Original village building restoration method and system based on 3D printing

Abstract

The invention discloses an original village building repairing method and system based on 3D printing, and relates to the field of ancient building repairing, the method comprises the steps that a to-be-repaired building is scanned, a numerical model of the to-be-repaired building is obtained, and the numerical model comprises the appearance, the internal structure and the specific condition of a damaged part of the building; based on the original materials and environmental conditions of the building to be repaired, 3D printing materials are selected, and the 3D printing materials have physical characteristics and chemical characteristics similar to those of the building materials to be repaired and are used for ensuring the compatibility of the repaired part and the original building; designing a repair scheme based on the constructed numerical model; the repair scheme comprises a complete replacement part and a reinforcement repair part; and based on the repairing scheme, a repairing component is printed and installed on the to-be-repaired building.

Description

Technical Field

[0001] The present invention relates to the field of ancient building restoration, and in particular to a method and system for restoring primitive village buildings based on 3D printing. Background Art

[0002] In the current field of cultural relics protection and historical building restoration, traditional restoration methods often face problems such as difficult material matching, complex processes, high costs, and unpredictable restoration effects. With the development of science and technology, especially the advancement of 3D printing technology, traditional restoration methods have been revolutionized. 3D printing technology, with its advantages of high precision, customizability, and rapid prototyping, provides a new solution for the restoration of historical buildings. However, how to effectively use 3D printing technology to restore historical buildings and ensure that the restored buildings are both safe and beautiful while retaining their historical value has become an urgent problem to be solved.

[0003] Traditional 3D printing restoration methods usually rely on simple geometric models and limited material selection, which limits the accuracy and effectiveness of restoration work. In addition, due to the lack of in-depth understanding of the complex structure of historical buildings, traditional methods often find it difficult to achieve perfect integration with the original buildings. Therefore, a more advanced restoration solution is needed that can comprehensively consider the structural characteristics, material characteristics and aesthetic requirements of the building to achieve efficient, accurate and beautiful restoration results. Summary of the invention

[0004] In order to solve the technical problems in the above background, the present invention uses high-precision 3D scanning technology to obtain a detailed numerical model of the building, including the appearance, internal structure and specific conditions of the damaged parts. Then, based on the materials and environmental conditions of the original building, 3D printing materials with similar physical and chemical properties are selected to ensure the compatibility of the repaired part with the original building. Through the constructed numerical model, a detailed repair plan is designed, including determining the completely replaced part and the reinforced repair part. Finally, according to the repair plan, the repair parts are printed and installed on the building to be repaired.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for restoring primitive village buildings based on 3D printing, characterized in that the steps include:

[0007] Scanning the building to be repaired to obtain a numerical model of the building to be repaired, wherein the numerical model includes: the appearance, internal structure and specific conditions of the damaged part of the building;

[0008] Based on the original materials and environmental conditions of the building to be repaired, 3D printing materials are selected. The 3D printing materials have similar physical and chemical properties to the building materials to be repaired to ensure the compatibility of the repaired part with the original building;

[0009] Designing a repair plan based on the constructed numerical model; the repair plan includes determining a completely replaced part and a reinforced repair part;

[0010] Based on the repair plan, a repair component is printed and installed on the building to be repaired.

[0011] Preferably, the method for scanning the building to be repaired includes: using 3D scanning equipment to comprehensively scan the building; during scanning, comprehensively scanning the building from multiple angles and directions; at the same time, scanning the internal structure of the building, including the inside of the walls, floor slabs, and beams and columns.

[0012] Preferably, after the scanning is completed, a 3D model is constructed using the scanning results; the steps include:

[0013] (1) Point cloud preprocessing

[0014] It is used to improve the quality of point cloud data, reduce noise, and optimize subsequent processing. It uses a local density-based method to calculate the local density of each point. If the density of a point is significantly lower than the average density of its neighboring points, the point is considered an outlier and is removed.

[0015] (2) Feature extraction

[0016] First, the feature descriptor is generated by calculating the local geometric features of each point. For each point p, its normal vector n is calculated. p and point p in the neighborhood i The normal vector n i ; Next, define the feature descriptor F(p) as the statistics of the relative direction and distance between these normal vectors. The descriptor is expressed as:

[0017] F(p)=∑ i∈N(p) w i ·(α i , β i , γ i ),

[0018] Where N(p) is the set of neighboring points of p; w i represents the distance-based weight; αi, βi, and γi are point p and its neighboring point p, respectively. i The relative direction angle between

[0019] Through the above steps, a feature descriptor can be generated for each point, which is used for point cloud matching and recognition, thereby providing rich geometric information in 3D modeling;

[0020] (3) Surface reconstruction

[0021] Poisson's equation is solved to generate a smooth 3D surface. The formula is as follows:

[0022]

[0023] In the formula, n represents the normal vector field of the point cloud, which is a vector field that describes the normal vector of each point on the surface of the point cloud; Represents the divergence of the normal vector field, which is a scalar field that describes the degree of divergence of the normal vector field at each point.

[0024] Preferably, after scanning is completed and the 3D model is constructed, a static analysis is performed for subsequent repair plan design, the steps of which include:

[0025] Stress Analysis:

[0026]

[0027] In the formula, σ represents stress; F represents the applied force; A represents the force-bearing area;

[0028] Evaluate material deformation:

[0029]

[0030] In the formula, ∈ represents strain, which is a dimensionless proportional value, indicating the deformation per unit length of an object when it is subjected to force; ΔL represents the deformation, that is, the change in the length of the object after it is subjected to force; L represents the length of the component;

[0031] Assess the overall deformation of the building:

[0032] u=K -1 F,

[0033] Where u represents the displacement vector; K -1 represents the inverse matrix of the overall stiffness matrix; F represents the load vector;

[0034] Assessing the stability of a building under load:

[0035]

[0036] Where P cr represents the critical load; E represents the elastic modulus; I represents the section moment of inertia; K represents the effective length coefficient; and L represents the member length.

