An AI-driven digital design method and system for intelligent ship coating engineering

Abstract

The application discloses an AI-driven ship intelligent coating engineering digital design method and system, and the method comprises the following steps: constructing a structured coating engineering definition library and a multi-dimensional dynamic knowledge base; analyzing and enhancing the semantics of a three-dimensional design model to generate an enhanced three-dimensional coating model; calling a clustering algorithm fused with process constraints to map a target coating component into a high-dimensional feature vector and automatically generate an intelligent work package; calculating the coating area based on the enhanced three-dimensional coating model and estimating the working hours; generating a structured digital work order for each intelligent work package and outputting an interactive document or a mobile terminal data package to directly reach a construction site terminal; receiving actual data fed back from the construction site, re-fitting the data by using a machine learning algorithm, and updating the optimized parameters to the multi-dimensional dynamic knowledge base; and the application solves the problems of relying on experience, data islands, extensive calculation and inability to continuously optimize in traditional coating design.

Description

Technical Field

[0001] This invention belongs to the fields of intelligent ship manufacturing, digital twins and industrial artificial intelligence, and in particular relates to an AI-driven digital design method and system for intelligent ship painting engineering. Background Technology

[0002] Currently, ship painting is a crucial part of shipbuilding, directly impacting a ship's corrosion resistance, pollution prevention, appearance, and overall lifecycle maintenance costs. With the increasing size and complexity of ships and rising demands for green and environmentally friendly practices, traditional painting processes face the following challenges in design, planning, and execution.

[0003] First, design planning relies heavily on manual experience, resulting in low efficiency and poor consistency. Coating process planning (such as coating area division, film thickness distribution, construction stage division, and material and time estimation) is highly dependent on the individual experience of process personnel. Planning schemes vary significantly between different projects and personnel, making standardization and knowledge accumulation difficult, leading to long planning cycles and difficulty in quickly responding to design changes or ship type variations. Second, design and construction information are disconnected, hindering accurate on-site guidance. Traditional coating process documents (such as process cards and drawings) are mostly two-dimensional and textual descriptions with weak correlation to three-dimensional design models. On-site construction personnel struggle to accurately map two-dimensional drawings onto the complex three-dimensional structure of the ship, easily causing misunderstandings and affecting construction accuracy and quality. Significant information attenuation occurs in the transmission from design intent to construction instructions. Third, the construction process is "black box," lacking real-time data-driven dynamic optimization. Current construction management relies heavily on manual reports and periodic inspections, resulting in delayed data collection and strong subjectivity. Data such as actual construction progress, material consumption, environmental conditions (temperature and humidity), equipment status, and quality inspection results are not fed back in real-time and fail to form a closed loop with the design and planning stages. Therefore, it is impossible to promptly identify schedule deviations, quality risks, or resource waste during construction and dynamically adjust subsequent plans. Fourth, process knowledge is tacit, making it difficult to iterate, optimize, and pass on. Excellent construction experience and process know-how specific to certain regions or environments often exist as tacit knowledge in individuals' minds, lacking effective digital and structured expression and management methods. This makes it difficult for enterprises to systematically accumulate process knowledge bases, hinders the continuous optimization of process parameters through data mining, and also exposes them to the risk of experience loss.

[0004] Therefore, existing digital tools have significant functional limitations. 3D design software is primarily used for static visualization and drawing, unable to achieve dynamic correlation and driving between process models, construction parameters, and management data. Manufacturing execution systems focus on task assignment and schedule management, lacking underlying process logic and knowledge base support deeply integrated with 3D design models, resulting in low levels of intelligent decision-making. The industry has yet to build a fully integrated digital twin environment encompassing 3D coating design, intelligent process planning, dynamic construction simulation, on-site execution, and quality feedback.

[0005] In summary, the existing technology system is static, fragmented, and passive, and cannot form an intelligent coating engineering digital twin that is centered on data and knowledge integration, uses 3D models as a carrier, is driven by AI for decision-making, and achieves closed-loop optimization of the entire process. Summary of the Invention

[0006] Purpose of the invention: The purpose of this invention is to provide an AI-driven digital design method and system for intelligent ship painting engineering that can improve construction efficiency and resource utilization.

