中文摘要：

工程场地地震安全性评价和地震小区划工作在我国日益普及，对适于工程使用的土层液化发生概率计算方法的需求日趋强烈。该文沿用我国建筑抗震设计规范液化判别模式并使用同样基础数据，以地下水位、埋深、标准贯入击数等土层常规指标为直接变量，采用较为成熟的二分类Logistic回归分析理论，构造砂土液化概率计算公式和不同概率水平下液化临界值计算公式，通过回归分析和近期地震液化调查新数据分项检验其合理性和可行性。构造公式时采用我国大陆以往159例液化数据，检验公式时采用近期地震液化调查358例新数据，来源于1995年阪神地震和1999年集集地震。以上述两方面数据，检验概率50％的该文公式与现有建筑抗震设计规范确定性方法，结果表明：二者对国内大陆液化资料回判成功率基本相当，但该文公式略为保守；就阪神和集集地震新数据，该文公式不同砂层埋深检验结果均可接受，而现有规范砂层埋深下小于10m结果可以接受，大于10m时显著保守，方法需要改进。以上述两方面数据检验该文公式不同液化概率水平下的表现，结果表明：不同液化概率下的该文公式不仅定性上符合现有认识，而且非液化和液化概率水平相同时，对大陆非液化和液化场地回判成功率基本相当，对阪神和集集地震液化与非液化场地判别成功率基本相当，不同埋深下液化与非液化场地判别成功率大体一致。两方面数据检验表明，该文公式对不同液化概率水平、各种地震强度、地下水位和砂层埋深均有较好的适用性。

英文摘要：

A reliable liquefaction probabilistic method in engineering activities is strongly required as the seismic safety evaluation for given sites and seismic microzonation performed widespread in China. The conventional soils parameters, used in the liquefaction evaluation method in current Chinese seismic design code such as water level, soil depth and SPT N-value, are selected to develop liquefaction probability formula and propose critical SFF N-value calculation equation through the logistic regress theory. The proposed probability formula and equation are verified by 358 ease histories collected from the 1995 Kobe earthquake and 1999 Chi-Chi earthquake. The proposed formula under given 50% liquefaction probability is compared with the Chinese seismic design code deterministic method developing from the China mainland liquefaction database. The comparison results show that： ①the proposed formula under given liquefaction probability of 50% is a little conservative having almost the same verification success ratio with the Chinese seismic design code method. ② The verification result of the proposed formula is acceptable, but the design code method is high conservative using the separated verification data in which the soils depth is greater than 10m. The proposed formula under different liquefaction probabilities are examined, indicating that： ① the qualitative results are agree with the current understanding. ② The verification success ratio using liquefied sites are almost the same with that of using non- liquefied sites deriving from China mainland data and the verification data with different soils depth. ③ The proposed formula is applicable to different conditions, i.e. different shaking level, different water table, and soils depth.