摘要
本文主要通过加窗傅里叶变换和功率谱样本熵提取地震波数据的特征，并分别建立支持向量机分类模型和随机森林回归模型，对地震事件进行分类和地震等级预测。结果表明，对于地震识别，测试集的准确率达到82.7%；对于等级预测，MAE达到0.58。因此，分类支持向量机和回归随机森林可以用于地震识别和等级预测。此外，波形信号的功率谱样本熵可以用来衡量波形的特征。基于傅里叶功率谱样本熵和机器学习算法的地震识别和震级预测方法在地震监测预警领域具有潜在的应用价值。这些研究成果为地震相关决策提供了可行的技术工具，有助于提高地震灾害的预测和响应能力。但仍需进一步的研究和验证，以进一步改进和优化方法的性能和稳定性。
关键词： 傅里叶变换；样本熵；支持向量机；随机森林
Abstract
In this paper, the characteristics of seismic wave data are extracted mainly through windowed Fourier transform and power spectrum sample entropy. A support vector machine classification model and random forest regression model are respectively established to classify seismic events and predict the grade of the earthquake. The results showed that for earthquake discrimination, the accuracy of the test set reached 82.7%; For grade prediction, the MAE reached 0.58. Therefore, classification support vector machine and regression random forest can be used for earthquake identification and grade prediction. Additionally, the power spectrum sample entropy of the waveform signal can be used to measure the characteristics of the waveform. The earthquake discrimination and magnitude prediction methods based on Fourier power spectrum sample entropy and machine learning algorithms have potential applications in the field of earthquake monitoring and early warning. These research results provide a feasible technical tool for earthquake-related decision-making and help improve the prediction and response capability of earthquake hazards. However, further research and validation are still needed to further improve and optimize the performance and stability of the method.
Key words： Fourier transform; Sample entropy; Support vector machine; Random forest
参考文献 References
[1] Renouard, A., Maggi, A., Grunberg, M., Doubre, C., & Hibert, C. (2021). Toward false event detection and quarry blast versus earthquake discrimination in an operational setting using semiautomated machine learning. Seismological Research Letters, 92(6), 3725-3742.
[2] Yang, D. H., Zhou, X., Wang, X. Y., & Huang, J. P. (2021). Mirco-earthquake source depth detection using machine learning techniques. Information Sciences, 544, 325-342.
[3] Saini, K., Kalra, S., & Sood, S. K. (2022). An Integrated Framework for Smart Earthquake Prediction: IoT, Fog, and Cloud Computing. Journal of Grid Computing, 20(2), 17.
[4] Xiong, P., Marchetti, D., De Santis, A., Zhang, X., & Shen, X. (2021). SafeNet: SwArm for earthquake perturbations identification using deep learning networks. Remote Sensing, 13(24), 5033.
[5] Toh, G., & Park, J. (2020). Review of vibration-based structural health monitoring using deep learning. Applied Sciences, 10(5), 1680.
[6] Mukherjee, T., Singh, C., & Biswas, P. K. (2021). A Novel Approach for Earthquake Early Warning System Design using Deep Learning Techniques. arXiv preprint arXiv:2101.06517.