An AI mapping high-precision generation method and system based on a correlation data set, a terminal, and a storage medium

Abstract

The application relates to the technical field of intelligent mapping, and discloses an AI mapping high-precision generation method and system based on an associated data set, a terminal and a storage medium.The method comprises the following steps: constructing a multi-layer associated data set, establishing a bidirectional mapping relationship between each layer of the data set through an associated index engine; obtaining product demand data, searching and recalling associated data records in the multi-layer associated data set, and combining the associated data records into a constraint parameter package; inputting the constraint parameter package into an AI mapping model to generate mold mapping data; extracting multi-dimensional parameters in the mold mapping data, comparing the multi-dimensional parameters with standard values in the data set, and calculating deviations; if there is a deviation, generating a correction instruction according to the deviation and feeding back the correction instruction to the AI mapping model for iterative correction until all parameters meet preset convergence conditions; and outputting target mold mapping data.The application can accurately match each dimension parameter of the target mold mapping data with a data set standard value, and improve the precision and process matching of AI mapping.

Description

Technical Field

[0001] This invention relates to the field of intelligent mapping technology, and in particular to a method, system, terminal, and computer-readable storage medium for generating high-precision AI maps based on associated datasets. Background Technology

[0002] Currently, using AI to automatically generate engineering drawings (such as CAD drawings) has become an important area of ​​exploration for the digital transformation of the manufacturing industry. However, existing AI drafting technologies still have the following shortcomings in practical industrial applications: First, general AI drafting models are mainly trained based on publicly available general drawing data, lacking the support of industry-specific structured datasets. As a result, the generated drawings have serious inaccuracies in terms of dimension annotation, material specifications, and process parameters.

[0003] Second, the existing AI drafting process lacks an automated closed-loop verification and correction mechanism. The drawings output by AI cannot be automatically compared and corrected with standard parameters, and manual verification is required for each item, resulting in poor process matching and long delivery cycle.

[0004] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0005] The main objective of this invention is to provide a method, system, terminal, and computer-readable storage medium for generating high-precision AI drawings based on associated datasets, aiming to solve the problems of insufficient accuracy, poor process matching, and long delivery cycle of drawings generated by existing AI drawing methods.

[0006] To achieve the above objectives, this invention provides a method for generating high-precision AI maps based on associated datasets. This method includes the following steps: Construct a multi-level associated dataset, wherein the associated datasets at each level in the multi-level associated dataset establish a bidirectional mapping relationship through an associated index engine; Obtain product demand data, perform multi-level matching retrieval in the multi-layered associated dataset based on the product demand data, recall the matching associated data records, and combine all the recalled associated data records into a constraint parameter package; The constraint parameter package is input into the AI ​​drawing model, and the AI ​​drawing model generates mold drawing data based on the product requirement data and the constraint parameter package. Extract multi-dimensional parameters from the mold drawing data, compare each multi-dimensional parameter with the standard value in the multi-layered associated dataset, and calculate the deviation of each item. Under the condition of the aforementioned deviation, a correction instruction is generated based on the deviation, and the correction instruction is fed back to the AI ​​drawing model. The AI ​​drawing model is controlled to iteratively correct the mold drawing data according to the correction instruction, and then jumps to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all parameters meet the preset convergence conditions. After all parameters meet the preset convergence conditions, the target mold drawing data is output, and a drawing data traceability list is generated simultaneously.

[0007] Furthermore, the construction of a multi-level associated dataset, wherein the various levels of associated datasets in the multi-level associated dataset establish a bidirectional mapping relationship through an associated index engine, specifically includes: Acquire multi-source industrial data, parse and extract fields from the multi-source industrial data to obtain structured data; The structured data is mapped to predefined standard fields in the multi-level associated dataset. Based on the primary key information in the structured data, the structured data is automatically associated with the records of each level of the multi-level associated dataset to form structured associated records. A unique association ID is assigned to the structured associated records, and the bidirectional mapping relationship is established between the levels of the multi-level associated dataset through the unique association ID.

[0008] Furthermore, the step of acquiring product demand data, performing multi-level matching retrieval based on the product demand data in the multi-layered associated dataset, recalling matching associated data records, and combining all recalled associated data records into a constraint parameter package specifically includes: Obtain product requirement data, and perform standardized verification and completion on the product requirement data to obtain target product requirement data; Using the target product demand data as the retrieval primary key, multi-level matching retrieval is performed in the multi-layered associated dataset, and all matching associated data records are retrieved. The retrieved associated data records are then combined into a constraint parameter package.

[0009] Furthermore, the step of inputting the constraint parameter package into the AI ​​drawing model, and generating mold drawing data through the AI ​​drawing model based on the product requirement data and the constraint parameter package, specifically includes: Encode each parameter in the constraint parameter package into a condition vector; The conditional vector is concatenated with the latent space features of the AI ​​mapping model to obtain concatenated features. The concatenated features are then input into the AI ​​mapping model, which controls the AI ​​mapping model to generate the mold drawing data based on the product requirement data within the feasible domain defined by the constraint parameter package.

