基于BP神经网络模型的多层砖房震害预测方法

Seismic damage prediction of multistory masonry buildings based on BP neural network model

针对传统的基于地震烈度的建筑物震害预测方法的不足,本文以地震动峰值加速度作为建筑物震害预测的地震动指标,结合几次大地震中多层砖房的震害实例,提出了一种基于BP神经网络模型的建筑物震害预测方法,模型的输入为反映结构抗震性能的各类物理参数,输出为给定地震动峰值加速度下建筑物破坏状态的概率。研究表明:基于BP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,本文的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。

The authors analyse the deficiencies of the traditional methods for predicting the seismic damage to multistory masonry buildings based on earthquake intensity and adopt the peak acceleration value as a ground motion index to predict the seismic damage to buildings, so a new predicting model based on BP neural network model is presented. Several seismic damage samples of multistory masonry buildings are induded. In this model, the input data are the structural physical parameters which can express the performance of the earthquake-resistance and the output results are the probabilities of the different failure status under different peak accelerations. The research shows that the prediction results are similar to the actual seismic damage to muhistory masonry buildings by the BP neural network model and the analytic method and process discussed in this paper can also be applied to the seismic damage prediction of other structures with different forms.