[0037] Preferably, after the static analysis, the damaged parts are marked to clarify the key areas and specific needs of the repair; at the same time, multiple verifications and corrections are carried out, including on-site comparisons and expert reviews.

[0038] Preferably, when designing a repair solution, a convolutional neural network is used to construct a prediction model to optimize the repair solution; wherein the prediction model includes: an input layer, a convolutional layer, a pooling layer, a flattening layer, a fully connected layer and an output layer;

[0039] The input layer is used to receive the image data of the restoration case;

[0040] The convolution layer includes three layers of convolution, namely, the first layer of convolution: using 32 3x3 convolution kernels, the step size is 1, and the activation function is ReLU, which is used to extract the basic features of the image; the second layer of convolution: using 64 3x3 convolution kernels, the step size is 1, and the activation function is ReLU, which is used to capture more complex features; the third layer of convolution: using 128 3x3 convolution kernels, the step size is 1, and the activation function is ReLU, which is used to identify more advanced features;

[0041] The pooling layer uses a 2x2 maximum pooling with a step size of 2 to reduce the feature dimension and reduce the amount of calculation;

[0042] The flattening layer is used to flatten the multi-dimensional feature map into one dimension so as to be input into the fully connected layer;

[0043] The fully connected layers include: the first fully connected layer: 128 neurons, the activation function is ReLU, which is used to further process the flattened features; the second fully connected layer: 64 neurons, the activation function is ReLU, which is used to reduce the complexity of the model and improve the generalization ability;

[0044] Output layer: contains multiple neurons, each neuron corresponds to a repair strategy or material selection, and the activation function is softmax, which is used for multi-classification problems.

[0045] Preferably, the loss function of the prediction model uses a cross entropy loss function; the training uses an Adam optimizer, and the learning rate is set to 0.001, the momentum is set to 0.9, and the learning rate is adjusted adaptively.

[0046] The present invention also provides a 3D printing-based original village building restoration system, the system is used to implement the above method, including: a scanning module, a selection module, a design module and a printing module;

[0047] The scanning module is used to scan the building to be repaired to obtain a numerical model of the building to be repaired, wherein the numerical model includes: the appearance, internal structure and specific conditions of the damaged part of the building;

[0048] The selection module is used to select 3D printing materials based on the original materials and environmental conditions of the building to be repaired. The 3D printing materials have physical and chemical properties similar to those of the building materials to be repaired, so as to ensure the compatibility of the repaired part with the original building;

[0049] The design module is used to design a repair plan based on the constructed numerical model; the repair plan includes determining a completely replaced part and a reinforced repair part;

[0050] The printing module is used to print repair parts based on the repair plan and install the repair parts on the building to be repaired.

[0051] Preferably, the workflow of the scanning module includes: using 3D scanning equipment to comprehensively scan the building; during scanning, comprehensively scanning the building from multiple angles and directions; at the same time, scanning the internal structure of the building, including the inside of the walls, floor slabs and beams.

[0052] Preferably, after the scanning is completed, a 3D model is constructed using the scanning results; the steps include:

[0053] (1) Point cloud preprocessing

[0054] It is used to improve the quality of point cloud data, reduce noise, and optimize subsequent processing. It uses a local density-based method to calculate the local density of each point. If the density of a point is significantly lower than the average density of its neighboring points, the point is considered an outlier and is removed.

[0055] (2) Feature extraction

[0056] First, the feature descriptor is generated by calculating the local geometric features of each point. For each point p, its normal vector n is calculated. p and point p in the neighborhood i The normal vector n i ; Next, define the feature descriptor F(p) as the statistics of the relative direction and distance between these normal vectors. The descriptor is expressed as:

[0057] F(p)=∑ i∈N(p) w i ·(α i , β i , γ i ),

[0058] Where N(p) is the set of neighboring points of p; w i represents the distance-based weight; αi, βi, and γi are point p and its neighboring point p, respectively. i The relative direction angle between

[0059] Through the above steps, a feature descriptor can be generated for each point, which is used for point cloud matching and recognition, thereby providing rich geometric information in 3D modeling;

[0060] (3) Surface reconstruction

[0061] Poisson's equation is solved to generate a smooth 3D surface. The formula is as follows:

[0062]

[0063] In the formula, n represents the normal vector field of the point cloud, which is a vector field that describes the normal vector of each point on the surface of the point cloud; Represents the divergence of the normal vector field, which is a scalar field that describes the degree of divergence of the normal vector field at each point.

[0064] Compared with the prior art, the present invention has the following beneficial effects:

[0065] This invention uses advanced 3D scanning technology to obtain a detailed numerical model of the building, and combines machine learning algorithms to optimize the restoration plan, making the restoration work more scientific and reasonable. At the same time, the use of 3D printing materials similar to the original building materials ensures the harmony and unity of the restored part and the original building, prolongs the service life of the building, and also opens up a new path for the protection and inheritance of traditional architecture. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0067] Figure 1 It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0069] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0070] Embodiment 1

[0071] like Figure 1 FIG. 1 is a schematic diagram of the method flow of this embodiment, and the steps include:

[0072] S1. Scan the building to be repaired to obtain a numerical model of the building to be repaired.

[0073] In this embodiment, 3D modeling is the basis and key step of the entire restoration process. The main goal of this stage is to use high-precision 3D scanning equipment to comprehensively scan the building to be restored and obtain a detailed digital model. This model not only includes the exterior features of the building, but also covers the specific conditions of the internal structure and damaged parts, providing accurate data support for subsequent restoration design and material selection.

[0074] First of all, it is crucial to choose the right 3D scanning equipment. The three common scanning equipment include: laser scanner, structured light scanner and photogrammetry system.

