Automatic Identification And Health Condition Assessment Of Timber Frame Ancient Building Elements

Chinese culture has a long history,of which timber-framed ancient buildings are an important part.This long-term stability depends on a number of factors,including building materials,workmanship,structure and maintenance.Currently,the state attaches great importance to the safety of existing buildings;Shan Jixiang also stated in his article ’Reflections on the establishment of a research-based conservation mechanism for ancient buildings in the Forbidden City’ that assessment work is disconnected from practical work and that special evaluations are generally inadequate.Therefore,assessing the safety of the wooden structures of ancient buildings and proposing reasonable repair solutions is an issue that needs to be properly addressed.The point cloud data generated using 3D laser scanning technology for the classification of wooden structural elements of ancient buildings and the assessment of safety provides comprehensive 3D data to support this paper focuses on how to achieve automatic 3D point cloud segmentation and health diagnosis of wooden elements in ancient buildings,including the efficient modelling of complete finite element models of the current state of ancient building elements.This study follows the principle of sustainable development of heritage restoration and aims to provide theoretical and technical support for the restoration and strengthening of the Forbidden City buildings,with the following main research content:(1)For 3D point cloud classification tasks targeting large scenes like ancient buildings,Point Net networks and Point Net++ networks in deep learning can be competent.In this paper,Point Net++ is improved to enhance the classification of column,beam,and square point clouds in ancient buildings.In addition to the data enhancement aspect,this paper also improves the classification effect by updating the Adam optimizer of Pointnet++ with the cross-entropy loss function,using Adam W with the cross-entropy loss function using smoothing with label(Cross Entropy with label smoothing),and forgoes adding depth to the network to address the "overfitting and poor generalization problems caused by the small amount of data" and to improve the accuracy and efficiency of the neural network for segmenting wooden components.Compared with the original Point Net++ network,the improved Point Net++ algorithm in this paper improved the Io U of wooden columns and beams by 39.3% and 12.6% respectively.(2)In this paper,after collating and analysing the historical and archaeological data of the building,referring to the "Building Method","Engineering Practice" and "Three Hundred Years of Taihe Hall",a complete finite element model of the current state of the ancient building components is generated for the 3D digitised point cloud model of the ancient building through a point cloud component segmentation algorithm and a standard component deformation algorithm guided by the point cloud.Using the finite element method,combined with computer numerical simulation and analysis methods to quantify the damage of the components,a static analysis is carried out on the simplified solid as-is model of the reconstructed Taihe Hall.The simplified solid model of the Hall of Peace consists of three components from top to bottom:the upper beam frame,the arch,and the beam frame and column network.The weight of the roof is converted into a concentrated load,and the concentrated mass is applied to the top of the upper beam frame and the golden pillar pickets,while the weight of the lower roof is converted and applied to the ground floor pickets;the connection of the beam,the square and the pillar is simplified to a rigid connection;the bottom of the pillar is connected by a hinged connection.(3)The quantified results of the damage condition of the components obtained from the numerical computer simulation analysis are used to determine the most at-risk components and,in conjunction with the cumulative damage model,to predict the remaining life of the building and carry out a health diagnosis of the ancient building.The health diagnosis includes the determination of the state of damage and the assessment of the remaining life of the building.The results of the FEA calculations were compared with GB50165-92: Technical Specification for the Maintenance and Strengthening of Wooden Structures in Ancient Buildings to determine the current state of the building damage.At the same time,timber members subjected to constant axial forces and moments are simulated by numerical analysis using a cumulative member damage model to complete the damage durability assessment,thereby assessing the remaining life of the members and the building.The Gerhards cumulative damage model used was modified for corrosion to account for changes in the strength of the timber due to corrosion.Combining the results of the damage analysis with the assessment of the remaining life of the building,a final health diagnosis of the Hall of Peace was completed.The results show that the Taihe Hall is in a good state of health,with the stress and strain levels of the elements well within the specification,and that it will be safe for at least 1044-1063 years without taking into account natural disasters such as earthquakes and floods,and human damage,and that the Taihe Hall columns and beams have survived for 328 years,which is much shorter than the estimated life span.