DOI: 10.13512/j.hndz.2021.03.16
备注
为实现自动检测地震噪声波形是否异常,提出应用BP神经网络技术进行地震噪声波形检测。选取福建地震台网88个测震台站2018-2019年的地震噪声原始波形,计算波形的加速度功率谱密度(PSD)值作为神经网络的输入特征值,在MATLAB中构建BP神经网络进行学习训练和仿真测试。测试验证了训练后的BP神经网络模型具备了可靠的地震噪声波形是否异常的检测能力。应用BP神经网络检测地震噪声波形免去了人工判断的工作,实现全自动处理,提高了检测效率,为今后地震噪声波形质量自动监控提供了新的技术方法。
In order to automatically detect whether the seismic noise waveform is normal,the paper proposes theapplication of BP neural network technology to detect seismic noise waveform.The paper chooses seismic noisewaveform of 88 seismic stations in Fujian Seismic Network from 2018 to 2019, calculates acceleration power spectral density(PSD)as input characteristic value of neural network, and constructs BP neural network in MATLAB for training and simulation.The result of simulation verifies that the trained BP neural network model hasreliable ability to detect whether the seismic noise waveform is abnormal.The method of detecting seismic noisewaveform using BP neural network avoids the work of manual judgment,realizes automatic processing,improves the detection efficiency,and provides a new technical method for automatic monitoring of seismic noise waveform quality in the future.