Structural seismic demand information entropy importance analysis based on kernel density estimation

8 input random variables including the selected 7 ground motion records and an El Centro one were sampled, respectively. The dynamic nonlinear time history analysis was conducted for a multi-floor building with the finite element software OpenSees. Its base shear, top point displacement, the maximum floor acceleration and the maximum interlayer displacement angle were obtained as the structure's seismic demands. The kernel density estimation was used to do the structural seismic demand information entropy importance analysis, and the results were compared with those using the variance importance analysis method. The results showed that the used sampling method can get better results when the sample number is hundreds, and this number is far less than thousands required by the existing sampling method, so it is an accurate and efficient sampling method; the information entropy importance sequencing of each input random variable based on the kernel density estimation and that based on the variance importance analysis are basically consistent, so the importance measure analysis method based on information entropy is a good method of importance analysis. © 2019, Editorial Office of Journal of Vibration and Shock. All right reserved.