A method and device for supporting propulsion determination for AI-assisted CAD modification

Abstract

The application discloses a kind of propelling judgment support method and device for AI assisted CAD modification, belong to computer aided design and intelligent interaction technical field.The method includes: first, local editing context is constructed and area constraint is divided, then matching operator is generated Modification direction and cluster screening, selected direction is constructed patch script and generates candidate scheme, preliminary and in-depth stage verification is carried out after gathering detection data, and integration context, direction and verification data, recommended total score is calculated using weighting algorithm, and structured result is output and encapsulated can be submitted modification object.The application constructs local editing context, divides target, protection and influence area, combines editing operator and cluster screening modification direction, improves targeting, generates candidate scheme by parameterized search, carries out multi-dimensional verification by geometry, assembly, manufacturing, integrates decision using weighting algorithm, outputs structured recommendation scheme, realizes scientific screening and standard execution of CAD modification, improves efficiency and controllability.

Description

Technical Field

[0001] This invention belongs to the field of computer-aided design and intelligent interaction technology, specifically relating to a method and apparatus for advancing judgment support for AI-assisted CAD modification. Background Technology

[0002] With the deep integration of industrial design and intelligent manufacturing, CAD models have become the core digital carrier for product design, assembly verification, and processing production. Market demands for product iteration efficiency continue to rise, leading to an increase in the frequency and complexity of local modifications to CAD models. The combination of AI technology and CAD editing has become a significant industry trend. Complex CAD models contain multi-level geometric topologies, assembly interfaces, adjacency relationships, and verification dependencies. Local modifications can easily have a ripple effect, requiring precise area division, constraint control, and logical judgment to achieve compliant editing. Currently, the industry urgently needs a complete AI-assisted CAD modification advancement judgment system to complete semantic parsing of user modification requests, construction of local editing scenarios, intelligent screening of modification directions, automatic generation of candidate solutions, and multi-dimensional verification. This system would form a systematic modification support logic, adapting to diverse model formats and editing request types, becoming a key technological direction for improving the efficiency and quality of CAD editing.

[0003] Traditional CAD model modification relies primarily on manual work, lacking an intelligent local editing context construction system. This makes it difficult to accurately define the modification execution area, the area to be protected, and the areas affected by the transmission of changes. It also hinders the establishment of complete boundary constraints and spatial association rules, easily leading to problems such as uncontrolled modification scope, altered protected objects, and interference with related structures. The generation and selection of modification directions depend entirely on the designer's experience, resulting in path redundancy, low selection efficiency, and a lack of automated optimization mechanisms. Solution generation only allows for simple editing operations, with weak local parametric search capabilities and a single, unobjective evaluation dimension. The verification process lacks a phased, comprehensive testing mechanism; geometric compliance, assembly consistency, and manufacturing adaptability verification are incomplete; decision outputs lack structured integration and quantitative scoring criteria; modification object encapsulation is not standardized; and the overall process is time-consuming and has a high error rate, failing to meet the needs of efficient and accurate modification of complex models.

[0004] Therefore, there is an urgent need for a new method to support the decision-making process for AI-assisted CAD modifications, in order to solve the problem that candidate solutions are easy to generate but difficult to determine whether they are worth continuing to pursue in the existing technology. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a method and apparatus for advancing judgment support in AI-assisted CAD modification.

