摘要
提出了一种基于稀疏正则化的方法识别地下水点污染源。首先,对地下水一维对流-弥散方程时域有限元格式进行拉普拉斯变换得到频域方程,然后建立以l1范数项为约束的地下水点污染源识别问题的目标函数,从而克服空间分布稀疏的点污染源识别问题的不适定性;接着,利用交替优化法进行迭代求解。研究结果表明,所提方法能在噪声条件下有效识别地下水点污染源的位置和强度变化。
This paper proposes a method based on sparse regularization to identify groundwater point pollution sources.Firstly,the time domain finite element discretized equation of groundwater one-dimensional convection-diffusion equation is used to obtain the frequency domain equation by Laplace transform,and then the objective function of the groundwater point pollution source identification problem constrained by the l1 norm term is established,thus overcoming the ill-posed problem of the point source identification due to sparse spatial distribution.The identification equation is then solved iteratively using the alternating optimization method.The research results show that the proposed method can effectively identify the position and intensity changes of groundwater pollution sources under noise conditions.
作者
杨方浩
吕中荣
汪利
YANG Fanghao;LV Zhongrong;WANG Li(School of Aeronautics and Astronautics,Sun Yat-sen University,Guangzhou 510006,China)
出处
《中山大学学报(自然科学版)》
CAS
CSCD
北大核心
2020年第5期40-48,共9页
Acta Scientiarum Naturalium Universitatis Sunyatseni
基金
国家自然科学基金(11972380)。
关键词
对流-弥散
稀疏正则化
污染源识别
交替优化法
正则化参数
convection-dispersion
sparse regularization
pollution source identification
alternating optimization method
regularization parameter