基于AlexNet卷积神经网络模型的宁夏天然地震和非天然地震识别研究

Identification of Natural and Unnatural Earthquakes in Ningxia Based on AlexNet Convolutional Neural Network Model

宁夏回族自治区及周边天然地震和非天然地震频发,精确快速识别天然和非天然地震有利于震后应急响应、科学研究、赈灾救援等工作。基于AlexNet卷积神经网络模型,选取宁夏境内及周边130个地震事件(天然地震80个、非天然地震50个),对其进行了单个台站波形记录地震事件的训练和多个台站波形记录地震事件的测试,并将模型结果与宁夏测震台网人工编目结果进行比对,结果表明单个台站波形记录地震事件的训练结果中,AlexNet卷积神经网络模型对宁夏天然地震和非天然地震的正确识别率为99%;多个台站波形记录地震事件的测试结果中,此模型对宁夏天然地震和非天然地震的正确识别率为97.01%。AlexNet卷积神经网络模型作为人工智能领域的成熟技术之一,可以很好的运用于宁夏天然地震和非天然地震的识别工作之中。

Natural and unnatural earthquakes occur frequently in and around Ningxia Hui Autonomous Region.Accurate and rapid identification of natural and unnatural earthquakes is beneficial to post-earthquake emergency response,scientific research,disaster relief and other work.AlexNet convolution based neural network model,this paper selected within the territory of ningxia and the surrounding 130seismic events(natural earthquake 80,unnatural 50),on the single stations waveform records of earthquake events of training and multiple stations waveform records of earthquake events of test,and the model results compare with the results of ningxia ride network artificial catalog,The results show that AlexNet convolution neural network model has a correct recognition rate of 99% for natural and non-natural earthquakes in Ningxia.In the test results of seismic events recorded by multiple stations,the correct recognition rate of natural and non-natural earthquakes by the model is 97.01%.AlexNet convolutional neural network model,as one of the mature technologies in the field of artificial intelligence,can be well applied to the identification of natural and unnatural earthquakes in Ningxia.