Prediction Of Rural Residence Seismic Damage In Huairou Area Based On BP Neural Network

Earthquake is a natural disaster with great harmfulness.Earthquake will cause huge casualties and property losses.According to the earthquake damage experience of every major earthquake in China,the earthquake damage is usually more serious than that of urban buildings due to many reasons,such as design,construction,and so on.Since 2008,one of the key tasks of the country has been to improve the earthquake resistance level of rural residence.Building damage prediction is an important part of urban earthquake prevention and disaster reduction planning at present,which can provide basic data support for earthquake disaster prevention,seismic reinforcement and post-earthquake emergency.On the basis of widely collecting the research results of earthquake damage prediction of buildings at home and abroad,in this paper,seismic hazard analysis method is used to analyze and study the seismic damage of farmhouses under different earthquake intensities in Huairou District of Beijing.On this basis,using BP neural network method,using seismic survey data of buildings in Shijiazhuang,Hengshui and Changping as training data set,the earthquake damage prediction model of buildings based on BP neural network is established.The model is used to predict the earthquake damage in Huairou area.On the basis of this model,an improved BP neural network earthquake disaster prediction model is developed by using the improved algorithm,whichcan be used to meet the requirements of practical functions.It lays a foundation for the subsequent research and software development.The main work and related conclusions are as follows:(1)The relevant achievements and data of earthquake disaster prediction of buildings at home and abroad are widely collected,and detailed research,classification,screening and so on are carried out.In this paper,we collected the earthquake data of rural housing in Huai District,and took the basic data as the research object.At the same time,the building survey data of Shijiazhuang,Hengshui and Changping districts are collected in order to prepare for the follow-up study.(2)Through the analysis of the survey data of the current situation of earthquake resistance in Huairou area,It can be found that the overall seismic level of rural housing in Huairou is not high due to the selection of structural forms,the age of construction,the adoption and construction of relevant technical standards.The ability to resist earthquakes is inadequate.(3)By using the method of seismic hazard factors,the earthquake damage of Huairou area was analyzed and studied,and the earthquake damage of the Farmhouses in Huairou area under the action of different earthquake intensity was obtained.The percentage of different damage levels of rural residence in Huairou under different conditions,and use the Google Earth tool to draw pictures.Compared with the national fortification intensity list,it is known that most of the villages and towns in Huairou area are 8 degrees of fortification.From the results of the earthquake damage prediction analysis,the fortification ability can't meet the requirements.(4)According to the neural network method,a BP neural network model suitable for earthquake damage prediction of buildings is established.Firstly,the building survey data of Shijiazhuang and Hengshui are used as training set to train and correct BP neural network model.Then,the BP neural network model was used to analyze the earthquake damage prediction of the rural residence in huairou area(calculated by the seismic hazard factor method),and the error rate is calculated.From the analysis results,we can find that the neural network model trained by City training data set has larger prediction error.(5)According to the research results in(4),the data of rural residence in changping district were used as training data to train and correct BP neural network model.And the BP neural network model was used to analyze the earthquake damage prediction of the rural residence huairou area.From the analysis results,it can be found that the neural network model established by the rural residence data set in changping district has a small error in the prediction results and can meet the practical needs.(6)According to the research results in(4)and(5),it is found that in the study of earthquake damage prediction of building by using BP neural network,the training data set must reach a certain number in order to ensure the reliability of the BP neural network model;Secondly,the training data set with comparability and similarity should be selected to reduce the error and improve the accuracy.(7)In order to improve the calculation efficiency and enhance the ease of use of BP neural network model,the L-M algorithm is used to optimize the BP neural network model,on this basis,a prediction model based on BP neural network is developed,which has a faster calculation speed,less computation time and more practicability.