摘要
基于上海地铁二号线的实测沉降数据,运用遗传算法(GA)和粒子群算法(PSO)对传统BP神经网络进行了优化,以弥补BP神经网络在网络结构、权值和阈值选择上的随机性以及容易局部收敛等缺陷,据此提出了两种新型隧道长期沉降预测模型,即GA-BP神经网络和PSO-BP神经网络模型;并对比研究了经验曲线、BP神经网络、GA-BP神经网络以及PSO-BP神经网络等模型方法的优缺点及预测效果.研究发现,以上各神经网络模型均取得了较为满意的预测结果,其中PSO-BP神经网络模型的预测精度最佳,且运算速度最快,是文中所提方法中最适用的盾构隧道长期沉降预测模型.
Based on in-situ monitoring data of Shanghai metro line 2,two new subsidence prediction models of tunnels,GA-BP neural network model and PSO-BP neural network model are proposed.These two models optimize the conventional BP neural network model by means of genetic algorithm(GA)and particle swarm optimization(PSO)in order to remedy the defects of BP neural network model,i.e.the randomness of selection on network structure,weight values and threshold values,as well as the inclination to local convergence.A comparative analysis is carried out between empirical curve,BP neural network model,GA-BP neural network model and PSO-BP neural network model,about their strengths,weaknesses and prediction effects.The results show that,PSO-BP neural network model turns to be the best model with optimal accuracy and fast operation speed,which is the most suitable prediction model for the long-term subsidence of shield tunnels,although the above prediction models have achieved appropriate prediction.
作者
李翔宇
李新源
李明宇
聂俊霞
冯晓波
LI Xiangyu;LI Xinyuan;LI Mingyu;NIE Junxia;FENG Xiaobo(State Key Laboratory of Building Safety and Built Environment,Beijing 100013,China;Institute of Foundation Engineering,China Academy of Building Research,Beijing 100013,China;School of Civil Engineering,Xuzhou Institute of Technology,Xuzhou 221018,China;School of Civil Engineering,Zhengzhou University,Zhengzhou 450001,China;Urban Rail Transit Engineering Company Limited of China Railway 15th Bureau,Luoyang 471499,China;Government Offices Administration,Xinhua News Agency,Beijing 100803,China)
出处
《西安建筑科技大学学报（自然科学版）》
北大核心
2021年第2期186-193,共8页
Journal of Xi'an University of Architecture & Technology(Natural Science Edition)
基金
中国建筑科学研究院有限公司青年科研基金资助项目(20161602331030048)
国家自然科学基金资助项目(51508520)
河南省住房城乡建设科技计划资金资助项目(K-1817、K-1818、K1816、K-1940)。
