基于BFO-LSSVM算法的爆破振动峰值速度预测  

Predicting Peak Velocity of Blasting Vibration Using BFO-LSSVM Algorithm

作　　者：张国鹏 赵根[1] 胡英国[1] 郑皓文 饶宇[1] 杨招伟 Zhang Guopeng;Zhao Gen;Hu Yingguo;Zheng Haowen;Rao Yu;Yang Zhaowei(Key Laboratory of Geotechnical Mechanics and Engineering of Ministry of Water Resources,Yangtze River Science Research Institute,Wuhan 430010,China;College of Civil and Transportation Engineering,Hohai University,Nanjing 210098,China)

机构地区：[1]长江科学院水利部岩土力学与工程重点实验室,武汉430010 [2]河海大学土木与交通学院,南京210098

出　　处：《长江技术经济》2022年第5期51-56,共6页Technology and Economy of Changjiang

基　　金：国家自然科学基金青年基金项目(52109148);中央级公益性科研院所基本科研业务费(CKSF2021461/YT)。

摘　　要：钻爆法施工隧洞开挖往往面临复杂的围岩地质条件,采用传统的爆破振速预测方法难以得到准确的峰值振速,为此利用细菌觅食算法(BFO)优化最小二乘支持向量机(LSSVM)参数,以炮孔进深、距爆心水平距离、高程、最大单响药量、总药量为输入因子,峰值振速为输出因子,运用BFO-LSSVM模型预测爆破振动峰值速度。结合滇中引水工程昆明段下游引水隧洞主洞爆破开挖过程中30组现场爆破振动监测数据,分别采用BFO-LSSVM模型、LS-SVM模型和萨道夫斯基公式进行爆破振动速度预测,预测结果与实测值的平均相对误差分别为4.02%、12.18%、27.85%。结果表明BFO-LSSVM模型具有更显著的数据拟合能力,对于复杂地质条件下的隧洞开挖爆破振动峰值速度预测有更强的适用性。Tunnel excavation by drilling and blasting is often faced with complex geological conditions of surrounding rock.It is difficult to obtain accurate peak vibration velocity by traditional prediction methods.To solve this problem,bacterial foraging optimization(BFO)algorithm was adopted to optimize the parameters of least squares support vector machine(LS-SVM).A BFO-LSSVM model was established and used to predict the peak velocity of blasting vibration with the depth of blasthole,horizontal distance from the blast center,elevation,maximum single blow charge and total charge as input factors,and peak vibration velocity as output factor.Thirty groups of on-site blasting vibration monitoring data during the blasting excavation of the main tunnel of the downstream diversion tunnel in Kunming segment of the Central Yunnan Water Diversion Project were taken as a case study.BFO-LSSVM,LS-SVM and the Sadowsky formula were employed to predict the blasting vibration velocity.The average relative errors between prediction results and measured values are 4.02%,12.18%,and 27.85%,respectively.The results demonstrate that the BFO-LSSVM model is of superior data fitting ability and is more applicable to predict the peak velocity of tunnel excavation blasting vibration under complex geological conditions.

关 键 词：菌群觅食算法 最小二乘法支持向量机 隧道开挖 爆破振动速度 预测精度 

分 类 号：TV213[水利工程—水文学及水资源]