Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Performance and Applicability of Recognizing Microseismic Waveforms Using Neural Networks in Tunnels
https://orcid.org/http://orcid.org/0000-0002-3144-2448
https://orcid.org/http://orcid.org/0000-0002-2645-1112
https://orcid.org/http://orcid.org/0000-0002-2607-8696
The sudden and harmful nature of rockbursts in tunnels necessitates an accurate and applicable method for automatically recognizing rock fracture signals during rockburst microseismic (MS) monitoring. In this paper, the performance and applicability of recognizing MS waveforms using an artificial neural network (ANN) and a deep neural network (DNN) were studied in tunnels excavated by different methods. The results show that ANN performs very well in recognizing rock fracturing waveforms with a signal-to-noise ratio (SNR) ≥ 3 but has a low accuracy for those with an SNR < 3. The DNN also performs well for waveforms with SNR ≥ 3, and has a relatively high accuracy for waveforms with SNR < 3. The ANN model can be used in tunnels excavated by drilling and blasting (D&B) since there are fewer “small” rock fracturing events. The DNN model is applicable in tunnels excavated by the tunnel boring machine (TBM), recognizing more “small” events. In addition, the ANN model is a better choice, with fewer training samples at the initial stage of monitoring working. With continuous monitoring, the DNN model can be used to ensure and improve the accuracy. These results lay a foundation for automatic rockburst MS monitoring techniques in tunnel engineering.
Performance and Applicability of Recognizing Microseismic Waveforms Using Neural Networks in Tunnels
https://orcid.org/http://orcid.org/0000-0002-3144-2448
https://orcid.org/http://orcid.org/0000-0002-2645-1112
https://orcid.org/http://orcid.org/0000-0002-2607-8696
The sudden and harmful nature of rockbursts in tunnels necessitates an accurate and applicable method for automatically recognizing rock fracture signals during rockburst microseismic (MS) monitoring. In this paper, the performance and applicability of recognizing MS waveforms using an artificial neural network (ANN) and a deep neural network (DNN) were studied in tunnels excavated by different methods. The results show that ANN performs very well in recognizing rock fracturing waveforms with a signal-to-noise ratio (SNR) ≥ 3 but has a low accuracy for those with an SNR < 3. The DNN also performs well for waveforms with SNR ≥ 3, and has a relatively high accuracy for waveforms with SNR < 3. The ANN model can be used in tunnels excavated by drilling and blasting (D&B) since there are fewer “small” rock fracturing events. The DNN model is applicable in tunnels excavated by the tunnel boring machine (TBM), recognizing more “small” events. In addition, the ANN model is a better choice, with fewer training samples at the initial stage of monitoring working. With continuous monitoring, the DNN model can be used to ensure and improve the accuracy. These results lay a foundation for automatic rockburst MS monitoring techniques in tunnel engineering.
Performance and Applicability of Recognizing Microseismic Waveforms Using Neural Networks in Tunnels
https://orcid.org/http://orcid.org/0000-0002-3144-2448
https://orcid.org/http://orcid.org/0000-0002-2645-1112
https://orcid.org/http://orcid.org/0000-0002-2607-8696
The sudden and harmful nature of rockbursts in tunnels necessitates an accurate and applicable method for automatically recognizing rock fracture signals during rockburst microseismic (MS) monitoring. In this paper, the performance and applicability of recognizing MS waveforms using an artificial neural network (ANN) and a deep neural network (DNN) were studied in tunnels excavated by different methods. The results show that ANN performs very well in recognizing rock fracturing waveforms with a signal-to-noise ratio (SNR) ≥ 3 but has a low accuracy for those with an SNR < 3. The DNN also performs well for waveforms with SNR ≥ 3, and has a relatively high accuracy for waveforms with SNR < 3. The ANN model can be used in tunnels excavated by drilling and blasting (D&B) since there are fewer “small” rock fracturing events. The DNN model is applicable in tunnels excavated by the tunnel boring machine (TBM), recognizing more “small” events. In addition, the ANN model is a better choice, with fewer training samples at the initial stage of monitoring working. With continuous monitoring, the DNN model can be used to ensure and improve the accuracy. These results lay a foundation for automatic rockburst MS monitoring techniques in tunnel engineering.
Performance and Applicability of Recognizing Microseismic Waveforms Using Neural Networks in Tunnels
KSCE J Civ Eng
Zhang, Wei (Autor:in) / Bi, Xin (Autor:in) / Hu, Lei (Autor:in) / Li, Pengxiang (Autor:in) / Yao, Zhibin (Autor:in)
KSCE Journal of Civil Engineering ; 28 ; 951-966
01.02.2024
16 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Performance and Applicability of Recognizing Microseismic Waveforms Using Neural Networks in Tunnels
| Springer Verlag | 2024
Sectional velocity model for microseismic source location in tunnels
| Online Contents | 2015
Sectional velocity model for microseismic source location in tunnels
| British Library Online Contents | 2015
Sectional velocity model for microseismic source location in tunnels
| British Library Online Contents | 2015