[0006] To achieve the above objectives, the present invention provides the following solution: A method for supporting decision-making in AI-assisted CAD modifications includes: S1. Local editing context construction: Obtain the CAD model to be modified and the user's modification request, parse the geometric topology, interface, assembly, adjacency and verification dependencies of the CAD model to be modified and the semantics of the user's modification request, divide the local region according to the parsing results, and use the analytic hierarchy process to determine the regional composition and constraints of the local editing context. The local editing context comprises a target region, a protected region, and a potential impact region. The target region is a designated local area in the CAD model to be modified, defined based on the analysis results, where modification operations will be performed. The protected region is a designated local area in the CAD model to be modified, defined based on the analysis results, where the geometry and assembly relationships remain unchanged. The potential impact region is a designated local area in the CAD model to be modified, defined based on the analysis results, that is related to the target region and is affected by the modification operations. The constraints of the local editing context include the modification range boundary constraints of the target region, the prohibited change boundary constraints of the protected region, the affected range boundary constraints of the potential impact region, the spatial position constraints of the target region and the protected region, the association constraints between the target region and the potential impact region, and the isolation constraints between the protected region and the potential impact region. S2. Modification Direction Generation and Filtering: Based on the local editing context matching preset editing operator in S1, multi-dimensional modification direction objects are generated. All modification direction objects are clustered using the K-means clustering algorithm and redundant objects are removed to determine the candidate modification direction set. The candidate modification direction set consists of all modification direction objects retained after the K-means clustering algorithm completes clustering iteration and removes redundant objects. The retained modification direction objects correspond to independent modification implementation paths of the target area of ​​the CAD model to be modified. Redundant objects are modification direction objects whose feature vectors are less than a preset judgment threshold distance from the corresponding cluster center vectors. S3. Generation of candidate solutions within the direction: In response to the target modification direction selected by the user, a local patch script is constructed based on the constraints of the target modification direction. Within the limited range, a parameterized local search is performed to generate candidate solutions within the direction. The candidate solutions within the direction are sorted according to preset evaluation indicators. S4. Phased verification execution: For the candidate solutions within the directions sorted in S3, perform preliminary verification and in-depth verification in sequence, and collect detection data of each dimension during the verification process to determine the verification score of each candidate solution; S5. Structured Result Output Encapsulation: Integrate the local editing context data of S1, the candidate modification direction data of S2, and the verification score data of S4. Use a weighted summation decision algorithm to calculate the recommended total score of each candidate solution as the advancement judgment result, output structured results, and encapsulate the selected candidate solution into a submittable modification object after the user selects an operation.

[0007] Optionally, in S1, the construction of the local editing context, the CAD model to be modified is at least one of a parametric native model, a boundary representation model, or a STEP model; the user modification request is a natural language request, a structured instruction, or a combination of both; the parsed content of the user modification request semantics includes the modification target, stability maintenance requirements, and downstream related constraints.

[0008] Optionally, in S2, the preset editing operator for generating the filter includes local thickening, local thinning, local expansion, local contraction, boundary reconstruction, hole adjustment, fillet adjustment, support enhancement, and transition shape adjustment.

[0009] Optionally, in S2, the preset editing operator for generating the filter includes local thickening, local thinning, local expansion, local contraction, boundary reconstruction, hole adjustment, fillet adjustment, support enhancement, and transition shape adjustment.

[0010] Optionally, in S3, during the generation of candidate schemes within the direction, the constructed local patch script includes a set of editable objects, a sequence of editable operators, a range of local free parameters, spatial constraints inherited from the local editing context, protection constraints inherited from the local editing context, and preset evaluation indicators including the degree of request satisfaction, the degree of preservation of the protected area, the expected verification readiness, the editing cost, and uncertainty.

[0011] Optionally, in the S4 phased verification execution, the preliminary verification includes geometric illegality detection, the preliminary verification includes protected area conflict detection, the preliminary verification includes local rule conflict detection, the in-depth verification includes geometric legality verification, the in-depth verification includes protection consistency verification, the in-depth verification includes manufacturing-related constraint verification, the in-depth verification includes structural-related constraint verification, the in-depth verification includes assembly-level verification, the assembly-level verification includes interface consistency verification, the assembly-level verification includes mating relationship maintenance verification, the assembly-level verification includes clearance risk verification, and the assembly-level verification includes interference risk verification.

[0012] Optionally, in the S5 structured result output encapsulation, the structured result includes candidate modification direction information, candidate scheme information within the direction, verification evidence information, and a continue-progress prompt information. The object that can be submitted for modification includes an executable patch, a complete verification report, parameter change records, candidate screening records, verification trigger records, and version rollback information.

[0013] The present invention also provides a progress judgment support device for AI-assisted CAD modification, comprising: Local editing context building module: used to obtain the CAD model to be modified and the user's modification request, parse the model relationship and request semantics, divide the local area and determine the regional composition and constraints; Modification direction generation and filtering module: used to match preset editing operators to generate modification direction objects, and remove redundant objects through K-means clustering to obtain a set of candidate modification directions; In-direction candidate solution generation module: In response to the user-selected target modification direction, it constructs a local patch script and generates and sorts in-direction candidate solutions; Phased verification execution module: Used to perform preliminary and in-depth verification of candidate solutions and calculate verification scores; Structured Result Output Encapsulation Module: This module integrates data and calculates the total recommendation score using a weighted summation decision algorithm, outputs structured results, and encapsulates them into a submittable and modifiable object.

