神经网络对结构地震反应的预测及试验研究

NEURAL NETWORK PREDICTION FOR SEISMIC RESPONSE OF STRUCTURE AND EXPERIMENTAL STUDY

本文基于神经网络对非线性系统具有辨识和预测功能 ,并结合具有二阶收敛效应的Levenberg_Marquardt算法 ,采用一多层前馈网络对建筑结构的地震反应进行了预测。首先以一三层钢筋混凝土结构的振动台试验数据对网络结构进行批量训练 ,然后用未曾训练的地震波数据对结构进行地震反应预测 ,并与试验数据进行对比 ,分析结果表明 :Lev enberg_Marquardt算法能快速收敛 ,神经网络能准确地预测结构的地震反应。

In the paper,based on identification and prediction ability of neural network for nonlinear systems,and combined Levenberg-Marquardt algorithm which has second-order-convergence effect,a multi-layer forward network is adopted to predict the seismic response of the structure.First of all,the network is trained in batch by the vibration table test data of three-floor reinforced concrete structure,then the seismic response of the structure is predicted with the raw earthquake data,and the predict response is compared with the experimenal one.It can be shown from the analysis that Levenberg-Marquardt algorithm is of very good convergence rate,and the neural network can predict the dynamic response of the structure well enough.