基于递归神经网络的减震结构模糊控制研究
|
摘要
根据神经-模糊控制理论和减震结构混合控制理论,利用基于弹塑性时程分析自动提取减震结构模糊控制规则的方法获取原始训练样本,能考虑不同地震动特性的影响.应用带偏差单元的递归神经网络(RN-NWBU)形成减震结构模糊控制规则的关系生成方法和推理合成算法,能够实现神经网络驱动的混合控制结构体系模糊推理,从而能实现神经-模糊控制器的设计,应用装设有神经-模糊控制器的系统能够在线控制结构地震反应.通过半主动调谐质量阻尼器(SATMD)与消能减震相结合的混合控制结构体系的数值仿真分析,可看出本文方法能有效地在线控制结构地震反应.本文对推动结构控制理论的发展具有重要意义.
According to the neural-fuzzy control theory and the hybrid control theory of seismic reduction structure,original training samples were acquired,using a method of generating fuzzy control rule automatically for aseismatic structure based on elastoplastic time-history analysis.Therefore the influence of different characteristics of earthquake ground motion can be considered.By means of recurrent neural network with bias unit(RNNWBU) to produce creating method of relation and synthetical algorithm of ratiocination of fuzzy control rules of seismic reduction structure.Fuzzy reasoning can be achieved driven by neural network of structure with hybrid control,accordingly the neural-fuzzy controller can be designed.The system equipped with neural-fuzzy controller can be used for online control of the seismic responses of structure.Numerical simulations were carried out for structure with hybrid control of semi-active tuned mass damper(SATMD) combined with energy dissipation system.Analysis results indicate that the method proposed is valid for online control of the seismic responses of structure.The work may promote the development of the theory of structural control.
According to the neural-fuzzy control theory and the hybrid control theory of seismic reduction structure,original training samples were acquired,using a method of generating fuzzy control rule automatically for aseismatic structure based on elastoplastic time-history analysis.Therefore the influence of different characteristics of earthquake ground motion can be considered.By means of recurrent neural network with bias unit(RNNWBU) to produce creating method of relation and synthetical algorithm of ratiocination of fuzzy control rules of seismic reduction structure.Fuzzy reasoning can be achieved driven by neural network of structure with hybrid control,accordingly the neural-fuzzy controller can be designed.The system equipped with neural-fuzzy controller can be used for online control of the seismic responses of structure.Numerical simulations were carried out for structure with hybrid control of semi-active tuned mass damper(SATMD) combined with energy dissipation system.Analysis results indicate that the method proposed is valid for online control of the seismic responses of structure.The work may promote the development of the theory of structural control.
引文
[1]黄崇福.模糊人工神经元网络在地震工程中的应用[J].地震工程与工程振动,1993,13(2):84-93.HUANG Chong-fu.Application of fuzzy neuron to earthquake engineering[J].Earthquake engineering and engi-neering vibration,1993,13(2):84-93.
[2]蒋式勤,陈海洪.模糊神经网络技术用于结构主动控制的研究[J].同济大学学报,1998,26(5):605-608.JIANG Shi-qin,CHEN Haihong.Si mulation study onfuzzy neutral network andits applicationto structural control[J].Journal of TongJi University,1998,26(5):605-608.
[3]陈维强,李成伟,李革臣.一种新型的神经网络模糊控制器[J].哈尔滨工业大学学报,1998,30(3):49-52.CHEN Wei-qiang,LI Cheng-wei,LI Ge-chen.A new fuzzy controller based on neural network[J].Journal ofHarBin Institute of Technology,1998,30(3):49-52.
[4]何波,吴广玉.基于模糊-神经融合的自适应模糊控制系统研究[J].哈尔滨工业大学学报,1999,31(4):1-4.HE Bo,WU Guang-yu.Aresearch on adaptive fuzzy control system based on fuzzy-neural fusion[J].Journal ofHarBin Institute of Technology,1999,31(4):1-4.
[5]周建中,赵鸿铁.SATMD与消能减震相结合的混合控制研究[J].世界地震工程,2003,19(1):136-140.ZHOUJian-zhong,ZHAO Hong-tie.Study on hybrid control of semi-active tuned mass damper combined with en-ergy dissipation system[J].World earthquake engineering,2003,19(1):136-140.
[6]周建中.基于模糊-神经元的抗震结构智能混合控制[D].西安:西安建筑科技大学,2003.ZHOUJian-zhong.Intelligent hybrid control of aseismatic structure based onfuzzy-neuron[D].Xi an:Xi an Univ.of Arch.&Tech.,2003.
[7]周建中,赵鸿铁,薛建阳.粘弹性阻尼结构的递归神经网络优化[J].西安建筑科技大学学报:自然科学版,2005,37(1):35-39.ZHOUJian-zhong,ZHAO Hong-tie,XUE Jian-yang.Opti mum of the viscoelastic structure based on recurrentneural network[J].J.Xi an Univ.of Arch.&Tech.(Natural Science Edition),2005,37(1):35-39.
[8]HECHT N R.Counterpropagation Networks[J].Applied Optics,1987,26(1):4979-4984.
[2]蒋式勤,陈海洪.模糊神经网络技术用于结构主动控制的研究[J].同济大学学报,1998,26(5):605-608.JIANG Shi-qin,CHEN Haihong.Si mulation study onfuzzy neutral network andits applicationto structural control[J].Journal of TongJi University,1998,26(5):605-608.
[3]陈维强,李成伟,李革臣.一种新型的神经网络模糊控制器[J].哈尔滨工业大学学报,1998,30(3):49-52.CHEN Wei-qiang,LI Cheng-wei,LI Ge-chen.A new fuzzy controller based on neural network[J].Journal ofHarBin Institute of Technology,1998,30(3):49-52.
[4]何波,吴广玉.基于模糊-神经融合的自适应模糊控制系统研究[J].哈尔滨工业大学学报,1999,31(4):1-4.HE Bo,WU Guang-yu.Aresearch on adaptive fuzzy control system based on fuzzy-neural fusion[J].Journal ofHarBin Institute of Technology,1999,31(4):1-4.
[5]周建中,赵鸿铁.SATMD与消能减震相结合的混合控制研究[J].世界地震工程,2003,19(1):136-140.ZHOUJian-zhong,ZHAO Hong-tie.Study on hybrid control of semi-active tuned mass damper combined with en-ergy dissipation system[J].World earthquake engineering,2003,19(1):136-140.
[6]周建中.基于模糊-神经元的抗震结构智能混合控制[D].西安:西安建筑科技大学,2003.ZHOUJian-zhong.Intelligent hybrid control of aseismatic structure based onfuzzy-neuron[D].Xi an:Xi an Univ.of Arch.&Tech.,2003.
[7]周建中,赵鸿铁,薛建阳.粘弹性阻尼结构的递归神经网络优化[J].西安建筑科技大学学报:自然科学版,2005,37(1):35-39.ZHOUJian-zhong,ZHAO Hong-tie,XUE Jian-yang.Opti mum of the viscoelastic structure based on recurrentneural network[J].J.Xi an Univ.of Arch.&Tech.(Natural Science Edition),2005,37(1):35-39.
[8]HECHT N R.Counterpropagation Networks[J].Applied Optics,1987,26(1):4979-4984.
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