基于单台加速度记录的混合全局优化HVSR反演场地浅层速度结构

摘要

The inversion of site velocity structure using Horizontal-to-Vertical Spectral Ratio (HVSR) is a rapidly developing research field.The essence of this inversion is a global optimization problem of model-space search for soil structure.When the complexity of the model-space search increases,the commonly used search algorithm is slow to converge and the calculation efficiency is low.To solve these problems,a hybrid global optimization algorithm combining genetic and simulated annealing algorithm is proposed and validated by a theoretical model and observation data of thevertical array,respectively.The results show that by this algorithm permits to enhance the convergenceof the Monte Carlo method,avoid premature convergence of the genetic algorithm and improve thesearch inefficiency of the simulated annealing algorithm and achieve good inversion results.Besides,this paper also discusses the applicability of S-wave HVSRs of accelerograms recordedby a single station for inversion of site velocity structure.These provide an accurate and highefficient inversion method for determining the site velocity structure based on observation of asingle station.%利用水平与竖向谱比(HVSR)方法反演场地速度结构是国际上迅速发展的研究领域.HVSR反演计算实质是一个土层场地模型空间搜索的全局优化问题,当模型搜索空间的复杂程度增大时,目前常用的搜索算法收敛速度慢,计算效率较低.本文实现了一种结合遗传和模拟退火方法优点的混合全局优化HVSR反演算法,通过理论模型和竖向台阵实测数据的检验,表明该算法能获得很好的反演效果,较好地解决了蒙特卡罗方法收敛速度慢,遗传算法收敛早熟和模拟退火算法搜索效率低的问题.本文在此基础上讨论了单台加速度S波记录用于场地速度结构HVSR反演的适用性,为基于单个地震台的地震观测记录反演浅层速度结构提供了一种高效且较为准确的反演方法.