[0010] Furthermore, the step of extracting multi-dimensional parameters from the mold drawing data, comparing these multi-dimensional parameters with the standard values ​​in the multi-layered associated dataset item by item, and calculating the deviations for each item specifically includes: Extract the geometric parameters from the mold drawing data, compare the geometric parameters with the standard geometric constraints in the multi-layer associated dataset, and calculate the geometric deviation; Extract the process parameters from the mold drawing data, compare the process parameters with the standard parameter ranges of corresponding materials and processes in the multi-layer associated dataset, and calculate the process deviation; Extract the structural parameters from the mold drawing data, verify the assembly relationship of the structural parameters, and calculate the structural deviation; Extract the annotation parameters from the mold drawing data, verify the completeness and consistency of the annotation parameters, and calculate the annotation deviation.

[0011] Furthermore, under the condition of the aforementioned deviation, a correction instruction is generated based on the deviation, and the correction instruction is fed back to the AI ​​drawing model. The AI ​​drawing model is then controlled to iteratively correct the mold drawing data according to the correction instruction, and the process jumps to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all parameters meet the preset convergence condition. Specifically, this includes: The deviations are obtained, including geometric deviations, process deviations, structural deviations, and annotation deviations; If any of the geometric deviation, process deviation, structural deviation, and annotation deviation is non-zero, a correction instruction for the corresponding dimension is generated based on the non-zero deviation. The correction instruction is fed back to the AI ​​drawing model, which then corrects the mold drawing data according to the correction instruction, generating a corrected version of the mold drawing data. The process then jumps to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all the geometric deviation, process deviation, structural deviation, and annotation deviation are reduced to zero.

[0012] Furthermore, after all parameters satisfy the preset convergence condition, the target mold drawing data is output, and a drawing data traceability list is generated simultaneously, specifically including: After all parameters meet the preset convergence condition, the current version of the mold drawing data is determined as the target mold drawing data, and the target mold drawing data is output. Obtain the unique association ID of all related data records called during this mapping process, the version number of the AI ​​mapping model, the deviation value generated in each round of comparison, and the correction record of each round of correction. Combine the obtained unique association ID, version number, deviation value, and correction record into a mapping data traceability list, and store the mapping data traceability list in association with the target mold mapping data.

[0013] Furthermore, to achieve the above objectives, the present invention also provides a high-precision AI mapping generation system based on associated datasets. This system is used to implement the high-precision AI mapping generation method based on associated datasets as described above. The high-precision AI mapping generation system based on associated datasets includes: The dataset construction module is used to construct a multi-level associated dataset, wherein the associated datasets in the multi-level associated dataset establish a bidirectional mapping relationship between each level through an associated index engine; The retrieval and matching module is used to obtain product demand data, perform multi-level matching retrieval in the multi-layered associated dataset based on the product demand data, recall the matched associated data records, and combine all the recalled associated data records into a constraint parameter package. The constraint generation module is used to input the constraint parameter package into the AI ​​drawing model, and generate mold drawing data through the AI ​​drawing model based on the product requirement data and the constraint parameter package; The multidimensional comparison module is used to extract multidimensional parameters from the mold drawing data, compare the multidimensional parameters with the standard values ​​in the multi-layer association dataset item by item, and calculate the deviations of each item. The iterative correction module is used to generate correction instructions based on the deviation when the deviation exists, feed the correction instructions back to the AI ​​drawing model, control the AI ​​drawing model to iteratively correct the mold drawing data according to the correction instructions, and jump to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all parameters meet the preset convergence conditions. The output archiving module is used to output the target mold drawing data and generate a drawing data traceability list simultaneously after all parameters meet the preset convergence conditions.

[0014] Furthermore, to achieve the above objectives, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and an AI mapping high-precision generation program based on an associated dataset stored in the memory and executable on the processor, wherein when the AI ​​mapping high-precision generation program based on an associated dataset is executed by the processor, it implements the steps of the AI ​​mapping high-precision generation method based on an associated dataset as described above.

[0015] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores an AI mapping high-precision generation program based on an associated dataset, and when the AI ​​mapping high-precision generation program based on the associated dataset is executed by a processor, it implements the steps of the AI ​​mapping high-precision generation method based on the associated dataset as described above.

[0016] The beneficial effects of this invention are as follows: By constructing a multi-layered associated dataset and establishing a bidirectional mapping relationship between each layer, multi-source heterogeneous industrial drawing data is integrated into a structured knowledge system; based on product demand data, corresponding constraint parameter packages are retrieved, matched, and combined from the dataset. Industrial standards and customer requirements are transformed into constraints that the AI ​​model can execute, thus limiting the AI ​​generation process to the feasible domain specified by the constraint parameter packages and eliminating parameter drift caused by free generation of general models. After extracting multi-dimensional parameters from the generated mold drawing data and comparing them item by item with the standard values ​​of the dataset, correction instructions generated based on the deviations are fed back to the AI ​​model for multiple rounds of closed-loop iterative correction until all parameters meet the preset convergence conditions. The parameters of each dimension of the target mold drawing data are accurately matched with the standard values ​​of the dataset, and the multi-dimensional synchronous correction effectively suppresses the deviation of other dimension parameters caused by single-dimensional correction, significantly improving the accuracy and process matching of AI drawing, greatly reducing manual intervention, and shortening the drawing delivery cycle. By synchronously generating and storing a traceability list of drafting data associated with the target mold drafting data, complete traceability of the entire drafting process is achieved, providing a data foundation for the rapid location of drawing quality issues and the continuous iteration of AI models. Attached Figure Description

[0017] Figure 1 This is a flowchart of a preferred embodiment of the AI ​​mapping high-precision generation method based on associated datasets of the present invention; Figure 2 This is a structural diagram of a preferred embodiment of the AI ​​mapping high-precision generation system based on associated datasets of the present invention; Figure 3 This is a structural diagram of a preferred embodiment of the terminal of the present invention. Detailed Implementation

[0018] This application provides a method, system, terminal, and storage medium for high-precision AI mapping generation based on associated datasets. To make the objectives, technical solutions, and effects of this application clearer and more explicit, the following detailed description is provided with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit the scope of this application.

