Citation Search Quick Search
  2023, Vol. 42 Issue (S1): 3458-3472    DOI: 10.13722/j.cnki.jrme.2022.0284
Artiles Current Issue | Archive | Adv Search |
Prediction method of fracture behaviors of rock samples with an infilled fracture based on machine learning
CHEN Jinfan1，SHANG Delei1，2，3，ZHAO Zhihong1，CHEN Zhaowei4
(1. Department of Civil Engineering，Tsinghua University，Beijing 100084，China；2. Institute of Deep Earth Sciences and Green Energy，Shenzhen University，Shenzhen，Guangdong 518060，China；3. Guangdong Provincial Key Laboratory of Deep Earth Sciences and Geothermal Energy Exploitation and Utilization，Shenzhen University，Shenzhen，Guangdong 518060，China；
4. CNPC Engineering Technology R&D Co.，Ltd.，Beijing 102206，China)
Download: PDF (3270 KB)   HTML (1 KB) 
Export: BibTeX | EndNote (RIS)      
Abstract  Fracture behaviors of rock samples with discontinuities is difficult to quantify and predict. In this study，machine learning method is used to predict mode I fracture toughness and crack propagation mode of rock samples with an infilled fracture. Firstly，the feasibility of the method was verified by comparison of results of notched semi-circle bend test and numerical simulation based on discrete element method，and the dataset which filtered the outliers was generated by a lot of random numerical simulations. Multiple techniques were used in the preprocessing of dataset. Then four classical machine learning models were established. The optimal parameters of the models were obtained by grid search and five-fold cross-validation method. Multiple indicators and graphs were used to verify the prediction performances of the models. Comparison between the four models shows that Multilayer Perceptron(MLP) is the optimal machine learning model. The sensitivity of the MLP model was analyzed and the stability is good. The MLP model was applied to the test samples，which has high prediction accuracy. And a user interface program was developed. The data driven model of rock mechanics based on the machine learning method reduces time cost compared with the traditional test and numerical simulation，which provides a new method and thought for solving traditional rock engineering problems.
Key words： rock mechanics      fracture toughness      machine learning      discrete element method      discontinuities
     
Service
E-mail this article
Add to my bookshelf
Add to citation manager
E-mail Alert
RSS
Articles by authors
CHEN Jinfan1
SHANG Delei1
2
3
ZHAO Zhihong1
CHEN Zhaowei4
Cite this article:   
CHEN Jinfan1,SHANG Delei1,2, et al. Prediction method of fracture behaviors of rock samples with an infilled fracture based on machine learning[J]. , 2023, 42(S1): 3458-3472.
URL:  
https://rockmech.whrsm.ac.cn/EN/10.13722/j.cnki.jrme.2022.0284      OR      https://rockmech.whrsm.ac.cn/EN/Y2023/V42/IS1/3458
Copyright © 2005-2015 Edited and Published by Editorial Office for Chinese Journal of Rock Mechanics and Engineering
Sponsored by : Chinese Society for Rock Mechanics and Engineering 
Published by : Science Press 
Address： Wuhan 430071, China  Tel: (027)87199250  Fax: (027)87199250  Email: rock@whrsm.ac.cn
DESIGNED BY: Beijing Magtech Science & Technolgy Development Co.,Ltd.
鄂公网安备 42010602003581号