[0007] Technical solution: The present invention provides an AI-driven digital design method for intelligent ship painting engineering, comprising:

[0008] Obtain ship specifications, historical data, and ship 3D design models; construct a structured coating engineering definition library and a multi-dimensional dynamic knowledge base, establish a multi-level coding system, and convert the time correction coefficient and environmental coefficient into machine-readable feature vectors;

[0009] The three-dimensional design model is analyzed and semantically enhanced. The geometric parameters of the model surface are calculated, and a deep learning model is used to perform semantic segmentation and classification on the model surface to generate an enhanced three-dimensional coating model.

[0010] A clustering algorithm that integrates process constraints is invoked to map the target coating components into high-dimensional feature vectors. A multi-objective optimization algorithm is used to comprehensively optimize the rationality of the process, construction efficiency, and resource balance, and an intelligent work package is automatically generated. The coating area is calculated based on the enhanced three-dimensional coating model, a dynamic loss coefficient is introduced to calculate the material consumption, and an AI regression model is used to estimate the working hours.

[0011] A structured digital work order is generated for each intelligent work package, integrating a three-dimensional visualization view and corresponding construction parameters, and outputting it as an interactive document or mobile data package, which is directly delivered to the construction site terminal.

[0012] The system receives actual data from the construction site, uses machine learning algorithms to refit the data, and updates the optimized parameters to the multidimensional dynamic knowledge base.

[0013] Furthermore, the process of parsing and semantically enhancing the three-dimensional design model, calculating the geometric parameters of the model surface, and using a deep learning model to perform semantic segmentation and classification of the model surface to generate an enhanced three-dimensional coating model includes: importing the ship's three-dimensional design model, performing lightweight processing and feature extraction; parsing the model geometry using boundary representation methods, calculating the surface area and curvature; using a deep learning model to perform semantic segmentation and classification of the model surface, automatically identifying and labeling the surface type and construction environment, and generating an enhanced three-dimensional coating model.

[0014] Furthermore, the deep learning model is a point cloud neural network model, which is used to perform semantic segmentation on the point cloud data of the 3D design model to identify the various painting areas of the hull structure, outfitting components and piping system.

[0015] Furthermore, the clustering algorithm that integrates process constraints maps the target coating components into high-dimensional feature vectors. A multi-objective optimization algorithm is then used to comprehensively optimize process rationality, construction efficiency, and resource balance, automatically generating intelligent work packages, including:

[0016] Map painted components to high-dimensional feature vectors;

[0017] An optimization objective function is defined, which includes minimizing process changeover costs, maximizing construction continuity, and balancing resource load.

[0018] Perform evolutionary computation using the improved non-dominated sorting genetic algorithm NSGA-II, and output the Pareto optimal solution set to generate the smart job package.

[0019] Furthermore, the calculation of material usage specifically includes:

[0020] Calculation of coating area based on triangular mesh integration;

[0021] The AI ​​prediction target model is invoked, and the curvature of the work package, environmental wind speed and humidity parameters are input to predict the dynamic loss coefficient.

[0022] The formula for calculating material requirements is as follows:

[0023] Material usage = (area × dry film thickness) / (10 × volume solids × dynamic loss coefficient).

[0024] Furthermore, the step of generating a structured digital work order for each intelligent work package, integrating a 3D visualization view and corresponding process parameters, and outputting as an interactive document or mobile data package includes: dynamically generating a structured digital work order for each intelligent work package; automatically extracting the corresponding area in the 3D model to generate a highlighted view, and outputting corresponding construction parameters based on a multi-dimensional dynamic knowledge base; supporting output as an interactive 3D PDF, a mobile APP data package, or AR spatial positioning instructions, directly reaching the construction site terminal.

[0025] Furthermore, the process of receiving actual data from the construction site, refitting it using machine learning algorithms, and updating the optimized parameters to the multidimensional dynamic knowledge base includes:

[0026] Collect feedback data from the construction site to construct a training dataset. The features of the training dataset include structural complexity and construction accessibility, and the labels are actual loss ratio or actual working hours.

[0027] The gradient boosting tree algorithm is used to train the training dataset to fit the mapping relationship between the design parameters and the construction results;

[0028] Update the dynamic loss coefficient and working hours in the multidimensional dynamic knowledge base based on the training results.

