| Typhoon disasters are frequent in the coastal areas of Beibu Gulf,and a large number of buildings are destroyed every year.In order to grasp the building vulnerability in time,this paper carried out a rapid assessment of the physical vulnerability of buildings in Longmen Port Town,providing scientific decision-making basis and technical support for disaster prevention and reduction and economic construction in the Beibu Gulf region in the future.Based on the research status and results of PTVA(Papathoma Vulnerability Assessment)model at home and abroad,this paper comprehensively considers the influence of typhoon downwind and water on building structure.An index system and evaluation model of building vulnerability under typhoon are proposed for longmen Port town.In this paper,geographic information system(GIS)technology,modified PTVA model,RESNET-50(Residual Network-50)convolutional neural Network,analytic hierarchy process and other methods were used to carry out case study,application and verification on the geo-spatial data of building disaster victims in Longmen Gang Town,Qinzhou City.The main research contents of this paper are as follows:(1)The deep learning model based on RESNET-50 was used to determine the types of buildings in the photos,and the buildings could be divided into three categories according to building materials: reinforced concrete structure,brick and concrete structure,and shed.The deep learning model can extract the structural attributes of corresponding buildings,reduce the workload of PTVA model building attribute collection,and also collect a large number of image data for the classification of building vulnerability in the future.This deep learning model does not take other factors into account and identifies the vulnerability level of buildings in a quick and emergency manner only from the structural vulnerability of the building itself.(2)This paper comprehensively considers the impact of typhoon,stroke and water on building structure,analyzes the regional architectural characteristics of Longmen Port Town,establishes the vulnerability evaluation index of single building,and revises PTVA model;The weight of each vulnerability index of the modified PTVA model was determined by using analytic hierarchy process(AHP)according to experts’ opinions,and then the vulnerability of buildings was calculated according to the modified MODEL,and the relative vulnerability assessment model of disaster bearing bodies of buildings in the coastal area of Beibu Gulf was established.(3)A geospatial interpolation analysis of 200 m×200 m grid was conducted on the relative vulnerability of single buildings on Longmen Island in Longmen Port Town of Rammasun typhoon.This paper studies the geographical spatial distribution characteristics of vulnerability level of building bearing body under Rammasun typhoon.The study reached the following conclusions:(1)in this paper,the field acquisition of 926 Zhang Jianzhu images of 50 were taken as training samples,obtained the depth of the learning training model building materials structure type discriminant accuracy can reach 91.7%,deep learning model can be used in the field to collect a large amount of image data for building types,rapid emergency identification construction level of vulnerability.(2)The PTVA model was modified to assess the vulnerability of 2168 buildings on Longmen Island,longmen Port Town,and the evaluation results were compared with the field survey results of buildings under two tropical cyclones "Rammasun" and "Whale".It was found that the relative vulnerability assessment results calculated by the model were basically consistent with the damage situation of buildings in the field survey.Under the typhoon disaster,the higher the vulnerability of the building,the more serious the damage of the building.This indicates that the modified PTVA model is more accurate in assessing the relative vulnerability of single buildings in the coastal area of Beibu Gulf.The model is very effective in evaluating the relative vulnerability of small scale single building and can provide reference for disaster prevention and mitigation in Longmen Port Town.(3)Based on the results of the relative vulnerability of individual buildings,interpolation analysis was conducted to obtain the interpolation results of 200 m×200m geospatial grid,and the proportion of each vulnerable building area and geospatial distribution characteristics were obtained.The building area of general fragile grade accounts for the highest proportion,about 60%;The next is about 30% in low vulnerability areas.The lowest is high and extremely high vulnerability region,about10%;The geographical and spatial distribution characteristics of building vulnerability are "low inside and high outside".The vulnerability evaluation system of single building under typhoon constructed in this paper comprehensively considers the vulnerability of the building itself and the impact of environmental factors on the building.The building vulnerability level calculated based on the modified PTVA model can reflect that buildings with high vulnerability are more likely to be damaged in typhoon disasters.This paper carried out a refined building vulnerability evaluation index system construction and building vulnerability evaluation research in Longmen Port Town,Qinzhou City,which not only made up for the shortcomings of the research on building vulnerability and typical disaster-bearing body,but also provided some guidance for the disaster prevention and mitigation work of Longmen Port town. |