[0019] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0020] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0021] The preferred embodiment of the present invention describes a high-precision AI mapping generation method based on associated datasets, such as... Figure 1 As shown, the high-precision AI mapping generation method based on associated datasets includes the following steps: S10. Construct a multi-level associated dataset, wherein the multi-level associated datasets in the multi-level associated dataset establish a bidirectional mapping relationship between each layer of associated datasets through an associated index engine.

[0022] The purpose of this step is to establish a structured data foundation covering all elements of industrial drafting, enabling the AI ​​drafting model to retrieve a complete set of related parameters as constraints during the subsequent generation stage. The multi-layered related dataset contains at least two data levels, and the specific number of levels can be flexibly increased or decreased according to the application scenario. In a preferred embodiment of the invention, the multi-layered related dataset specifically includes five layers: a drawing geometry layer dataset, a metal material layer dataset, a processing technology layer dataset, a mold matching layer dataset, and a customer customization layer dataset. A bidirectional mapping relationship is established between each layer dataset through a related index engine, forming a unified related index system.

[0023] Furthermore, the construction of the multi-level associated dataset, wherein the various levels of the associated dataset in the multi-level associated dataset establish a bidirectional mapping relationship through an associated index engine, specifically includes: S11. Obtain multi-source industrial data, parse and extract fields from the multi-source industrial data to obtain structured data.

[0024] In this embodiment, the system receives batches of source data, including historical CAD drawings in formats such as DWG, DXF, and STEP, material BOMs (Bill of Materials) from the enterprise ERP (Enterprise Resource Planning) system, process cards, mold detail sheets, and customer order specifications, through a multi-format data import interface. Then, the system uses a data parsing engine to identify and extract fields from the historical CAD drawings. The extracted content includes outline geometric coordinates, dimension values, technical condition text, and title block attributes (such as drawing number, material, plate thickness, and customer code). Through parsing and extraction, the heterogeneous data originally scattered across multiple business systems is transformed into a structured, standard data format.

[0025] S12. Map the structured data to predefined standard fields in the multi-level associated dataset. Based on the primary key information in the structured data, automatically associate the structured data with the records of each level of the multi-level associated dataset to form structured associated records. Assign a unique association ID to the structured associated records and establish the bidirectional mapping relationship between each level of the multi-level associated dataset through the unique association ID.

[0026] In this embodiment, the system first establishes a mapping rule base, uniformly mapping data fields from different sources and in different formats to predefined standard fields in the multi-level associated dataset. For example, the "Material" field from the title block of a CAD drawing and the "Material Name" field from the ERP system are both mapped to the "Material Grade" standard field in the material dataset. Then, based on the primary key information such as the material name, plate thickness, processing method, and customer code in the drawing, the system automatically matches the drawing data with the corresponding records in the material dataset, process dataset, mold dataset, and customer-customized dataset. After matching, a complete structured associated record containing five dimensions—drawing geometry, material properties, processing technology, mold parameters, and customer specifications—is formed, and a unique associated ID is assigned to this record.

[0027] Through this unique association ID, the system establishes a bidirectional mapping relationship between the layers of the multi-layered associated dataset. Starting from a data record in any layer, the system can retrieve associated data records in all other layers using this association ID. For example, given an association ID, the system can simultaneously obtain the corresponding drawing geometry data, material parameters, process parameters, mold parameters, and customer specifications. After the association is completed, the system stores the structured associated records separately in a basic general read-only database (storing industry-standard data such as GB / T and JIS standards) and an updatable customizable database (storing enterprise-specific data and customer-specific data). The two databases can be searched together.

[0028] It should be noted that in other embodiments of the present invention, the number of layers in the multi-layer associated dataset can be adjusted according to the actual application scenario. For example, for scenarios involving only sheet metal cutting and not mold design, the mold matching dataset can be omitted, and a four-layer associated dataset can be constructed. For complex scenarios involving more dimensions, new data layers can also be added, such as equipment parameter layers, detection standard layers, etc.

[0029] S20. Obtain product demand data, perform multi-level matching retrieval in the multi-layered associated dataset based on the product demand data, recall the matching associated data records, and combine all the recalled associated data records into a constraint parameter package.

[0030] The purpose of this step is to transform the product requirements input by the user into structured constraints that can be used in the AI ​​drawing model, ensuring that the AI-generated drawings are based on data in all dimensions, including dimensions, materials, processes, molds, and customer specifications.

[0031] Furthermore, the step of acquiring product demand data, performing multi-level matching retrieval based on the product demand data in the multi-layered associated dataset, recalling matching associated data records, and combining all recalled associated data records into a constraint parameter package specifically includes: S21. Obtain product requirement data, perform standardization verification and completion on the product requirement data, and obtain target product requirement data.

