一种基于Pred-RNN模型的地震预测方法———以圣安德烈斯断层区域为例  

Earthquake prediction based on the Pred-RNN model-a case study in the San Andreas fault zone

作　　者：王雪娇[1] 陈雨[1] WANG Xuejiao;CHEN Yu(Sichuan University,Chengdu 610000,China)

机构地区：[1]四川大学,成都610000

出　　处：《电子测试》2023年第4期86-90,共5页Electronic Test

摘　　要：地震预测具有十分重要的现实意义和社会价值。本文提出一种利用时空预测模型的方法来进行地震预测,基于加利福尼亚圣安德烈斯断层区域的地震数据,采用Pred-RNN模型预测未来地震事件的区域、震级和趋势。实验表明,Pred-RNN模型在MSE、SSIM等评价指标上都具有很好的表现。分析结果表明,本文提出的方法在强化地震预测方面具有显著潜力,为地震预测研究提供了新思路。Earthquake prediction is a critical problem due to its obviously practical and societal implications.We propose an approach using spatiotemporal prediction models for earthquake prediction.The model is trained and verified on the seismic data in California.It is found that our proposed method successfully detects spatiotemporal correlations between major earthquakes.The experiments demon-strate that the Pred-RNN model performs well on evaluation metrics such as MSE(Mean Squared Error)and SSIM(Structural Similar-ity Index).These results suggest that our method has significant potential for enhancing earthquake prediction and provides a new way for earthquake prediction research.

关 键 词：地震预测 Pred-RNN 圣安德烈斯断层 时空预测 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]