An accurate estimation of the applied load pattern is an essential component in each pushover procedure. Recently, a number of adaptive pushover methods have been proposed in which the effects of the higher modes as w...An accurate estimation of the applied load pattern is an essential component in each pushover procedure. Recently, a number of adaptive pushover methods have been proposed in which the effects of the higher modes as well as the progressive changes in the dynamic characteristics of structures are taken into account to compute the applied load pattern. The basic shortcoming of these advanced pushover methods is related to employing the quadratic modal combination rule, whereby the sign reversals of the modal load vectors are suppressed. In this study, an improved displacement-based adaptive pushover method is developed in which the applied load pattern is computed using the factor modal combination rule(FMC). In the proposed procedure, multiple load patterns, depending on the number of the modes considered, are determined in order to take into account the sign reversals of different modal load vectors. The accuracy of the proposed method is verifi ed for seven moment resisting frame buildings of 3, 9 and 20 stories with regularity or vertically geometric and mass irregularities subjected to 60 earthquake ground motion records. The results showed that the proposed methodology is capable of reproducing the peak dynamic responses with very good accuracy.展开更多
小世界神经网络具有较快的收敛速度和优越的容错性,近年来得到广泛关注.然而,在网络构造过程中,随机重连可能造成重要信息丢失,进而导致网络精度下降.针对该问题,基于Watts-Strogatz(WS)型小世界神经网络,提出了一种基于突触巩固机制的...小世界神经网络具有较快的收敛速度和优越的容错性,近年来得到广泛关注.然而,在网络构造过程中,随机重连可能造成重要信息丢失,进而导致网络精度下降.针对该问题,基于Watts-Strogatz(WS)型小世界神经网络,提出了一种基于突触巩固机制的前馈小世界神经网络(Feedforward small-world neural network based on synaptic consolidation,FSWNN-SC).首先,使用网络正则化方法对规则前馈神经网络进行预训练,基于突触巩固机制,断开网络不重要的权值连接,保留重要的连接权值;其次,设计重连规则构造小世界神经网络,在保证网络小世界属性的同时实现网络稀疏化,并使用梯度下降算法训练网络;最后,通过4个UCI基准数据集和2个真实数据集进行模型性能测试,并使用Wilcoxon符号秩检验对对比模型进行显著性差异检验.实验结果表明:所提出的FSWNN-SC模型在获得紧凑的网络结构的同时,其精度显著优于规则前馈神经网络及其他WS型小世界神经网络.展开更多
The syndromes have theirself spontaneous process from production,development to end,and the compositing symptoms and signs of syndrome is from no to be,from light to heavy,from simple to multiple.The different essenti...The syndromes have theirself spontaneous process from production,development to end,and the compositing symptoms and signs of syndrome is from no to be,from light to heavy,from simple to multiple.The different essential forms of the syndrome are composed with the different symptoms and signs.And the different essential forms of the syndrome and their sub-assembly of compositing symptoms and signs are the reflection of objective rule existing within the internal evolving changes of spontaneous process of syndromes.展开更多
文摘An accurate estimation of the applied load pattern is an essential component in each pushover procedure. Recently, a number of adaptive pushover methods have been proposed in which the effects of the higher modes as well as the progressive changes in the dynamic characteristics of structures are taken into account to compute the applied load pattern. The basic shortcoming of these advanced pushover methods is related to employing the quadratic modal combination rule, whereby the sign reversals of the modal load vectors are suppressed. In this study, an improved displacement-based adaptive pushover method is developed in which the applied load pattern is computed using the factor modal combination rule(FMC). In the proposed procedure, multiple load patterns, depending on the number of the modes considered, are determined in order to take into account the sign reversals of different modal load vectors. The accuracy of the proposed method is verifi ed for seven moment resisting frame buildings of 3, 9 and 20 stories with regularity or vertically geometric and mass irregularities subjected to 60 earthquake ground motion records. The results showed that the proposed methodology is capable of reproducing the peak dynamic responses with very good accuracy.
文摘小世界神经网络具有较快的收敛速度和优越的容错性,近年来得到广泛关注.然而,在网络构造过程中,随机重连可能造成重要信息丢失,进而导致网络精度下降.针对该问题,基于Watts-Strogatz(WS)型小世界神经网络,提出了一种基于突触巩固机制的前馈小世界神经网络(Feedforward small-world neural network based on synaptic consolidation,FSWNN-SC).首先,使用网络正则化方法对规则前馈神经网络进行预训练,基于突触巩固机制,断开网络不重要的权值连接,保留重要的连接权值;其次,设计重连规则构造小世界神经网络,在保证网络小世界属性的同时实现网络稀疏化,并使用梯度下降算法训练网络;最后,通过4个UCI基准数据集和2个真实数据集进行模型性能测试,并使用Wilcoxon符号秩检验对对比模型进行显著性差异检验.实验结果表明:所提出的FSWNN-SC模型在获得紧凑的网络结构的同时,其精度显著优于规则前馈神经网络及其他WS型小世界神经网络.
文摘The syndromes have theirself spontaneous process from production,development to end,and the compositing symptoms and signs of syndrome is from no to be,from light to heavy,from simple to multiple.The different essential forms of the syndrome are composed with the different symptoms and signs.And the different essential forms of the syndrome and their sub-assembly of compositing symptoms and signs are the reflection of objective rule existing within the internal evolving changes of spontaneous process of syndromes.
