K-means clustering based model reduction method for seismic damage analysis of urban buildings
Refined seismic damage simulation technology for urban buildings based on nonlinear time history analy-sis provides necessary basis and reference for decision-making for pre-earthquake disaster prevention and post-earth-quake emergency rescue.In order to reduce calculation amount and calculation time of seismic damage simulation,a model reduction method based on clustering algorithm for urban buildings is proposed in this paper.This method firstly clusters urban buildings based on their attribute vectors using K-means clustering algorithm,classifying the similar buildings into the same group,then selects a representative building from each group to form a reduction model.In this way,this method reduces the calculation amount of seismic damage simulation for urban buildings by reducing the number of building structures needed to be analyzed.In this study,traditional K-means algorithm is improved to automatically adjust the number of clusters based on the preset upper limit of the difference among buildings in the same group.The structural seismic damage index is used as the attribute vector of the building structure for clustering for the first time,and reduction model of urban buildings based on damage index is compared with that based on structural model parameter,and the results show that the damage index-based grouping is significantly better than the grouping based on model parameters,using model reduction method can significantly reduce calculation work in the premise of ensuring sufficient accuracy.
urban buildingsK-means algorithmmodel reductionstructural model parameterseismic damage index
万方数据
维普
