华南地震

立足于一线地震科研人员进行人工智能技术应用研究的实际需求，从地震监测预警技术系统在线生产的角度出发，设计了一个“智能+”地震监测预警技术平台。该平台在现有地震监测预警技术系统的基础上，增加了人工智能技术的核心模块，即智能模型训练平台。通过提供该平台的逻辑架构设计、物理架构设计、网络架构设计和人工智能框架选择，希望该设计方案能给同业内人工智能技术的应用提供一个新的思路。

Based on the actual needs of the first-line earthquake researchers for the research and application of artificial intelligence(AI)technology and from the perspective of online production of earthquake monitoring and early warning technology system，the paper promotes a design of “AI+” earthquake monitoring and early warning technical platform. On the basis of the existing earthquake monitoring and early warning technology system，the core module of AI technology，namely AI model training platform，is added to the technical system. This paper provides the logical architecture design，physical architecture design，network architecture design and artificial intelligence framework selection of the AI model training platform，hoping that this design can provide a new idea for the application of AI technology in the industry.

引言
1 设计思路
2 设计方案
3 讨论与总结

图1 现有地震监测预警技术系统框架<br/>Fig.1 The framework of the existing earthquake monitoring and early warning technology system

图1 现有地震监测预警技术系统框架
Fig.1 The framework of the existing earthquake monitoring and early warning technology system

图2 “智能+”地震监测预警技术系统设计框架<br/>Fig.2 The framework of “AI+” earthquake monitoring and early warning technology system

图2 “智能+”地震监测预警技术系统设计框架
Fig.2 The framework of “AI+” earthquake monitoring and early warning technology system

图3 智能模型训练平台逻辑架构设计图<br/>Fig.3 The logical architecture design diagram of the intelligent model training platform

图3 智能模型训练平台逻辑架构设计图
Fig.3 The logical architecture design diagram of the intelligent model training platform

图4 智能模型训练平台物理架构设计图<br/>Fig.4 Physical architecture design diagram of the intelligent model training platform

图4 智能模型训练平台物理架构设计图
Fig.4 Physical architecture design diagram of the intelligent model training platform

图5 智能模型训练平台网络架构设计图<br/>Fig.5 Network architecture design diagram of the intelligent model training platform

图5 智能模型训练平台网络架构设计图
Fig.5 Network architecture design diagram of the intelligent model training platform

表1 主流人工神经网络框架热度对比[数据来源Github 2020-06-08①]<br/>Table 1 Comparison of the popularity of mainstream artificial neural network frameworks [Data source Github 2020-06-08①]

表1 主流人工神经网络框架热度对比[数据来源Github 2020-06-08①]
Table 1 Comparison of the popularity of mainstream artificial neural network frameworks [Data source Github 2020-06-08①]

[1] LeCun Y，Bengio Y，Hinton G. Deep learning[J]. Nature，2015，521(7553):436-444.
[2] Hinton G，Deng L，Yu D，et al. Deep neural networks for acoustic modeling in speech recognition:the shared views of four research groups[J]. IEEE Signal Processing Magazine，2012，29(6):82-97.
[3] 赵刚，黄汉明，卢欣欣，等. 基于BP-Adaboost方法的天然地震和人工爆炸事件波形信号分类识别研究[J]. 地震工程学报，2017，39(3):0557-0562.
[4] Perol T，Gharbi M，Denolle M. Convolutional neural network for earthquake detection and location[J]. Science Advances，2018，doi:10.1126/sciadv.1700578.
[5] Ross Z E，Meier M A，Hauksson E，et al. Generalized seismic phase detection with deep learning[J]. Bulletin of the Seismological Society of America，2018，doi:10.1785/0120180080.
[6] 赵明，陈石，Dave Yuen. 基于深度学习卷积神经网络的地震波形自动分类与识别[J].地球物理学报，2019，62(1):374-382.
[7] 蔡振宇，盖增喜. 人工智能在拾取地震 P 波初至中的应用——以汶川地震余震序列为例[J]. 北京大学学报(自然科学版)，2019，55(3):151-460.

备注

引言

1 设计思路

2 设计方案

3 讨论与总结

期刊信息