Recognition Of Seismic Signals Research Articles

DiTing: A large-scale Chinese seismic benchmark dataset for artificial intelligence in seismology

In recent years, artificial intelligence technology has exhibited great potential in seismic signal recognition, setting off a new wave of research. Vast amounts of high-quality labeled data are required to develop and apply artificial intelligence in seismology research. In this study, based on the 2013–2020 seismic cataloging reports of the China Earthquake Networks Center, we constructed an artificial intelligence seismological training dataset (“DiTing”) with the largest known total time length. Data were recorded using broadband and short-period seismometers. The obtained dataset included 2,734,748 three-component waveform traces from 787,010 regional seismic events, the corresponding P- and S-phase arrival time labels, and 641,025 P-wave first-motion polarity labels. All waveforms were sampled at 50 Hz and cut to a time length of 180 s starting from a random number of seconds before the occurrence of an earthquake. Each three-component waveform contained a considerable amount of descriptive information, such as the epicentral distance, back azimuth, and signal-to-noise ratios. The magnitudes of seismic events, epicentral distance, signal-to-noise ratio of P-wave data, and signal-to-noise ratio of S-wave data ranged from 0 to 7.7, 0 to 330 km, –0.05 to 5.31 dB, and –0.05 to 4.73 dB, respectively. The dataset compiled in this study can serve as a high-quality benchmark for machine learning model development and data-driven seismological research on earthquake detection, seismic phase picking, first-motion polarity determination, earthquake magnitude prediction, early warning systems, and strong ground-motion prediction. Such research will further promote the development and application of artificial intelligence in seismology.

A Recognition Algorithm of Seismic Signals Based on Wavelet Analysis

In order to meet the requirements of mobile marine seismometers to observe and record seismic signals, a study of fast and accurate seismic signal recognition was carried out. This paper introduces the use of the wavelet analysis method for seismic signal processing and recognition, and compares and analyzes the abilities of different wavelet basis functions to detect the seismic signal. By denoising and reconstructing the signal, the distribution law of the wavelet coefficients of seismic signal at different scales was obtained. On this basis, this paper proposes an identification model of seismic signals based on wavelet analysis and thereby solves the conflict between high speed and high accuracy of seismic signal recognition methods. In this study, the simulation was carried out in the Matlab2020b environment, and the feasibility of wavelet recognition algorithm was proven by applying this algorithm to the seismic signal database for experimental verification.

Seismic signal recognition and interpretation of the 2019 “7.23” Shuicheng landslide by seismogram stations

A systematic study of the physical and mechanical processes of landslide development and evolution is important for forecasting, early warning, and prevention of landslide hazards. In the absence of on-site monitoring data, seismic networks can be employed to continuously record ground seismicity generated during landslides. However, landslide seismic signals are relatively weak and inevitably affected by noise interference. Furthermore, systematic characterization and reconstruction of the landslide evolution process remain poorly reported. An evaluation method to recognize landslide events based on seismic signal characteristics is therefore important. This study analyzes the 2019 “7.23” Shuicheng landslide based on data from nearby seismic stations. A landslide seismic signal recognition method is developed based on short-time Fourier transform (STFT) and band-pass filter (BP-filter) analysis. Data from 14 stations near the landslide were reviewed and the landslide data from one station was selected for analysis. The landslide seismic signal was noise-attenuated by using the empirical mode decomposition (EMD) and BP-filter methods. Fast Fourier transform (FFT), STFT, and power spectral density analyses were applied to the landslide seismic signal with higher signal-to-noise ratio (SNR) to obtain the time–frequency signal characteristics of the landslide process. Finally, combined with landslide field survey data, the dynamic process of the landslide was reconstructed based on the seismic signal, and the landslide was divided into four stages: the fracture-transition stage, the accelerated initiation stage, the bifurcation-scraping stage, and the deposition stage. The dynamic characteristics of each stage of the landslide are presented. The results indicate that the initial fracture point of the landslide is located between the bottom of the sliding source area and the top of the acceleration zone, not as traditionally thought, at the top of the sliding source area; this would be difficult to determine through field survey and analysis only. These results provide theoretical guidance for the study of seismic signal extraction, identification of landslide dynamic parameters, and characterization and reconstruction of landslide processes.

