{"title":"A physics-guided hierarchical deep learning framework for underground rock reinforcement compliance check based on 4D point cloud data","authors":"Zhen Han , Qian Li , Xiangyu Wang , Danqi Li","doi":"10.1016/j.tust.2025.106829","DOIUrl":null,"url":null,"abstract":"<div><div>Rock bolts have been extensively used for rock reinforcement in underground mines. The compliance check for rock bolts installation pattern becomes significantly important for ensuring an optimal balance between the cost and performance of rock reinforcement. The current manual compliance check process requires tremendous manpower and inevitably introduces human errors and data bias issues. In order to address this challenge, in this study, a novel physics-guided hierarchical deep learning framework for underground rock reinforcement compliance check based on 4D point cloud data, termed PGHDFramework, was proposed. In this framework, a physics-based forward approach for bolt-level classification named Intensity-based Forward Classification (IBFC) model was introduced first, which requires no training process. Then a hierarchical deep neural network based on PointNet++ that can concatenate spatial information (i.e., x, y, z coordinates) and physical property information (i.e., intensity value) at different levels of abstraction, termed 4D Bolt Detection Neural Network (4DBDNet), was developed. The SLAM-LiDAR point cloud data from five sections of an underground mine with different conditions containing 25,146,657 points were used for validating the framework as well as comparing its performance with the existing methods. The precision, recall, F1 score and IoU of the proposed approach at point level are 0.92, 0.94, 0.93 and 0.73 separately, and at bolt-level are 0.60, 0.84, 0.70 respectively, showing a much higher promising performance than other methods. The accuracy and effectiveness of the proposed framework was further confirmed by compliance checking a fresh underground mine drive to autonomously generate the rock bolts spacing and row spacing. This study ultimately provides the framework of a universal-applicable digital approach to the mining industry for a more cost-effective and accurate rock reinforcement compliance check in practice.</div></div>","PeriodicalId":49414,"journal":{"name":"Tunnelling and Underground Space Technology","volume":"164 ","pages":"Article 106829"},"PeriodicalIF":7.4000,"publicationDate":"2025-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Tunnelling and Underground Space Technology","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0886779825004675","RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"CONSTRUCTION & BUILDING TECHNOLOGY","Score":null,"Total":0}
引用次数: 0
Abstract
Rock bolts have been extensively used for rock reinforcement in underground mines. The compliance check for rock bolts installation pattern becomes significantly important for ensuring an optimal balance between the cost and performance of rock reinforcement. The current manual compliance check process requires tremendous manpower and inevitably introduces human errors and data bias issues. In order to address this challenge, in this study, a novel physics-guided hierarchical deep learning framework for underground rock reinforcement compliance check based on 4D point cloud data, termed PGHDFramework, was proposed. In this framework, a physics-based forward approach for bolt-level classification named Intensity-based Forward Classification (IBFC) model was introduced first, which requires no training process. Then a hierarchical deep neural network based on PointNet++ that can concatenate spatial information (i.e., x, y, z coordinates) and physical property information (i.e., intensity value) at different levels of abstraction, termed 4D Bolt Detection Neural Network (4DBDNet), was developed. The SLAM-LiDAR point cloud data from five sections of an underground mine with different conditions containing 25,146,657 points were used for validating the framework as well as comparing its performance with the existing methods. The precision, recall, F1 score and IoU of the proposed approach at point level are 0.92, 0.94, 0.93 and 0.73 separately, and at bolt-level are 0.60, 0.84, 0.70 respectively, showing a much higher promising performance than other methods. The accuracy and effectiveness of the proposed framework was further confirmed by compliance checking a fresh underground mine drive to autonomously generate the rock bolts spacing and row spacing. This study ultimately provides the framework of a universal-applicable digital approach to the mining industry for a more cost-effective and accurate rock reinforcement compliance check in practice.
期刊介绍:
Tunnelling and Underground Space Technology is an international journal which publishes authoritative articles encompassing the development of innovative uses of underground space and the results of high quality research into improved, more cost-effective techniques for the planning, geo-investigation, design, construction, operation and maintenance of underground and earth-sheltered structures. The journal provides an effective vehicle for the improved worldwide exchange of information on developments in underground technology - and the experience gained from its use - and is strongly committed to publishing papers on the interdisciplinary aspects of creating, planning, and regulating underground space.