基于自适应BP神经网络的桥梁结构荷载识别
|
摘要
在传统的BP神经网络中引入学习速率自适应调整算法,通过多次数值模拟计算确定学习速率和动量系数等网络关键参数的取值;分析了学习速率、动量系数等网络参数对网络收敛速度和输出精度的影响;探讨了训练样本容量与网络识别效果的关系.分别使用挠度、挠度曲率、应变和应变曲率作为输入参数对桁架桥梁荷载进行识别.结果显示以挠度曲率或应变曲率为输入参数的网络识别效果明显优于以挠度或应变为输入参数的网络,以应变为输入的网络识别效果优于挠度的情况;学习速率自适应调整算法有效避免了网络训练过程中误差曲线振荡现象的产生,提高了网络的学习效率,网络关键参数的最优取值改善了网络的收敛速度和输出精度.
The self-adaptive algorithm of learning rate for BP neural networks was implemented to identify loads exerted on bridge structures.Values of the network's key parameters such as initial learning rate and momentum factor were optimized by numerical simulation for several times.Training specimen capacity was varied to obtain different output precision,and the relationship between the specimen capacity and the identification effect was discussed.The displacement(or displacement curvature) and strain(or strain curvature) were used as input data of the BP networks respectively.The identification effect of network using displacement curvature or strain as input data; networks using strain(strain curvature) as input data performed better than networks using displacement(displacement curvature)as input data.The developed algorithm can avoid oscillation phenomenon during the training process and improve learning rate effectively. Optimum values of network's key parameters enhance error convergence velocity and output precision of the BP networks.
The self-adaptive algorithm of learning rate for BP neural networks was implemented to identify loads exerted on bridge structures.Values of the network's key parameters such as initial learning rate and momentum factor were optimized by numerical simulation for several times.Training specimen capacity was varied to obtain different output precision,and the relationship between the specimen capacity and the identification effect was discussed.The displacement(or displacement curvature) and strain(or strain curvature) were used as input data of the BP networks respectively.The identification effect of network using displacement curvature or strain as input data; networks using strain(strain curvature) as input data performed better than networks using displacement(displacement curvature)as input data.The developed algorithm can avoid oscillation phenomenon during the training process and improve learning rate effectively. Optimum values of network's key parameters enhance error convergence velocity and output precision of the BP networks.
引文
[1]高宝成,刘红霞,杨叔子.神经网络用于结构动荷载识别的研究[J].郑州工学院院报,1996,17(2):91-94.GAO Bao-cheng,LIU Hong-xia,YANG Shu-zi.An in-vestigation on the identification method of dynamic loadsof constructure[J].Journal of Zhengzhou Institute ofTechnology,1996,17(2):91-94.
[2]吴大宏,赵人达.基于神经网络的混凝土桥梁荷载识别方法研究[J].中国铁道科学,2002,23(1):25-28.WU Da-hong,ZHAO Ren-da.Research on method of loadidentification for concrete bridge based on neural network[J].China Railway Science,2002,23(1):25-28.
[3]CAO X,SUGIYAMA Y,MITSUI Y.Application of ar-tificial neural networks to load identification[J].Com-puters&Structures,1998,69:63-78.
[4]CHEN Q,CHAN Y W,WORDEN K.Structural faultdiagnosis and isolation using neural networks based onresponse-only data[J].Computers&Structures,2003,81:2165-2172.
[5]SZEWCZYK Z,HAJELA P.Neural network baseddamage dectection in structure[J].ASCE Journal ofEngineering mechanics and civil engineering,1994,8(2):163-178.
[6]飞思科技产品研发中心.MATLAB 6.5辅助神经网络分析与设计[M].北京:电子工业出版社,2003:65-66.
[7]PANDEY P C,BARAI S V.Mutilayer perceptron indamage detection of bridge structures[J].Computers&Structures,1995,54(4):597-608.
[8]朱宏平,张源.基于自适应神经网络的结构损伤检测[J].力学学报,2003,35(1):110-116.ZHU Hong-ping,ZHANG Yuan.Application of self-adap-tive BP neural networks to the detection of structure damages[J].Acta Mechanica Sinica,2003,35(1):110-116.
[9]温利明,黄奕辉.神经网络用于结构损伤识别的几个关键问题研究[J].工业建筑,2002,32(8):39-41.WEN Li-ming,HUANG Yi-hui.Study of several keyproblems of neural networks using for damage detection[J].Industrial Construction,2002,32(8):39-41.
[10]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.
[11]瞿伟廉,黄东梅.高耸塔架结构节点损伤基于神经网络的两步诊断法[J].地震工程与工程震动,2003,23(2):143-149.QU Wei-lian,HUANG Dong-mei.Two-step detection ap-proach to joint damage in high-rise tower structures by u-sing artificial neural network[J].Earthquake Engineeringand Engineering Vibration,2003,23(2):143-149.下期论文摘要预登
[2]吴大宏,赵人达.基于神经网络的混凝土桥梁荷载识别方法研究[J].中国铁道科学,2002,23(1):25-28.WU Da-hong,ZHAO Ren-da.Research on method of loadidentification for concrete bridge based on neural network[J].China Railway Science,2002,23(1):25-28.
[3]CAO X,SUGIYAMA Y,MITSUI Y.Application of ar-tificial neural networks to load identification[J].Com-puters&Structures,1998,69:63-78.
[4]CHEN Q,CHAN Y W,WORDEN K.Structural faultdiagnosis and isolation using neural networks based onresponse-only data[J].Computers&Structures,2003,81:2165-2172.
[5]SZEWCZYK Z,HAJELA P.Neural network baseddamage dectection in structure[J].ASCE Journal ofEngineering mechanics and civil engineering,1994,8(2):163-178.
[6]飞思科技产品研发中心.MATLAB 6.5辅助神经网络分析与设计[M].北京:电子工业出版社,2003:65-66.
[7]PANDEY P C,BARAI S V.Mutilayer perceptron indamage detection of bridge structures[J].Computers&Structures,1995,54(4):597-608.
[8]朱宏平,张源.基于自适应神经网络的结构损伤检测[J].力学学报,2003,35(1):110-116.ZHU Hong-ping,ZHANG Yuan.Application of self-adap-tive BP neural networks to the detection of structure damages[J].Acta Mechanica Sinica,2003,35(1):110-116.
[9]温利明,黄奕辉.神经网络用于结构损伤识别的几个关键问题研究[J].工业建筑,2002,32(8):39-41.WEN Li-ming,HUANG Yi-hui.Study of several keyproblems of neural networks using for damage detection[J].Industrial Construction,2002,32(8):39-41.
[10]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.
[11]瞿伟廉,黄东梅.高耸塔架结构节点损伤基于神经网络的两步诊断法[J].地震工程与工程震动,2003,23(2):143-149.QU Wei-lian,HUANG Dong-mei.Two-step detection ap-proach to joint damage in high-rise tower structures by u-sing artificial neural network[J].Earthquake Engineeringand Engineering Vibration,2003,23(2):143-149.下期论文摘要预登
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心