Search

In this paper, a scalable end-to-end tree boosting system called XGBoost has been applied for predicting the magnitude of an earthquake from the early part of earthquake waveform data. This model uses the features extracted from the early P wave phase of the records as an input. The model's effectiveness has been verified by using data on earthquakes occurring in the Eurasian plate of Japan Islands from 1996 to 2021. Feature engineering has given 29 new features identified from the early P wave phase of the record, which show a high correlation with the magnitude of an earthquake. The comparison of predicted and actual magnitude shows that a trained XGboost model, which uses a single input record for magnitude prediction, gives an average prediction error of 0.004 ± 0.57 for earthquakes in the test dataset. In contrast, the average prediction error of –1.1 ± 0.80 and –0.65 ± 0.69 has been obtained for the magnitude estimated from conventional τc and Pd methods using the same test dataset. It is further seen that the average predicted magnitude of a single earthquake of magnitude 4.5 and 6.1 (MJMA) obtained by using multiple nearfield records using XGBoost model is 4.58 ± 0.33 and 6.32 ± 0.29, which is close to the actual magnitude of the earthquake. The results presented in this paper clearly show that the structured data can be effectively used by complex machine learning or deep learning models to predict earthquake magnitude from single or multiple records. [ABSTRACT FROM AUTHOR]

This paper presents a method for quantitatively analyzing the influence of earthquake history on the pile-soil dynamic interaction for offshore wind turbines, and then modifying the dynamic p-y curve model to account for this effect. The analysis consists of three stages. Stage I involves centrifuge tests to investigate the impact of earthquake history on physical processes. These tests yield data, including dynamic p-y curves and earthquake history-related features. Stage II introduces a dynamic p-y curve model-guided deep learning model, where the p-y model was embedded in the loss function. This method specifically processes test data without requiring generalization. Stage III focuses on a quantitative analysis, where earthquake history-related features are reconstructed based on the analysis of earthquake history factors. Formulas are derived to fit the extracted earthquake history factors and then incorporated into the dynamic p-y curve model, resulting in a modified model considering the earthquake history effect. Compared to the test results, the modified dynamic p-y curve model shows improved predictions of soil resistance force p than the traditional p-y model. Quantitative analysis and validation results also demonstrate that deep learning models can help analyze experimental data automatically and intelligently, providing more quantitative evidence and explanations for observed phenomena. • The dynamic p-y curve and its key parameters undergo changes due to the earthquake history effect. • The dynamic p-y curve model guided-deep learning model can help extract earthquake history factors from test data. • Quantitative analysis shows that earthquake history can increase the relative density of saturated sand up to 85%–95%. • A modified dynamic p-y curve model shows predictive ability considering the earthquake history effect. [ABSTRACT FROM AUTHOR]