The Deep Learning-driven Identification And Repair Design Of Protection Cracks In Historical Buildings

Historic buildings are the carriers of spiritual civilization of mankind.As time changes,the surface layer of historic buildings has been damaged to different degrees,which seriously affects their appearance integrity and structural safety.Therefore,their protection problems are becoming more and more prominent.The Historic buildings in China are diversified and widely distributed.The lack of efficiency in damage detection,incomplete disease analysis,and unreasonable design of repair measures are the prominent problems in the conservation of historic buildings at present,and the intelligent detection of surface damage and quantitative assessment of damage degree are the main technical bottlenecks in this field.Studies have shown that the traditional surface damage detection methods,which are mainly based on manual inspection and sensor detection,are inefficient and not as objective as quantitative analysis,and have serious defects in practice.Therefore,based on deep learning theory and the advanced feature of cracks learned,this paper proposes a rapid detection method that can realize automatic recognition and feature extraction of cracks on the surface of historic buildings,constructs a model for assessing cracks on the surface of historic buildings with crack occurrence element and scale elements as calculation factors,and proposes repair strategies and building reuse design solutions for crack damage.The research work is as follows: First,the theories such as the basic deep learning theory,the theory of building crack damage and crack management technology were introduced.Then,the feasibility of the research was demonstrated from four aspects:technology,social value,recycling and equipment,and the basis for the development of strategies for surface crack detection,damage assessment and repair and reuse of historical buildings was clarified.Further,Convolutional Neural Network(CNN)was used to construct a crack identification and feature extraction model,and the model was tested on a data set with a real environment as the background to acquire crack information.Finally,a damage assessment model was established by using the nature,position,average width and number of branches of cracks as factors to evaluate the damage of cracks in historical buildings.According to the degree of damage,crack management scheme was proposed and the best repair effect was achieved,hoping to provide a new reference for the conservation of historic buildings in China.