摘要
地震信号可视为时间序列,采用在时间序列数据处理中有较好表现的LSTM神经网络实现天然地震和爆破的分类。对原始信号进行短时傅里叶变换,得到时频谱,拆分为多通道时间序列,作为LSTM神经网络的输入,实现时间序列到标签的分类。通过反复的实验选取了最优的输入数据尺度和模型超参数,选取的模型包含两个堆叠的双向LSTM层的神经网络。训练集上的5折交叉验证结果显示验证集平均准确率达到98.36%,测试集准确率为97.5%。测试结果表明,采用的模型在地震事件分类中有较好的效果。
As seismic signal can be regarded as time series,this paper used LSTM neural network which had good performance in time series data processing to realize the classification of natural earthquake and blasting.The time-frequency spectral density map obtained by short-time Fourier transform of original signal was divided into multi-channel time series,which was used as the input of LSTM neural network to realize the classification of time series to tags.Through repeated experiments,the optimal input data scale and model hyperparameters were determined,and the neural network model consisted of two bi-directional LSTM layers.The 5 folds cross-validation results on the training set show that the average accuracy of the verification set is 98.36%,and the accuracy of the test set is 97.5%.The test results show that this model has a good effect in the classification of seismic events.
作者
张帆
杨晓忠
王树波
Zhang Fan;Yang Xiaozhong;Wang Shubo(School of Control and Computer Engineering,North China Electric Power University,Beijing 102206,China;School of Mathematics and Physics,North China Electric Power University,Beijing 102206,China;Seismological Bureau of Inner Mongolia Autonomous Region,Hohhot 010051,Inner Mongolia,China)
出处
《计算机应用与软件》
北大核心
2023年第4期75-79,共5页
Computer Applications and Software
基金
中国地震局地震科技星火计划项目(XH20014)
内蒙古自治区地震局局长基金项目(2019JC27)。
关键词
天然地震
爆破
分类
机器学习
LSTM
Earthquake
Blasting
Classification
Machine learning
LSTM