[0029] Based on the same inventive concept, this invention also provides an AI-driven digital design system for intelligent ship painting engineering, comprising:

[0030] The knowledge base construction module is used to acquire ship specifications, historical data and ship 3D design models; to build a structured coating engineering definition library and a multi-dimensional dynamic knowledge base, to establish a multi-level coding system, and to convert the time correction coefficient and environmental coefficient into machine-readable feature vectors;

[0031] The model processing module is used to parse and semantically enhance the three-dimensional design model, calculate the geometric parameters of the model surface, and use a deep learning model to perform semantic segmentation and classification on the model surface to generate an enhanced three-dimensional coating model.

[0032] The intelligent planning module is used to call a clustering algorithm that integrates process constraints to map the target coating components into high-dimensional feature vectors. Through a multi-objective optimization algorithm, it comprehensively optimizes the rationality of the process, construction efficiency, and resource balance, and automatically generates intelligent work packages. Based on the enhanced three-dimensional coating model, the coating area is calculated, a dynamic loss coefficient is introduced to calculate the material consumption, and an AI regression model is used to estimate the working hours.

[0033] The instruction output module is used to generate a structured digital work order for each intelligent work package, integrate a three-dimensional visualization view and corresponding construction parameters, and output as an interactive document or mobile data package, which is directly delivered to the construction site terminal.

[0034] The closed-loop optimization module is used to receive actual data from the construction site, refit the data using machine learning algorithms, and update the optimized parameters to the multidimensional dynamic knowledge base.

[0035] Based on the same inventive concept, the present invention also provides a computing device, comprising: one or more processors, one or more memories, and one or more programs, the programs being stored in the memory and configured to be executed by the processor, wherein when the programs are loaded onto the processor, they implement the steps of the AI-driven digital design method for intelligent ship painting engineering according to any of the preceding claims.

[0036] Based on the same inventive concept, the present invention also provides a storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, cause the processor to perform the steps of the AI-driven digital design method for intelligent ship painting engineering according to any one of the preceding claims.

[0037] Beneficial effects: Compared with existing technologies, this invention achieves intelligent decomposition and multi-objective optimization of painting operations through AI algorithms, significantly improving construction efficiency and resource utilization; adopts dynamic loss coefficients and AI regression models to greatly improve the accuracy of material and time calculations; establishes a construction feedback closed loop to achieve continuous self-field information barriers for the system and reduce management costs. Attached Figure Description

[0038] Figure 1 This is a flowchart of the method according to the first embodiment of the present invention;

[0039] Figure 2 This is a flowchart of the method according to the second embodiment of the present invention. Detailed Implementation

[0040] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0041] Example 1, as Figure 1 As shown in this embodiment, the AI-driven digital design method for intelligent ship painting engineering includes:

[0042] Step (1) Obtain ship specifications, historical data and ship 3D design model; construct a structured coating engineering definition library and a multi-dimensional dynamic knowledge base, establish a multi-level coding system, and convert the time correction coefficient and environmental coefficient into machine-readable feature vectors;

[0043] Step (2) Analyze and semantically enhance the three-dimensional design model, calculate the geometric parameters of the model surface, and use a deep learning model to perform semantic segmentation and classification on the model surface to generate an enhanced three-dimensional coating model;

[0044] Step (3) calls the clustering algorithm that integrates process constraints to map the target coating component into a high-dimensional feature vector. Through a multi-objective optimization algorithm, the rationality of the process, construction efficiency and resource balance are optimized in a coordinated manner, and an intelligent work package is automatically generated. The coating area is calculated based on the enhanced three-dimensional coating model, the material consumption is calculated by introducing a dynamic loss coefficient, and the working hours are estimated by using an AI regression model.

[0045] Step (4) Generate a structured digital work order for each intelligent work package, integrate a three-dimensional visualization view and corresponding construction parameters, and output it as an interactive document or mobile data package, which is directly delivered to the construction site terminal.

[0046] Step (5) Receive actual data from the construction site, refit using machine learning algorithms, and update the optimized parameters to the multidimensional dynamic knowledge base.

[0047] Specifically, in step (1), a structured coating engineering definition library and a multi-dimensional dynamic knowledge base are constructed. Based on ship specifications, contract specifications and historical project data, a parameterized coating engineering definition library is established to define elements such as coating system, dry / wet film thickness, and rust removal grade, and to establish a strong association with the ship's three-dimensional structure tree. At the same time, a multi-dimensional dynamic knowledge base is constructed to encapsulate material properties, construction process windows and verified mathematical calculation models, and to convert time correction coefficients and environmental coefficients into machine-readable feature vectors.