[0075] Laser scanners are characterized by high precision and fast scanning, and are suitable for scanning large buildings and complex structures; structured light scanners obtain three-dimensional information of objects by projecting specific light patterns, and are suitable for medium-sized buildings; photogrammetry systems calculate the three-dimensional model of objects by taking photos from multiple angles and using software, and are suitable for small buildings and areas with rich details. According to the scale and complexity of the building, choose the most appropriate scanning equipment to ensure that the acquired data has high precision and high resolution.

[0076] After selecting the scanning equipment, it is necessary to conduct a comprehensive scan of the building. This process usually requires multi-angle and multi-directional scanning to ensure that the acquired data is complete and without omission. When scanning, special attention should be paid to the details of the building, including doors, windows, columns, roofs, walls, etc., as well as any visible damaged parts. For complex structures, such as carvings and murals, high-precision local scanning is required to ensure that these details are accurately presented in the digital model. In addition, the internal structure of the building needs to be scanned, including the inside of the wall, floor slabs, beams and columns, etc., in order to fully understand the structural condition of the building.

[0077] After the scan is completed, the 3D model is constructed using the scan results. 3D point cloud technology is widely used in 3D modeling, especially when obtaining accurate 3D models from objects in the real world. 3D point cloud data is collected by devices such as 3D scanners. These data represent a series of points on the surface of an object, each of which has its coordinates (x, y, z) in three-dimensional space and possible color information. Generating a 3D model from a 3D point cloud includes the following steps: point cloud preprocessing, feature extraction, and surface reconstruction.

[0078] (1) Point cloud preprocessing

[0079] The purpose of preprocessing is to improve the quality of point cloud data, reduce noise, and optimize subsequent processing. This embodiment uses statistical analysis to remove outliers; at the same time, the number of points is reduced by downsampling to improve processing efficiency. If the point cloud data comes from multiple perspectives, alignment is required to ensure that all point clouds are in the same coordinate system. This embodiment uses a local density-based method. Specifically, the local density of each point is calculated. If the density of a point is significantly lower than the average density of its neighboring points, the point is considered an outlier and is removed.

[0080] (2) Feature extraction

[0081] Feature extraction is to extract information from point cloud data that is helpful for subsequent processing, such as edges, planes, etc. This embodiment uses feature descriptors to perform feature extraction, and the specific steps include:

[0082] First, the feature descriptor is generated by calculating the local geometric features of each point. For each point p, its normal vector n is calculated. p and point p in the neighborhood i The normal vector n i Next, the feature descriptor F(p) is defined as the statistics of the relative directions and distances between these normal vectors. Specifically, the FPFH descriptor can be expressed as:

[0083] F(p)=∑ i∈N(p) w i ·(α i , β i , γ i ),

[0084] Where N(p) is the set of neighboring points of p; w i represents the distance-based weight; αi, βi, and γi are point p and its neighboring point p, respectively. i The relative direction angle between .

[0085] Through the above steps, a feature descriptor can be generated for each point, which can be used for point cloud matching and recognition, thereby providing rich geometric information in 3D modeling.

[0086] (3) Surface reconstruction

[0087] Surface reconstruction is the process of converting point cloud data into a continuous 3D surface. This embodiment uses Poisson's equation to solve and generate a smooth 3D surface. The formula is as follows:

[0088]

[0089] In the formula, n represents the normal vector field of the point cloud, which is a vector field that describes the normal vector of each point on the surface of the point cloud; Represents the divergence of the normal vector field, which is a scalar field that describes the degree of divergence of the normal vector field at each point.

[0090] After the 3D model is completed, it needs to be analyzed and evaluated in detail. The main task of this stage is to check the integrity and accuracy of the model to ensure that all key parts and details are accurately presented. At the same time, the model needs to be structurally analyzed to evaluate the building's load-bearing capacity, stability, etc., so that these factors can be considered in the subsequent restoration design. Structural analysis usually includes static analysis and dynamic analysis. For the original village buildings, due to their historicity and structural complexity, static analysis methods are mainly used. Static analysis can evaluate the load-bearing capacity and stability of the building in a static state (such as self-weight, wind load, snow load, etc.).

[0091] This embodiment uses finite element analysis (FEA) to perform structural analysis. Finite element analysis is a numerical method that divides the structure into multiple small units (finite elements), then performs mechanical analysis on each unit, and finally integrates the analysis results of the overall structure. FEA can accurately simulate the stress, strain and deformation of the building and is suitable for the analysis of complex structures. The specific steps include:

[0092] Divide the 3D model into multiple small units (finite elements). The fineness of the mesh directly affects the accuracy of the analysis. For key parts and details, finer meshes are required. Define material properties for each finite element, including elastic modulus (E), Poisson's ratio (ν), and density (ρ). Define the fixed and free points of the building, as well as the applied loads (such as self-weight, wind load, snow load, etc.).

[0093] After the finite element model is established, a static analysis is performed to prepare for the subsequent repair plan design. Based on the material properties and geometry, the stiffness matrix of each finite element is calculated; and the stiffness matrices of all finite elements are assembled into an overall stiffness matrix. After that, the load vector is applied to the overall stiffness matrix; finally, the linear equations are solved to obtain the node displacement, stress, and strain. Stress analysis uses the following formula to evaluate whether the material exceeds the yield strength:

[0094]

[0095] In the formula, σ represents stress; F represents the applied force; and A represents the force-bearing area.

[0096] By calculating the strain value of each finite element, the deformation of the material is evaluated:

[0097]

[0098] In the formula, ∈ represents strain, which is a dimensionless proportional value, indicating the deformation per unit length of an object when subjected to force; ΔL represents the deformation, that is, the change in the length of the object after being subjected to force; and L represents the length of the component.

[0099] Calculate the displacement value of each node and evaluate the overall deformation of the building:

[0100] u=K -1 F,

[0101] Where u represents the displacement vector; K -1 represents the inverse matrix of the overall stiffness matrix; F represents the load vector.

