Rapid Prediction Model And System For 7 Degrees Of RC Structure Earthquake Damage Based On LM-BP Neural Network

In recent years,China has been increasing its investment in disaster prevention and mitigation engineering.Based on various research results and theoretical support,the Code for Seismic Design of Buildings in China has been constantly innovating,and the safety of new buildings has been greatly improved.However,the old buildings built according to the old version of the code have a higher rate of severe damage in the earthquake,and the proportion of casualties and economic losses is relatively heavy.Therefore,it is of great significance to implement reinforcement and reconstruction measures for such buildings.How to accurately identify such buildings with poor seismic capacity is the first task.At present,the methods of earthquake damage prediction for single building in our country can be divided into two categories: empirical summary method and direct statistics method.The methods of earthquake damage prediction for building group can be divided into empirical statistics method,theoretical calculation method,semiempirical and semi-theoretical method and other methods.Most of these methods need the support of sufficient basic data,which is time-consuming and laborious,and can not be applied to the general survey of earthquake damage prediction of buildings in large areas.For this reason,this paper takes reinforced concrete frame structure and high-rise frame-shear wall structure in 7 degree fortification area as the research object,and uses a small number of earthquake damage impact factors to quickly and accurately predict earthquake damage of buildings.This method is fast and effective.It can quickly conduct preliminary earthquake damage survey of large area buildings.It can ensure accuracy and save manpower and material resources greatly.The main work of this paper is as follows:(1)Taking 40 frame structures as the research object,12 key data such as story number,story height,building height and column area ratio,which are easily accessible and highly correlated,are taken as the impact factors of earthquake damage.Taking full advantage of the good visibility of MATLAB and the remarkable advantages of LM algorithm,a BP neural network earthquake damage prediction model based on LM algorithm is trained.The data of single simulation are compared by using multi-group data tests.The accuracy and stability of the model are verified by randomly selecting data samples for many times.At the same time,it also proves the feasibility of using a small number of earthquake damage impact factors to predict earthquake damage,which lays a theoretical foundation for the expansion of the following chapters.(2)Expanding the data samples of the frame structure database and classifying them according to residential and non-residential categories.Based on the purpose of fast prediction and on the premise of ensuring the accuracy of the model,the "wall area ratio" is discarded after comparative study.Using the same method,earthquake damage prediction models for residential and non-residential buildings are established respectively by using 11 earthquake damage influence factors.The data of error variation curve and regression curve of training data in the process of modeling are analyzed.The actual simulation and result comparison of relevant examples prove that the two models can make accurate judgment for earthquake damage prediction of frame structures.(3)Analyzing the advantages and disadvantages of various structural forms of high-rise reinforced concrete buildings and their application status in China,taking frame-shear wall structure as the research object,and selecting 11 corresponding data as the impact factors of earthquake damage to carry out the modeling of the neural network earthquake damage prediction model.The modeling process and results of the model are analyzed,and the results of two examples show that the model can achieve the desired results.(4)Using the GUI function of MATLAB software to create a graphical user interface,the earthquake damage prediction models of three types of structures,namely residential frame structure,non-residential frame structure and high-rise frame-shear wall structure,are embedded in a visual program.The program can interact with users in real time,input the data of the impact factors of earthquake damage to the building in the input part,select the structure type and calculate by the program,then the simulated earthquake damage index and the simulated earthquake damage degree of the building under different earthquake intensity can be obtained.