[0048] Among them, the structured coating engineering definition library adopts a graph structure association, inheriting the multi-level coding system of "region-cabin-part-component" to ensure a strong association between the coating system and the ship's structure tree.

[0049] In step (2), the three-dimensional design model is analyzed and semantically enhanced, the geometric parameters of the model surface are calculated, and the model surface is semantically segmented and classified using a deep learning model to generate an enhanced three-dimensional coating model. This includes: importing the ship three-dimensional design model, performing lightweight processing and feature extraction; analyzing the model geometry using boundary representation method, calculating the surface area and curvature; using a deep learning model to perform semantic segmentation and classification on the model surface, automatically identifying and labeling the surface type, construction accessibility and corrosion environment level, and generating an enhanced three-dimensional coating model.

[0050] Among them, the deep learning model uses PointNet++ or graph neural network (GNN) to achieve accurate semantic segmentation of complex curved surface components.

[0051] In step (3), the clustering algorithm that integrates process constraints is invoked to map the target coating component into a high-dimensional feature vector. A multi-objective optimization algorithm is then used to comprehensively optimize process rationality, construction efficiency, and resource balance, automatically generating an intelligent work package, including:

[0052] The coated components are mapped as high-dimensional feature vectors that include geometric features, process features, and resource features.

[0053] An optimization objective function is defined, which includes minimizing process changeover costs, maximizing construction continuity, and balancing resource load.

[0054] Perform evolutionary computation using the improved non-dominated sorting genetic algorithm NSGA-II, and output the Pareto optimal solution set to generate the smart job package.

[0055] The target coating components include hull plating, bulkheads, stiffeners, piping systems, outfitting components, etc.

[0056] The calculation of material usage specifically includes:

[0057] Calculation of coating area based on triangular mesh integration;

[0058] The AI ​​prediction target model is invoked, and the curvature of the work package, environmental wind speed and humidity parameters are input to predict the dynamic loss coefficient.

[0059] The formula for calculating material requirements is as follows:

[0060] Material usage = (area × dry film thickness) / (10 × volume solids × dynamic loss coefficient).

[0061] The initial value of the dynamic loss coefficient is derived from the knowledge base and is dynamically updated in step (5) based on construction feedback data.

[0062] Step (4) includes: dynamically generating a structured digital work order for each smart work package; automatically extracting the corresponding area in the 3D model to generate a highlighted view; and outputting the corresponding construction parameters based on the multi-dimensional dynamic knowledge base; supporting output as an interactive 3D PDF, a mobile APP data package, or an AR spatial positioning command, which can be directly delivered to the construction site terminal.

[0063] The digital work order integrates a 3D visualization view, engineering specifications, bill of materials, and recommended process parameters, and supports access from multiple terminals.

[0064] In step (5), the construction site transmits real-time data on actual membrane thickness, temperature and humidity, actual working hours, and material consumption via IoT devices and mobile terminals; the system automatically compares the design predictions with the actual measured values, uses machine learning algorithms to refit the data, and updates the optimized parameters to the multidimensional dynamic knowledge base, including:

[0065] Collect feedback data from the construction site to construct a training dataset. The features of the training dataset include structural complexity and construction accessibility, and the labels are actual loss ratio or actual working hours.

[0066] The gradient boosting tree algorithm is used to train the training dataset to fit the mapping relationship between the design parameters and the construction results;

[0067] Update the dynamic loss coefficient and working hours in the multidimensional dynamic knowledge base based on the training results.

[0068] Example 2, as Figure 2 As shown in this embodiment, the AI-driven digital design method for intelligent ship painting engineering includes:

[0069] Step 1: Construct a structured coating engineering definition library and a multi-dimensional dynamic knowledge base;

[0070] Step 1-1: In the Engineering Definition and Knowledge Management module, the user builds a parameterized coating engineering definition library based on ship contracts, technical specifications and historical project data, defines elements such as coating system, dry / wet film thickness, and rust removal grade, and establishes a strong association with the ship's three-dimensional structure tree;

[0071] Steps 1-2 involve simultaneously building a multi-dimensional dynamic knowledge base, encapsulating material characteristic parameters (volume solids content, theoretical coating rate, etc.), typical construction cases, and verified process rules, and transforming time correction coefficients and environmental / quality / safety management items into machine-readable structured rules and feature vectors.