[0102] Common methods for evaluating the stability of a building under load include buckling analysis and vibration analysis. Buckling analysis can evaluate the stability of a building under compressive loads:

[0103]

[0104] Where P cr represents the critical load; E represents the elastic modulus; I represents the section moment of inertia; K represents the effective length coefficient; and L represents the member length.

[0105] In addition, the damaged parts need to be marked in detail to identify the key areas and specific needs of the repair, so as to provide an accurate reference for the subsequent repair design. In order to ensure the quality of 3D modeling, multiple verifications and corrections are required. This step includes on-site comparison and expert review. On-site comparison refers to comparing the 3D model with the actual building to check the matching degree between the model and the actual building to ensure the accuracy and reliability of the model. The expert review is to invite experts in the fields of architecture, history, materials science, etc. to evaluate the model and put forward improvement opinions and suggestions. Through these verifications and corrections, the quality of the 3D model can be further improved to ensure that it plays an important role in the subsequent repair process.

[0106] After completing 3D modeling, the model needs to be exported to a standard format for use in subsequent repair design and 3D printing. Common formats include STL, OBJ, FBX, etc. These formats can be supported by most 3D modeling software and 3D printers. When exporting, it is necessary to ensure that the accuracy and resolution of the model meet the requirements to avoid data loss or loss of accuracy in subsequent steps.

[0107] S2. Select 3D printing materials based on the original materials and environmental conditions of the building to be restored.

[0108] The 3D printing materials have physical and chemical properties similar to those of the building materials to be repaired, which is used to ensure the compatibility of the repaired part with the original building.

[0109] When performing 3D printing repairs on original village buildings, it is crucial to choose the right 3D printing materials. The choice of materials not only affects the beauty and durability of the repair effect, but also the structural safety and environmental adaptability of the building. The following will explain in detail how to choose materials suitable for 3D printing based on the original materials of the building and environmental conditions. Before choosing 3D printing materials, you first need to understand the original materials of the building in detail. The type, performance and characteristics of the original materials will directly affect the choice of repair materials. Common original building materials include:

[0110] Stone: such as limestone, sandstone, granite, etc. Stone has high strength and durability, but is heavy and difficult to process.

[0111] Brick: such as red brick, blue brick, etc. Brick has good thermal insulation and fire resistance, but its strength is relatively low.

[0112] Wood: such as pine, oak, etc. Wood has good toughness and processing properties, but is easily affected by moisture, insects and fire.

[0113] Concrete: such as ordinary concrete, lightweight concrete, etc. Concrete has high strength and durability, but is prone to cracking.

[0114] Metal: such as steel, aluminum alloy, etc. Metal has high strength and corrosion resistance, but the cost is high.

[0115] Environmental conditions also have an important impact on the selection of 3D printing materials. Factors that need to be considered include:

[0116] Climate conditions: such as temperature, humidity, rainfall, etc. Different climate conditions have different requirements for the weather resistance and durability of materials.

[0117] Geological conditions: such as soil type, groundwater level, etc. Geological conditions affect the foundation stability of the building and the corrosion resistance of the materials.

[0118] Pollution conditions: such as air pollution, soil pollution, etc. Pollution conditions may have an adverse effect on the performance of materials.

[0119] Biological factors: such as insects, microorganisms, etc. Biological factors may affect the durability and safety of materials.

[0120] Choose appropriate 3D printing materials based on the original materials and environmental conditions of the building.

[0121] S3. Design a repair plan based on the constructed numerical model.

[0122] In this example, these options include determining which parts need to be completely replaced and which parts can be reinforced or repaired. The design should fully consider the historical value and aesthetic requirements of the building.

[0123] Based on the above stress analysis, determine which parts need to be completely replaced and which parts can be reinforced or repaired to restore their functions. For example, for key parts of load-bearing walls, we may choose to use high-strength, high-durability 3D printing materials to manufacture replacement parts to ensure that they can withstand future loads. For non-load-bearing walls or decorative elements, more economical materials that are closer in appearance to the original building will be used for repair. When designing the repair plan, the historical value and aesthetic requirements of the building must also be taken into account to ensure that the repaired building is both safe and maintains its original appearance.

[0124] In addition, machine learning algorithms will be used to optimize the restoration plan. By analyzing a large number of historical building restoration cases, the effects of different restoration strategies are predicted, so as to select the best restoration path. Specifically, a convolutional neural network (CNN) is used to build a prediction model to analyze image data of historical building restoration cases, such as photos of buildings before and after restoration, and detailed descriptions of the materials and techniques used for restoration. In this way, CNN can learn the association between different types of damage and restoration strategies, and can predict the most effective restoration method under given damage conditions. In this embodiment, the prediction model includes: an input layer, a convolutional layer, a pooling layer, a flattening layer, a fully connected layer, and an output layer.

[0125] The input layer is used to receive image data of restoration cases, which are usually photos of buildings before and after restoration, as well as detailed images of the restoration parts.

[0126] The convolution layer includes three layers of convolution, namely: the first convolution layer: using 32 3x3 convolution kernels, the step size is 1, and the activation function is ReLU, which is used to extract basic features of the image, such as edges and textures. The second convolution layer: using 64 3x3 convolution kernels, the step size is 1, and the activation function is also ReLU, which is used to capture more complex features. The third convolution layer: using 128 3x3 convolution kernels, the step size is 1, and the activation function is ReLU, which is used to identify more advanced features, such as local patterns.

[0127] The pooling layer uses a 2x2 maximum pooling with a step size of 2 to reduce the feature dimension and reduce the amount of calculation.

[0128] The flattening layer flattens the multi-dimensional feature map into one dimension so as to be input into the fully connected layer.

[0129] The fully connected layers include: the first fully connected layer: 128 neurons, the activation function is ReLU, which is used to further process the flattened features; the second fully connected layer: 64 neurons, the activation function is ReLU, which is used to reduce the complexity of the model and improve the generalization ability.

