基于灰关联与人工神经网络综合评价模型的多层砖房震害预测

Forecasting seismic damage of multistory masonry building based on grey relation and artificial neural network model

结合几次大地震中多层砖房的实际震害资料,基于灰关联识别方法,解析了各影响因子对多层砖房抗震性能的影响程度。以反映结构抗震性能的各类物理参数作为输入数据,以给定地震动峰值加速度下建筑物破坏状态的概率作为输出数据,采用8-6-5层结构,建立了基于BP人工神经网络的非线性模型,并对震害样本进行了训练。结果表明:利用灰关联分析,可得出各因子对多层砖房抗震性能影响程度的大小排序,有利于实际的工程抗震设计;基于BP人工神经网络模型的多层砖房的震害预测结果与震害实例的实际情况比较吻合,其思路和方法可推广于其他不同类型的建筑结构的震害预测。

Based on the seismic damage samples of multistory masonry buildings in several strong earthquakes, the factors that affect the performance of the earthquake-resistance are analyzed. And these seismic damage samples are trained by BP artificial neural network model which adopts 8 -6 -5 framework. The research shows that： to some extent effect of the factors on the earthquake-resistance performance of the buildings can be obtained by grey relation analysis, and it can be applied the earthquake-resistance design of actual engineering; the prediced results are agreed with to the actual seismic damage of multistory masonry buildings by the grey relation and BP neural network model. The analytic method and process discussed in this paper can also be applied to the seismic damage prediction of the other structures with different forms.