[0014] The present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory, wherein the computer program, when executed by the processor, implements the above-described method.

[0015] The present invention also provides a storage medium storing a computer program, which executes the above-described method when running.

[0016] Compared with the prior art, the present invention has at least the following beneficial effects: Firstly, this invention constructs a local editing context and divides it into three categories: target, protected, and potential impact areas. It clarifies boundary constraints and spatial association rules, adapting to various CAD models and diverse user modification requests, providing a clear operational basis and constraint framework for AI-assisted CAD modification. By combining preset editing operators to generate multi-dimensional modification direction objects and relying on clustering algorithms to eliminate redundant content, it selects a set of independent and feasible candidate modification directions, avoiding repetitive and ineffective modification paths and reducing the time spent exploring modification directions. This method significantly improves the targeting of AI-assisted modification, eliminating the need for manual analysis of model relationships. It not only stabilizes the geometry and assembly relationships of the protected area but also precisely controls the transmission impact of modification operations, making the scope of local CAD modifications controllable and logically clear. This effectively reduces the operational difficulty and error probability of editing complex models, optimizing the rationality and controllability of the modification process.

[0017] Secondly, this invention constructs local patch scripts for selected modification directions and conducts parametric local searches to generate compliant candidate solutions, which are then sorted according to multi-dimensional indicators. These solutions undergo phased verification to complete multi-dimensional testing of geometry, assembly, and manufacturing, comprehensively collecting data to form an objective verification score. Relying on a weighted summation decision algorithm, the invention integrates context, modification direction, and verification data to calculate the total score for recommended solutions and output structured results, encapsulating the preferred solutions into directly submitable modification objects. This process achieves scientific screening and rigorous verification of candidate solutions, reducing the number of rework attempts and improving feasibility and compliance. Simultaneously, it presents the entire modification process information in a structured manner, simplifying user selection and operation, making AI-assisted CAD modification more standardized and efficient, significantly improving the overall quality of complex model modifications, and enhancing the practical value and ease of use of AI-assisted editing. Attached Figure Description

[0018] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart of a method for advancing judgment support for AI-assisted CAD modification; Figure 2 A schematic diagram illustrating data transmission between steps in a method for advancing judgments and supporting AI-assisted CAD modifications; Figure 3 A framework diagram of a propulsion judgment support device for AI-assisted CAD modification. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] Example 1: Reference Figure 1 One embodiment of the present invention proposes a method for advancing judgment support for AI-assisted CAD modification. It adopts a data processing mechanism of hierarchical construction of local editing context, K-means clustering iterative screening of modification direction, parameterized local search to generate scheme, two-stage progressive verification, and weighted summation decision scoring. In the scenario of AI-assisted CAD model local modification, it can accurately define the model editing scope and constraint boundary, generate independent modification paths without redundancy, batch generate compliant modification schemes, complete multi-dimensional compliance verification and output directly executable modification objects, significantly reduce the manual intervention cost of CAD model local modification and improve the accuracy and execution efficiency of modification operations.

[0023] The method described in this embodiment specifically includes: S1 Local Editing Context Construction This step is used to complete a full-dimensional analysis of the CAD model to be modified and the user's modification request. The analytic hierarchy process is used to divide the local area and quantify the constraints, thus defining clear execution boundaries for subsequent modification operations.

[0024] Specifically, this step consists of five sub-steps executed sequentially: Model and Request Acquisition: Load the CAD model to be modified and receive modification requests from the user. The CAD model to be modified is loaded using at least one of the following formats: parametric native model, boundary representation model, or STEP model. The user's modification request is input in the form of natural language requests, structured instructions, or a combination of both. Figure 2 As shown.

[0025] Model Relationship Analysis: Perform full-dimensional structural analysis on the CAD model to be modified, and fully extract the geometric topological relationships, interface mating relationships, assembly constraint relationships, adjacent component relationships, and verification dependency relationships of the model to form the model relationship relationship analysis results.

[0026] Request semantic parsing: Semantically decompose user modification requests, extract modification goals, stability requirements, and downstream related constraints, and form request semantic parsing results.

[0027] Local region segmentation: Combining the model association analysis results and request semantic analysis results, the CAD model to be modified is divided into three categories of designated local regions: the target region is the designated local region in the CAD model to be modified where the modification operation is directly performed; the protected region is the designated local region in the CAD model to be modified where the geometry and assembly relationship must remain unchanged throughout the process; and the potentially affected region is the designated local region in the CAD model to be modified that has a direct or indirect relationship with the target region and will be affected by the transmission of the modification operation.