[0130] Output layer: contains multiple neurons, each neuron corresponds to a repair strategy or material selection, and the activation function is softmax, which is used for multi-classification problems.

[0131] When training the above prediction model, the loss function uses the cross entropy loss function, the Adam optimizer is used, the learning rate is set to 0.001, the momentum is set to 0.9, and the learning rate is adjusted adaptively.

[0132] S4. Based on the repair plan, print the repair parts and install them on the building to be repaired.

[0133] After the restoration plan is formulated, the actual printing and installation phase begins. First, the corresponding parameters are set on the 3D printer according to the design plan, including the choice of printing materials, printing speed, layer height, etc., to ensure that the printed restoration parts can meet the structural strength and appearance requirements. During the printing process, we will closely monitor the printing quality to ensure that the accuracy and quality of each part meet the design standards. After printing, these parts will be post-processed, such as polishing, coloring or distressing, to match the style and material of the original building. Next, we will organize a professional team to go to the site of the building to be restored and install the printed parts one by one. During the installation process, we will use precision measuring tools to ensure the accurate position of each part, and use appropriate fixing methods to ensure the stability and durability of the parts. Finally, a comprehensive inspection is carried out to ensure that all restoration parts have been correctly installed and perfectly integrated with the original building.

[0134] Embodiment 2

[0135] The present embodiment also provides a primitive village building restoration system based on 3D printing, comprising: a scanning module, a selection module, a design module and a printing module; the scanning module is used to scan the building to be restored and obtain a numerical model of the building to be restored, and the numerical model includes: the appearance of the building, the internal structure and the specific conditions of the damaged part; the selection module is used to select 3D printing materials based on the original materials and environmental conditions of the building to be restored, and the 3D printing materials have physical and chemical properties similar to those of the building materials to be restored, and are used to ensure the compatibility of the restored part with the original building; the design module is used to design a restoration plan based on the constructed numerical model; the restoration plan includes determining the completely replaced part and the reinforced repair part; the printing module is used to print the restoration parts based on the restoration plan and install the restoration parts on the building to be restored.

[0136] The following will describe in detail how the present invention solves the technical problems in practical work in conjunction with this embodiment.

[0137] First, the building to be repaired is scanned using a scanning module to obtain a numerical model of the building to be repaired.

[0138] In this embodiment, 3D modeling is the basis and key step of the entire restoration process. The main goal of this stage is to use high-precision 3D scanning equipment to comprehensively scan the building to be restored and obtain a detailed digital model. This model not only includes the exterior features of the building, but also covers the specific conditions of the internal structure and damaged parts, providing accurate data support for subsequent restoration design and material selection.

[0139] The scanning devices selected in the scanning module of this embodiment include: laser scanner, structured light scanner and photogrammetry system.

[0140] Laser scanners are characterized by high precision and fast scanning, and are suitable for scanning large buildings and complex structures; structured light scanners obtain three-dimensional information of objects by projecting specific light patterns, and are suitable for medium-sized buildings; photogrammetry systems calculate the three-dimensional model of objects by taking photos from multiple angles and using software, and are suitable for small buildings and areas with rich details. According to the scale and complexity of the building, choose the most appropriate scanning equipment to ensure that the acquired data has high precision and high resolution.

[0141] After selecting the scanning equipment, it is necessary to conduct a comprehensive scan of the building. This process usually requires multi-angle and multi-directional scanning to ensure that the acquired data is complete and without omission. When scanning, special attention should be paid to the details of the building, including doors, windows, columns, roofs, walls, etc., as well as any visible damaged parts. For complex structures, such as carvings and murals, high-precision local scanning is required to ensure that these details are accurately presented in the digital model. In addition, the internal structure of the building needs to be scanned, including the inside of the wall, floor slabs, beams and columns, etc., in order to fully understand the structural condition of the building.

[0142] After the scan is completed, the 3D model is constructed using the scan results. 3D point cloud technology is widely used in 3D modeling, especially when obtaining accurate 3D models from objects in the real world. 3D point cloud data is collected by devices such as 3D scanners. These data represent a series of points on the surface of an object, each of which has its coordinates (x, y, z) in three-dimensional space and possible color information. Generating a 3D model from a 3D point cloud includes the following steps: point cloud preprocessing, feature extraction, and surface reconstruction.

[0143] (1) Point cloud preprocessing

[0144] The purpose of preprocessing is to improve the quality of point cloud data, reduce noise, and optimize subsequent processing. This embodiment uses statistical analysis to remove outliers; at the same time, the number of points is reduced by downsampling to improve processing efficiency. If the point cloud data comes from multiple perspectives, alignment is required to ensure that all point clouds are in the same coordinate system. This embodiment uses a local density-based method. Specifically, the local density of each point is calculated. If the density of a point is significantly lower than the average density of its neighboring points, the point is considered an outlier and is removed.

[0145] (2) Feature extraction

[0146] Feature extraction is to extract information from point cloud data that is helpful for subsequent processing, such as edges, planes, etc. This embodiment uses feature descriptors to perform feature extraction, and the specific steps include:

[0147] First, the feature descriptor is generated by calculating the local geometric features of each point. For each point p, its normal vector n is calculated. p and point p in the neighborhood i The normal vector n i Next, the feature descriptor F(p) is defined as the statistics of the relative directions and distances between these normal vectors. Specifically, the FPFH descriptor can be expressed as:

[0148] F(p)=∑ i∈N(p) w i ·(α i , β i , γ i ),

[0149] Where N(p) is the set of neighboring points of p; w i represents the distance-based weight; αi, βi, and γi are point p and its neighboring point p, respectively. i The relative direction angle between .

[0150] Through the above steps, a feature descriptor can be generated for each point, which can be used for point cloud matching and recognition, thereby providing rich geometric information in 3D modeling.

