Steel structure paint area intelligent statistical method and system based on artificial intelligence and three-dimensional model fusion

Abstract

The invention discloses a steel structure paint area intelligent statistical method and system based on artificial intelligence and three-dimensional model fusion. The method comprises the following steps: firstly, extracting an initial surface area of a component from a BIM design model; on the basis of a built-in geometric rule base, coating-free areas such as welding seams, bolt holes, concrete wrapping sections and assembly binding surfaces are automatically recognized and deducted, and a theoretical net area is obtained; after the component is manufactured, a real-time curved surface model is reconstructed through three-dimensional laser scanning, and the area is corrected based on Gaussian curvature integration; in the coating process, an RGB-D camera and a U-Net + + visual model are adopted to recognize defects such as missed coating and sagging, and back projection is carried out on the defects to the quantized area and position of the three-dimensional model; and finally, a multi-dimensional dosage report and a quality early warning work order are automatically generated in combination with paint volume solid component parameters and are pushed to an MES system. According to the method, full-process automatic area statistics and closed-loop quality management are realized, and the precision and the construction digitization level are remarkably improved.

Description

Technical Field

[0001] This invention relates to the fields of intelligent manufacturing and digital twin technology for steel structures, and more specifically to a dynamic statistical method and system for paint area of ​​steel structures that integrates BIM model analysis, 3D scanning reconstruction, computer vision and rule reasoning. Background Technology

[0002] In steel structure manufacturing and construction, accurately calculating the paint (including anti-corrosion paint and fire-retardant paint) coating area is a crucial step in material procurement, cost accounting, construction planning, and worker performance evaluation. However, traditional calculation methods have significant shortcomings.

[0003] Firstly, in current steel structure engineering practice, the calculation of paint usage generally relies on the theoretical surface area of ​​components output by design software (such as Tekla). However, the theoretical area includes a large number of areas that do not require painting (such as the 150mm range on both sides of the weld, the bolt connection surface, and the concrete-encased section), which need to be manually marked, resulting in low efficiency and easy omissions.

[0004] Secondly, for irregularly shaped components with free-form surfaces, variable cross-sections, or cast steel nodes, there are often geometric deviations between the design model and the actual machined parts, leading to inaccurate area calculations. For example, the cast steel nodes of an airport terminal building experienced local curvature changes due to casting shrinkage, resulting in a difference between the measured surface area and the Tekla model. If paint is purchased based on the theoretical value, it will lead to material waste or delays in on-site material replenishment.

[0005] Furthermore, traditional methods lack real-time quality monitoring and closed-loop feedback mechanisms for the coating process. Defects such as missed areas, runs, and uneven thickness that occur during construction usually rely on manual inspections, which makes it difficult to achieve comprehensive coverage, precise location, and quantitative statistics. Moreover, it is impossible to effectively link defect information with three-dimensional component models and the responsible construction party, thus hindering the improvement of quality traceability and rectification efficiency.

[0006] In summary, current industry practice has yet to find an automated, high-precision, and fully closed-loop solution for paint area statistics that can systematically solve the aforementioned problems. Summary of the Invention

[0007] The purpose of this invention is to provide an intelligent statistical method and system for steel structure paint area based on the integration of artificial intelligence and three-dimensional model, so as to realize dynamic area correction, defect identification and cost control throughout the entire process from design to construction.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: On one hand, it provides an intelligent statistical method for the paint area of ​​steel structures based on the fusion of artificial intelligence and three-dimensional models, comprising the following steps:

[0009] Step 1: Initial theoretical area extraction. Parse the boundary representation model of the components from steel structure BIM design software (such as Tekla), extract the part numbers, materials, and spatial locations, and calculate the initial theoretical outer surface coating area Q of the components.

[0010] Step Two: Automatic Identification and Subtraction of Unpainted Areas. Based on the built-in geometric rule library, and through spatial adjacency relationships and Boolean operations, the total area Q′ of unpainted areas is automatically identified and subtracted to obtain the theoretical net painted area. The paint-free area includes:

[0011] (1) Weld area: Extract all welding symbol lines in the model, and offset them 150mm outward along the surface normal of the two associated base material patches to generate a ring-shaped deduction area;

[0012] (2) Bolt shielding surface: For each bolt hole, extend 70mm outward along the three orthogonal directions X / Y / Z to form a hexahedral shielding area, and find its intersection with the surface of the component;