[0028] Constraint determination: The Analytic Hierarchy Process (AHP) is used to calculate and quantify the weights of boundaries, spatial relationships, and associations of the divided local areas, and to determine six types of constraints for the local editing context: the modification range boundary constraint of the target area, the prohibited change boundary constraint of the protected area, the impact range boundary constraint of the potential impact area, the spatial location constraint between the target area and the protected area, the association constraint between the target area and the potential impact area, and the isolation constraint between the protected area and the potential impact area.

[0029] For example, when modifying the load-bearing support CAD model in a mechanical assembly, the loaded CAD model to be modified is a parametric native model, and the received user modification request is in the form of natural language: perform local shape adjustment on the stress-bearing parts of the support structure, keep the geometry and assembly relationship of the assembly docking parts unchanged, and the impact of the modification operation shall not affect the mating clearance of the surrounding related parts. Structural analysis was performed on the load-bearing CAD model, extracting the following: geometric topology as an integrated molding structure; interfaces as docking and mating structures with the assembly body; assembly relationships as a combination of positioning and fastening connections; adjacency relationships as adjacent arrangement with surrounding functional components; and verification dependencies as requiring the modification to meet both structural stress and assembly compatibility requirements. Semantic analysis was performed on user modification requests, extracting the following: modification target as local morphological adjustment of the load-bearing parts of the support structure; stability maintenance requirement as all parameters of the assembly docking parts remaining unchanged; and downstream correlation constraints as the modification's impact range not affecting the mating gaps of surrounding related components. Based on the analysis results, local regions were divided: the load-bearing parts of the support structure were designated as the target region; the assembly docking parts as the protected region; and regions directly connected to the target region and susceptible to modification transmission were designated as potential impact regions. Constraint quantification was performed using the analytic hierarchy process (AHP) to determine the modification range boundaries of the target region, the prohibited change boundaries of the protected region, and the impact range boundaries of the potential impact region. Simultaneously, the spatial relationship between the target region and the protected region, the correlation transmission rules between the target region and the potential impact region, and the isolation restriction rules between the protected region and the potential impact region were clarified.

[0030] S2, Modify direction to generate filters: This step is used to match the editing operator based on the local editing context, generate modification direction objects, and perform iterative clustering and redundancy removal using the K-means clustering algorithm to obtain an independent and non-repeating set of candidate modification directions.

[0031] Specifically, this step includes four sub-steps executed sequentially: Editing operator matching and modification direction object generation: Based on the local editing context determined by S1, it is accurately matched with the preset editing operator library to generate multi-dimensional modification direction objects; the preset editing operators include local thickening, local thinning, local expansion, local contraction, boundary reconstruction, hole adjustment, rounded corner adjustment, support enhancement, and transition form adjustment, and each type of operator corresponds to independent modification implementation logic.

[0032] Feature vector construction: Extract features from all generated modification direction objects and construct feature vectors containing modification type, parameter range, constraint fit, and propagation effect degree for subsequent clustering calculations.

[0033] K-means clustering iteration: The K-means clustering algorithm is used to iteratively cluster the feature vectors of all modified objects. The mathematical expression of the algorithm is: ,in For the sum of squared clustering errors, To preset the number of clusters, To modify the total number of direction objects, For the first The feature vector of a modified direction object. For the first The center vectors of each cluster; Cluster Center The mathematical expression for vector calculation is: , For the first The number of modification direction objects contained within each cluster. For the first The set of modification direction objects corresponding to each cluster; during the clustering process, the cluster center vector is continuously updated until the sum of squared clustering errors converges to the preset range, thus completing the clustering.

[0034] Redundant object removal and candidate set determination: Calculate the Euclidean distance between the feature vector of each modification direction object and the corresponding cluster center vector. Modification direction objects with a distance less than a preset judgment threshold are judged as redundant objects and removed. All modification direction objects retained after removing redundant objects form a candidate modification direction set. Each retained modification direction object corresponds to an independent modification implementation path in the target region.

