Abstract: Reservoir earthquake has different characteristics of damage for its shallower depth and smaller magnitude to natural moderate earthquakes.In actual reservoir earthquakes,the damage to buildings is much heavier as the high frequency part of ground motion attenuates slowly.So existing statistical vulnerability matrix could not be used to estimate damage losses.In mainland China,scientists have monitored the reservoir earthquake since 1960s,and as many as over 20 seismic events were considered to be reservoir earthquakes.However there is not enough quantitative information to analyze the building vulnerability.Therefore,an approach is proposed in the paper,in which the natural earthquake vulnerability matrix is used to deduce the suitable vulnerability for reservoir earthquakes.Magnitude, epicenter depth and intensity are regarded as classification standard while K-means clustering adopted. Then the selected matrixes are modified by a set of regional macroeconomic factors.The method was developed in early 1990s based on the relationship between regional macroeconomic development level and seismic damage.Three parameters,population density,per capita GDP and proportion between the three levels of industries are divided into three brackets.Each bracket has weighting factor from AHP(Analytic Hierarchy Process).After data preparation,one important parameter named closeness tolerance is calculated by the Hamming distance algorithm of fuzzy mathematics,in which the parameter reflects difference of two regions in macroeconomic development as well as the reliability or vulnerability of buildings to earthquake.For a group of closeness tolerances among selected matrixes and aimed area,the summation of multiplication between vulnerability probabilities and closeness tolerances is the aimed area's vulnerability value under certain intensity.Compared with the standard matrix and building vulnerability to natural earthquakes in southwestern China,the new vulnerability value is higher and buildings in developed region have better reliability than those in poverty areas.

Key words: reservoir earthquake, buildings vulnerability, earthquake damage prediction

摘要： 通过类比的方法,选择有现场调查资料并有建筑物破坏比结果的天然地震,基于地震中建筑物震害表现的一致性,以震级、震源深度和极震区烈度为标准,聚类得到与水库地震震害相似的天然地震震例。考虑到宏观经济水平与地震经济损失已有的统计模型,采用了人口密度、人均GDP和三产比例等3个宏观经济指标来反映不同地区的建筑物总体抗震水平,以加权海明(Hamming)距离来定义已知矩阵与目标矩阵的近似度,最终的加权结果即为待求地区的易损性矩阵。通过实际检验,所得的建筑物易损性关系能够反映水库地震的破坏特点,较中强天然地震的易损性要高。这种工作思路也可以用于水库地震其它方面的研究中。

关键词: 水库地震, 建筑物易损性, 震害预测