Abstract:The field distribution of pulse characteristics is an important issue for the probabilistic seismic hazard analysis (PSHA). However, as the basis of its establishment, current identification methods for pulse-like ground motions are not precise enough. In this paper, an intelligent identification method is proposed, and empirical formulas are established to quantify the field distribution of three-dimensional orientation-independent pulse characteristics. Firstly, Convolutional Neural Network (CNN) models are enhanced by S-Transform to efficiently capture the pulses in the time-frequency domain, and the Issue-Targeted Fusion Learning (ITFL) training framework is proposed to learn the existing knowledge and solve the critical identification issues for pulse characteristics, including inconsistent identification results, limited applicability, and difficulties in concurrently identifying robust pulse period. Then, the worldwide near-fault ground motions from the NGA-West2 database are identified using the proposed identification method in all possible nonredundant orientations in space, where pulse-like ground motions are defined as the ground motions with pulses in at least one orientation, and TPmDRotnn is defined to represent the pulse period of ground motions in all orientations. Finally, detailed analysis of the field distribution of directivity pulse characteristics indicates that the vertical components of near-fault pulses and the directionality/polarization effects are remarkable in near-fault ground motions, whereas current research may misidentify the pulse-like ground motions and further underestimate the seismic hazard because of ignoring these two important factors. Therefore, the proposed identification method could be provided as a useful tool for practitioners to accurately identify the pulse-like ground motions, and the established empirical formulas could be used to predict the three-dimensional orientation-independent pulse characteristics in near-fault regions.

An Easy-To-Update Pulse-Like Ground Motion Identification Method Based on Siamese Convolutional Neural Networks

Active Source Seismic Identification and Automatic Picking of the P ̄wave First Arrival Using a Convolutional Neural Network 1

Spectrum-matched Ground Motion Selection Method Based on Siamese Convolutional Neural Networks

A Simple and Quantitative Algorithm for Identifying Pulse‐Like Ground Motions Based on Zero Velocity Point Method

Intelligent identification method for pulse-like ground motions and field distribution of three-dimensional orientation-independent pulse characteristics

An Efficient Algorithm to Identify Strong Pulse-Like Ground Motions Based on the Smoothed Significant Velocity Half-Cycles

Automatic Waveform Classification and Arrival Picking Based on Convolutional Neural Network

An improved energy‐based approach for selecting pulse‐like ground motions

1D Convolutional Seismic Event Classification Method Based on Attention Mechanism and Light Inception Block

Improved Mine Microseismic Event Recognition Method Combining Neural Network and Transfer Learning.

Vehicle Classification Based on Seismic Signatures Using Convolutional Neural Network

Quantitative identification of near-fault pulse-like ground motions based on energy

Automated Heartbeat Classification Exploiting Convolutional Neural Network with Channel-Wise Attention

A nonlinear structural pulse-like seismic response prediction method based on pulse-like identification and decomposition learning

Identifying microseismic events using a dual-channel CNN with wavelet packets decomposition coefficients

Full-Azimuth Quantitative Identification of Near-Fault Pulse-Like Ground Motions Based on M&P Wavelet

Early identification of potential loess landslide using convolutional neural networks with skip connection: a case study in northwest Lvliang City, Shanxi Province, China

The Identification of ECG Signals Using Wavelet Transform and WOA-PNN

Shapelet-informed machine learning classifiers: A path towards precise identification of pulse-like ground motions

Earthquake magnitude estimation using a two-step convolutional neural network

Identification of Intensive Oscillation Parameters of Power System Based on CS-CNN-GRU Network