[0035] For example, in the above-mentioned modification scenario of the supporting CAD model, multiple modification direction objects are generated based on the local editing context and the preset editing operator library; a feature vector containing modification type, parameter range, constraint fit, and degree of propagation influence is constructed for each modification direction object; K-means clustering iterative calculation is started according to the preset number of clusters, and the cluster center vector is continuously updated until the sum of squared clustering errors reaches a convergent state, completing the clustering of all modification direction objects; the distance between each modification direction object and the corresponding cluster center vector is calculated one by one, and all objects that meet the redundancy judgment condition are removed. The remaining modification direction objects form a candidate modification direction set, and all modification directions in the set are independent and non-repeating modification implementation paths.

[0036] Generation of candidate solutions within the S3 direction: This step is used to respond to the user's selected target modification direction, build a standardized local patch script, generate solutions in batches through parameterized local search, and sort the solutions according to multi-dimensional evaluation indicators.

[0037] Specifically, this step includes four sub-steps executed sequentially: Target modification direction response: Receive the target modification direction selected by the user from the candidate modification direction set, and extract all constraints and parameter adjustment ranges corresponding to that direction.

[0038] Local patch script construction: Based on the constraints of the target modification direction, a standardized local patch script is generated. The script contains five categories of content: set of editable objects, sequence of editable operators, range of local free parameters, spatial constraints inherited from the local editing context, and protection constraints inherited from the local editing context. It clarifies the execution object of the modification operation, the available operators, parameter boundaries and restriction rules.

[0039] Parametric local search execution: Within the parameter range defined by the local patch script, a traversal parametric local search is performed with a fixed step size. Each set of valid parameter combinations corresponds to a candidate solution in one direction. After the traversal is completed, all candidate solutions in all directions are obtained.

[0040] Candidate scheme evaluation and ranking: The candidate schemes in all directions are quantitatively scored using preset evaluation indicators, which include the degree of request satisfaction, the degree of preservation of the protected area, the expected verification readiness, the editing cost, and the uncertainty. After scoring, all candidate schemes are ranked from high to low according to the total score to form an ordered sequence of candidate schemes in each direction.

[0041] For example, after the user selects a target modification direction from the set of candidate modification directions, all constraints and parameter adjustment rules corresponding to that direction are extracted; based on the above information, a local patch script is constructed, specifying the executable editable model components, the allowed editing operators, the adjustable parameter boundaries, and the spatial constraints and protection constraints to be followed; within the scope defined by the script, a parameterized local search is performed, traversing all valid parameter combinations to generate candidate solutions within multiple directions; each candidate solution is scored according to five indicators: request satisfaction level, protection area retention level, expected verification readiness, editing cost, and uncertainty; the solutions are ranked according to the scoring results to obtain a sequence of candidate solutions with clear priorities.

[0042] S4, phased verification execution: This step is used to perform a two-stage progressive verification of the candidate solutions within the sorted directions, collect full-dimensional detection data and calculate the verification score to ensure that the solutions meet the full-dimensional requirements of geometry, manufacturing and assembly.

[0043] Specifically, the verification process follows the order of preliminary verification followed by in-depth verification. The ranked candidate solutions are verified in sequence, and solutions that fail the preliminary verification are eliminated and do not proceed to the in-depth verification stage.

[0044] Preliminary verification: The preliminary verification is a rapid compliance test, which includes three types of test items. If all test items are passed, the preliminary verification is deemed successful. Geometric illegality detection: Detects whether the modified model of the candidate solution has geometric anomalies such as broken surfaces, overlaps, gaps, and topological errors.

[0045] Conflict detection in protected areas: Detects whether there are any conflicts with protection requirements, such as changes in size, positional shifts, or shape deformations in the protected area.

[0046] Local rule conflict detection: Detects whether candidate solutions violate rules such as boundary constraints, spatial constraints, and association constraints set in the local editing context.

[0047] In-depth verification: In-depth verification is a comprehensive compliance check, including five types of verification items. If all verification items meet the standards, the in-depth verification is deemed successful. Geometric legality verification: Verify the compliance and completeness of the modified topology, geometry, and dimensional parameters of the model.

[0048] Consistency verification: Verify the consistency of geometric parameters, assembly relationships, and morphological features of the protected area throughout the entire modification process.

[0049] Manufacturing-related constraint verification: Verify whether the candidate solutions meet the constraint requirements of conventional machining processes such as cutting, stamping, and casting.

[0050] Structural constraint verification: Verify whether the candidate schemes meet the structural performance constraints such as structural stress, stiffness adaptation, and load bearing.

