基于改进的基本粒子群算法的Van Genuchten方程参数优化

Parameter Calibration of the Van Genuchten Equation based on Modified Particle Swarm Optimization

针对目前土壤水分特征曲线Van Genuchten方程参数优化的不足,引入动态调整惯性权重对基本粒子群算法进行改进,使惯性权重随着迭代次数以及不同粒子与最优粒子之间的距离大小而变化,并将其运用到Van Genuchten方程参数识别,最后进行了模型验证和误差分析。结果表明改进惯性权重的粒子群算法计算精度高,适用性强。从3种准则函数的优化结果可以看出,加权耦合的绝对与相对标准差最小准则在参数优化理论上值得进一步研究,在Van Genuchten方程参数拟合问题中值得推荐,从而为Van Genuchten方程参数的优化求解提供了一条新途径。

Focusing on the issue of parameter calibration of the Van Genuchten equation,an innovated dynamic regulation of inertia weight was introduced into the normal particle swarm optimization（ PSO） to ensure inertia weight change both with the number of iterations and with the distance between various particles to optimal particles. Furthermore the new method was applied in the comparisons of criterion functions in parameters identification of the Van Genuchten equation. Finally the new method was verified in model validation and error analysis. The results showed that the new PSO method with improved inertia weights displayed higher accuracy and wider applicability. Besides,optimization results based on the three criterion functions revealed that the criterion of minimum absolute and relative standard deviations in weighted coupling proved to be an optimization criterion deserving further study and prior selection in parameter estimation for the Van Genuchten equation in soil science. The new method proposed an innovated approach for the calibration of Van Genuchten equation parameters.