场地PGA放大系数与场地特征参数相关性及地震动快速评估方法研究

Correlation analysis of PGA amplification factor with site characteristic parameters and fast assessment of ground motions

场地地震动PGA放大系数是重大工程地震预警、烈度速报技术及结构抗震设计地震动输入的基础。搜集整理日本Kik-net台网239个Ⅱ场地台站的地震记录,以场地覆盖层厚度D、场地特征周期T、场地v_(s30)以及场地等效剪切波速v_(se)为场地特征参数,以井下记录PGA为地震动强度指标,研究场地PGA放大系数F_(PGA)与场地特征参数之间的相关性;以地表PGA值40 Gal作为预警阈值,采用回归决策树CART方法建立了地震预警的特征参数组合标准。分析结果表明:场地F_(PGA)与场地覆盖层厚度D及场地特征周期T负相关,与场地vs30正相关,与vse相关性可忽略;F_(PGA)与场地特征参数的相关性随着场地地震动强度增大而增大;通过搜集的地震数据对基于CART分类方法建立的场地地震动预警参数指标进行回判检验,检验的总体成功率达到90.09%,其中不预警的成功率为93.64%,预警成功率82.68%。虽然地震动快速评估方法总体成功率达到90%以上,但仍存在5.61%偏于危险的"漏判",所以本文方法仍需要进一步的改进和完善。

The PGA amplification factor FpGA is a basic index for seismic prewarning of important engineering struc- tures and the seismic input of engineering structures design. Selecting 239 seismic stations from Kik-Net, Japan, and the records recorded at the stations which are on site class Ⅱ , the correlation of FpGA with respect to site charac- teristic parameters, i. e. , the site thickness D, site characteristic period T, Vs30 and equivalent shear wave velocity Vse, are studied under different seismic intensity which is represented by the input PGA values. Taking ground mo- tion PGA 40gal as the target prewarning threshold, the parameters combinations are presented using the classifica- tion and regression tree （CART） method. The analytical results show： FPGAiS negatively correlated with D and T, but positively correlated with Vs30; Vse has a quite weak correlation with FpGA; the correlation of site characteristic parameters and FpGA is strengthened with increasing input intensity ; by means of the CART method, the tion prewarning criterion is tested on the selected Kik-net data that the total success rates keep 90.09% 93.64% of non-prewarning and 82.68% of prewarning. Albeit the total success rates of the criterion than 90%, 5.61% of the prewarning data are omitted that the results could be dangerous. Therefore, presented herein should be improved.