用随机启发式优化算法反演2003年云南大姚地震断层面参数的比较研究

The comparison of inversion study in estimating the fault plane parameters of Yunnan Dayao Earthquake in 2003 based on stochastic heuristics optimization algorithms

基于余震分布确定主震断层面的数学模型,以确定断层面的走向和倾角参数进行计算,研究了遗传算法、模拟退火算法、差分演化算法、粒子群算法等4种最优化反演方法的反演效果和可靠性。结果显示,在涉及到的反演参数较少和非线性不太严重时,4种方法都有较好的表现,差分演化算法、粒子群算法速度快,精度高,遗传算法速度较慢,精度较低,模拟退火由于缺乏并行机制,速度较慢,精度高于遗传算法。余震在求出的断层附近分布图直观地反映出4种方法的效果和可靠性。

Based upon the model of estimating the main earthquake＇s fault plane parameters, the numerical inversion example is made and the inversion effect and reliability of four different stochastic optimization algorithms such as genetic algorithm, simulated annealing algorithm, particle swarm optimization algorithms, differential evolution algorithm are studied as well. The good results show that the inversed parameters obtained by the four algorithms are all of satisfied when the parameters is not too many and the nonlinearity is not too serious, the differential evolution and particle swarm algorithms have more high velocity and precision than others, the genetic algorithm has less velocity and precision, the simulated annealing algorithm＇s precision is higher than genetic algorithm, and its velocity is lower than others because of lacking parallel mechanism. The effect and reliability of four stochastic algorithms can also be seen directly from the figure of seismic data around the fault plane. Thus all these methods are simple, efficient and feasible inversion algorithms for geodesy inversion.