粘弹性阻尼结构的递归神经网络优化

Optimum of the viscoelastic structure based on recurrent neural network

根据粘弹性阻尼结构减震理论及神经网络控制理论,提出应用带有偏差单元的递归神经网络(RNNWBU)进行粘弹性阻尼结构优化的新方法.该法通过对粘弹性阻尼结构进行神经网络优化设置,从而可以考虑不同地震动特性的影响.通过数值仿真分析,可看出该方法是对粘弹性阻尼结构的传统优化方法的进一步发展,因而具有更大的优越性,对推动结构控制理论的发展具有重要意义.

Based on the seismic reduction theory of structure with viscoelastic damper and the neural network control theory, this paper put forward a new optimum method of structure with viscoelastic damper applying recurrent neural network with bias unit (RNNWBU). By optimizing installation of structure with viscoelastic damper applying neural network technique, and therefore the influence of different characters of earthquake ground motion can be considered. The numerical simulation analysis indicates that the method of this paper has made further development on the traditional optimum method of structure with viscoelastic damper and therefore has some more superiority. This paper played an important role in promoting the development of the theory of structural control.