting accuracy of soil characteristic curve was better, the saturated and residual water content were higher, and the corresponding accuracy of soil parameter estimation Lower; sample size has little effect on the fitting mean of parameters.