[0051] Assembly-level verification includes four sub-verifications: interface consistency verification, fit relationship maintenance verification, gap risk verification, and interference risk verification, to ensure the compatibility of the modified model with the complete machine assembly.

[0052] After verification, test data from various dimensions of preliminary and in-depth verification are collected, and the verification score of each candidate solution is calculated based on the percentage of qualified test items.

[0053] For example, the candidate solutions in the top-ranked directions are sequentially verified in stages. First, preliminary verification is carried out, and geometric illegality detection, protection area conflict detection, and local rule conflict detection are completed one by one. After confirming that there are no abnormalities, the in-depth verification stage is entered. In the in-depth verification, geometric legality, protection consistency, manufacturing-related constraints, and structural-related constraints are verified sequentially. Then, the four sub-verifications of assembly-level interfaces, mating relationships, clearances, and interference are carried out. The result data of all detection items are collected throughout the process, and the verification score of the candidate solution is calculated based on the proportion of qualified items.

[0054] S5, encapsulation of structured result output: This step integrates data from the entire process, calculates the recommended total score using a weighted summation decision algorithm, outputs a structured decision result, and encapsulates the user-selected solution into a modification object that can be directly submitted.

[0055] Specifically, this step includes four sub-steps executed sequentially: Multi-dimensional data integration: Integrate the local editing context data generated by S1, the candidate modification direction data generated by S2, and the verification score data calculated by S4 to form a complete scheme evaluation dataset.

[0056] Weighted summation decision calculation: The weighted summation decision algorithm is used to calculate the total recommended score for each candidate solution, and the total recommended score is used as the result of the advancement decision. The mathematical expression of the algorithm is: Where Score is the total recommended score for the candidate solution. The verification score is calculated in S4. Modify the preset direction feature score corresponding to the direction selected by the user for the target. The score represents the matching degree between the local editing context and the preset standard context in S1. To verify the score weighting coefficients, The weighting coefficients for directional feature scores. Match the score weight coefficients to the context, and ; To calculate the weight coefficients using the eigenvalue method of the judgment matrix in the Analytic Hierarchy Process (AHP), the AHP first constructs the judgment matrix. The judgment matrix is ​​normalized column-wise to obtain: Then sum by row to get Finally, the weight coefficients are obtained by normalizing the rows and vectors. ,in The judgment matrix constructed for the analytic hierarchy process. To determine the first element in the matrix Line number Assigning importance values ​​to columns To determine the order of a matrix, To determine the elements of a matrix after column normalization, The intermediate vector elements obtained by summing rows. For the final calculated number Weight coefficients That is, the weight coefficients corresponding to the three evaluation dimensions. .

[0057] Structured results generation: Structured results are generated based on the total recommendation score. The structured results include four categories of content: candidate modification direction information, candidate solution information within the direction, verification evidence information, and prompts for further progress.

[0058] Encapsulation of Modifiable Objects: In response to the user's scheme selection operation, the selected candidate scheme is encapsulated into modifiable objects. Modifiable objects include six categories of content: executable patch, complete verification report, parameter change record, candidate screening record, verification trigger record, and version rollback information.

[0059] For example, after integrating three types of data—local editing context, candidate modification directions, and verification scores—these are substituted into a weighted summation decision algorithm to calculate the recommended total score for each candidate solution. Based on the total score, a structured result is generated, clearly displaying details of the candidate modification directions, the solution ranking results, full-process verification evidence, and execution prompts. After the user selects a target solution, the executable modification script, full-process verification data, parameter adjustment records, screening process records, verification execution records, and version rollback basis corresponding to that solution are integrated and encapsulated to form a modification object that can be directly submitted.

[0060] Example 2: Based on the same inventive concept, such as Figure 3 As shown, the present invention also provides a progress judgment support device for AI-assisted CAD modification, the device being used to execute the progress judgment support method for AI-assisted CAD modification described in any of the above claims, the device comprising: Local editing context building module: This module serves as the fundamental parsing unit of the device, responsible for performing full-dimensional parsing of the model and requests, region partitioning, and constraint quantification. It acquires the CAD model to be modified and the user's modification request, parses the model's geometric topology, interfaces, assembly, adjacency, and verification dependencies, and parses the semantic information of the user's modification request. Based on the parsing results, it partitions the target region, protected region, and potential impact region, determines the regional composition and constraints of the local editing context using the analytic hierarchy process (AHP), and outputs complete local editing context data.

