基于智能算法推求Van Genuchten方程的参数

引入遗传算法、模拟退火算法、粒子群算法等智能算法,应用于土壤水分特征曲线Van Genuchten方程参数确定的非线性拟合问题,分别针对粉壤土、细砂土、砾石的吸湿和脱湿曲线得到了Van Genuchten方程的相关参数。结果表明:三种算法的模拟结果非常接近,模拟精度较高;不同算法在细砂土和砾石上的模拟结果十分接近,而遗传算法在粉壤土上应用的误差稍大;与实测结果相比,模拟结果在土壤水吸力高于1000cm的高吸力范围和曲线由快速下降转为平稳阶段的过渡阶段(土壤水吸力约为200cm)模拟误差较大。

Estimating parameters of Van Genuchen equation based on intelligent algorithms

Intelligent algorithms including genetic algorithm,simulated annealing algorithm and particle swarm optimization algorithm were applied in the parameters determination of Van Genuchten equation.Case studies were carried out on both the imbibition and drainage curves of silt loam soil,fine sandy soil and gravel.The accuracy of simulation was high;no significant differences were founded between results by different methods.Comparison between soils illustrate that the results of different methods are high consistent with each other on fine sandy soil and gravel,but when it comes to the silt loam soil,genetic algorithm methods result in a less accurate one.Comparison between simulation and observation show large error occurs both at the high suction phrase with suction over 1000 cm and the transition phrase from quickly descended to stable with suction about 200cm.