[0032] In this embodiment, the user inputs product requirement parameters through the interactive interface of the AI-powered visual operation terminal. Product requirement data includes, but is not limited to: product type (e.g., sheet metal parts, stamped parts, pipe fittings), material grade (e.g., SECC, SPCC, SUS304), sheet thickness, target processing technology (e.g., stamping, laser cutting, wire cutting), customer name or customer number, order number, etc. After obtaining the above product requirement data, the system first performs standardization verification and completion on the input information: checking whether each field conforms to the preset data format requirements, attempting to complete missing fields from historical orders or default configurations, and converting or prompting the user to correct non-standard data. After verification and completion, the system obtains target product requirement data with a complete structure and uniform format.

[0033] S22. Using the target product demand data as the retrieval primary key, perform multi-level matching retrieval in the multi-level associated dataset, and recall all matching associated data records, and combine the recalled associated data records into a constraint parameter package.

[0034] In this embodiment, the system uses the verified and completed target product requirement data as the retrieval primary key and performs multi-level matching retrieval in a multi-layered associated dataset. The retrieval process matches step by step according to the priority order of "material, plate thickness, process, customer, and product type": First, it matches the corresponding material record in the metal material dataset based on the material grade; then, it matches the corresponding plate thickness series and physical property parameters in the material record based on the plate thickness value; next, it matches the corresponding process parameters and tolerance thresholds in the processing process dataset based on the target processing technology; then, it matches the corresponding drawing template and acceptance standard in the customer customization dataset based on the customer name; and finally, it matches whether there is a reusable historical geometric template in the drawing geometry dataset based on the product type.

[0035] After the retrieval is complete, the system recalls all matching associated data records and combines these records into a constraint parameter package for the drafting task. This constraint parameter package includes at least the following: a baseline geometry template (if reusable historical drawings exist), material property parameters (such as tensile strength and yield strength) and bending springback coefficient, available process routes and corresponding tolerance windows, mold selection parameters and recommended clearance values, and a client-specific drafting specification template. Each parameter in the constraint parameter package is derived from a standard value in a multi-layered associated dataset, representing the reference benchmark for the drafting task across all dimensions.

[0036] It should be noted that if no identical historical template for the same product type is found in the geometric data layer of the drawing, the system will perform similarity matching based on the semantics of the product type and recall the closest geometric template as a reference. If there are no matching historical geometric templates, the constraint parameter package will not contain a baseline geometric template, and the AI ​​model will generate a zero-sample structure based on the semantics of the product type. However, the generated geometric features are still limited by material parameters, process parameters, and mold parameters.

[0037] S30. Input the constraint parameter package into the AI ​​drawing model, and generate mold drawing data through the AI ​​drawing model based on the product requirement data and the constraint parameter package.

[0038] The purpose of this step is to input the retrieved and assembled constraint parameter package as a hard constraint into the AI ​​drawing model, so that the AI ​​generation process changes from free and random drawing to controllable generation under conditional constraints, fundamentally avoiding the parameter drift and accuracy deviation problems caused by the lack of constraints in general AI models.

[0039] Further, the step of inputting the constraint parameter package into the AI ​​drawing model, and generating mold drawing data through the AI ​​drawing model based on the product requirement data and the constraint parameter package, specifically includes: S31. Encode each parameter in the constraint parameter package into a condition vector.

[0040] In this embodiment, the system converts all parameters in the constraint parameter package (including geometric template features, material property parameters, process tolerance windows, recommended die clearance values, customer specification identifiers, etc.) into numerical condition vectors acceptable to the AI ​​drafting model. Specifically, for numerical parameters (such as plate thickness, bending springback coefficient, and tolerance threshold), the system directly normalizes these parameters into vector elements. For categorical parameters (such as material grade, process type, and customer number), the system converts the parameters into vector representations through one-hot encoding or embedding. After all parameters are encoded, they are combined into a unified condition vector, which fully represents the constraints of this drafting task across all dimensions.

[0041] S32. The condition vector is concatenated with the latent space features of the AI ​​mapping model to obtain concatenated features. The concatenated features are then input into the AI ​​mapping model, which controls the AI ​​mapping model to generate the mold drawing data within the feasible domain defined by the constraint parameter package, based on the product requirement data.

[0042] In this embodiment, the AI ​​mapping model adopts an architecture based on a diffusion model or a generative adversarial network, with its core generation process completed in the latent space. The system concatenates the conditional vector encoded in step S31 with the latent space features currently sampled by the AI ​​mapping model to obtain concatenated features. The system inputs these concatenated features into the decoder or denoising network of the AI ​​mapping model, controlling the model to generate samples within the feasible region defined by the constraint parameter package.

[0043] By concatenating the conditional vector with the latent space features and inputting it into the decoder, the spatial distribution of the generated result is strictly restricted within the feasible domain defined by the constraint parameter package, thereby achieving controllable generation under conditional constraints.

[0044] In a specific embodiment of the present invention, a sheet metal chassis panel of a certain model (material SECC galvanized sheet, thickness 1.2mm, customer A specification) is taken as an example. The system encodes the retrieved constraint parameter package into a condition vector. This constraint parameter package includes the mechanical parameters of SECC 1.2mm (tensile strength not less than 270MPa, yield strength not less than 140MPa), bending springback compensation angle (0.5 degrees to 1.2 degrees), stamping process recommended tolerance (IT10 grade), and customer A's exclusive drawing frame, annotation font, and layer color specifications. The system concatenates the encoded condition vector with the latent space features of the AI ​​drawing model and inputs it into the model. The AI ​​model generates pre-version mold drawing data within the feasible domain defined by the constraint parameter package. The dimensions of all geometric contours, the angle of bending lines, the numerical values ​​of annotations, and the font styles are all strictly limited by the constraint parameter package, and no dimensional drift or annotation format errors common in general AI models occur.

