工程上常将规范反应谱拟合的确定性取值作为地震动演变功率谱参数,这虽然便于工程应用,但同时存在明显弊端:一方面,参数并非从实测强震记录中获得,具有较强的经验性;另一方面,确定性参数生成的地震动样本过于规则,且工程特性单一,无法全面地反映地震动的随机性.为了克服上述难题,从PEER的NGAGWest2地震动数据库中筛选1 766条主轴方向的实测强震记录,并根据《中国地震动参数区划图》建议的场地类别和聚类方法划分为15组;随后,识别实测强震记录的演变功率谱参数,并结合K-S检验和BIC信息准则,确定每个参数最优概率模型;最后,根据演变功率谱参数统计建模结果,结合降维方法,生成了 Ⅱ类场地与Ⅲ类场地的代表性时程.与《建筑抗震设计规范》中建议取值不同,研究参数来源于实测地震动,且具有随机性,避免地震动样本工程特性的单一化和规则化.
In engineering practices,the deterministic values of the response spectrum in accord-ance with seismic design codes are generally applied as evolutionary power spectral density(EPSD)parameters of ground motions.Although this treatment is convenient for engineering ap-plications,it comes with two evident disadvantages.On the one hand,parameters are not ob-tained from measured strong motion records,which makes them highly empirical.On the other hand,the ground motion samples generated by deterministic parameters are extremely regular and exhibit unitary engineering characteristics,which cannot fully reflect the randomness of ground motions.To overcome the abovementioned challenges,we selected 1 766 strong motion records in the main axis direction from PEER and divided them into 15 groups based on the site categories suggested by the Seismic Ground Motion Parameters Zonation Map of China.Subse-quently,the EPSD parameters of strong motion records were identified.Combined with the K-S test and BIC information criterion,the optimal probability models for each parameter were determined.Finally,based on statistical modeling results of EPSD parameters,representative time histories of site categories Ⅱ and Ⅲ were generated through the dimension reduction meth-od.In contrast to the recommended values from the Code for Seismic Design of Buildings,the parameters suggested in this study were derived from measured records and possessed significant randomness,thus avoiding the simplification and regularization of engineering characteristics of ground motion samples.