[0072] Step 2: Lightweighting of the 3D model and intelligent enhancement of semantic information;

[0073] Step 2-1: Import the ship's 3D design model through the 3D model intelligent processing module for lightweighting and feature extraction;

[0074] Step 2-2: The system uses boundary representation to analyze the model geometry, calculates the surface area and curvature, and uses a deep learning model to perform semantic segmentation and classification on the model surface. It automatically identifies and labels the flat outer plate, curved bulkhead, dense reinforcement material area, etc., and marks its construction accessibility level to generate an enhanced three-dimensional coating model.

[0075] Step 3: AI-driven intelligent generation and optimization of painting work packages;

[0076] Step 3-1: The AI ​​intelligent design engine module is started, calling the clustering model and multi-objective optimization algorithm that integrates business rules. It not only considers geometric adjacency, but also optimizes multiple objectives such as "coating consistency", "balanced daily workload of construction teams" and "reducing the number of robot turns", automatically dividing the entire ship's coating surface into several optimal work packages.

[0077] Java

[0078] / / Pseudocode for intelligent job package partitioning based on NSGA-II

[0079] public class CoatingJobPackOptimizer {

[0080] public List <jobpackage>optimize(List <coatingelement>elements) {

[0081] / / Map painted components to high-dimensional feature vectors [geometric features, process features, resource features]

[0082] double[][] featureMatrix = extractFeatures(elements);

[0083] / / Set multi-objective optimization parameters: minimize process changeover, maximize continuity, balance load

[0084] Problem problem = new CoatingJobPackProblem(featureMatrix);

[0085] Algorithm algorithm = new NSGA2(problem);

[0086] / / Perform evolutionary calculations

[0087] SolutionSet solutions = algorithm.execute(200); / / 200 generations of evolution

[0088] / / Output the Pareto optimal solution set and generate 3 job packages (such as parallel midships, head surface, and dense structure region).

[0089] return selectBestJobPackages(solutions);

[0090] }

[0091] }

[0092] Step 3-2: Subsequently, calculate the area for each work package, associate the coating requirements with the structured coating engineering definition library, and call the material calculation model [actual paint usage = area × dry film thickness / (10 × volume solids × 100 × current loss factor)] from the knowledge base to calculate accurate material requirements and time estimates;

[0093] Java

[0094] / / Pseudocode for dynamic loss coefficient calculation and resource estimation

[0095] public class ResourceCalculator {

[0096] private DynamicLossModel lossModel;

[0097] public CalculationResult calculate(JobPackage jobPackage) {

[0098] / / Calculate coating area based on precise integration using a triangular mesh

[0099] double area = calculateMeshArea(jobPackage.getGeometry());

[0100] / / Use an AI model to predict dynamic loss coefficients (considering curvature, wind speed, humidity, etc.)

[0101] double dynamicLossRatio = lossModel.predict(

[0102] jobPackage.getCurvature(),

[0103] jobPackage.getWindSpeed(),

[0104] jobPackage.getHumidity() );

[0106] / / Accurately calculate material usage

[0107] double materialVolume = (area * jobPackage.getDFT()) /

[0108] (10 * jobPackage.getVolumeSolid() * dynamicLossRatio);

[0109] / / AI regression model estimates working time (considering area, complexity, and correction factors)

[0110] double manHours = estimateManHours(area, jobPackage.getComplexity());

[0111] return new CalculationResult(materialVolume, manHours);

[0112] }

[0113] }

[0114] Step 4: Digital process instruction generation and multimodal output;

[0115] Step 4-1: The process planning and output module generates a digital process card for each work package and automatically extracts the corresponding area from the 3D model to generate a highlighted view.

[0116] Step 4-2: The card integrates the 3D highlight map of the area and lists the AI ​​recommended sequence of "sandblasting and rust removal → cleaning → primer spraying → intermediate paint spraying → topcoat spraying" in a structured table. Each process is automatically associated with material specifications, calculated usage, recommended spraying equipment (such as robot model) and process parameters (pump pressure, speed) matched from the robot knowledge base.

[0117] Step 4-3 supports output as interactive 3D PDF, mobile APP data package, and AR spatial positioning instructions, which can be directly delivered to the construction site terminal.