[0045] In another specific embodiment of the present invention, a trapezoidal spiral coil forming mold in a refrigeration component is taken as an example. The system encodes the retrieved constraint parameter package into a condition vector. This constraint parameter package includes the historical record of the spiral coil mold cavity structure, the recommended value of the punch-die clearance (1.05 times the material thickness), the mold base specifications, the springback coefficient and friction coefficient of SUS304 in a semi-hard state, and the cold stamping forming speed limit. The AI ​​model completes the pre-generation of mold drawing data for mold assembly drawings and part drawings within the feasible domain defined by the constraint parameter package.

[0046] S40. Extract the multi-dimensional parameters from the mold drawing data, compare the multi-dimensional parameters with the standard values ​​in the multi-layer association dataset item by item, and calculate the deviations of each item.

[0047] The purpose of this step is to extract and standardize the parameters of the AI-generated mold drawing data in all dimensions, quantitatively calculate the deviation between each parameter in the mold drawing data and the standard value in the associated dataset, and provide a data basis for subsequent automatic iterative correction.

[0048] Furthermore, the step of extracting multi-dimensional parameters from the mold drawing data, comparing these multi-dimensional parameters with the standard values ​​in the multi-layered associated dataset item by item, and calculating the deviations for each item specifically includes: S41. Extract the geometric parameters from the mold drawing data, compare the geometric parameters with the standard geometric constraints in the multi-layer associated dataset, and calculate the geometric deviation.

[0049] In this embodiment, the system extracts all geometric elements from the AI-pre-generated mold drawing data, including the contour coordinate sequence, the center coordinates and diameter of each hole, the values ​​of each side length, and the bending angles and bending line positions. After extraction, the system compares the above geometric parameters with the standard geometric template in the multi-layer associated dataset item by item: if the constraint parameter package contains a reference geometric template, it directly compares with the corresponding geometric elements in the template; if there is no reference geometric template, it compares with the preset geometric constraint rules (such as minimum hole diameter limit, maximum bending angle limit, etc.). The geometric deviation is calculated as follows: the deviation between the actual value and the standard value of each geometric parameter is calculated one by one, and then the deviations of all geometric parameters are comprehensively processed to obtain the comprehensive geometric deviation. When the comprehensive geometric deviation is zero, it means that the geometric parameter is completely consistent with the standard value; the larger the comprehensive geometric deviation value, the more serious the geometric deviation.

[0050] S42. Extract the process parameters from the mold drawing data, compare the process parameters with the standard parameter ranges of the corresponding materials and processes in the multi-layer associated dataset, and calculate the process deviation.

[0051] In this embodiment, the system extracts process-related parameters from AI-pre-generated mold drawing data, including bending radius, blanking clearance, and threaded hole diameter. After extraction, the system compares these process parameters with the standard recommended value ranges for the corresponding material, thickness, and process combination in the multi-layer associated dataset. The system checks whether each process parameter falls within the standard parameter range for that material, plate thickness, and process condition. The process deviation is calculated as follows: for each process parameter, it checks whether it is below the lower limit or above the upper limit of the standard range. If the parameter falls within the range, the deviation of that parameter is zero. If the parameter exceeds the range, the deviation value of that parameter is calculated according to the extent of the exceedance. Finally, the deviation values ​​of all process parameters are comprehensively processed to obtain the comprehensive process deviation.

[0052] S43. Extract the structural parameters from the mold drawing data, verify the assembly relationship of the structural parameters, and calculate the structural deviation.

[0053] In this embodiment, the system extracts structural parameters from AI-pre-generated mold drawing data, including the position coordinates, dimensions, and mating surface information of each component in three-dimensional space. After extraction, the system performs assembly relationship verification in three-dimensional space, detecting whether there is overlap (interference) between components, whether the gap is within the allowable range, and whether there are obstacles on the assembly path. The structural deviation is calculated by detecting the maximum interference, maximum gap deviation, and maximum assembly path offset in the assembly, and taking the maximum value of the three as the comprehensive structural deviation. When there is no interference, the gap is within the allowable range, and the path is unobstructed, the structural deviation is zero.

[0054] S44. Extract the annotation parameters from the mold drawing data, verify the completeness and consistency of the annotation parameters, and calculate the annotation deviation.

[0055] In this embodiment, the system extracts annotation parameters from AI-pre-generated mold drawing data, including various dimensional annotations, roughness symbols, welding symbols, heat treatment requirements, and other technical annotations. After extraction, the system checks whether all annotations are complete, whether the annotation positions conform to specifications, and whether the annotation values ​​are consistent with the corresponding geometric elements. The annotation deviation is calculated as follows: the number of missing items, the number of items with non-compliant positions, and the number of items with inconsistent values ​​are counted among all items that should be annotated. These three types of errors are weighted and then divided by the total number of items that should be annotated to obtain the comprehensive annotation deviation. A comprehensive annotation deviation of zero indicates that the annotation is completely correct.

