Machine learning-based classification of rock discontinuity trace: SMOTE oversampling integrated with GBT ensemble learning

Chen, Jiayao; Huang, Hongwei; Cohn, Anthony G.; Zhang, Dong Ming; Zhou, Mingliang

doi:10.1016/j.ijmst.2021.08.004

Cited by 61 publications

(15 citation statements)

References 70 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similar to image pixels, the points that are sampled along joint or crack openings are typically sparser than their background materials. Chen et al [27] use the synthetic minority over-sampling technique (SMOTE) to generate more sampling points along joint traces in a point cloud of a tunnel face. Azhari et al implement voxelisation followed by down-sampling to reduce the total number of points in the point cloud for deep learning input, but maintains the sparse nature of the points along rock discontinuities [21].…”

Section: Data Imbalancementioning

confidence: 99%

Potential applications of deep learning in automatic rock joint trace mapping in a rock mass

Chiu

Mengshoel

2023

IOP Conf. Ser.: Earth Environ. Sci.

View full text Add to dashboard Cite

In blasted rock slopes and underground openings, rock joints are visible in different forms. Rock joints are often exposed as planes confining rock blocks and visible as traces on a well-blasted, smooth rock mass surface. A realistic rock joint model should include both visual forms of joints in a rock mass: i.e., both joint traces and joint planes. Imaged-based 2D semantic segmentation using deep learning via the Convolutional Neural Network (CNN) has shown promising results in extracting joint traces in a rock mass. In 3D analysis, research studies using deep learning have demonstrated outperforming results in automatically extracting joint planes from an unstructured 3D point cloud compared to state-of-the-art methods. We discuss a pilot study using 3D true colour point cloud and their source and derived 2D images in this paper. In the study, we aim to implement and compare various CNN-based networks found in the literature for automatic extraction of joint traces from laser scanning and photogrammetry data. Extracted joint traces can then be clustered and connected to potential joint planes as joint objects in a discrete joint model. This can contribute to a more accurate estimation of rock joint persistence. The goal of the study is to compare the efficiency and accuracy between using 2D images and 3D point cloud as input data. Data are collected from two infrastructure projects with blasted rock slopes and tunnels in Norway.

show abstract

Section: Data Imbalancementioning

confidence: 99%

Potential applications of deep learning in automatic rock joint trace mapping in a rock mass

Chiu

Mengshoel

2023

IOP Conf. Ser.: Earth Environ. Sci.

View full text Add to dashboard Cite

show abstract

“…Several studies have been conducted using convolutional neural networks (CNNs), which show decent performance in image recognition among machine learning techniques, to evaluate the rock grade of a tunnel face [1,2,3]. In addition, a machine learning model for recognising joint traces from images of rock has been IOP Publishing doi:10.1088/1755-1315/1124/1/012007 2 developed through a combination of machine learning algorithms such as gradient boosting tree (GBT), random forest (RF), decision tree (DT), and multi-layer perceptron (MLP) [4].…”

Section: Introductionmentioning

confidence: 99%

Predicting the geological condition beyond the tunnel excavation face using MSP monitoring data and LSTM algorithm

Lee

2023

IOP Conf. Ser.: Earth Environ. Sci.

View full text Add to dashboard Cite

The ground conditions beyond an excavation face, especially discontinuities in rock masses, have a significant influence on tunnel construction. However, the actual ground conditions observed during tunnel construction are often different from the ground conditions predicted in the geotechnical site explorations carried out in the design stage. Changes in ground conditions may require alterations in tunnel design, leading to substantial disruptions in the construction schedule and budget. In this regard, accurate ground evaluation prior to the design and construction stages are essential for successful tunnel construction projects. Machine learning models have been developed in order to evaluate the condition of rock discontinuities within 50 m of the tunnel face. Machine data (rotational pressure, feed pressure, and drilling (advance) speed) obtained from a large boring hole machine, called MSP, at an NATM construction site in a granite formation located in South Korea were logged, and the actual ground Discontinuity Score (DS) was appraised by analysing internal bore hole images taken after drilling. Then, the LSTM algorithm was applied to develop the machine learning model to determine DS based on the logged machine data. DS was most accurately predicted when the drilling speed was included in the input data, whereas those cases using only the rotational and feed pressure in the input data showed low prediction accuracy. Therefore, the drilling speed seems to have a higher correlation than hydraulic pressure with regard to ground conditions, including discontinuities. Once additional data is collected from various tunnel sites, the machine learning model could be further enhanced to become more robust and provide solutions to various engineering problems.

show abstract

“…In this paper, we aim to carry out fine-grained identification and quantification of rock linear discontinuities (cracks) on various types of rock surfaces and segment each crack from the background at pixel level. Current batch-based segmentation methods have provided excellent pixel-wise segmentation methods for rock defect detection using deep learning, but they mostly consider simple weak interlayer segmentation or tunnel face defect identification [27][28][29]. These methods can barely handle the fine-grained segmentation problem of multiple linear discontinuities at the pixel level.…”

Section: Introductionmentioning

confidence: 99%

Multi-CrackNet: A Fast Segmentation and Quantification Combined Method for Rock Discontinuity Traces by Capturing an Image

Lan

Deng

et al. 2022

Geofluids

View full text Add to dashboard Cite

Discontinuity investigation and characterization onsite is a labor-dependent work because current techniques cannot precisely handle multiple discontinuity identifications automatically under different work conditions. This paper proposes the multi-CrackNet which enables us to identify and segment linear discontinuities (joints and cracks) for random types of rock surface. A modified feature extraction network called the multiscale feature fusion pyramid network (MFFPN) has been developed based on FPN to capture and fuse more sensitive texture features of cracks across different types of background. With the help of a new training scheme by setting up 3 stages of training to simulate the human-based learning process, the established model can learn more features steadily and robustly from well-labelled databases. Additionally, a hybrid pixel-level quantification method is proposed to automatically compute the length, width, and inclination of cracks. Results show that the proposed method can achieve a detection accuracy of 87.1% for 1 to 9 sets of cracks on the rock surface across different types of rock. Case studies in Anshan West are provided to verify the reliability and accuracy of our method in macrolinear discontinuity identification and quantification, which sees great potentials in site investigation by saving a large amount of labor force.

show abstract

Machine learning-based classification of rock discontinuity trace: SMOTE oversampling integrated with GBT ensemble learning

Cited by 61 publications

References 70 publications

Potential applications of deep learning in automatic rock joint trace mapping in a rock mass

Potential applications of deep learning in automatic rock joint trace mapping in a rock mass

Predicting the geological condition beyond the tunnel excavation face using MSP monitoring data and LSTM algorithm

Multi-CrackNet: A Fast Segmentation and Quantification Combined Method for Rock Discontinuity Traces by Capturing an Image

Contact Info

Product

Resources

About