[0151] (3) Surface reconstruction

[0152] Surface reconstruction is the process of converting point cloud data into a continuous 3D surface. This embodiment uses Poisson's equation to solve and generate a smooth 3D surface. The formula is as follows:

[0153]

[0154] In the formula, n represents the normal vector field of the point cloud, which is a vector field that describes the normal vector of each point on the surface of the point cloud; Represents the divergence of the normal vector field, which is a scalar field that describes the degree of divergence of the normal vector field at each point.

[0155] After the 3D model is completed, it needs to be analyzed and evaluated in detail. The main task of this stage is to check the integrity and accuracy of the model to ensure that all key parts and details are accurately presented. At the same time, the model needs to be structurally analyzed to evaluate the building's load-bearing capacity, stability, etc., so that these factors can be considered in the subsequent restoration design. Structural analysis usually includes static analysis and dynamic analysis. For the original village buildings, due to their historicity and structural complexity, static analysis methods are mainly used. Static analysis can evaluate the load-bearing capacity and stability of the building in a static state (such as self-weight, wind load, snow load, etc.).

[0156] This embodiment uses finite element analysis (FEA) to perform structural analysis. Finite element analysis is a numerical method that divides the structure into multiple small units (finite elements), then performs mechanical analysis on each unit, and finally integrates the analysis results of the overall structure. FEA can accurately simulate the stress, strain and deformation of the building and is suitable for the analysis of complex structures. The specific steps include:

[0157] Divide the 3D model into multiple small units (finite elements). The fineness of the mesh directly affects the accuracy of the analysis. For key parts and details, finer meshes are required. Define material properties for each finite element, including elastic modulus (E), Poisson's ratio (ν), and density (ρ). Define the fixed and free points of the building, as well as the applied loads (such as self-weight, wind load, snow load, etc.).

[0158] After the finite element model is established, a static analysis is performed to prepare for the subsequent repair plan design. Based on the material properties and geometry, the stiffness matrix of each finite element is calculated; and the stiffness matrices of all finite elements are assembled into an overall stiffness matrix. After that, the load vector is applied to the overall stiffness matrix; finally, the linear equations are solved to obtain the node displacement, stress, and strain. Stress analysis uses the following formula to evaluate whether the material exceeds the yield strength:

[0159]

[0160] In the formula, σ represents stress; F represents the applied force; and A represents the force-bearing area.

[0161] By calculating the strain value of each finite element, the deformation of the material is evaluated:

[0162]

[0163] In the formula, ∈ represents strain, which is a dimensionless proportional value, indicating the deformation per unit length of an object when subjected to force; ΔL represents the deformation, that is, the change in the length of the object after being subjected to force; and L represents the length of the component.

[0164] Calculate the displacement value of each node and evaluate the overall deformation of the building:

[0165] u=K -1 F,

[0166] Where u represents the displacement vector; K -1 represents the inverse matrix of the overall stiffness matrix; F represents the load vector.

[0167] Common methods for evaluating the stability of a building under load include buckling analysis and vibration analysis. Buckling analysis can evaluate the stability of a building under compressive loads:

[0168]

[0169] Where P cr represents the critical load; E represents the elastic modulus; I represents the section moment of inertia; K represents the effective length coefficient; and L represents the member length.

[0170] In addition, the damaged parts need to be marked in detail to identify the key areas and specific needs of the repair, so as to provide an accurate reference for the subsequent repair design. In order to ensure the quality of 3D modeling, multiple verifications and corrections are required. This step includes on-site comparison and expert review. On-site comparison refers to comparing the 3D model with the actual building to check the matching degree between the model and the actual building to ensure the accuracy and reliability of the model. The expert review is to invite experts in the fields of architecture, history, materials science, etc. to evaluate the model and put forward improvement opinions and suggestions. Through these verifications and corrections, the quality of the 3D model can be further improved to ensure that it plays an important role in the subsequent repair process.

[0171] After completing 3D modeling, the model needs to be exported to a standard format for use in subsequent repair design and 3D printing. Common formats include STL, OBJ, FBX, etc. These formats can be supported by most 3D modeling software and 3D printers. When exporting, it is necessary to ensure that the accuracy and resolution of the model meet the requirements to avoid data loss or loss of accuracy in subsequent steps.

[0172] The selection module then chooses the 3D printing material based on the original material and environmental conditions of the building to be restored.

[0173] The 3D printing materials have physical and chemical properties similar to those of the building materials to be repaired, which is used to ensure the compatibility of the repaired part with the original building.

[0174] When performing 3D printing repairs on original village buildings, it is crucial to choose the right 3D printing materials. The choice of materials not only affects the beauty and durability of the repair effect, but also the structural safety and environmental adaptability of the building. The following will explain in detail how to choose materials suitable for 3D printing based on the original materials of the building and environmental conditions. Before choosing 3D printing materials, you first need to understand the original materials of the building in detail. The type, performance and characteristics of the original materials will directly affect the choice of repair materials. Common original building materials include:

[0175] Stone: such as limestone, sandstone, granite, etc. Stone has high strength and durability, but is heavy and difficult to process.

[0176] Brick: such as red brick, blue brick, etc. Brick has good thermal insulation and fire resistance, but its strength is relatively low.

[0177] Wood: such as pine, oak, etc. Wood has good toughness and processing properties, but is easily affected by moisture, insects and fire.

[0178] Concrete: such as ordinary concrete, lightweight concrete, etc. Concrete has high strength and durability, but is prone to cracking.

[0179] Metal: such as steel, aluminum alloy, etc. Metal has high strength and corrosion resistance, but the cost is high.

[0180] Environmental conditions also have an important impact on the selection of 3D printing materials. Factors that need to be considered include:

[0181] Climate conditions: such as temperature, humidity, rainfall, etc. Different climate conditions have different requirements for the weather resistance and durability of materials.

[0182] Geological conditions: such as soil type, groundwater level, etc. Geological conditions affect the foundation stability of the building and the corrosion resistance of the materials.

[0183] Pollution conditions: such as air pollution, soil pollution, etc. Pollution conditions may have an adverse effect on the performance of materials.