[0056] It should be noted that in other embodiments of the present invention, the deviation calculations of the above four dimensions can be added, deleted, or the priority of each item can be adjusted according to the actual application scenario. For example, for scenarios that only focus on geometric accuracy, only geometric accuracy verification can be performed while the verification of other dimensions is omitted. For complex mold design scenarios that focus on assembly accuracy, the priority of structural interference verification can be increased.

[0057] S50. Under the condition that the deviation exists, a correction instruction is generated according to the deviation, and the correction instruction is fed back to the AI ​​drawing model. The AI ​​drawing model is controlled to iteratively correct the mold drawing data according to the correction instruction, and jumps to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all parameters meet the preset convergence conditions.

[0058] The purpose of this step is to establish a closed-loop automatic correction mechanism for AI drawing. Through iterative cycles of comparison, correction, and re-comparison, the mold drawing data generated by AI gradually approaches the standard values ​​in the multi-layered correlated dataset, ultimately achieving the accuracy requirements for industrial mass production.

[0059] Furthermore, under the condition of the aforementioned deviation, a correction instruction is generated based on the deviation, and the correction instruction is fed back to the AI ​​drawing model. The AI ​​drawing model is then controlled to iteratively correct the mold drawing data according to the correction instruction, and the process jumps to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all parameters meet the preset convergence condition. Specifically, this includes: S51. Obtain the deviation, which includes geometric deviation, process deviation, structural deviation and labeling deviation.

[0060] In this embodiment, the system obtains all deviation values ​​calculated in steps S41 to S44, including geometric deviation, process deviation, structural deviation, and annotation deviation. These four deviation values ​​together constitute a complete deviation profile between the current AI-generated drawing and the standard values ​​of the multi-layered associated dataset.

[0061] S52. If any one of the geometric deviation, process deviation, structural deviation, and annotation deviation is non-zero, a correction instruction for the corresponding dimension is generated based on the non-zero deviation. The correction instruction is fed back to the AI ​​drawing model, which then controls the AI ​​drawing model to correct the mold drawing data according to the correction instruction, generating a corrected version of the mold drawing data. The process then jumps to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all the geometric deviation, process deviation, structural deviation, and annotation deviation are reduced to zero.

[0062] In this embodiment, the system first determines whether the geometric deviation, process deviation, structural deviation, and annotation deviation are all zero. If all are zero, it means that the currently generated mold drawing data completely matches the standard values ​​of the multi-layer associated dataset in all dimensions, and no correction is required. The system then proceeds directly to step S60.

[0063] If any of the four deviations mentioned above is non-zero, the system enters an iterative correction process. For each non-zero deviation dimension, the system generates a correction instruction for that dimension based on the specific value of the deviation and the direction of deviation (too large or too small). For example: If there is a geometric deviation, the system generates a geometric correction command based on the direction of deviation of each geometric parameter, such as increasing the bending angle by a specified value or adjusting the hole position coordinates by a specified distance in a certain direction.

[0064] If there is a process deviation, the system generates a process correction instruction based on the direction of deviation of each process parameter, such as adjusting the bending radius to the recommended value or adjusting the punching gap to a specified multiple of the material thickness.

[0065] If structural deviations exist, the system generates structural correction instructions based on the specific location of the interference or deviation. For example, it may translate a component a specified distance in a specified direction to eliminate the interference, or increase the clearance of the mating surfaces of a component by a specified value.

[0066] If there is a labeling deviation, the system generates a labeling correction instruction based on the missing or inconsistent labeling items, such as adding dimension labels at specified locations or correcting the roughness symbol from the current value to the target value.

[0067] The system feeds back the correction instructions to the AI ​​drawing model, controlling the AI ​​drawing model to partially redraw or adjust parameters of the current version of the mold drawing data according to the correction instructions. After the correction is completed, the system generates the corrected version of the mold drawing data and automatically jumps back to step S40 to perform a new round of multi-dimensional parameter extraction and comparison.

[0068] The above iterative process continues until all geometric deviations, process deviations, structural deviations, and annotation deviations are reduced to zero, meaning that all parameters meet the preset convergence conditions.

[0069] In one specific embodiment of the present invention, AI drawing generation of a sheet metal chassis panel is taken as an example. When the correction unit performs the comparison, it detects that the springback compensation angle at the bend is too low (deviation of 0.8 degrees). Based on this deviation, the system generates a geometric correction instruction to "increase the bending angle by 0.8 degrees". After the system feeds this correction instruction back to the AI ​​model, the AI ​​model corrects the bending line angle and redraws the data of the bend development diagram. After the correction is completed, the system jumps back to step S40 to perform a new round of comparison to confirm that the deviation has been eliminated. After two rounds of iteration, all parameters are completely matched, and all deviations are reduced to zero.

[0070] In another specific embodiment of the present invention, the AI ​​drawing generation of a trapezoidal spiral cooling coil mold is taken as an example. During the correction process, it was found that the fillet radius of the punch deviated from the recommended value and the spiral helix angle annotation was missing. The system generates geometric correction instructions and annotation correction instructions respectively, and the AI ​​model corrects the fillet radius of the punch and completes the spiral helix angle annotation accordingly.

