Forecasting Seismic Damage In The Frame Structure Of Bottom By FNN

Urbanization construction in China is progressing steadily; however, once earthquake disaster occurs, it may have significant impact on urban architecture. The Chinese urban construction market is very brisk and quite a lot of engineering projects are masonry building with bottom framework. As the building of such structure can be flexibly arranged with lower construction difficulty; moreover, the project cost is not high and contractors prefer to adopt it. The construction of bottom framework is typical“upper rigid lower flexible” building. In the seismic performance view, the material used in the upper part and lower part of such building has clear difference; and rigidity and deformation among layers are also distinct obviously. In other words, the integrate force conditions are complex and when the earthquake happens, the seismic load has some uncertainties, which increase the difficulty for seismic precaution and reduction of such building; so that, the major factor analysis for such earthquake damage and the result prediction are needed. By combining with the current earthquake data, it is extremely essential and meaningful to do damage prediction through secondary development; in this way, disaster preparedness and emergency plan can be well done prior to earthquake occurs.There are some merits for BP network mode in this study, which is built based on fuzzy theory. It can clearly describe the lessons learned from the earthquakes, analyze the variable features of earthquake damage under multiple factor joint actions, estimate damage and optimize conventional neural network against its shortcomings, which evidently improve its performance and make certain contribution for earthquake disaster mitigation.In this study, damage prediction was completed by employing neural network algorithm. Firstly, the analysis mode, which combined BP net with fuzzy theory, was presented; the design of BP network mode was designed, including confirming the predictive factor and membership degree; and then fuzzy BP model was established,neural network algorithm and its improvement was made; besides, conventional neural network was optimized and so on. After finishing the above steps, the optimal BP neutral network was built on the basis of genetic algorithm, which mainly discussed steps of damage prediction for bottom framework building, the selection and quantification of earthquake damage factor and the division of damage level. Finally,FNN neutral network was created. The case test method was adopted in this study and the test was proceeded by collecting building samples. Except that, the influential factors for building damage were illustrated, which the impacts of each factor on building damage were explored respectively, including the effect of aspect ratio, number of layers, horizontal spanning, longitudinal spanning, seismic intensity, and number of seismic resistant cross walls. The leading factor of building damage was confirmed and building damage estimation based on optimal BP neutral network was introduced. Theresult showed that the effect of neutral network estimation is better.