人工神经网络和GIS在城市防震减灾中的应用

Application of ANN and GIS to Urban Seismic Prevention and Disaster Mitigation

基于组件式GIS(ComponentObjectModel,COMGIS)技术,讨论了BP神经网络和GIS耦合模型在城市防震减灾中的应用。研究表明:商用程序SHAKE91在VB菜单下可直接调用,实现了任意设定地震下地震动影响场计算的模块化;BP神经网络应用于城市建筑物震害预测中,能达到较理想的效果,其计算模型在系统菜单下可直接调用;GIS的空间分析功能可使震害预测结果与建筑物信息进行空间匹配,实现在假定地震的情况下,快速、有效地得出城市建筑物潜在地震灾害的空间分布图,有利于城市防震减灾工作的开展。

COMGIS (Component Object Model Geography Information System) technique is used to develop the 'Seismic Damage Evaluation System for Building' combined with BP ANN (Artificial Neural Network). The research indicates that:1 SHAKE91 can be called from the system menu and the earthquake influence field will be simulated. Therefore, the obtained peak ground surface acceleration can be an input factor of the BP ANN model. 2. The relatively perfect results can be obtained when the BP ANN model is utilized to predict the seismic damage of building. Also, this model is integrated into the COMGIS system by importing the ANN toolbox from Matlab. 3. By GIS technique, the BP ANN evaluated results can be perfectly matched with the building information spatially. Thus the critical seismic damage area can also be identified.