结构损伤识别的耦合神经网络方法

英文篇名：A Coupled Neural Networks Method for Structural Damage Identification
中文刊名：沈阳建筑大学学报(自然科学版)
英文刊名：Journal of Shenyang Jianzhu University(Natural Science)
作者：焦莉 ; 李宏男
英文作者：JIAO Li~(1 ; 2) ; LI Hong-nan~1(1.State Key Laboratory of Costal and offshore Engineering ; Dalian University of Technology ; Dalian ; China ; 116024 ; 2.School of Science ; Shenyang Jianzhu University ; Shenyang ; China ; 110168)
中文关键词：神经网络 ; 遗传算法 ; 耦合神经网络 ; 损伤识别 ; 基准结构
英文关键词：neural networks ; genetic algorithm ; coupled neural networks ; damage identification ; benchmark structure
出版日期：2006-02-25
机构：大连理工大学海岸及近海工程国家重点实验室,大连理工大学海岸及近海工程国家重点实验室辽宁大连116024,沈阳建筑大学理学院,辽宁沈阳110168,辽宁大连116024
年：2006
期：01
出版单位：沈阳建筑大学学报(自然科学版)

摘要

目的基于误差传播算法的BPNN和基于自适应共振理论的ART神经网络,提出一种耦合神经网络的三级识别模型,以实现对结构损伤的自主识别.方法采用分步识别的思想,利用ART神经网络首先识别出有损伤的层,然后用遗传算法搜索最佳的BP神经网络结构来分别识别结构损伤的具体位置和损伤程度.结果通过对结构健康监测基准问题的计算表明,提出的耦合神经网络的识别模型能够自主识别结构损伤的发生,正确识别结构损伤的具体位置和损伤程度.结论基于误差传播算法的BPNN和基于自适应共振理论的ART神经网络组成的耦合神经网络识别模型具有自主识别结构损伤发生的能力,且识别速度快,能够正确识别结构损伤发生的具体位置和损伤程度,适宜于在线监测.
A three-level identification model is proposed in the damage identification of structures.The BPNN and ART neural networks,which are based on error propagation algorithm and self-adaptive resonance theory respectively,are coupled in this model to improve its performance.The damage of structures could be identified step by step.First ART neural networks are employed to identify the occurrence of the damage,and then the location and grade of the damage is confirmed by using the genetic algorithm to search for the optimal BP neural networks.It can be concluded from the numerical analyses of the benchmark structures that satisfactory results could be obtained by using this coupled neural networks method.The coupled method based on BPNN and ART could identify the occurrence,location and grade of the damage and it can efficiently monitor the status of on-line structures.

引文

[1]李国强,李杰.工程结构动力检测理论与应用[M].北京:科学出版社,2002.
    [2]沈清,汤霖.模式识别导论[M].长沙:国防科技大学出版社,1991.
    [3]Carpenter G A,Grossberg S.ART-2:Self-organi-zation of stable category recognition codes for analoginput pattern[J].Applied Optics,1987,26:4919-4930.
    [4]冯夏庭,张治强,杨成祥等.位移反分析的进化神经网络方法研究[J].岩石力学与工程学报,1999,18(5):497-502.
    [5]Johnson E A,Lam H F,Katafygiotis L S and Beck JL.A Benchmark Problem for Structural Health Moni-toring and Damage Detection[A].Proceedings of The14th ASCE Engineering Mechanics Conference[C].USA:Austin,2000,21-24.
    [6]黄石生,钱迎雪.基于ART人工神经网络的焊缝跟踪检测算法[J].机械工程学报,1994,30(2):93-98.
    [7]李宏男,孙鸿敏.基于小波分析和神经网络的框架结构损伤诊断方法[J].地震工程与工程振动,2003,23(5):141-148.
    [8]Maartje E J,Raijmakers Peter,Molenaar C M.ExactART:A complete implementation of an ART net-work[J].Neural Networks,1997,10(4):649-669.
    [9]LIANG Y C,FFENG D P,LIU G R,et al.Neural i-dentification of rock parameters using fuzzy adaptivelearning parameters[J].Computers&Structures,2003,81:2373-2382.
    [10]Nakamura M,Masri S F,Chassiakos A G and Caugh-ey T K.A method for non-parametric damage de-tection through the use of neural networks[J].Earth-quake Engineering&Structural Dynamics,1998,27(9):997-1010.
    [11]焦莉,李宏男.基于自适应共振理论的结构损伤识别[J].沈阳建筑大学学报:自然科学版,2005,21(5):486-490.