摘要
地震动衰减关系系数的回归是地震动衰减关系研究中一项关键的技术,由于地震动衰减公式具有震级与距离相关性和地震动参数本身的不确定性较大等特点,一般迭代回归方法难以满足回归精度的要求。目前常用的回归方法有单随机变量加权最小二乘法、多随机变量加权最小二乘法、两步法、新两步法等;在上述方法中,两步法和新两步法虽然考虑了震级和距离相关性的特点,但回归系数的稳定性较差。因此,在两步法思想的基础上提出了既能考虑震级和距离相关性,又能提高回归系数的稳定性的一种新方法—改进的两步法。
The regression of ground motion attenuation relationship＇ s coefficient is a key technology in the study of ground motion attenuation relationship. As a result of the relevance between magnitude and distance and ground motion data characteristics with great uncertainty, general regression method can not meet the requirements of the regression ac- curacy. The commonly used methods include the single variable weighted least square regression method, the multi-vari- able weighted least square regression method, the two-step method, and the new two-step method etc. The regression methods mentioned above without considering the relevance between magnitude and distance of an epicenter beside the two-step method and the new two - step method. However, they all have a common character of instability in determining coefficients. Therefore, based on the ideas of the two-step method, a new regression method that is named the improved two-step method is put forward. This method not only considered the relevance between the magnitude and the distance of the epicenter but also improved the stability in determining the~ coefficients.
出处
《世界地震工程》
CSCD
北大核心
2009年第3期46-48,共3页
World Earthquake Engineering
基金
国家科技支撑计划项目(2006BAC13B01)
关键词
衰减关系
改进的两步法
两步法
回归
attenuation relationship
improved two-step method
two-step method
regression
