引用本文: 王乐洋, 靳锡波, 许光煜. 断层参数反演的动态惯性因子的粒子群算法[J]. 武汉大学学报 ( 信息科学版), 2021, 46(4): 510-519.
WANG Leyang, JIN Xibo, XU Guangyu. Particle Swarm Optimization Algorithm with Dynamic Inertia Factors for Inversion of Fault Parameters[J]. Geomatics and Information Science of Wuhan University, 2021, 46(4): 510-519.
 Citation: WANG Leyang, JIN Xibo, XU Guangyu. Particle Swarm Optimization Algorithm with Dynamic Inertia Factors for Inversion of Fault Parameters[J]. Geomatics and Information Science of Wuhan University, 2021, 46(4): 510-519.

## Particle Swarm Optimization Algorithm with Dynamic Inertia Factors for Inversion of Fault Parameters

• 摘要: 采用大地测量观测数据进行地震断层参数反演是大地测量反演的研究热点，也是研究地震发生机制的重点。针对目前在断层参数反演中所用粒子群算法反演精度较低的问题, 分析了地震断层参数反演的非线性特点和基本粒子群算法的特征。考虑到基本粒子群算法在处理高度非线性问题时易陷入局部最优解，且求解过程中局部最优解与全局最优解之间会相互影响，通过分段调整影响粒子速度的惯性因子和影响全局最优解与局部最优解的加速因子，得到了一种适用于地震断层参数反演的分段动态调整参数的粒子群算法，应用于模拟地震与拉奎拉真实地震的断层参数反演。模拟地震实验结果表明，所提算法具有稳定性，且求得的断层倾角、滑动角较多峰值粒子群算法更接近真值。在2009年拉奎拉地震实例中，所提算法求得的断层参数正演后所得形变量与地表观测值的均方根误差为5.2 mm，优于多峰值粒子群算法的6.7 mm。以上结果表明，所提算法获得的断层模型更符合真实的断裂条件，具有一定的实际应用价值。

Abstract:
Objectives  Inversion of seismic fault parameters using geodetic observation data is a hotspot in geodetic inversion, and it is also the focus of studying the mechanism of earthquake occurrence. Aiming at the low accuracy of particle swarm optimization (PSO) currently used in fault parameter inversion, this paper analyzes the nonlinear characteristics of seismic fault parameter inversion and the characteristics of basic PSO. Basic PSO is easy to fall into a local optimal solution in highly nonlinear problems, and the local optimal solution and the global optimal solution may affect each other during the PSO solution process. This paper proposed a new particle swarm algorithm for inversion of fault parameters to solve the local optimization.
Methods  In this paper, we adopted the strategy of segmentally and dynamically adjusting the parameters, including the inertia factor that affects the particle velocity and the acceleration factors that affect the local and global optimal solutions.
Results  The proposed algorithm was applied to inverse the fault parameters for the simulation earthquake and the L?Aquila earthquake. The results of the simulation earthquake show that the proposed algorithm is stable, and the inclination and sliding angle obtained by the proposed algorithm are closer to the true value. The results of the L?Aquila earthquake show that the root mean squared error (RMSE) of surface observations and deformation variables obtained by the proposed algorithm is 5.2 mm, which is better than 6.7 mm obtained by multi-peak particle swarm optimization (MPSO).
Conclusions  The experiment results show that the fault model obtained by the proposed algorithm is more consistent with the true fracture condition, and the proposed algorithm has practical application value.

/

• 分享
• 用微信扫码二维码

分享至好友和朋友圈