[0013] (3) Concrete-encased section: The area below the column base is automatically cut off according to the burial depth parameters;

[0014] (4) Assembly mating surfaces: Traverse all connection pairs, calculate the Boolean intersection of the contact surfaces, and mark the areas that are not visible in the final assembly model;

[0015] Step 3: Real-world geometric scanning and dynamic area correction. After the component fabrication and painting are completed, a 3D laser scanner is used to acquire real-world point cloud data of the steel structure. A high-precision real-world surface model is constructed using point cloud processing algorithms (such as denoising and registration). This real-world model is intelligently compared with the design model to detect geometric changes caused by processing errors or design modifications. Subsequently, a closed mesh is generated using Poisson surface reconstruction, and the actual surface area is calculated using discrete Gaussian curvature integral. For areas with a deviation >3mm from the design model, the theoretical value is replaced with the real-world area, and the corrected area is output. .

[0016] Step Four: Vision-Driven Coating Status Monitoring and Defect Quantification. During the coating process, images of the steel structure surface are acquired in real time using image acquisition equipment deployed on-site. A trained AI vision model (such as a convolutional neural network) is used to analyze the images, achieving two main functions: first, identifying coated and uncoated areas to achieve visualized and automatic statistical analysis of construction progress; second, detecting coating defects (such as missed areas, uneven thickness, and runs), and feeding back the defect location and area to the statistical system, linking it to the corresponding responsible unit or guiding repairs.

[0017] Step 5: Data Integration and Intelligent Output. Integrate the dynamic coating area data calculated in the above steps with other project management systems (such as ERP and MES). Automatically generate coating area statistical reports by component, batch, type of work, and time dimension. Furthermore, based on parameters such as the paint's "volume solids content," the theoretical paint usage can be further calculated more precisely to guide procurement and cost control. The formula for calculating the theoretical paint usage M is:

[0018] Among them, A final The final dynamic coating area (m²) is... The effective area after deducting the area of ​​visually identified missed coatings is T, where T is the dry film thickness (μm). The report engine supports generating statistical views by component, production batch, work team, or date, and pushes them to the ERP / MES system via API. If the area of ​​missed coating exceeds the threshold, a rework order is automatically generated.

[0019] Secondly, an intelligent statistical system for implementing the above method is provided, comprising:

[0020] a) Model parsing module, used to interface with BIM design software, parse the boundary representation of the topology, and output the geometric and semantic information of the components;

[0021] b) Rule reasoning module, with a built-in paint-free area recognition rule engine, executes space deduction logic based on Boolean operations;

[0022] c) Point cloud processing and fusion module, used to process 3D scan data, realize the comparison between the design model and the actual model and the area correction calculation;

[0023] d) AI visual analysis module, integrating computer vision algorithms, supports back projection of coating status and defect quantification;

[0024] e) Data warehouse and reporting engine, storing full-process area data and generating multi-dimensional statistical reports.

[0025] Optionally, the system also includes an interface module for integration with ERP and MES systems, supporting the automatic push of quality warning work orders and usage reports.

[0026] The present invention has the following beneficial effects:

[0027] 1. Significantly improves the accuracy of coating area calculation: By integrating 3D scanning reconstruction and surface integration technology, it effectively corrects area errors caused by processing errors and design deviations, and realizes accurate area calculation for irregular components and complex nodes.

[0028] 2. Automated identification and subtraction of paint-free areas: Based on the built-in geometric rule library and spatial topology analysis, it automatically identifies various paint-free areas such as weld seams, bolt-covered surfaces, and assembly mating surfaces, greatly reducing the workload of manual annotation and improving the identification coverage and calculation efficiency.

[0029] 3. Establish a closed-loop traceability mechanism for coating quality: Align the visual recognition results with the three-dimensional model space to achieve accurate location and area quantification of coating defects, and support visual monitoring and accountability of construction quality.

[0030] 4. Supports refined cost control throughout the entire process: By dynamically updating coating area data and integrating it with paint parameters, it enables accurate prediction and statistics of paint usage, significantly improving the accuracy of material procurement and cost control.

[0031] 5. Promote the digitalization and intelligentization of construction management: The system is seamlessly integrated with management platforms such as MES and ERP to achieve full-process data connectivity and intelligent decision support from design, processing, painting to acceptance. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the system architecture according to an embodiment of the present invention;

[0033] Figure 2 This is a schematic diagram showing the total area of ​​the components and the deduction of unpainted areas in an embodiment of the present invention. Detailed Implementation

[0034] The following is in conjunction with the appendix Figures 1 to 2 The present invention provides a more detailed description of the specific implementation of a method and system for intelligent statistical analysis of paint area on steel structures based on the fusion of artificial intelligence and three-dimensional models.