[0118] Step 5: Construction feedback and AI optimization closed loop;

[0119] Step 5-1: The instruction is sent to the tablet computer on site. After construction, the quality inspector reports the measured film thickness, actual working hours and consumable data on the terminal. The data is then sent back to the construction feedback and data analysis module.

[0120] Step 5-2: The AI ​​optimization and knowledge learning module analysis found that for specific structural areas (such as densely reinforced material areas), the actual working hours were consistently higher than the initial system estimate. The machine learning model was automatically retrained to analyze the correlation between the characteristics of the area (structural complexity, accessibility labels) and the actual working hours. The working hour estimation sub-model for this type of characteristic area was dynamically adjusted, and the corresponding correction rules in the knowledge base were updated.

[0121] Java

[0122] / / Pseudocode for construction feedback and online learning of AI models

[0123] public class DynamicLossModelTrainer {

[0124] public void updateModel(List <constructionfeedback>feedbackData) {

[0125] / / Prepare the training dataset (features: structural complexity, reachability; labels: actual loss ratio)

[0126] Dataset <row>trainingData = prepareTrainingData(feedbackData);

[0127] / / Configure a gradient boosting tree regression model (XGBoost / LightGBM)

[0128] GBTRegressor gbt = new GBTRegressor()

[0129] .setLabelCol("actual_loss_ratio")

[0130] .setFeaturesCol("features")

[0131] .setMaxIter(100);

[0132] / / Train the model to fit the hidden patterns in the "design parameters - construction results" relationship.

[0133] GBTRegressionModel model = gbt.fit(trainingData);

[0134] / / Update the dynamic loss coefficient and working time correction rules in the knowledge base

[0135] updateKnowledgeBase(model);

[0136] }

[0137] }

[0138] Step 5-3 involves feeding the optimized parameters back into the knowledge base to achieve continuous closed-loop improvement in design quality and autonomous evolution of system intelligence, such as... Figure 1 The feedback arrows in S5 point to S1 and S3 as shown.

[0139] Example 3: The following describes this implementation method in detail with reference to the coating engineering design of a ballast tank of a certain type of bulk carrier. In the engineering definition and model processing stage, the user selects and configures a coating system conforming to the PSPC standard for "No.1 Ballast Tank" in the engineering definition and knowledge management module, and structurally inputs parameters such as the dry film thickness and surface cleanliness level of each coating layer. The 3D model of the tank segment is imported through the 3D model intelligent processing module. The system uses algorithms to automatically identify and label the flat outer plates, curved bulkheads, and densely reinforced areas inside the tank, and marks their accessibility levels.

[0140] In the AI-powered intelligent design phase, the AI ​​intelligent design engine module is activated. The algorithm library calls a clustering model that integrates business rules, considering not only geometric adjacency but also optimizing multiple objectives such as "coating consistency," "balanced daily workload for construction teams (approximately 150 square meters)," and "reducing robot turning frequency." This automatically divides the ballast tank into three optimal work packages (e.g., "parallel midships straight area," "head curved surface area," and "dense structure area"). Subsequently, the area of ​​each work package is calculated, and coating requirements are correlated with the engineering definition library. Material calculation models are retrieved from the knowledge base to calculate accurate material requirements and work time estimates.

[0141] During the instruction generation and output phase, the process planning and output module generates digital process cards for the "parallel midships straight area operation package". The card integrates a 3D highlighted map of the area and lists the AI ​​recommended sequence of "sandblasting and rust removal → cleaning → primer spraying → intermediate paint spraying → topcoat spraying" in a structured table. Each process is automatically associated with material specifications, calculated usage, recommended spraying equipment (such as robot model), and process parameters (pump pressure, speed) matched from the robot knowledge base.

[0142] In the construction feedback and AI optimization closed-loop phase, instructions are issued to on-site tablets. After construction, quality inspectors report the measured membrane thickness, actual working hours, and material consumption data on their terminals. This data is then fed back to the construction feedback and data analysis module. The AI ​​optimization and knowledge learning module analyzes the data and finds that for "dense reinforcement areas," the actual working hours consistently exceed the system's initial estimates. This module automatically initiates retraining of the machine learning model, analyzes the correlation between the area's characteristics (structural complexity, accessibility labels) and the actual working hours, dynamically adjusts the working hour estimation sub-model for this type of characteristic area, and updates the corresponding correction rules in the knowledge base. Subsequently, for new designs with similar structural characteristics, the system will automatically use the optimized model for estimation, achieving continuous closed-loop improvement in design quality and autonomous evolution of system intelligence.

