Neural Network-Based Strong Motion Prediction for On-Site Earthquake Early Warning

Abstract

Developing on-site earthquake early warning systems has been a challenging problem because of time limitations and the amount of information that can be collected before the warning needs to be issued. A potential solution that could prevent severe disasters is to predict the potential strong motion using the initial P-wave signal and provide warnings before serious ground shaking starts. In practice, the accuracy of prediction is the most critical issue for earthquake early warning systems. Traditional methods use certain criteria, selected through intuition or experience, to make the prediction. However, the criteria thresholds are difficult to select and may significantly affect the prediction accuracy. This paper investigates methods based on artificial intelligence for predicting the greatest earthquake ground motion early, when the P-wave arrives at seismograph stations. A neural network model is built to make the predictions using a small window of the initial P-wave acceleration signal. The model is trained by seismic waves collected from 1991 to 2019 in Taiwan and is evaluated by events in 2020 and 2021. From these evaluations, the proposed scheme significantly outperforms the threshold-based method in terms of its accuracy and average leading time.

Analysis and Simulation of Improved Seismic Data using Adaptive Processing by National Data Centre Iraq

“…These features are then processed by pooling layers, reducing the spatial dimensions of the input, and finally passed through fully connected layers to generate a final output classification (Lin & Lin, 2020). CNNs have also been used in earthquake detection, particularly for detecting seismic signals indicating an impending earthquake (Chiang et al, 2022). These signals are usually very weak and buried in background noise, making them difficult to detect using traditional methods .…”

“…However, training a CNN on seismic data makes it possible to extract and distinguish these signals from noise. The CNN can then be used to detect the occurrence of an earthquake (Chiang et al, 2022) based on the strength and pattern of the seismic signals.…”

Breaking the Silence: Men's Experience of Miscarriage. An Interpretative Phenomenological Analysis

“…Previously, Dhanya and Raghukanth [18] adopted an artificial neural network (ANN) in combination with a genetic algorithm (GA) to train a data-driven model to predict PGA, PGV, and spectral accelerations at 26 periods between 0.01 and 4 s. However, in that study, the authors utilized conventional feed-forward neural networks, which do not explicitly account for the internal cross-IM dependencies and only attempt to ensure that individual mean IM predictions agree well with observed IM values. Apart from this, studies like Datta et al [19], Fayaz and Galasso [20], Chiang et al [21], Hyun-Su [22], Liu and Dai [23], etc., have explored the utilization of convolutional neural networks (CNNs), variational autoencoders (VAE), and RNNs for the real-time estimation of source parameters (e.g earthquake magnitude and location), ground-motion intensity measures (e.g., PGA), and structural response parameters (e.g., drift levels). However, such research studies and their related computational tools aim to advance/improve earthquake early warning and are not particularly suitable for ground motion modelling and seismic hazard/risk analysis.…”

Shaking Table Design for Testing Earthquake Early Warning Systems

“…The latest types of seismic warning systems operate based on convolutional neural networks, trained and validated for earthquakes with many strong motions. Their role is to select relevant features from P-waves to predict whether the peak ground acceleration of subsequent waves usually surpasses 80 cm/s 2 -a threshold acceleration value from which people strongly feel the ground motion [6]. As the time window is critical for EEWS triggering, these artificial methods for ground motion prediction require a period parameter from which to follow a reasonable leading time.…”