Hibrit Makine Öğrenmesi ve Derin Öğrenme Modelleri Kullanılarak Geliştirilmiş Deprem Büyüklüğü Tahmini

Öz

Bu çalışma, tarihsel sismik verileri kullanarak deprem büyüklüklerini tahmin etmek için makine öğrenimi ve hibrit derin öğrenme modellerinin performansını değerlendirmektedir. Rastgele Orman (RF), ARIMA, Uzun Kısa Dönem Bellek (LSTM), CNN+LSTM ve Transformatör + Gauss Süreçleri (GP) dahil olmak üzere beş model, Ortalama Karesel Hatanın Kökü (RMSE) ve R^2 gibi ölçütler kullanılarak karşılaştırılmıştır. RF modeli, 0,072 RMSE ve 0,30 R^2 ile oldukça verimliydi. Ancak, zamansal analizi içermiyordu. ARIMA da 0.065 RMSE ve 0.42 R^2 ile daha iyiydi ve doğrusal ilişkiler için en uygun modeldi. LSTM sıralı ilişkileri iyi tanımlamış ve 0.097 RMSE ve 0.51 R^2 sağlamıştır. Hibrit CNN+LSTM modeli, uzamsal ve zamansal özellikleri birleştirerek 0,090 RMSE ve 0,58 R^2 ile bağımsız yaklaşımlardan daha iyi performans göstermiştir. Transformatör + GP modeli, 0,063 RMSE ve 0,62 R^2 ile en yüksek doğruluğu elde etmiş ve güven aralıkları aracılığıyla sağlam belirsizlik ölçümü sunmuştur. Bu sonuçlar, sismik tahminde hibrit modellerin üstünlüğünü vurgulamakta, tahmin doğruluğunu artırma ve daha iyi risk yönetimi stratejilerini destekleme potansiyellerini göstermektedir.

Anahtar Kelimeler

• Derin Öğrenme
• Yapay Zeka
• Transformers
• Deprem Tahmini

Enhanced Earthquake Magnitude Prediction Using Hybrid Machine Learning and Deep Learning Models

Abstract

This study evaluates the performance of machine learning and hybrid deep learning models for predicting earthquake magnitudes using historical seismic data. Five models, including Random Forest (RF), ARIMA, Long Short-Term Memory (LSTM), CNN+LSTM, and Transformer + Gaussian Processes (GP), were compared using metrics such as Root Mean Squared Error (RMSE) and R2. The RF model was quite efficient, with an RMSE of 0.072 and an R2 of 0.30. However, it did not incorporate temporal analysis. ARIMA was also better, with an RMSE of 0.065 and R2 of 0.42, which is best suited for linear relationships. LSTM identified the sequential relations well and provided an RMSE of 0.097 and R2 of 0.51. The hybrid CNN+LSTM model outperformed standalone approaches with an RMSE of 0.090 and R2 of 0.58 by combining spatial and temporal features. The Transformer + GP model achieved the highest accuracy, with an RMSE of 0.063 and R2 of 0.62, offering robust uncertainty quantification through confidence intervals. These results highlight the superiority of hybrid models in seismic forecasting, demonstrating their potential to improve predictive accuracy and support better risk management strategies.

Keywords

• Deep Learning
• Artificial Intelligence
• Transformers
• Earthquake Prediction

References

1. [1] Z. Y. İlerisoy, F. Gökşen, and A. Soyluk, “Deprem Kaynaklı İkincil Afetler ve Türkiye Örneklemi / Secondary Disasters Caused by Earthquakes and Turkey Sample,” 2022. [Online]. Available: https://www.researchgate.net/publication/357635207
2. [2] A. C. Yoloğlu and F. Zorlu, “Regional Agglomeration and Disparities in Relation to Earthquake Risks in Türkiye,” Journal of City and Regional Planning, vol. 06, no. 01, pp. 78–96, 2024.
3. [3] E. Demirelli, H. İ. Solak, and İ. Tirkyakioğlu, “Makine öğrenmesi algoritmaları ile deprem katalogları kullanılarak deprem tahmini,” Gümüşhane Üniversitesi Fen Bilimleri Enstitüsü Dergisi, Sep. 2023, doi: 10.17714/gumusfenbil.1268504.
4. [4] N. Yelboğa, “Kahramanmaraş Depremi Özelinde Travmatik Yas Ve Sosyal Hizmetin Yas Danişmanliği Müdahalesi,” Tobider - International Journal of Social Sciences, vol. 7, no. 1, Mar. 2023, doi: 10.30830/tobider.sayi.13.6.
5. [5] T. Bhandarkar, V. K, N. Satish, S. Sridhar, R. Sivakumar, and S. Ghosh, “Earthquake trend prediction using long short-term memory RNN,” International Journal of Electrical and Computer Engineering (IJECE), vol. 9, no. 2, p. 1304, Apr. 2019, doi: 10.11591/ijece.v9i2.pp1304-1312.
6. [6] A. Doğan and E. Demir, “Earthquake Prediction in Turkey with Structural Recurrent Neural Networks,” in 2020 28th Signal Processing and Communications Applications Conference (SIU), IEEE, Oct. 2020, pp. 1–4. doi: 10.1109/SIU49456.2020.9302400.
7. [7] R. Li, X. Lu, S. Li, H. Yang, J. Qiu, and L. Zhang, “DLEP: A Deep Learning Model for Earthquake Prediction,” in 2020 International Joint Conference on Neural Networks (IJCNN), IEEE, Jul. 2020, pp. 1–8. doi: 10.1109/IJCNN48605.2020.9207621.
8. [8] A. Berhich, F.-Z. Belouadha, and M. I. Kabbaj, “LSTM-based Models for Earthquake Prediction,” in Proceedings of the 3rd International Conference on Networking, Information Systems & Security, New York, NY, USA: ACM, Mar. 2020, pp. 1–7. doi: 10.1145/3386723.3387865.

