利用改进的粒子群算法反演视震源时间函数

Retrieving apparent source time function by improved PSO algorithm

视震源时间函数的提取是研究震源参数的重要途径.本文提出了利用改进的粒子群(PSO)算法反演视震源时间函数的方法,以水平线方法得到的结果作为PSO算法的初值,并对PSO算法的惯性因子和学习因子进行改进,提高计算效率.采用改进的PSO算法对模拟数据进行了反演计算,并与映射Landweber反褶积(PLD)方法和遗传算法(GA)进行了对比分析.结果表明,相对于PLD方法,改进的PSO算法反演结果与真实结果误差更小;相对于遗传算法,改进的PSO算法计算效率提高了5倍以上.最后,利用改进的算法对2005年10月8日巴基斯坦克什米尔MW7.6地震的P波视震源时间函数进行了提取,结果表明此次地震P波视震源时间函数在25s之内,震源沿西北向破裂.该结果与张勇等的结果一致.

Retrieving apparent source time function （ASTF） is an important way to study earthquake source parameters. This paper proposes an improved particle swarm optimization （PSO） algorithm for retrieving ASTF. The result from water-level method is used as the initial global best value of PSO, and the inertia factor and learning factor of PSO are modified to increase computing efficiency. The results of simulation experiments using the improved PSO method are compared with those by using projected Landweber deconvolution （PLD） and generic algorithm （GA）. The comparison shows that the inversion result by using improved PSO has smaller residuals relative to that by using PLD, and the improved PSO has increased computation efficiency by more than 500% relative to GA. Finally, the improved PSO method is used to retrieve P-wave ASTFs of the 8 October 2005 Kashmir MwT. 6 earthquake. The results show that the duration of P-wave ASTF of the earthquake is within 25 s, and the rupture extends in northwest direction, which are consistent with those by Zhang et al.