地下水污染源反演的Hooke Jeeves吸引扩散粒子群混合算法

根据污染物质量浓度监测数据进行地下水污染源反演是一类典型的地下水逆问题，该问题可转化为决策变量为污染源位置和强度的最优化问题进行求解。基于Hooke-Jeeves粒子群混合算法，引入吸引扩散粒子群(ARPSO)算法的粒子群发散算子,保证混合算法的种群多样性，并提出HJ-ARPSO混合算法,再结合地下水污染物迁移模型MT3DMS反演地下水污染源的位置和强度信息。在已知污染源位置和未知污染源位置两种情形下,分别利用HJ-ARPSO算法、HJ-PSO算法和GA算法进行地下水污染源反演。在两种情形下,HJ-ARPSO算法均具有较高的寻优成功率(分别对应为100%和90%)；与之相比，未引入粒子群发散算子的HJ-PSO算法在未知污染源位置情形下其寻优成功率迅速降为60%；GA算法寻优效率则最低。算例结果表明，HJ-ARPSO算法是一种有效的地下水污染源反演优化算法。

Groundwater Contaminant Source Identification by Hybrid HookeJeeves and Attractive Repulsive Particle Swarm Optimization Method

Identification and estimation of the groundwater contaminant source based on the obtained monitoring data is a groundwater inverse deduction problem. These problems can be defined as optimization problems which the location and strength of contaminant sources are taken as the decision variables. In this study, a new method named hybrid Hooke-Jeeves attraction and repulsion particle swarm optimization method (HJ-ARPSO), based on the hybrid Hooke Jeeves particle swarm optimization method, is proposed and incorporated with the contaminant transport model MT3DMS to identify the location and strength of groundwater contaminant source. In HJ-ARPSO method, the population diversity is guaranteed by combined with repulsive operator in ARPSO. Afterwards, HJ-ARPSO method, HJ-PSO method, and GA method are adopted to identify two source identification cases (under known source position and under unknown source position). Under both cases, HJ-ARPSO method achieved high optimization rate (100 percent and 90 percent respectively); while the optimization rate of HJ-PSO method (without coupling the repulsive operator) is reduced to 60 percent under unknown source position case; among these three methods, the efficiency of the GA method achieved the lowest efficiency. The results indicate that HJ-ARPSO is applicable to groundwater contaminant source identification.