9. [9] M. Karcı and İ. Şahin, “Derin Öğrenme Yöntemleri Kullanılarak Deprem Tahmini Gerçekleştirilmesi,” ARTIFICIAL INTELLIGENCE STUDIES, vol. 5, no. 1, pp. 23–34, 2022, doi: 10.30855/AIS.2022.05.01.03.
10. [10] B. Bhargava and S. Pasari, “Earthquake Prediction Using Deep Neural Networks,” in 2022 8th International Conference on Advanced Computing and Communication Systems (ICACCS), IEEE, Mar. 2022, pp. 476–479. doi: 10.1109/ICACCS54159.2022.9785011.
11. [11] A. Doğan, “Earthquake prediction for the northwest of Turkey with machine learning methods,” Yerbilimleri/ Earth Sciences, vol. 44, no. 2, pp. 166–178, Aug. 2023, doi: 10.17824/yerbilimleri.1325321.
12. [12] P. Kavianpour, M. Kavianpour, E. Jahani, and A. Ramezani, “A CNN-BiLSTM model with attention mechanism for earthquake prediction,” J Supercomput, vol. 79, no. 17, pp. 19194–19226, Nov. 2023, doi: 10.1007/s11227-023-05369-y.
13. [13] N. S. M. Ridzwan and S. H. Md. Yusoff, “Machine learning for earthquake prediction: a review (2017–2021),” Earth Sci Inform, vol. 16, no. 2, pp. 1133–1149, Jun. 2023, doi: 10.1007/s12145-023-00991-z.
14. [14] S. Gücek and İ. Zorluer, “Bir Boyutlu Analiz Yöntemiyle Sahaya Özel Sıvılaşma Risk Haritalarının Oluşturulması: Afyonkarahisar Örneği,” Afyon Kocatepe University Journal of Sciences and Engineering, vol. 21, no. 4, pp. 908–921, Aug. 2021, doi: 10.35414/akufemubid.930999.
15. [15] S. Baruah et al., “Moment Magnitude (M W) and Local Magnitude (M L) Relationship for Earthquakes in Northeast India,” Pure Appl Geophys, vol. 169, no. 11, pp. 1977–1988, Nov. 2012, doi: 10.1007/s00024-012-0465-9.
16. [16] T. C. Hanks and H. Kanamori, “A moment magnitude scale,” J Geophys Res Solid Earth, vol. 84, no. B5, pp. 2348–2350, May 1979, doi: 10.1029/JB084iB05p02348.
17. [17] E. M. Scordilis, “Empirical Global Relations Converting M S and m b to Moment Magnitude,” J Seismol, vol. 10, no. 2, pp. 225–236, Apr. 2006, doi: 10.1007/s10950-006-9012-4.
18. [18] J.Brownlee, “Long Short-term Memory Networks with Pyt,” 2017.
19. [19] G. Tunnicliffe Wilson, “Time Series Analysis: Forecasting and Control,5th Edition, by George E. P. Box, Gwilym M. Jenkins, Gregory C. Reinsel and Greta M. Ljung, 2015. Published by John Wiley and Sons Inc., Hoboken, New Jersey, pp. 712. ISBN: 978-1-118-67502-1,” J Time Ser Anal, vol. 37, p. n/a-n/a, Mar. 2016, doi: 10.1111/jtsa.12194.
20. [20] S. Hochreiter and J. Schmidhuber, “Long Short-Term Memory,” Neural Comput, vol. 9, no. 8, pp. 1735–1780, Nov. 1997, doi: 10.1162/neco.1997.9.8.1735.
21. [21] A. Halbouni, T. S. Gunawan, M. H. Habaebi, M. Halbouni, M. Kartiwi, and R. Ahmad, ‘CNN-LSTM: Hybrid Deep Neural Network for Network Intrusion Detection System’, IEEE Access, vol. 10, pp. 99837–99849, 2022.
22. [22] C. E. Rasmussen and C. K. I. Williams, Gaussian Processes for Machine Learning. The MIT Press, 2005. doi: 10.7551/mitpress/3206.001.0001.
23. [23] T. Chai and R. R. Draxler, “Root means square error (RMSE) or mean absolute error (MAE)?,” Feb. 28, 2014. doi: 10.5194/gmdd-7-1525-2014.
24. [24] L. Bouzid, M. A. Yallese, S. Belhadi, and A. Haddad, “Modelling and optimization of machining parameters during hardened steel AISID3 turning using RSM, ANN and DFA techniques: Comparative study,” Journal of Mechanical Engineering and Sciences, vol. 14, no. 2, pp. 6835–6847, Jun. 2020, doi: 10.15282/JMES.14.2.2020.23.0535.
25. [25] Ö. Kas, “Turkey Earthquake Prediction with Deep Learning Algorithms,” MSc Thesis, National College of Ireland, 2023. [Online]. Available: https://norma.ncirl.ie/7197/1/omerkas.pdf
26. [26] W. Li, M. Chakraborty, J. Köhler, C. Quinteros-Cartaya, G. Rümpker, and N. Srivastava, “Earthquake monitoring using deep learning with a case study of the Kahramanmaraş Turkey earthquake aftershock sequence,” Solid Earth, vol. 15, pp. 197–210, 2024. doi: 10.5194/se-15-197-2024.
27. [27] S. Shah, A. Lin, S. Lin, and J. Patel, “Turkey’s Earthquakes: Damage Prediction and Feature Significance Using a Multivariate Analysis,” arXiv preprint, arXiv:2411.08903, 2024. [Online]. Available: https://arxiv.org/abs/2411.08903