基于微粒群–BP神经网络算法的堆石坝坝体变形监控模型研究  

STUDY OF MONITORING MODEL OF FACED ROCKFILL DAM'S DEFORMATION BASED ON PARTICLE SWARM OPTIMIZATION-BP ALGORITHM

作　　者：罗丹[1,2] 李昌彩 吴长彬[3] 

机构地区：[1]武汉大学水资源与水电工程科学国家重点实验室,湖北武汉430072 [2]湖北能源集团股份有限公司,湖北武汉430062 [3]中国水电工程顾问集团昆明勘测设计研究院,云南昆明650051

出　　处：《岩石力学与工程学报》2012年第A01期2926-2931,共6页Chinese Journal of Rock Mechanics and Engineering

摘　　要：将微粒群算法与BP神经网络算法相结合,建立基于微粒群–BP神经网络算法的大坝安全监控模型,解决传统神经网络存在的网络学习速度慢,且易于收敛到局部最优点等问题。经实例分析,该模型在用于建立大坝安全监控预报模型过程中,能更好地反映大坝应力变形的非线性问题,且在精度和速度上都有较大的提高,为实现实时在线分析评价大坝的安全性态提供了有力的技术支持。A dam safety monitoring model is presented by combining a BP neural network algorithm and particle swarm optimization（PSO） algorithm. This model solves existing the learning and traditional neural network to speed of neural network is slow, and it＇s easy to converge to local optimum. The application shows that this method is effective for reflect the stress deformation of nonlinear problems during establishment of neural network forecast model. This model significantly promotes the speed and precision of the calculation and provides the real time on line assessment of dam safety with a useful technical support.

关 键 词：水利工程 微粒群算法 BP神经网络 堆石坝 变形监控模型 

分 类 号：TV31[水利工程—水工结构工程]