[0184] Biological factors: such as insects, microorganisms, etc. Biological factors may affect the durability and safety of materials.

[0185] Choose appropriate 3D printing materials based on the original materials and environmental conditions of the building.

[0186] The module involved designs a repair solution based on the constructed numerical model.

[0187] In this example, these options include determining which parts need to be completely replaced and which parts can be reinforced or repaired. The design should fully consider the historical value and aesthetic requirements of the building.

[0188] Based on the above stress analysis, determine which parts need to be completely replaced and which parts can be reinforced or repaired to restore their functions. For example, for key parts of load-bearing walls, we may choose to use high-strength, high-durability 3D printing materials to manufacture replacement parts to ensure that they can withstand future loads. For non-load-bearing walls or decorative elements, more economical materials that are closer in appearance to the original building will be used for repair. When designing the repair plan, the historical value and aesthetic requirements of the building must also be taken into account to ensure that the repaired building is both safe and maintains its original appearance.

[0189] In addition, machine learning algorithms will be used to optimize the restoration plan. By analyzing a large number of historical building restoration cases, the effects of different restoration strategies are predicted, so as to select the best restoration path. Specifically, a convolutional neural network (CNN) is used to build a prediction model to analyze image data of historical building restoration cases, such as photos of buildings before and after restoration, and detailed descriptions of the materials and techniques used for restoration. In this way, CNN can learn the association between different types of damage and restoration strategies, and can predict the most effective restoration method under given damage conditions. In this embodiment, the prediction model includes: an input layer, a convolutional layer, a pooling layer, a flattening layer, a fully connected layer, and an output layer.

[0190] The input layer is used to receive image data of restoration cases, which are usually photos of buildings before and after restoration, as well as detailed images of the restoration parts.

[0191] The convolution layer includes three layers of convolution, namely: the first convolution layer: using 32 3x3 convolution kernels, the step size is 1, and the activation function is ReLU, which is used to extract basic features of the image, such as edges and textures. The second convolution layer: using 64 3x3 convolution kernels, the step size is 1, and the activation function is also ReLU, which is used to capture more complex features. The third convolution layer: using 128 3x3 convolution kernels, the step size is 1, and the activation function is ReLU, which is used to identify more advanced features, such as local patterns.

[0192] The pooling layer uses a 2x2 maximum pooling with a step size of 2 to reduce the feature dimension and reduce the amount of calculation.

[0193] The flattening layer flattens the multi-dimensional feature map into one dimension so as to be input into the fully connected layer.

[0194] The fully connected layers include: the first fully connected layer: 128 neurons, the activation function is ReLU, which is used to further process the flattened features; the second fully connected layer: 64 neurons, the activation function is ReLU, which is used to reduce the complexity of the model and improve the generalization ability.

[0195] Output layer: contains multiple neurons, each neuron corresponds to a repair strategy or material selection, and the activation function is softmax, which is used for multi-classification problems.

[0196] When training the above prediction model, the loss function uses the cross entropy loss function, the Adam optimizer is used, the learning rate is set to 0.001, the momentum is set to 0.9, and the learning rate is adjusted adaptively.

[0197] Finally, the printing module prints the repair parts based on the repair plan and installs the repair parts on the building to be repaired.

[0198] After the restoration plan is formulated, the actual printing and installation phase begins. First, the corresponding parameters are set on the 3D printer according to the design plan, including the choice of printing materials, printing speed, layer height, etc., to ensure that the printed restoration parts can meet the structural strength and appearance requirements. During the printing process, we will closely monitor the printing quality to ensure that the accuracy and quality of each part meet the design standards. After printing, these parts will be post-processed, such as polishing, coloring or distressing, to match the style and material of the original building. Next, we will organize a professional team to go to the site of the building to be restored and install the printed parts one by one. During the installation process, we will use precision measuring tools to ensure the accurate position of each part, and use appropriate fixing methods to ensure the stability and durability of the parts. Finally, a comprehensive inspection is carried out to ensure that all restoration parts have been correctly installed and perfectly integrated with the original building.

[0199] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for restoring primitive village buildings based on 3D printing, characterized in that: The steps include: Scanning the building to be repaired to obtain a numerical model of the building to be repaired, wherein the numerical model includes: the appearance, internal structure and specific conditions of the damaged part of the building; Based on the original materials and environmental conditions of the building to be repaired, 3D printing materials are selected. The 3D printing materials have similar physical and chemical properties to the building materials to be repaired to ensure the compatibility of the repaired part with the original building; Designing a repair plan based on the constructed numerical model; the repair plan includes determining a completely replaced part and a reinforced repair part; Based on the repair plan, a repair component is printed and installed on the building to be repaired.

2. The original village building restoration method based on 3D printing according to claim 1 is characterized in that: Methods for scanning the building to be restored include: using 3D scanning equipment to conduct a comprehensive scan of the building; during the scanning, the building is comprehensively scanned from multiple angles and directions; at the same time, the internal structure of the building is scanned, including the inside of the walls, floor slabs, and beams and columns.

3. The original village building restoration method based on 3D printing according to claim 2 is characterized in that: After scanning is completed, the 3D model is constructed using the scan results; the steps include: (1) Point cloud preprocessing It is used to improve the quality of point cloud data, reduce noise, and optimize subsequent processing. It uses a local density-based method to calculate the local density of each point. If the density of a point is significantly lower than the average density of its neighboring points, it will be considered an outlier and removed. (2) Feature extraction First, the feature descriptor is generated by calculating the local geometric features of each point. For each point p, its normal vector n is calculated. p and point p in the neighborhood i The normal vector n i ; Next, define the feature descriptor F(p) as the statistics of the relative direction and distance between these normal vectors. The descriptor is expressed as: F(p)=∑ i∈N(p) w i ·(a i ,b i ,c i ), Where N(p) is the set of neighboring points of p; w i represents the distance-based weight; αi, βi, and γi are point p and its neighboring point p, respectively. i The relative direction angle between Through the above steps, a feature descriptor can be generated for each point, which is used for point cloud matching and recognition, thereby providing rich geometric information in 3D modeling; (3) Surface reconstruction Poisson's equation is solved to generate a smooth 3D surface. The formula is as follows: In the formula, n represents the normal vector field of the point cloud, which is a vector field that describes the normal vector of each point on the surface of the point cloud; Represents the divergence of the normal vector field, which is a scalar field that describes the degree of divergence of the normal vector field at each point.

