• 欢迎访问ERC!
引用本文:
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
过刊浏览    高级检索
本文已被:浏览 91次   下载 0  
分享到: 微信 更多
基于深度学习的地电场信号异常检测应用
卫雷1, 安张辉1, 范莹莹1, 陈全1, 刘君1, 元丽华2
1.中国地震局兰州地震研究所;2.兰州理工大学 理学院
摘要:
荐于深度学习方法在时间序列异常检测领域已经取得了大量成果,本文引入了人工智能领域的语言生成模型,通过添加恒等映射将普通卷积神经网络的卷积层改为残差元结构,使用扩张因果卷积扩大了模型的感受域,选择对数概率密度函数方法进行异常检测,利用模拟数据对方法有效性和正确性进行了验证,并将其应用于甘肃平凉地电场观测台站的实际观测资料中。其结果表明:在汶川M_s 8.0、芦山M_s 7.0、岷漳M_s 6.6等地震前1个月,平凉台预测结果的每日累加误差对数概率密度值出现了突然变小的现象,异常出现时间与已有相关研究结果相吻合。结合空间电磁环境和震前微裂隙变化情况,对震前地电场出现异常的可能原因进行了初步理论解释。从本文工作可以看出,深度学习方法在地电场观测资料的顺利应用将有助于观测资料利用率和异常检测效率的提高,为观测资料更好地服务地震研究提供技术支持。
关键词:  深度学习  时间序列  扩张因果卷积  地电场  异常检测
DOI:
分类号:
基金项目:
Application on Anomaly Detection of Geoelectric Field Signal Based on Deep Learning
Wei Lei1, An Zhang Hui1, Fan YingYing1, Chen Quan1, Liu Jun1, Yuan li Hua2
1.Lanzhou Institute of Seismology, China Earthquake Administration;2.School of Science, Lanzhou University of Technology, Lanzhou
Abstract:
The deep learning method has made a lot of achievements in the field of time series about anomaly detection, this paper introduces the language generation model in the field of artificial intelligence, changing the convolution layer of the general convolution neural network to the residual element structure by adding identity mapping, expanding the receptive domain of the model by using the dilated causal convolution. Based on the dilated causal convolution network and the method of log probability density function, the anomaly events are detected according to the anomaly scores, and the validity of the method is verified by the simulation data, which is applied to the actual observation data of Pingliang earth electric field observation station in Gansu Province. The results show that one month before the Wenchuan M_s 8.0, Lushan M_s 7.0, and Minzhang M_s 6.6 earthquakes, the daily cumulative error of log-normal probabililty density of Pingliang station’s prediction results suddenly decreases, which is consistent with the actual earthquake anomalies in a certain time range. Combined with the spatial electromagnetic environment and the variation of micro fissures before the earthquake, the possible causes of the anomalies of the geoelectric field before the earthquake are preliminarily explained. The successful application of deep learning method in geoelectric field observation data will help to improve the utilization rate of observation data and the efficiency of anomaly detection, and provide technical support for observation data to serve the earthquake research better.
Key words:  Deep learning  Time series  Dilated causal convolution  Geoelectric field  Abnormal detection