[0143] Example 4, based on the same inventive concept, also provides an AI-driven digital design system for intelligent ship painting engineering, including:

[0144] The knowledge base construction module is used to acquire ship specifications, historical data and ship 3D design models; construct a structured coating engineering definition library and a multi-dimensional dynamic knowledge base, establish a multi-level coding system, and convert the time correction coefficient and environmental coefficient into machine-readable feature vectors;

[0145] The model processing module is used to parse and semantically enhance the three-dimensional design model, calculate the geometric parameters of the model surface, and use a deep learning model to perform semantic segmentation and classification on the model surface to generate an enhanced three-dimensional coating model.

[0146] The intelligent planning module is used to call a clustering algorithm that integrates process constraints to map the target coating components into high-dimensional feature vectors. Through a multi-objective optimization algorithm, it comprehensively optimizes the rationality of the process, construction efficiency, and resource balance, and automatically generates intelligent work packages. Based on the enhanced three-dimensional coating model, the coating area is calculated, a dynamic loss coefficient is introduced to calculate the material consumption, and an AI regression model is used to estimate the working hours.

[0147] The instruction output module is used to generate a structured digital work order for each intelligent work package, integrate a three-dimensional visualization view and corresponding construction parameters, and output as an interactive document or mobile data package, which is directly delivered to the construction site terminal.

[0148] The closed-loop optimization module is used to receive actual data from the construction site, refit the data using machine learning algorithms, and update the optimized parameters to the multidimensional dynamic knowledge base.

[0149] Example 5, based on the same inventive concept, also provides a computing device, including: one or more processors, one or more memories, and one or more programs, the programs being stored in the memory and configured to be executed by the processor, the programs being loaded onto the processor to implement the steps of the AI-driven digital design method for intelligent ship painting engineering according to any of the above.

[0150] Example 6, based on the same inventive concept, this embodiment also provides a storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, cause the processor to perform the steps of the AI-driven digital design method for intelligent ship painting engineering according to any of the above claims.

[0151] The above embodiments are merely illustrative of the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Therefore, any modifications made to the technical solution of the present invention based on the technical concept proposed in this invention shall fall within the scope of protection of the present invention.< / row> < / constructionfeedback> < / coatingelement> < / jobpackage>

Claims

1. An AI-driven digital design method for intelligent ship painting engineering, characterized in that, include: Obtain ship specifications, historical data, and 3D design models of ships; Construct a structured coating engineering definition library and a multi-dimensional dynamic knowledge base, establish a multi-level coding system, and transform the time correction coefficient and environmental coefficient into machine-readable feature vectors; The three-dimensional design model is analyzed and semantically enhanced. The geometric parameters of the model surface are calculated, and a deep learning model is used to perform semantic segmentation and classification on the model surface to generate an enhanced three-dimensional coating model. By invoking a clustering algorithm that integrates process constraints, the target coating components are mapped into high-dimensional feature vectors. Through a multi-objective optimization algorithm, the rationality of the process, construction efficiency, and resource balance are optimized in a coordinated manner, and intelligent work packages are automatically generated. The coating area is calculated based on the enhanced three-dimensional coating model, a dynamic loss coefficient is introduced to calculate the material consumption, and an AI regression model is used to estimate the working hours. A structured digital work order is generated for each intelligent work package, integrating a three-dimensional visualization view and corresponding construction parameters, and outputting it as an interactive document or mobile data package, which is directly delivered to the construction site terminal. The system receives actual data from the construction site, uses machine learning algorithms to refit the data, and updates the optimized parameters to the multidimensional dynamic knowledge base.

2. The AI-driven digital design method for intelligent ship painting engineering according to claim 1, characterized in that, The process of parsing and semantically enhancing the 3D design model, calculating the geometric parameters of the model surface, and using a deep learning model to perform semantic segmentation and classification of the model surface to generate an enhanced 3D coating model includes: importing the ship's 3D design model, performing lightweight processing and feature extraction; parsing the model geometry using boundary representation methods, calculating the surface area and curvature; using a deep learning model to perform semantic segmentation and classification of the model surface, automatically identifying and labeling the surface type and construction environment, and generating an enhanced 3D coating model.