4. The original village building restoration method based on 3D printing according to claim 3 is characterized in that: After scanning is completed and the 3D model is built, static analysis is performed for subsequent repair plan design. The steps include: Stress Analysis: In the formula, σ represents stress; F represents the applied force; A represents the force-bearing area; Evaluate material deformation: In the formula, ∈ represents strain, which is a dimensionless proportional value, indicating the deformation per unit length of an object when it is subjected to force; ΔL represents the deformation, that is, the change in the length of the object after it is subjected to force; L represents the length of the component; Assess the overall deformation of the building: u=K -1 F, Where u represents the displacement vector; K -1 represents the inverse matrix of the overall stiffness matrix; F represents the load vector; Assessing the stability of a building under load: Where P cr represents the critical load; E represents the elastic modulus; I represents the section moment of inertia; K represents the effective length coefficient; and L represents the member length.

5. The original village building restoration method based on 3D printing according to claim 4 is characterized in that: After the static analysis, the damaged parts are marked to clarify the key areas and specific needs of repair; at the same time, multiple verifications and corrections are carried out, including on-site comparisons and expert reviews.

6. The original village building restoration method based on 3D printing according to claim 1 is characterized in that: When designing a repair plan, a convolutional neural network is used to build a prediction model to optimize the repair plan. The prediction model includes: input layer, convolution layer, pooling layer, flattening layer, fully connected layer and output layer. The input layer is used to receive the image data of the restoration case; The convolution layer includes three layers of convolution, namely, the first layer of convolution: using 32 3x3 convolution kernels, the step size is 1, and the activation function is ReLU, which is used to extract the basic features of the image; the second layer of convolution: using 64 3x3 convolution kernels, the step size is 1, and the activation function is ReLU, which is used to capture more complex features; the third layer of convolution: using 128 3x3 convolution kernels, the step size is 1, and the activation function is ReLU, which is used to identify more advanced features; The pooling layer uses a 2x2 maximum pooling with a step size of 2 to reduce the feature dimension and reduce the amount of calculation; The flattening layer is used to flatten the multi-dimensional feature map into one dimension so as to be input into the fully connected layer; The fully connected layers include: the first fully connected layer: 128 neurons, the activation function is ReLU, which is used to further process the flattened features; the second fully connected layer: 64 neurons, the activation function is ReLU, which is used to reduce the complexity of the model and improve the generalization ability; Output layer: contains multiple neurons, each neuron corresponds to a repair strategy or material selection, and the activation function is softmax, which is used for multi-classification problems.

7. The original village building restoration method based on 3D printing according to claim 6 is characterized in that: The loss function of the prediction model uses the cross entropy loss function; the training adopts the Adam optimizer, sets the learning rate to 0.001, the momentum to 0.9, and the adaptive learning rate adjustment.

8. A 3D printing-based original village building restoration system, the system being used to implement the method described in any one of claims 1 to 7, characterized in that: include: Scanning module, selection module, design module and printing module; The scanning module is used to scan the building to be repaired to obtain a numerical model of the building to be repaired, wherein the numerical model includes: the appearance, internal structure and specific conditions of the damaged part of the building; The selection module is used to select 3D printing materials based on the original materials and environmental conditions of the building to be repaired. The 3D printing materials have physical and chemical properties similar to those of the building materials to be repaired, so as to ensure the compatibility of the repaired part with the original building; The design module is used to design a repair plan based on the constructed numerical model; the repair plan includes determining a completely replaced part and a reinforced repair part; The printing module is used to print repair parts based on the repair plan and install the repair parts on the building to be repaired.

9. The original village building restoration system based on 3D printing according to claim 8 is characterized in that: The workflow of the scanning module includes: using 3D scanning equipment to comprehensively scan the building; during scanning, comprehensively scanning the building from multiple angles and directions; at the same time, scanning the internal structure of the building, including the inside of the wall, floor slabs and beams.

10. The original village building restoration system based on 3D printing according to claim 9 is characterized in that: After scanning is completed, the 3D model is constructed using the scan results; the steps include: (1) Point cloud preprocessing It is used to improve the quality of point cloud data, reduce noise, and optimize subsequent processing. It uses a local density-based method to calculate the local density of each point. If the density of a point is significantly lower than the average density of its neighboring points, the point is considered an outlier and is removed. (2) Feature extraction First, the feature descriptor is generated by calculating the local geometric features of each point. For each point p, its normal vector n is calculated. p and point p in the neighborhood i The normal vector n i ; Next, define the feature descriptor F(p) as the statistics of the relative direction and distance between these normal vectors. The descriptor is expressed as: F(p)=∑ i∈N(p) w i ·(a i ,b i ,c i ), Where N(p) is the set of neighboring points of p; w i represents the distance-based weight; αi, βi, and γi are point p and its neighboring point p, respectively. i The relative direction angle between Through the above steps, a feature descriptor can be generated for each point, which is used for point cloud matching and recognition, thereby providing rich geometric information in 3D modeling; (3) Surface reconstruction Poisson's equation is solved to generate a smooth 3D surface. The formula is as follows: In the formula, n represents the normal vector field of the point cloud, which is a vector field that describes the normal vector of each point on the surface of the point cloud; Represents the divergence of the normal vector field, which is a scalar field that describes the degree of divergence of the normal vector field at each point.

Images