[0071] It should be noted that, in other embodiments of the present invention, the preset convergence condition can be adjusted according to the actual accuracy requirements. For example, for scenarios where the accuracy requirements are not high, the convergence condition can be set to ensure that all deviations are less than their respective tolerance thresholds, rather than forcibly requiring all deviations to be zero.

[0072] S60. After all parameters meet the preset convergence conditions, output the target mold drawing data and simultaneously generate a drawing data traceability list.

[0073] The purpose of this step is to output the mold drawing data, which has reached the industrial accuracy requirements after multiple rounds of iterative corrections, in a standardized format, while also fully recording the related data and correction process of the entire drawing process.

[0074] Furthermore, after all parameters satisfy the preset convergence condition, the target mold drawing data is output, and a drawing data traceability list is generated simultaneously, specifically including: S61. After all parameters meet the preset convergence conditions, the mold drawing data of the current version is determined as the target mold drawing data, and the target mold drawing data is output.

[0075] In this embodiment, when all parameters meet the preset convergence conditions (i.e., all deviations are zero or all deviations are less than the preset tolerance threshold), the system determines the latest version of the mold drawing data as the target mold drawing data. Then, the system renders the target mold drawing data according to the customer-specific drawing specification template or industry standard template in the constraint parameter package, including applying the correct drawing frame, layer color, annotation font, line type scale, etc. After rendering, the system outputs the target mold drawing data in the specified file format (including displaying the target mold drawing data in CAD drawing or 3D drawing format).

[0076] S62. Obtain the unique association ID of all related data records called in this drawing process, the version number of the AI ​​drawing model, the deviation value generated in each round of comparison, and the correction record of each round of correction. Combine the obtained unique association ID, version number, deviation value, and correction record into a drawing data traceability list, and store the drawing data traceability list in association with the target mold drawing data.

[0077] In this embodiment, while outputting the target mold drawing data, the system automatically collects key data from the entire drawing process, including: First, the unique association ID of all related data records called in this mapping task records all data sources on which the drawing was based.

[0078] Second, the version number of the AI ​​drawing model, which records the model version used to generate the drawing.

[0079] Third, the deviation value generated in each round of comparison. This data records the complete change trajectory of each deviation obtained in each round of comparison in step S40.

[0080] Fourth, the correction record for each round of correction, which records the specific content and execution results of each round of correction instructions in step S52.

[0081] The system combines the above four types of data into a complete traceability list of cartographic data. This traceability list is stored in a structured data format (such as JSON, XML, or database records) and associated with the target mold drawing data.

[0082] In one specific embodiment of the present invention, AI drawing generation of sheet metal chassis panels is taken as an example. While outputting the target mold drawing data, the system generates a traceability list containing an association ID (e.g., "ASSOC-2024-0087"), an AI model version number (e.g., "AIDraw-v3.2.1"), a sequence of deviation values ​​from two rounds of comparison (geometric deviation exists in the first round, all deviations are zero in the second round), and correction records from both rounds ("First round, bending angle increased by 0.8 degrees"). The total time taken is 4 minutes and 20 seconds. After this traceability list is stored in association with the target mold drawing data, when quality problems occur in the target mold drawing data in the future, engineers can quickly locate the source of the problem through the traceability list, determining whether it is due to missing data in the associated dataset, association errors, or deviations in the AI ​​model version.

[0083] It should be noted that the map data traceability list not only serves to trace the quality of the current drawings, but can also be used to supplement multi-level related datasets, so that the datasets can be continuously expanded, the relationships can be continuously optimized, and the AI ​​model can be continuously iterated.

[0084] Furthermore, such as Figure 2 As shown, based on the above-mentioned high-precision AI mapping generation method based on associated datasets, the present invention also provides a high-precision AI mapping generation system based on associated datasets, the high-precision AI mapping generation system based on associated datasets comprising: Data set construction module 51 is used to construct a multi-level associated dataset, wherein the multi-level associated datasets in the multi-level associated dataset establish a bidirectional mapping relationship between each layer of associated datasets through an associated index engine; The retrieval and matching module 52 is used to obtain product demand data, perform multi-level matching retrieval in the multi-layered associated dataset based on the product demand data, recall the matched associated data records, and combine all the recalled associated data records into a constraint parameter package. The constraint generation module 53 is used to input the constraint parameter package into the AI ​​drawing model, and generate mold drawing data through the AI ​​drawing model based on the product requirement data and the constraint parameter package; The multidimensional comparison module 54 is used to extract multidimensional parameters from the mold drawing data, compare the multidimensional parameters with the standard values ​​in the multi-layer association dataset item by item, and calculate the deviations of each item. The iterative correction module 55 is used to generate a correction instruction based on the deviation when the deviation exists, feed the correction instruction back to the AI ​​drawing model, control the AI ​​drawing model to iteratively correct the mold drawing data according to the correction instruction, and jump to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all parameters meet the preset convergence conditions. The output archiving module 56 is used to output the target mold drawing data and generate a drawing data traceability list simultaneously after all parameters meet the preset convergence conditions.

[0085] Furthermore, such as Figure 3 As shown, based on the above-mentioned AI mapping high-precision generation method and system based on associated datasets, the present invention also provides a terminal, which includes a processor 10, a memory 20 and a display 30. Figure 3 Only some of the terminal components are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0086] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory. In other embodiments, the memory 20 may be an external storage device of the terminal, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc. Further, the memory 20 may include both internal and external storage devices. The memory 20 is used to store application software and various types of data installed on the terminal, such as program code installed on the terminal. The memory 20 can also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores an AI mapping high-precision generation program 40 based on a correlated dataset, which can be executed by the processor 10 to implement the AI ​​mapping high-precision generation method based on a correlated dataset in this application.

