A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Assessing the Probability of Strainburst Potential Via an Integration of Monte Carlo Simulation and Machine Learning Algorithms
Abstract Strainburst poses significant risks to the safety of workers in deep underground engineering. Assessing strainburst potential plays a crucial role in disaster prevention. Due to the anisotropy and heterogeneity of rock mass, it is essential to quantify the strainburst potential from the perspective of probability. This study aims to propose a methodology that integrates Monte Carlo simulation (MCS) and machine learning (ML) algorithms to evaluate the probability of strainburst potential. First, a numerical model based on damage initiation and spalling limit (DISL) method is established. Then, a novel indicator, absolute local energy release rate (ALERR), is proposed to assess strainburst potential, which describes the absolute value of elastic strain energy released after rock mass failure. The location of the maximum value of ALERR is used to determine the depth of outburst pit, which is adopted to evaluate the strainburst potential level. So, the quantitative relationship between strainburst potential and ALERR can be established. Thereafter, a strainburst dataset including six variables and the depth of outburst pit is obtained based on the MCS method and numerical model. To improve the computational efficiency, seven different ML algorithms are integrated with MCS to calculate the probability of strainburst potential, respectively. Finally, the proposed methodology is used for the strainburst potential evaluation in the 3# headrace tunnel of Jinping II hydropower station. Results show that random forest (RF) and gradient boosting (GB) algorithms have better evaluation performance and can be combined with MCS to calculate the probability of strainburst potential reliably. The proposed methodology can provide valuable guidance for the probability assessment of strainburst potential.
Highlights A new methodology that integrates Monte Carlo simulation (MCS) and machine learning (ML) algorithms is proposed to evaluate the probability of strainburst potential.A novel indicator, absolute local energy release rate (ALERR), is proposed to assess strainburst potential.A numerical model, which combines MCS and damage initiation and spalling limit (DISL) approaches, is established to simultaneously simulate the brittle failure behavior of hard rock and variability of rock mechanics parameters.
Assessing the Probability of Strainburst Potential Via an Integration of Monte Carlo Simulation and Machine Learning Algorithms
Abstract Strainburst poses significant risks to the safety of workers in deep underground engineering. Assessing strainburst potential plays a crucial role in disaster prevention. Due to the anisotropy and heterogeneity of rock mass, it is essential to quantify the strainburst potential from the perspective of probability. This study aims to propose a methodology that integrates Monte Carlo simulation (MCS) and machine learning (ML) algorithms to evaluate the probability of strainburst potential. First, a numerical model based on damage initiation and spalling limit (DISL) method is established. Then, a novel indicator, absolute local energy release rate (ALERR), is proposed to assess strainburst potential, which describes the absolute value of elastic strain energy released after rock mass failure. The location of the maximum value of ALERR is used to determine the depth of outburst pit, which is adopted to evaluate the strainburst potential level. So, the quantitative relationship between strainburst potential and ALERR can be established. Thereafter, a strainburst dataset including six variables and the depth of outburst pit is obtained based on the MCS method and numerical model. To improve the computational efficiency, seven different ML algorithms are integrated with MCS to calculate the probability of strainburst potential, respectively. Finally, the proposed methodology is used for the strainburst potential evaluation in the 3# headrace tunnel of Jinping II hydropower station. Results show that random forest (RF) and gradient boosting (GB) algorithms have better evaluation performance and can be combined with MCS to calculate the probability of strainburst potential reliably. The proposed methodology can provide valuable guidance for the probability assessment of strainburst potential.
Highlights A new methodology that integrates Monte Carlo simulation (MCS) and machine learning (ML) algorithms is proposed to evaluate the probability of strainburst potential.A novel indicator, absolute local energy release rate (ALERR), is proposed to assess strainburst potential.A numerical model, which combines MCS and damage initiation and spalling limit (DISL) approaches, is established to simultaneously simulate the brittle failure behavior of hard rock and variability of rock mechanics parameters.
Assessing the Probability of Strainburst Potential Via an Integration of Monte Carlo Simulation and Machine Learning Algorithms
Liang, Weizhang (author) / Liu, Huanxin (author) / Zhao, Guoyan (author) / Chen, Ying (author) / Ma, Ju (author) / Lan, Ming (author)
2022
Article (Journal)
Electronic Resource
English
BKL:
38.58
Geomechanik
/
56.20
Ingenieurgeologie, Bodenmechanik
/
38.58$jGeomechanik
/
56.20$jIngenieurgeologie$jBodenmechanik
RVK:
ELIB41
Efficient Monte-Carlo probability integration
| TIBKAT | 1984
Influence of loading rate on strainburst: an experimental study
| Online Contents | 2018
Influence of loading rate on strainburst: an experimental study
| Online Contents | 2018
Numerical Analysis of Unloading-Induced Rock Failure: Insight into Strainburst Mechanism
| Online Contents | 2018