[0035] The example uses the fireproof coating project of the steel structure of a large sports center as an application scenario.

[0036] Figure 1 This diagram illustrates the system's processing flow. The entire process begins with "All parts under the current component," followed by the "Concrete: Deduct the surface area of ​​the part embedded in the concrete" processing step. Next, the system categorizes the component based on its "part profile" and executes corresponding processing logic for different profile cross-sections. This flowchart clearly presents the different processing steps and key considerations for various cross-sectional parts during area extraction and paint-free area identification, fully demonstrating the system's automated processing path from parsing to deduction.

[0037] Figure 2The process of identifying unpainted areas is illustrated in the form of a geometric diagram. The diagram uses geometric features such as the boundaries of bounded polygons, extreme regions, and diagonal lengths to assist in spatial analysis and support the accurate identification and subtraction of unpainted areas such as bolt hole occlusion surfaces and assembly mating surfaces.

[0038] Specifically, Ωmax and Ωmin, marked in the figure, are located at the top and bottom of the polygon, respectively, representing the upper and lower boundaries of the region in the vertical direction; Xmin and Xmax are located on the left and right sides, representing the minimum and maximum boundary values ​​in the horizontal direction, together defining the spatial range of the polygon. In addition, the figure also marks the lengths Lmax and Lmin of the two diagonals inside the polygon. These parameters can be used to assist in calculating the area, analyzing shape features, and providing a basis for verifying geometric analysis and optimization algorithms.

[0039] This diagram clearly presents the mathematical characteristics of bounded regions through geometric figures and markings, making it suitable for the verification and optimization of spatial subtraction rules, thereby improving the accuracy and efficiency of unpainted region recognition.

[0040] Implementation steps:

[0041] Step 1: The Tekla design model is read through the model parsing module. The rule reasoning module automatically identifies all areas completely isolated from the outside air due to component assembly, the steel column base areas encased in concrete, and the contact areas of all high-strength bolted connection plates, and deducts these from the total area (corresponding to...). Figure 2 (Geometric subtraction logic in the context of geometry).

[0042] Step two: After the components are manufactured in the factory, a handheld 3D scanner is used to scan the complex node components. The point cloud processing and fusion module calculates the surface micro-variation area caused by the processing and updates the data.

[0043] Step 3, the painting stage: High-definition cameras are deployed in the painting area. The AI ​​visual analysis module automatically analyzes the images daily, counts the area painted that day, and detects minor missed areas. The system automatically generates an early warning work order.

[0044] Step four: All data is aggregated in the data warehouse, and project managers can view the painting area, progress percentage, and estimated paint usage for each component in real time, achieving refined management.

[0045] Implementation results:

[0046] The rule engine automatically covers over 98% of paint-free scenarios, requiring manual review to only involve two special nodes, reducing manual intervention workload by approximately 90%.

[0047] The traditional Tekla method had an error of +8.7% in calculating the area of ​​this node. After point cloud correction, the error was reduced to -1.1%, meeting the engineering accuracy control requirement of ±2% (error data comes from a third-party inspection report of a sports center project).

[0048] The recall rate for missing coating identification reached 98.6% (the recall rate was calculated based on statistics from a total of 1,000 images in the test set), and the defect localization achieved three-source alignment of "area-image-model", supporting accurate quality traceability;

[0049] Compared to the traditional method of purchasing based on theoretical area, the deviation in actual paint usage has decreased from an average of 7.5% to 2.8%, resulting in an estimated material cost saving of approximately RMB 1.08 million (annual savings calculated based on a total project area of ​​100,000 square meters). The reporting engine automatically generates multi-dimensional statistical views at the component, batch, and work team levels, and pushes them to the project's MES system via an interface, enabling real-time visual management of painting progress and quality.

[0050] This embodiment verifies the feasibility, accuracy improvement benefits, and management value of the present invention in real engineering scenarios. Through full-process data fusion and intelligent analysis, it achieves a leap from "manual estimation and static theory" to "automatic calculation and dynamic real-time data" in steel structure coating area statistics, providing a replicable digital solution for similar engineering projects.