Seismic signal recognition by unsupervised machine learning

SUMMARYSeismic signal recognition can serve as a powerful auxiliary tool for analysing and processing ever-larger volumes of seismic data. It can facilitate many subsequent procedures such as first-break picking, statics correction, denoising, signal detection, events tracking, structural interpretation, inversion and imaging. In this study, I propose an automatic technique of seismic signal recognition taking advantage of unsupervised machine learning. In the proposed technique, seismic signal recognition is considered as a problem of clustering data points. All the seismic sampling points in time domain are clustered into two clusters, that is, signal or non-signal. The hierarchical clustering algorithm is used to group these sampling points. Four attributes, that is, two short-term-average-to-long-term-average ratios, variance and envelope are investigated in the clustering process. In addition, to quantitatively evaluate the performance of seismic signal recognition properly, I propose two new statistical indicators, namely, the rate between the total energies of original and recognized signals (RTE), and the rate between the average energies of original and recognized signals (RAE). A large number of numerical experiments show that when the signal is slightly corrupted by noise, the proposed technique performs very well, with recognizing accuracy, precision and RTE of nearly 1 (i.e. 100 per cent), recall greater than 0.8 and RAE about 1–1.3. When the signal is moderately corrupted by noise, the proposed technique can hold recognizing accuracy about 0.9, recognizing precision nearly to 1, RTE about 0.9, recall around 0.6 and RAE about 1.5. Applications of the proposed technique to real microseismic data induced from hydraulic fracturing and reflection seismic data demonstrate its feasibility and encouraging prospect.

Seismic signal recognition using improved BP neural network and combined feature extraction method

Seismic signal is generally employed in moving target monitoring due to its robust characteristic. A recognition method for vehicle and personnel with seismic signal sensing system was proposed based on improved neural network. For analyzing the seismic signal of the moving objects, the seismic signal of person and vehicle was acquisitioned from the seismic sensor, and then feature vectors were extracted with combined methods after filter processing. Finally, these features were put into the improved BP neural network designed for effective signal classification. Compared with previous ways, it is demonstrated that the proposed system presents higher recognition accuracy and validity based on the experimental results. It also shows the effectiveness of the improved BP neural network.

Landslide seismic signal recognition and mobility for an earthquake-induced rockslide in Tsaoling, Taiwan

Landslide seismic signal processing and the Newmark's sliding block analysis are used in this research to analyze the initiation time and mobility of the Tsaoling rockslide triggered by the 1999 Chi-Chi Earthquake in Central Taiwan. A new technique, the time–frequency spectrum analysis, is employed to identify the rockslide initiation and landing times from four real time seismic records of strong ground motion stations. The results suggest that higher frequencies ranging from 30 to 70Hz correspond to the rockslide initiation and sliding process, which are opposed to the low signal frequency of earthquake. The landslide began at 32.5s, the sliding block sped up at 37.5s, and impacted on the opposite riverbank at 76s on the seismic time record. The mobile scenario of Tsaoling rockslide induced by Chi-Chi Earthquake is simulated using Newmark's sliding block analysis. The maximum sliding velocity reaches approximately 75.1m/s. Despite being a simple conceptual model, the Newmark's sliding block analysis provides consistent results with the real time seismic signal analysis and details of the rockslide kinematics.

Recognition of moving ground targets by measuring and processing seismic signal

Because vehicles moving over ground generate a succession of impacts, the soil disturbances propagate away from the source as seismic waves. Thus, in the battlefield environment, we can detect moving ground vehicles by means of measuring seismic signals using a seismic velocity transducer, and automatically classify and recognize them by advance signal processing method. Because seismic sensor is easy to be developed by emerging micro-electro-mechanical system (MEMS) technology, seismic detection that will be low cost, low power, small volume and light weight is a promising method for moving ground targets. Such a detection method can be used in many different fields, such as battlefield surveillance, traffic monitoring, law enforcement and so on. The paper researches seismic signals of typical vehicle targets in order to extract features of seismic signal and to recognize targets. As a data fusion method, the technique of artificial neural networks (ANN) is applied to recognize seismic signals for vehicle targets. An improved BP algorithm and ANN data fusion architecture have been presented to improve learning speed and avoid local minimum points in error curve. The algorithm had been used for classification and recognition of seismic signals of vehicle targets in the outdoor environment. It can be proven that moving ground vehicles can be detected by measuring seismic signal, feature extraction of target seismic signal is correct and ANN data fusion is effective to solve the recognition and classification problem for moving ground targets.