[0061] Modify direction to generate filter module: This module is the path generation unit of the device, responsible for generating modification direction objects, iterative clustering, and redundancy removal. It generates multi-dimensional modification direction objects based on a preset editing operator matched to the local editing context; extracts the feature vectors of the modification direction objects; performs iterative clustering calculations using the K-means clustering algorithm; calculates the distance between the feature vectors and the cluster center vectors; removes redundant objects; and outputs a set of candidate modification directions without duplicates.

[0062] In-direction candidate solution generation module: This module serves as the device's scheme generation unit, responsible for script construction, parameter search, and scheme ranking. It responds to the user-selected target modification direction, constructs a local patch script based on constraints, including editable objects, editable operators, parameter ranges, spatial constraints, and protection constraints; performs a parameterized local search within a defined range to generate candidate schemes within the direction; and scores and ranks the schemes according to preset evaluation indicators, outputting an ordered sequence of candidate schemes.

[0063] Phased verification execution module: This module serves as the compliance verification unit for the device, responsible for two-stage verification and score calculation. It is used to sequentially perform preliminary and in-depth verification on candidate solutions within the sorted directions. The preliminary verification covers geometric illegality, protected area conflict, and local rule conflict detection, while the in-depth verification covers full-dimensional verification at the geometric, protection, manufacturing, structural, and assembly levels. It collects full-dimensional detection data during the verification process and calculates and outputs the verification score for each candidate solution.

[0064] Structured result output encapsulation module: This module serves as the device's output unit, responsible for data integration, decision calculation, and object encapsulation. It integrates local editing context data, candidate modification direction data, and verification score data; calculates the total recommended score for candidate solutions using a weighted summation decision algorithm; generates structured results containing direction information, solution information, verification evidence, and progress prompts; and, in response to user-selected operations, encapsulates a submitable modification object containing executable patches, verification reports, parameter records, filtering records, trigger records, and rollback information.

[0065] It should be noted that the data interaction between the various modules described above does not necessarily represent direct data transmission. Indirect data interaction methods are also acceptable as long as they achieve the objectives of this invention. The above descriptions are merely exemplary embodiments of this invention and should not be construed as limiting the scope of the invention.

[0066] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and practicing the disclosure of this invention. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

[0067] Example 3 Based on the same technical concept, in some specific embodiments of this application, a terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and a method that the processor can use to execute when executing the program.

[0068] Based on the same technical concept, in some specific embodiments of this application, a computer-readable storage medium is provided on which a computer program is stored, which can be used to execute a method when the program is executed by a processor.

[0069] Optionally, the memory is used to store programs; the memory may include volatile memory, such as random-access memory (RAM), such as static random-access memory (SRAM), double data rate synchronous dynamic random-access memory (DDR SDRAM), etc.; the memory may also include non-volatile memory, such as flash memory. The memory is used to store computer programs (such as application programs, functional modules, etc. that implement the above methods), computer instructions, etc., and the aforementioned computer programs, computer instructions, etc., can be partitioned and stored in one or more memories. Furthermore, the aforementioned computer programs, computer instructions, data, etc., can be accessed by the processor.

[0070] The aforementioned computer programs, computer instructions, etc., can be stored in partitions within one or more memory locations. Furthermore, the aforementioned computer programs, computer instructions, data, etc., can be accessed by a processor.

[0071] A processor is used to execute a computer program stored in memory to implement the various steps of the methods involved in the above embodiments. For details, please refer to the relevant descriptions in the preceding method embodiments.

[0072] The processor and memory can be separate structures or integrated structures. When the processor and memory are separate structures, they can be coupled together via a bus.

[0073] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0074] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0075] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0076] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0077] The foregoing has described some specific embodiments of this application. It should be understood that this application is not limited to the specific embodiments described above, and those skilled in the art can make various modifications or variations within the scope of the claims, which do not affect the substantive content of this application. The above-described preferred features can be used in any combination without conflict.

