Rapidly and accurately predicting on-site peak ground velocity (PGV) is important for earthquake hazard mitigation. Traditional methods used to predict PGV involve a single physics-based parameter, like the displacement (Pd) or squared integral (IV2) techniques; deep-learning neural network model, convolutional (CNN) recurrent (RNN) models, extract feature estimating PGV. Here, based on training dataset from events occurred in Japan, we construct hybrid (HybridNet) PGV, which consist of CNN...

SUMMARY Peak ground acceleration (PGA) is a key parameter used in earthquake early warning systems to measure the motion strength and initiate emergency protocols at major projects. The traditional P-wave peak displacement-dependent PGA prediction model (Pd-PGA model) tends underestimate for large earthquakes because it cannot make full use of fault continuity rupture information hidden time-varying process motion. In this paper, continuous long short-term memory (LSTM) neural network...

Abstract Rapid and accurate earthquake magnitude estimations are essential for early warning (EEW) systems. The distance information between the seismometers hypocenter can be important to estimation. We designed a deep-learning, multiple-seismometer-based estimation method using three heterogeneous multimodalities: three-component acceleration seismograms, differential P-arrivals, seismometer locations, with specific transformer architecture introduce implicit information. Using...

Magnitude estimation is a vital task within earthquake early warning (EEW) systems (EEWSs). To improve the magnitude determination accuracy after P-wave arrival, we introduce an advanced prediction model that uses deep convolutional neural network for (DCNN-M). In this paper, use inland strong-motion data obtained from Japan Kyoshin Network (K-NET) to calculate input parameters of DCNN-M model. The 12 extracted 3 s seismic recorded arrival as input, four layers, pooling batch normalization...

SUMMARY To rapidly and accurately provide alerts at target sites near the epicentre, we develop an on-site alert-level earthquake early warning (EEW) strategy involving P-wave signals machine-learning-based prediction equations. These equations are established for magnitude estimation peak ground velocity (PGV) accounting multiple feature inputs support vector machine (SVM). called SVM-M model estimating SVM-PGV predicting PGV, respectively. According to comparison between predicted PGV...

In recent years, although a variety of deep learning models have been developed for magnitude estimation, the complex and variable nature earthquakes limits generalizability accuracy these models. this study, we selected waveform data Japan earthquake. We applied four techniques (MagNet combined with bidirectional long- short-term memory network Bi-LSTM, DCRNN deepened CNN layers, DCRNNAmp introduction global scale factor, Exams multilayered architecture) real-time estimation. By comparing...

On April 14, 2010, a devastating Ms 7.1 earthquake occurred in Yushu, China. In the most severely struck area of Jiegu Town, approximately 94 % structures were damaged. A seismic intensity map was obtained based on field investigation data structural damage 63 residential areas, and differences four regions quantitatively compared using capacity index index. The typical for five types epicentral Town described summarized detail. Some recommendations improving buildings are provided; they...

Abstract Accurately estimating the magnitude within initial seconds after P-wave arrival is of great significance in earthquake early warning (EEW). Over past few decades, single-parameter approaches such as τc and Pd methods have been applied to EEW estimation studies considering first 3 s onset. However, these present considerable scatter are affected by signal-to-noise ratio (SNR) epicentral distance. In this study, using Japanese K-NET strong-motion data, we propose a machine-learning...

Abstract Rapid epicentral distance estimation is of great significance for earthquake early warning (EEW). To rapidly and reliably predict distance, we developed machine learning models with multiple feature inputs using a single station explored the feasibility three methods, namely, Random Forest, eXtreme Gradient Boosting, Support Vector Machine, estimation. We used strong-motion data recorded by Japanese Kyoshin network within range 1° (∼112 km) from epicenter to train models. 30...

Abstract The Sichuan–Yunnan region is one of the most seismically vulnerable areas in China. Accordingly, an earthquake early warning (EEW) system for essential to reduce future hazards. This research analyses utility two parameters (τ c and P d ) magnitude estimation using 273 events that occurred during 2007–2015. We find τ can more reliably predict high-magnitude a short P-wave time window (PTW) but produces greater uncertainty low-magnitude range, whereas highly correlated with event...

ABSTRACT The Sichuan–Yunnan region is a seismically active area. To explore the feasibility of using support vector machine (SVM) method for magnitude estimation in area and to improve rapid accuracy, we construct an SVM model transfer learning (TLSVM-M model) based on single-station record this study. We find that single station shows test dataset, within 3 s time window after P-wave arrival, average absolute error (which reflects size estimated as whole) standard deviation scatter error)...

Abstract Earthquake early warning (EEW) is of great significance in mitigating seismic disasters. Traditional EEW algorithms, which are knowledge‐driven approaches, rely on seismologists' analysis. The limited intensity measures were extracted by seismologists from P‐wave signals. And there considerable uncertainty for predicting epicentral distance, magnitude, peak ground acceleration (PGA), and velocity (PGV). Currently, data‐driven deep learning methods with the strong abilities do not...

In this paper, a nonlinear regression method called support vector (SVR) is presented to establish the relationship between engineering ground motion parameters and macroseismic intensity (MSI). Sixteen parameters, including peak acceleration (PGA), velocity (PGV), Arias intensity, Housner spectrum others, are considered as candidates for feature selection generate optimal SVR models. The datasets with both useable strong records corresponding investigated MSIs in Sichuan–Yunnan region,...

The traditional magnitude estimation method, which establishes a linear relationship between single warning parameter and the magnitude, exhibits considerable scatter underestimation. In addition, extraction of features from raw waveforms by deep learning network is black box. To provide more robust to construct with an interpretable input, in light earthquake rupture physics, we have established model (MEANet) via physics-based time series, attention mechanism, neural networks. We use...

ABSTRACT The accurate and reliable discrimination of earthquakes from background noise is a primary task earthquake early warning (EEW); however, ubiquitous complex microtremor signals substantially complicate this task. To mitigate problem, generative adversarial network (GAN) adopted to distinguish between microtremors in study. We train GAN based on 52,537 K-NET KiK-net strong ground motion records Japan, use the well-trained discriminator identify 5373 P waves testing set. results...

To observe the curative effects of platelet-rich plasma (PRP) combined with negative-pressure wound therapy (NPWT) on patients sternal osteomyelitis and sinus tract after thoracotomy.Sixty-two thoracotomy, hospitalized from March 2011 to June 2015, were retrospectively analyzed. Based whether receiving PRP or not, divided into two groups, group NPWT ( 22 December 2012) combination treatment (CT, 40 January 2013 2015). After debridement, in treated continuous (negative pressure values -15.96...

In this work we propose and apply a straightforward methodology for the automatic characterization of extended earthquake source, based on progressive measurement P-wave displacement amplitude at available stations deployed around source. Specifically, averaged peak measurements among all corrected observed distance attenuation effect to build logarithm vs. time function, named LPDT curve. The curves have an exponential growth shape, with initial increase final plateau level. By analyzing...