Seismic Landslide Susceptibility Assessment Based on Seismic Ground Motion and Earthquake Disaster Analysis
Proceedings of the 4th International Conference on Performance Based Design in Earthquake Geotechnical Engineering (Beijing 2022)(2022)
摘要
Seismic motion is one of the significant factors triggering slope instability. Landslides induced by intense earthquakes pose a great threat on the public security and traffic safety. About 720 landslides were caused by the 2014 Ms6.5 Ludian earthquake. The seismic ground motion records of Ludian earthquake and the influencing factors of seismic landslides are analyzed, and the susceptibility of landslides was evaluated in combination with machine learning models (BP neural network and SVM). The characteristics of Ludian earthquake motion records in time domain, frequency domain and time-frequency domain reveal that, the maximum value of PGA reached 949.2 cm/s2 (EW), and the duration of seismic wave within the nearest station to epicenter was 5.0 s (EW), 4.9 s (NS) and 4.3 s (UD), respectively. The result of Hilbert Huang Transform suggests that large instantaneous and cumulative energy concentrated in low frequency region (0–5 Hz), which can be considered to be related to large area landslides. Twelve factors are selected as the influencing factors of seismic landslides under Ludian earthquake, and the spatial correlation analysis indicates that the relation between factors and seismic landslide is characterized by strong nonlinear properties, of which ground motion parameters show a strong positive correlation with landslide distribution. The receiver operating characteristic (ROC) curve is adopted to compare the performance of two models. The results show that the AUC values of two curves are 90.1% and 89.5%, respectively, showing that BP neural network has higher accuracy of seismic landslide susceptibility results, and illustrating dependency on spatial distribution of seismic ground motion parameters.
更多查看译文
关键词
Ludian earthquake, Seismic landslides, Hilbert Huang Transform, BP neural network, Support vector machine, Landslide susceptibility
AI 理解论文
溯源树
样例
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要