Claims

1. A method for advancing judgment support in AI-assisted CAD modification, characterized in that, include: S1. Local editing context construction: Obtain the CAD model to be modified and the user's modification request, parse the geometric topology, interface, assembly, adjacency and verification dependencies of the CAD model to be modified and the semantics of the user's modification request, divide the local region according to the parsing results, and use the analytic hierarchy process to determine the regional composition and constraints of the local editing context. S2. Modification Direction Generation and Filtering: Based on the local editing context matching preset editing operator in S1, multi-dimensional modification direction objects are generated. All modification direction objects are clustered using the K-means clustering algorithm and redundant objects are removed to determine the candidate modification direction set. S3. Generation of candidate solutions within the direction: In response to the target modification direction selected by the user, a local patch script is constructed based on the constraints of the target modification direction. Within the limited range, a parameterized local search is performed to generate candidate solutions within the direction. The candidate solutions within the direction are sorted according to preset evaluation indicators. S4. Phased verification execution: For the candidate solutions within the directions sorted in S3, perform preliminary verification and in-depth verification in sequence, and collect detection data of each dimension during the verification process to determine the verification score of each candidate solution; S5. Structured Result Output Encapsulation: Integrate the local editing context data of S1, the candidate modification direction data of S2, and the verification score data of S4. Use a weighted summation decision algorithm to calculate the recommended total score of each candidate solution as the advancement judgment result, output structured results, and encapsulate the selected candidate solution into a submittable modification object after the user selects an operation.

2. The propulsion judgment support method according to claim 1, characterized in that, In S1, the local editing context construction, the CAD model to be modified is at least one of a parametric native model, a boundary representation model, or a STEP model; the user modification request is a natural language request, a structured instruction, or a combination of both; the semantic parsing content of the user modification request includes the modification target, stability maintenance requirements, and downstream related constraints.

3. The advancement judgment support method according to claim 1, characterized in that, In the S2 modification direction generation filtering, the matched preset editing operators include local thickening, local thinning, local expansion, local contraction, boundary reconstruction, hole adjustment, rounded corner adjustment, support enhancement, and transition shape adjustment.

4. The advancement judgment support method according to claim 1, characterized in that, In the S2 modification direction generation filtering, the matched preset editing operators include local thickening, local thinning, local expansion, local contraction, boundary reconstruction, hole adjustment, rounded corner adjustment, support enhancement, and transition shape adjustment.

5. The advancement judgment support method according to claim 1, characterized in that, In the generation of candidate schemes within direction S3, the constructed local patch script includes a set of editable objects, a sequence of editable operators, a range of local free parameters, spatial constraints inherited from the local editing context, protection constraints inherited from the local editing context, and preset evaluation indicators including the degree of request satisfaction, the degree of preservation of the protected area, the expected verification readiness, the editing cost, and uncertainty.

6. The advancement judgment support method according to claim 1, characterized in that, In the S4 phased verification process, the preliminary verification includes geometric illegality detection, the preliminary verification includes protected area conflict detection, the preliminary verification includes local rule conflict detection, the in-depth verification includes geometric legality verification, the in-depth verification includes protection consistency verification, the in-depth verification includes manufacturing-related constraint verification, the in-depth verification includes structural-related constraint verification, the in-depth verification includes assembly-level verification, the assembly-level verification includes interface consistency verification, the assembly-level verification includes mating relationship maintenance verification, the assembly-level verification includes clearance risk verification, and the assembly-level verification includes interference risk verification.

7. The advancement judgment support method according to claim 1, characterized in that, In the S5 structured result output encapsulation, the structured result includes candidate modification direction information, candidate scheme information within the direction, verification evidence information, and prompts for continued progress. The object to be modified includes an executable patch, a complete verification report, parameter change records, candidate screening records, verification trigger records, and version rollback information.

8. A progress judgment support device for AI-assisted CAD modification, the device being applicable to the progress judgment support method for AI-assisted CAD modification as described in any one of claims 1-7, characterized in that, include: Local editing context building module: used to obtain the CAD model to be modified and the user's modification request, parse the model relationship and request semantics, divide the local area and determine the regional composition and constraints; Modification direction generation and filtering module: used to match preset editing operators to generate modification direction objects, and remove redundant objects through K-means clustering to obtain a set of candidate modification directions; In-direction candidate solution generation module: In response to the user-selected target modification direction, it constructs a local patch script and generates and sorts in-direction candidate solutions; Phased verification execution module: Used to perform preliminary and in-depth verification of candidate solutions and calculate verification scores; Structured Result Output Encapsulation Module: This module integrates data and calculates the total recommendation score using a weighted summation decision algorithm, outputs structured results, and encapsulates them into a submittable and modifiable object.

9. An electronic device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as claimed in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores a computer program that, when executed, performs the method as described in any one of claims 1 to 7.

Images