节点文献
基于逻辑回归分析的强震地表破裂预测方法
A prediction method of surface rupture in strong earthquakes based on logistic regression analysis
【摘要】 强震引起的地表破裂通常会对工程结构造成严重的危害,工程建设中要对建设场地是否存在发震断裂以及能否产生地表破裂做出评价。由于地震的发生具有随机性,因此,强震引起的地表破裂也存在着不确定性,对其给予概率的表达符合当前人们对这一问题的认识水平和工程建设的风险选择要求。本文在总结强震地表破裂及震害的基础上,将逻辑回归这一研究多变量参与、预测某一事件发生概率的数学模型引入到强震地表破裂的预测研究中,介绍了逻辑回归方法的基本原理;论证了这一方法用于强震地表破裂概率预测的可行性和适宜性;提出了评价因子以及分级方法;给出了具体的数据计算方法并进行了显著性检验;以汶川特地震为例对模型的适宜性进行了检验和具体应用。本文的研究工作对从事这一领域研究的科技人员有重要的参考价值。
【Abstract】 Surface rupture in strong earthquakes usually causes serious damage to engineering structures. In engineering construction, it is necessary to evaluate the existence of seismogenic faults and the occurrence of surface ruptures inconstruction sites. The occurrence of earthquakes is random, so the surface rupture caused by strong earthquakes is also uncertain. Therefore, given its probability expression, it meets the current people’s understanding level of this problem and the requirement of risk selection in engineering construction. On the basis of summarizing the surface rupture and damage caused by strong earthquakes, this paper introduces the mathematical model of logistic regression, which is used to study multi-variable participation and predict the probability of occurrence of an event, into the prediction of surface rupture caused by strong earthquakes, meanwhile, the basic principle of logistic regression method is also introduced; the feasibility and suitability of this method for predicting the probability of surface rupture of strong earthquakes have been demonstrated; evaluation factors and classification methods were proposed; the specific data calculation method is given and the significance test is carried out; taking Wenchuan Earthquake as an example, the suitability of the model is tested and applied concretely. The work of this paper has important reference value for scientific researchers in this field.
【Key words】 strong earthquakes; surface rupture; logistic regression; prediction method;
- 【文献出处】 地震工程与工程振动 ,Earthquake Engineering and Engineering Dynamics , 编辑部邮箱 ,2019年04期
- 【分类号】P315.7
- 【被引频次】4
- 【下载频次】303