基于粒子群优化的多目标体中梯电阻率异常反演  

Inversion of multi-anomalies in resistivity profiling based on particle swarm optimization

作　　者：崔益安[1] 纪铜鑫[1] 李溪阳[1] 朱肖雄[1] 

机构地区：[1]中南大学地球科学与信息物理学院,长沙410083

出　　处：《地球物理学进展》2013年第4期2164-2170,共7页Progress in Geophysics

基　　金：国家自然科学基金项目(41274122);中南大学中央高校基本科研业务费专项资金项目(2011QNZT014)联合资助

摘　　要：在采用中梯装置的电阻率剖面法应用中,特别是在环境、水文和工程等领域,经常遇到需要对多个异常目标体进行快速定位以便及时进行相应处理的情况.利用倾斜椭球体来近似模拟这些电阻率异常目标体,使得正演计算可以采用解析表达式来实现,提高正演计算的时效性.相应的地球物理模型也可简化为由椭球体个数、中心点位置、倾角、轴径以及电阻率等参数构成的粒子,多个粒子组成的粒子群在粒子群优化算法的控制下在给定的搜索空间中并行地搜索最优模型.通过对粒子群优化算法参数的合理设计,利用其良好的全局与局部均衡的搜索能力实现对多个异常目标体的同时反演.数值实验结果表明该反演方法能有效实现对多个目标体的同时反演,计算速度快、反演拟合精度较高、同时具有一定的抗噪音能力.快速的多目标体反演,可以实时准确的定量解译中梯剖面法圈定的异常目标体,较好地满足工程等领域的高时效性要求.In the applications of direct current resistivity profiling with central gradient array, especially in environmental, hydrological or engineering fields, locating several anomalies in a profile as quickly as possible is a very common requirement. Based on simulating the anomalies by inclined ellipsoids and using superposition principle, the forward can be performed by analytic expressions quickly and effectively. Then, the geophysical model can be reduced to a particle including a set of parameters of the ellipsoids, such as the number of ellipsoids, positions of each ellipsoid＇s central point, dip angle, radius and resistivity. The particles form a particle swarm, and search the best model in a given model space in parallel by the means of particle swarm optimization algorithm. To be set with reasonable parameters, the particle swarm optimization algorithm is able to invert multi-anomalies simultaneously by taking advantages of its good balance between global and local searching capability. The results of numerical experiments show that the algorithm can invert multi-objects effectively and has some advantages such as fast, accurate and tolerant to noise. The proposed inversion is a very useful tool to interpret the anomalies in resistivity profiling quickly and quantitatively, and able to satisfy the urgent demands in many fields.

关 键 词：粒子群优化 反演 椭球异常体 中间梯度法 

分 类 号：P631[天文地球—地质矿产勘探]