摘要
隧道爆破参数不当会严重影响隧道施工的安全和质量。因此,确定合适的爆破参数是隧道施工中一项重要的工作。为了解决此问题,基于深度学习模型鲸鱼优化深度置信网络(whale optimization deep belief network,WO-DBN)及多目标优化算法非支配排序遗传算法Ⅱ(non-dominated sorting genetic algorithm Ⅱ,NSGA-Ⅱ),提出了一种进行隧道爆破参数优化的智能算法。首先,使用开发的深度学习模型WO-DBN构建了基于地质参数及爆破参数进行隧道爆破施工安全及质量预测的智能模型,以隧道拱顶下沉和超欠挖面积作为施工安全及质量评价的指标。其次,基于建立的隧道爆破施工安全及质量评价模型,采用NSGA-Ⅱ以控制拱顶下沉和超欠挖面积为目标,提出进行隧道爆破参数优化的智能算法。最后,以蟠龙山公路隧道爆破施工为例,对提出的新算法进行工程应用验证。结果表明,采用新算法得到的施工参数,可以使得隧道拱顶下沉和超欠挖面积分别降低27.05%和60.30%,施工效果得到极大提高。因此,提出的智能算法可以为隧道爆破参数的实时优化控制提供技术支持,为隧道施工的顺利进行提供有力保障。
Improper tunnel blasting parameters will seriously affect the safety and quality of tunnel construction.Therefore,the determination of appropriate blasting parameters is an important work in tunnel construction.In order to solve this problem,based on deep learning model-whale optimization deep belief network(WO-DBN)and multi-objective optimization algorithm-non-dominated sorting genetic algorithm Ⅱ(NSGA-Ⅱ),an intelligent algorithm for tunnel blasting parameters optimization was proposed.Firstly,using the developed deep learning model WO-DBN,an intelligent model for predicting the safety and quality of tunnel blasting construction based on geological parameters and blasting parameters was constructed.The tunnel crown subsidence and overbreak and underbreak area were taken as the index of construction safety and quality evaluation.Secondly,based on the established tunnel blasting construction safety and quality evaluation model,an intelligent algorithm for tunnel blasting parameter optimization was proposed by using NSGA-Ⅱ to control crown subsidence,overbreak and underbreak area.Finally,taking the blasting construction of Panlongshan highway tunnel as an example,the proposed new algorithm was verified by engineering application.The results show that the construction parameters obtained by the new algorithm can reduce the tunnel crown subsidence and the overbreak and underbreak area by 27.05%and 60.30%,respectively,and the construction effect is greatly improved.Therefore,the proposed intelligent algorithm can provide technical support for the real-time optimization control of tunnel blasting parameters and provide a strong guarantee for the smooth progress of tunnel construction.
作者
奚灵智
葛双双
李晨
高玮
张强
胡少斌
杨槐
陈新
赵志浩
XI Ling-zhi;GE Shuang-shuang;LI Chen;GAO Wei;ZHANG Qiang;HU Shao-bin;YANG Huai;CHEN Xin;ZHAO Zhi-hao(PowerChina Huadong Engineering Corporation Limited,Hangzhou 310022,China;School of Civil and Transportation Engineering,Hohai University,Nanjing 210098,China)
出处
《科学技术与工程》
北大核心
2025年第21期8841-8850,共10页
Science Technology and Engineering
基金
华东院重大科技计划(201计划)(KY2021-ZD-04)。
关键词
隧道爆破开挖
深度学习模型
拱顶下沉
超欠挖
多目标优化
tunnel blasting excavation
deep learning model
crown subsidence
overbreak and underbreak
multi-objective optimization
作者简介
第一作者:奚灵智(1982-),男,汉族,浙江台州人,硕士,正高级工程师。研究方向:交通与市政工程设计咨询及EPC总承包管理。E-mail:xi_lz@hdec.com;通信作者:高玮(1971-),男,汉族,陕西富平人,博士,教授。研究方向:岩石力学理论、岩土工程稳定分析及智能大数据技术的工程应用。E-mail:wgaowh@163.com。
