首页|改进K-SMOTE随机森林算法在地震信息发布安全风险评估中的应用研究

改进K-SMOTE随机森林算法在地震信息发布安全风险评估中的应用研究

扫码查看
为有效地提高地震监测、预警、灾情评估等信息发布的安全性提出一种改进型K-SMOTE随机森林(RF)方法构建地震信息安全风险评估模型.该模型通过改进K-SMOTE算法中运用混合采样获得均衡度高的样本集,然后使用随机K折交叉验证方法进行样本划分与模型优化,最终实现目标安全风险等级评估.对实际地震信息发布案例进行评测,文章所提方法构建模型评估准确率为92%,模型精确率和查全率分别为0.81和0.92,模型泛化能力强,能有效用于地震信息发布安全风险评估.本研究为完善地震信息发布安全评估体系、改进地震信息发布环境、降低安全风险提供了参考.
Application of the improved K-SMOTE random forest algorithm to the safety risk assessment of seismic information release
The safe,timely,and effective dissemination of special information such as earthquake monitoring,early warning,disaster assessment,and emergency is crucial for maintaining social safety and stability.This paper proposed an improved K-means synthetic minority oversampling technique(K-SMOTE)combined with the random forest method to effectively assess the safety risks associated with the release of this information and address potential weak links.A seismic information safety risk assessment model was then established,using the improved K-SMOTE algorithm to generate a well-balanced sample set.Finally,the random K-fold cross-validation method is applied to divide the samples and optimize the model,enabling the assessment of risk levels for target safety.Using an actual earthquake information release case as an example,the model proposed in this paper achieves an evaluation accuracy of 92%.The precision and recall of the model are 0.81 and 0.92,respectively,showing its strong generalization capability.These results indicate that the model is highly effective for assessing the safety risks associated with earthquake information release.This research provides valuable insights for improving the seismic information safety assessment system,enhancing the environment for publishing seismic infor-mation,and reducing safety risks.

seismic information releaserisk level assessmentimproved K-SMOTErandom forestrandom K-fold cross validation

李亚龙、何琳、万杰、潘丹、孙静

展开 >

安徽省地震局,安徽 合肥 230031

安徽蒙城地球物理国家野外科学观测研究站,安徽蒙城 233527

地震信息发布 风险等级评估 改进K-SMOTE 随机森林 随机K折交叉验证

2025

地震工程学报
中国地震局兰州地震研究所,中国地震学会,清华大学,中国土木工程学会

地震工程学报

北大核心
影响因子:1.191
ISSN:1000-0844
年,卷(期):2025.47(1)