Application of BP neural network for the design of isolated structureAuthor(s): Yan Yan, Shan Gu, Zhao Yingna, Ding Qingpeng
The effects of local construction of isolation layer on the overall architecture are described in this thesis. Nowadays, the earthquake is a great problem in the world, for it can not be predicted. The only thing we can do is to enhance the ability to resist the earthquake. So, the first thing of all is to improve the seismic performance of buildings. Based onBP neural network, a preliminary design system for isolation is established with the seismic fortification type, seismic fortification intensity, site classification, seismic grouping, the depth-width ratio, the length-width ratio, ground floor stiffness, mass and area as the main influencing factors, and the largest layer shear force ratio of the structure after isolation and the largest displacement as output results.After the network training with 25 training samples, the network test is done by 15 test samples. By the comparison of the test results with the actual design results, it is acknowledged that the average accuracy rate of neural network reaches 96%, which shows the analysis on the damping effect that the system of preliminary isolation design based on BP neural network has on isolated structure has high efficiency and accuracy.