A Novel Application of a Microaccelerometer for Target Classification

This paper presents a novel method of target classification by means of a microaccelerometer. Its principle is that the seismic signals from moving vehicle targets are detected by a microaccelerometer, and targets are automatically recognized by the advanced signal processing method. The detection system based on the microaccelerometer is small in size, light in weight, has low power consumption and low cost, and can work under severe circumstances for many different applications, such as battlefield surveillance, traffic monitoring, etc. In order to extract features of seismic signals stimulated by different vehicle targets and to recognize targets, seismic properties of typical vehicle targets are researched in this paper. A technique of artificial neural networks (ANNs) is applied to the recognition of seismic signals for vehicle targets. An improved back propagation (BP) algorithm and ANN architecture have been presented to improve learning speed and avoid local minimum points in error curve. The improved BP algorithm has been used for classification and recognition of seismic signals of vehicle targets in the outdoor environment. Through experiments, it can be proven that target seismic properties acquired are correct, ANN is effective to solve the problem of classification and recognition of moving vehicle targets, and the microaccelerometer can be used in vehicle target recognition.

Seismic signal detection by fractal dimension analysis

AbstractWe introduce a new detection algorithm with improved local and regional seismic signal recognition. The method is based on the difference between seismic signals and background random noise in terms of fractal dimension D. We compare the new method extensively with standard methods currently in use at the Seismic Network of the Istituto Nazionale di Geofisica. Results from the comparisons show that the new method recognizes seismic phases detected by existing procedures, and in addition, it features a greater sensitivity to smaller signals, without an increase in the number of false alarms. The new method was tested on real continuous data and artificially simulated high-noise conditions and demonstrated a capability to recognize seismic signals in the presence of high noise. The efficiency of the method is due to a radically different approach to the topic, in that the assertion that a signal is fractal implies a relationship between the spectral amplitude of different frequencies. This relationship allows, for the fractal detector, a complete analysis of the entire frequency range under consideration.

Three-component signal recognition using time, time-frequency, and polarization information-application to seismic detection of avalanches

A method for automatic signal recognition, applied to seismic signals classification, is presented. It is based on the fusion of data derived from the analysis of the signal in three domains: time, time-frequency, and polarization. In the time domain, two techniques are used for envelope shape parametrization. In the time-frequency domain, the autoregressive and Capon (ARCAP) time-frequency method is used on a gliding time-window to estimate the spectral components of the signal versus time. For each window, the frequencies are estimated using AR modelization. The power at each frequency and the corresponding filtered signal are estimated using Capon's (1969) method. A comparison with Fourier's narrow-bandpass filtering shows that Capon's method produces a better filtering. In the polarization domain, two original methods are proposed: one for checking the linear polarization of a signal and one for localizing the linear waves in the time-frequency plane. A system for automatic recognition of seismic signals associated with avalanches is then presented as an application. Signal features are derived from the analysis to sum up the characteristics of the signal in each domain. These features are combined using fuzzy logic and credibility factors, according to rules derived from physical knowledge (generating processes and propagation rules), in order to decide whether a signal comes from an avalanche or not. The global rate of correct recognition is over 90% for that first version of the system.

First results from a pre-operational system for automatic detection and recognition of seismic signals associated with avalanches

AbstractA system for automatic, reliable, semi-instantaneous estimation of avalanche activity is presented in an attempt to check deterministic models and improve the surveillance of risk zones. The principle is the seismic detection of avalanches. Two experimental sites equipped with standard seismological equipment are described. Avalanche seismic signals are recorded, as well as many extraneous signals of natural or human origin. After several years of a posteriori identification of the signals, we are now able to set up criteria for the automatic recognition of non-avalanche signals. We have therefore developed an automatic analysis/decision system to discriminate between avalanche signals and others. This system worked satisfactorily in pre-operational conditions during the winter of 1995-96. The results are presented and compared with other data related to avalanche activity. Although there is still room for improvement, our system seems to be able to estimate avalanche activity better than human visual observations.

First results from a pre-operational system for automatic detection and recognition of seismic signals associated with avalanches

AbstractA system for automatic, reliable, semi-instantaneous estimation of avalanche activity is presented in an attempt to check deterministic models and improve the surveillance of risk zones. The principle is the seismic detection of avalanches. Two experimental sites equipped with standard seismological equipment are described. Avalanche seismic signals are recorded, as well as many extraneous signals of natural or human origin. After several years of a posteriori identification of the signals, we are now able to set up criteria for the automatic recognition of non-avalanche signals. We have therefore developed an automatic analysis/decision system to discriminate between avalanche signals and others. This system worked satisfactorily in pre-operational conditions during the winter of 1995-96. The results are presented and compared with other data related to avalanche activity. Although there is still room for improvement, our system seems to be able to estimate avalanche activity better than human visual observations.