鞍形屋盖极值风压系数模糊聚类分区研究

Study on wind pressure zones of saddle roof based on Fuzzy C-Means

It is known that destruction mainly begins with the edge of the saddle roof and now there is no basis for saddle roof wind pressure partition. A saddle roof has been chosen as the research object and been divided into multiple regions by the Fuzzy c-means clustering method according to the wind vulnerability degree. In the process of partitioning, the Fuzzy C-means clustering is improved in the area of initial variable including dependency. Improvement measures include limit maximum of clustering number in advance and use the validity index to determine the optimal clustering number. Based on these measures, the most unfavorable mean and peak wind pressure coefficient distribution of saddle roof has been given and the wind pressure coefficient of each region has been calculated.