3. The AI-driven digital design method for intelligent ship painting engineering according to claim 1, characterized in that, The deep learning model is a point cloud neural network model, which is used to perform semantic segmentation on the point cloud data of the 3D design model to identify the various painting areas of the hull structure, outfitting components and piping system.

4. The AI-driven digital design method for intelligent ship painting engineering according to claim 1, characterized in that, The clustering algorithm that integrates process constraints maps the target coating components into high-dimensional feature vectors. A multi-objective optimization algorithm comprehensively optimizes process rationality, construction efficiency, and resource balance, automatically generating intelligent work packages, including: Map painted components to high-dimensional feature vectors; An optimization objective function is defined, which includes minimizing process changeover costs, maximizing construction continuity, and balancing resource load. Perform evolutionary computation using the improved non-dominated sorting genetic algorithm NSGA-II, and output the Pareto optimal solution set to generate the smart job package.

5. The AI-driven digital design method for intelligent ship painting engineering according to claim 1, characterized in that, The calculation of material usage specifically includes: Calculation of coating area based on triangular mesh integration; The AI ​​prediction target model is invoked, and the curvature of the work package, environmental wind speed and humidity parameters are input to predict the dynamic loss coefficient. The formula for calculating material requirements is as follows: Material usage = (area × dry film thickness) / (10 × volume solids × dynamic loss coefficient).

6. The AI-driven digital design method for intelligent ship painting engineering according to claim 1, characterized in that, The process of generating a structured digital work order for each intelligent work package, integrating a 3D visualization view and corresponding process parameters, and outputting as an interactive document or mobile data package includes: dynamically generating a structured digital work order for each intelligent work package; automatically extracting the corresponding area from the 3D model to generate a highlighted view; and outputting corresponding construction parameters based on a multi-dimensional dynamic knowledge base; supporting output as an interactive 3D PDF, a mobile APP data package, or AR spatial positioning instructions, directly reaching the construction site terminal.

7. The AI-driven digital design method for intelligent ship painting engineering according to claim 1, characterized in that, The process of receiving actual data from the construction site, refitting it using machine learning algorithms, and updating the optimized parameters to the multidimensional dynamic knowledge base includes: Collect feedback data from the construction site to construct a training dataset. The features of the training dataset include structural complexity and construction accessibility, and the labels are actual loss ratio or actual working hours. The gradient boosting tree algorithm is used to train the training dataset to fit the mapping relationship between the design parameters and the construction results; Update the dynamic loss coefficient and working hours in the multidimensional dynamic knowledge base based on the training results.

8. An AI-driven digital design system for intelligent ship painting engineering, characterized in that, include: The knowledge base construction module is used to acquire ship specifications, historical data, and ship 3D design models; Construct a structured coating engineering definition library and a multi-dimensional dynamic knowledge base, establish a multi-level coding system, and transform the time correction coefficient and environmental coefficient into machine-readable feature vectors; The model processing module is used to parse and semantically enhance the three-dimensional design model, calculate the geometric parameters of the model surface, and use a deep learning model to perform semantic segmentation and classification on the model surface to generate an enhanced three-dimensional coating model. The intelligent planning module is used to call a clustering algorithm that integrates process constraints to map the target coating components into high-dimensional feature vectors. Through a multi-objective optimization algorithm, it comprehensively optimizes the rationality of the process, construction efficiency and resource balance, and automatically generates intelligent work packages. The coating area is calculated based on the enhanced three-dimensional coating model, a dynamic loss coefficient is introduced to calculate the material consumption, and an AI regression model is used to estimate the working hours. The instruction output module is used to generate a structured digital work order for each intelligent work package, integrate a three-dimensional visualization view and corresponding construction parameters, and output as an interactive document or mobile data package, which is directly delivered to the construction site terminal. The closed-loop optimization module is used to receive actual data from the construction site, refit the data using machine learning algorithms, and update the optimized parameters to the multidimensional dynamic knowledge base.

9. A computing device, characterized in that, include: One or more processors, one or more memories, and one or more programs, said programs being stored in the memory and configured to be executed by the processor, said programs being loaded onto the processor to implement the steps of the AI-driven digital design method for intelligent ship painting engineering according to any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores a computer program, which includes program instructions that, when executed by a processor, cause the processor to perform the steps of the AI-driven digital design method for intelligent ship painting engineering according to any one of claims 1 to 7.

Images