随着受干扰地磁观测仪器数量的不断增多，现有半人工识别干扰事件的方法存在效率低、工作量大、识别结果因人而异等问题。本文利用2012年1月1日至2014年12月31日全国地磁台网原始观测数据和地磁专家标注的2小时内干扰事件记录，分别构建干扰事件样本和正常样本各51 357条，基于卷积神经网络和自注意力机制提出一种新的干扰事件识别模型，实现干扰事件的自动、快速分类。实验结果显示，该模型在验证集的准确率达到92.93%，在测试集的准确率达到93.37%。与MLP、FCN、ResNet三种模型相比，本模型在测试集上的准确率平均提高近8.76%，表明卷积神经网络和自注意力机制等深度学习算法在地磁观测数据干扰事件识别领域具有巨大潜力，为进一步精确识别各类干扰事件探索了一种新思路。
With the continuous increase in the number of disturbed geomagnetic observation instruments, the existing semi-manual methods for identifying interference events have problems such as low efficiency, large workload, and different identification results among individuals. This paper uses the original observation data of the national geomagnetic network from January 1, 2012, to December 31, 2014, and the disturbance event records marked by geomagnetic experts to construct 51 357 disturbance event samples and normal samples, respectively,and proposes a new interference event recognition model based on the convolutional neural networks and self-attention mechanism to realize automatic and fast classification of interference events. Experimental results show that the accuracy of the model proposed in this paper reached 92.93% in the validation set and 93.37% in the test set. Compared with the three models of MLP, FCN, and ResNet, the accuracy of this model on the test set is increased by nearly 8.76% on average, indicating that deep learning algorithms such as convolutional neural networks and self-attention mechanism have great potential in the field of identification of disturbance events of geomagnetic observation data. This is a novel idea for further accurate identification of various interference events of geomagnetic observation data.
2022,43（5）: 49-63 收稿日期：2022-06-10
DOI：10.3969/j.issn.1003-3246.2022.05.007
基金项目：中央高校基本科研业务费专项（项目编号：ZY20180119）；地震科技星火计划（项目编号：XH20024）；河北省地震科技星火计划项目红山野外站科研专项（项目编号：DZ2021110500003）；河北省自然科学基金（项目编号：D2022512001）；国家重点研发计划（项目编号：2018YFC1503806）
作者简介：单维锋,男,博士,教授,主要研究地震大数据技术、人工智能算法及其在地震领域的应用。E-mail:shwf@163.com
参考文献：
鲍海英,蒋延林,樊晓春,等. 高压直流输电对地电场观测的影响探讨[J]. 中国地震,2020,36(3):607-619.
程文凯,杜劲松,陈超,等. 基于XGBoost机器学习的地磁日变重构方法研究[J]. 地震学报,2021,43(1):100-112.
樊俊屹,刘高川,王晓,等. 地球物理台网观测环境干扰现状分析[J]. 地震地磁观测与研究,2021,42(3):242-249.
韩明亮,邹志辉,马锐. 利用反射地震资料和多尺度训练集的深度学习速度建模[J]. 石油地球物理勘探,2021,56(5):935-946.
李章,李本有,严吉,等. 地磁数据缺失重构方法的研究[J]. 科技资讯,2021,19(17):39-41.
李正媛,熊道慧,刘高川,等. 基于大数据挖掘的地震前兆台网观测数据跟踪分析[J]. 地震地磁观测与研究,2016,37(3):1-6.
廖绍欢,李雪浩,魏嘉曦. 成都地震基准台及其下属台站地磁干扰类型和数据预处理分析[J]. 四川地震,2021,(2):36-41.
刘军,陈磊,李文灿,等. 基于混沌蚁群优化小波阈值法地磁信号噪声压制[J]. 科学技术与工程,2020,20(25):10 177-10 181.
王静,贺巍,汪伟明. 基于离散Gabor变换的磁暴识别[J]. 地震地磁观测与研究,2020,41(4):64-69.
王军,刘春国,樊俊屹. 基于CNN的地震前兆台网观测数据异常图像识别方法[J]. 地震工程学报,2021,43(1):28-32.
王立辉,刘庆雅,许宁徽. 基于干扰检测和CEEMD的地磁信号降噪方法[J]. 中国惯性技术学报,2020,28(4):474-479.
杨晶,丁仁伟,林年添,等. 基于深度学习的地震断层智能识别研究进展[J]. 地球物理学进展,2022,37(1):298-311.
杨学慧,杨正纲,姚休义,等. 地磁观测数据中典型干扰识别与处理[J]. 高原地震,2020,32(3):30-36.
应允翔,车濛琪,刘红飞,等. 安徽省地磁仪器观测质量分析[J]. 科技资讯,2021,19(22):60-64.
张帆,杨晓忠,吴立飞,等. 基于短时傅里叶变换和卷积神经网络的地震事件分类[J]. 地震学报,2021,43(4):463-473.
周飞燕,金林鹏,董军. 卷积神经网络研究综述[J]. 计算机学报,2017,40(6):1 229-1 251.
Geng Y, Su L L, Jia Y H, et al. Seismic events prediction using deep temporal convolution networks[J]. Journal of Electrical and Computer Engineering, 2019, 2019:7343784.
Krizhevsky A, Sutskever I, Hinton G E. Imagenet classification with deep convolutional neural networks[C]//Proceedings of the 25th International Conference on Neural Information Processing Systems. Lake Tahoe, Nevada:Curran Associates Inc., 2012:1 097-1 105.
Ku B, Kim G, Ahn JK, et al. Attention-based convolutional neural network for earthquake event classification[J]. IEEE Geoscience and Remote Sensing Letters, 2021, 18(12):2 057-2 061.
Liu G C, Shan W F, Chen J, et al. Automatic classification and recognition of geomagnetic interference events based on machine learning[J]. Journal of Computational Methods in Sciences and Engineering, 2022, 22(4):1 157-1 170.
Perol T, Gharbi M, Denolle M. Convolutional neural network for earthquake detection and location[J]. Science Advances, 2018, 4(2):e1700578.
Purwins H, Li B, Virtanen T, et al. Deep learning for audio signal processing[J]. IEEE Journal of Selected Topics in Signal Processing, 2019, 13(2):206-219.
Qian B, Xiao Y, Zheng Z J, et al. Dynamic multi-scale convolutional neural network for time series classification[J]. IEEE Access, 2020, 8:109 732-109 746.
Ramachandran P, Parmar N, Vaswani A, et al. Stand-alone self-attention in vision models[C]//Proceedings of the 33rd International Conference on neural Information Processing Systems. Vancouver, Canada:Curran Associates, Inc., 2019:68-80.
Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[C]//Proceedings of the 31st International Conference on neural Information Processing Systems. Long Beach, California, USA:Curran Associates Inc., 2017:6 000-6 010.
Wang W H, Xie E Z, Li X, et al. PVT v2:Improved baselines with Pyramid Vision Transformer[J]. Computational Visual Media, 2022, 8(3):415-424.
Xu X, Huang L, Liu X J, et al. Deepmad:Deep learning for magnetic anomaly detection and denoising[J]. IEEE Access, 2020, 8:121 257-121 266.
Yin C Y, Zhang S, Wang J, et al. Anomaly detection based on convolutional recurrent autoencoder for IoT time series[J]. IEEE Transactions on Systems, Man, and Cybernetics:Systems, 2022, 52(1):112-122.