[0051] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligently calculating the paint area of ​​steel structures based on the fusion of artificial intelligence and 3D models, characterized in that, Includes the following steps: Step 1: Parse the boundary representation model of the components from the steel structure BIM design software, and extract the part number, material, spatial location, and initial theoretical surface area Q of each part; Step 2: Based on the built-in geometric rule library, through spatial adjacency relationships and Boolean operations, automatically identify and deduct the total area Q′ of the unpainted area to obtain the theoretical net painted area. ; Step 3: After the component is fabricated, a laser scanner is used to collect surface point cloud data, reconstruct and generate a mesh model, and calculate the actual surface area based on Gaussian curvature numerical integration. The actual area is then used to replace the corresponding area in the design model, and the model is dynamically updated. Step 4: Deploy RGB-D cameras in the painting work area to acquire surface images containing depth information in real time and output pixel-level painting status images; back-project the images onto the 3D component model through camera calibration parameters to identify unpainted areas, missed areas, and drip defects, and quantify their actual painting area and spatial coordinates; Step 5: Associate the dynamically updated painting area and defect data with the paint volume solids sub-parameters to automatically generate statistical reports and theoretical paint usage by component, batch, type of work, or time dimension, and push quality warnings to the MES system.

2. The intelligent statistical method for steel structure paint area based on the fusion of artificial intelligence and three-dimensional model as described in claim 1, characterized in that: In step two, the unpainted area includes: a) the base material area within 150mm on both sides of the weld centerline; b) the shielding surface formed by extending 70mm outward from the high-strength bolt holes along three orthogonal directions; c) the embedded section completely encased in concrete; and d) the contact surface that is tightly fitted with other components and has no airflow in the final assembled state.

3. The intelligent statistical method for steel structure paint area based on the fusion of artificial intelligence and three-dimensional model as described in claim 2, characterized in that: The identification of the weld area is achieved by extracting the welding symbol line in the BIM model and offsetting it along the normal of the associated base material surface to generate a ring-shaped deduction area; the identification of the assembly mating surface is achieved by calculating the Boolean intersection of the contact surfaces of adjacent components and determining their visibility attributes in the final assembly model.

4. The intelligent statistical method for paint area of ​​steel structures based on the fusion of artificial intelligence and three-dimensional models according to claim 1, characterized in that: In step three, point cloud data processing includes denoising, ICP registration, and Poisson surface reconstruction, with the surface reconstruction resolution being no less than 2mm. For areas where the design model deviation is greater than 3mm, the actual area is used to replace the theoretical value.

5. The intelligent statistical method for paint area of ​​steel structures based on the fusion of artificial intelligence and three-dimensional models according to claim 1, characterized in that, In step four, the visual model used is the U-Net++ architecture, the training dataset includes synthetic missed paint samples and real drip images collected on site, the loss function is Focal Loss, and the missed paint recognition recall rate is not less than 98%.

6. The intelligent statistical method for paint area of ​​steel structures based on the fusion of artificial intelligence and three-dimensional models according to claim 1, characterized in that, In step five, the formula for calculating the theoretical paint consumption M is: ， Among them, A final The final dynamic coating area is in square meters, T is the dry film thickness in micrometers, and Vs is the paint volume solids content in percentage.

7. An intelligent statistical system implementing the method as described in any one of claims 1-6, characterized in that, include: a) Model parsing module, used to interface with BIM design software, parse the boundary representation topology, and output the geometric and semantic information of components; b) Rule reasoning module, with a built-in rule engine for identifying paint-free areas, executes spatial subtraction logic based on Boolean operations; c) Point cloud processing and fusion module, used to process 3D scan data, realize the comparison between the design model and the actual model and the area correction calculation; d) AI visual analysis module, integrating computer vision algorithms, supports back projection of painting status and defect quantification; e) Data warehouse and reporting engine, storing full-process area data and generating multi-dimensional statistical reports.

8. The intelligent statistical system according to claim 7, characterized in that, The BIM design software is Tekla software, and the system reads model data through Open API.

9. The system according to claim 7, characterized in that, It also includes interface modules for integration with ERP and MES systems, supporting the automatic push of quality early warning work orders and usage reports.

10. The intelligent statistical system according to claim 7, characterized in that, When processing steel component nodes, the point cloud processing and fusion module has a surface reconstruction resolution of no less than 2mm, ensuring the area capture accuracy of tiny cast fillets.

Images