基于机器学习的极震区烈度快速预测方法  

Machine learning-based fast prediction method for the seismic intensity in meizoseismal area

作　　者：王茂岑 张令心[1,2] 钟江荣 张云霞[3] 张鹏 WANG Maocen;ZHANG Lingxin;ZHONG Jiangrong;ZHANG Yunxia;ZHANG Peng(Key Laboratory of Earthquake Engineering and Engineering Vibration,Institute of Engineering Mechanics,China Earthquake Administration,Harbin 150080,China;Key Laboratory of Earthquake Disaster Mitigation,Ministry of Emergency Management,Harbin 150080,China;National Disaster Reduction Center of China,Beijing 100124,China)

机构地区：[1]中国地震局工程力学研究所地震工程与工程振动重点实验室,哈尔滨150080 [2]地震灾害防治应急管理部重点实验室,哈尔滨150080 [3]应急管理部国家减灾中心,北京100124

出　　处：《振动与冲击》2025年第2期235-244,共10页Journal of Vibration and Shock

基　　金：国家自然科学基金(U2139209);中国地震局专项(CEA202112)。

摘　　要：极震区烈度的快速准确评估对于震后的应急响应至关重要。针对现有的极震区烈度预测精度差的问题,首先,整理了1949年—2021年的406次震级大于5.0且极震区烈度大于Ⅴ度的历史震例;然后,基于输入参数可在震后快速易于获取的原则,选择震级和震源深度作为输入参数,分别建立了基于随机森林、k近邻、逻辑回归以及决策树4种机器学习模型的极震区烈度快速预测方法;最后,对这几种方法的性能进行比较,并与已有的统计回归方法进行对比。结果显示:基于随机森林模型的预测方法性能更好,预测的准确率也很高;与仅选用震级作为输入参数的预测方法相比,该方法的准确率得到了较大提高;与现有的统计回归方法相比,该方法在准确率上有明显优越性。The rapid and accurate assessment of the intensity in the meizoseismal area is crucial for the post-earthquake emergency response.In order to solve the problem of poor accuracy of the intensity prediction in the meizoseismal area,firstly,406 historical earthquake cases with magnitude greater than 5.0 and intensity greater than V degree in meizoseismal area during 1949—2021 were collected and sorted out.Then,in consideration of the principle that the input parameters can be obtained quickly and easily after the earthquake,the magnitude and focal depth were selected as the input parameters.The fast prediction methods for the meizoseismal area intensity based on the random forest,k-nearest neighbor,logical regression and decision tree were established respectively.Finally,the performances of these methods were compared and also compared with the existing statistical regression methods.The results show that the method based on the random forest model has better performance and higher accuracy,and the accuracy of this method is greatly improved compared with the method using only magnitude as the input parameter.Compared with the existing statistical regression methods,this method has obvious advantage in accuracy.

关 键 词：震级 震源深度 极震区烈度 机器学习 快速预测方法 

分 类 号：P315.71[天文地球—地震学]