[0087] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, such as executing the AI ​​mapping high-precision generation method based on the associated dataset.

[0088] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the terminal and to display a visual user interface. The components of the terminal communicate with each other via a system bus.

[0089] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores an AI mapping high-precision generation program based on an associated dataset, and the AI ​​mapping high-precision generation program based on the associated dataset, when executed by a processor, implements the steps of the AI ​​mapping high-precision generation method based on the associated dataset as described above.

[0090] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal. Unless otherwise specified, an element defined by the phrase "comprising one" does not exclude the presence of other identical elements in the process, method, article, or terminal that includes that element.

[0091] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0092] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A high-precision AI mapping generation method based on associated datasets, characterized in that, The AI ​​mapping high-precision generation method based on associated datasets includes the following steps: Construct a multi-level associated dataset, wherein the associated datasets at each level in the multi-level associated dataset establish a bidirectional mapping relationship through an associated index engine; Obtain product demand data, perform multi-level matching retrieval in the multi-layered associated dataset based on the product demand data, recall the matching associated data records, and combine all the recalled associated data records into a constraint parameter package; The constraint parameter package is input into the AI ​​drawing model, and the AI ​​drawing model generates mold drawing data based on the product requirement data and the constraint parameter package. Extract multi-dimensional parameters from the mold drawing data, compare each multi-dimensional parameter with the standard value in the multi-layered associated dataset, and calculate the deviation of each item. Under the condition of the aforementioned deviation, a correction instruction is generated based on the deviation, and the correction instruction is fed back to the AI ​​drawing model. The AI ​​drawing model is controlled to iteratively correct the mold drawing data according to the correction instruction, and then jumps to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all parameters meet the preset convergence conditions. After all parameters meet the preset convergence conditions, the target mold drawing data is output, and a drawing data traceability list is generated simultaneously.

2. The high-precision AI mapping generation method based on associated datasets according to claim 1, characterized in that, The construction of a multi-level associated dataset, wherein the various levels of the associated dataset in the multi-level associated dataset establish a bidirectional mapping relationship through an associated index engine, specifically includes: Acquire multi-source industrial data, parse and extract fields from the multi-source industrial data to obtain structured data; The structured data is mapped to predefined standard fields in the multi-level associated dataset. Based on the primary key information in the structured data, the structured data is automatically associated with the records of each level of the multi-level associated dataset to form structured associated records. A unique association ID is assigned to the structured associated records, and the bidirectional mapping relationship is established between the levels of the multi-level associated dataset through the unique association ID.

3. The high-precision AI mapping generation method based on associated datasets according to claim 1, characterized in that, The process of acquiring product demand data, performing multi-level matching retrieval based on the product demand data in the multi-layered associated dataset, recalling matching associated data records, and combining all recalled associated data records into a constraint parameter package specifically includes: Obtain product requirement data, and perform standardized verification and completion on the product requirement data to obtain target product requirement data; Using the target product demand data as the retrieval primary key, multi-level matching retrieval is performed in the multi-layered associated dataset, and all matching associated data records are retrieved. The retrieved associated data records are then combined into a constraint parameter package.

4. The high-precision AI mapping generation method based on associated datasets according to claim 1, characterized in that, The step of inputting the constraint parameter package into the AI ​​drawing model, and generating mold drawing data by the AI ​​drawing model based on the product requirement data and the constraint parameter package, specifically includes: Encode each parameter in the constraint parameter package into a condition vector; The conditional vector is concatenated with the latent space features of the AI ​​mapping model to obtain concatenated features. The concatenated features are then input into the AI ​​mapping model, which controls the AI ​​mapping model to generate the mold drawing data based on the product requirement data within the feasible domain defined by the constraint parameter package.

5. The high-precision AI mapping generation method based on associated datasets according to claim 1, characterized in that, The step of extracting multi-dimensional parameters from the mold drawing data, comparing each multi-dimensional parameter with the standard value in the multi-layered association dataset, and calculating the deviations for each parameter specifically includes: Extract the geometric parameters from the mold drawing data, compare the geometric parameters with the standard geometric constraints in the multi-layer associated dataset, and calculate the geometric deviation; Extract the process parameters from the mold drawing data, compare the process parameters with the standard parameter ranges of corresponding materials and processes in the multi-layer associated dataset, and calculate the process deviation; Extract the structural parameters from the mold drawing data, verify the assembly relationship of the structural parameters, and calculate the structural deviation; Extract the annotation parameters from the mold drawing data, verify the completeness and consistency of the annotation parameters, and calculate the annotation deviation.

6. The high-precision AI mapping generation method based on associated datasets according to claim 1, characterized in that, Under the condition of the aforementioned deviation, a correction instruction is generated based on the deviation, and the correction instruction is fed back to the AI ​​drawing model. The AI ​​drawing model is then controlled to iteratively correct the mold drawing data according to the correction instruction, and the process jumps to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all parameters meet the preset convergence condition. Specifically, this includes: The deviations are obtained, including geometric deviations, process deviations, structural deviations, and annotation deviations; If any of the geometric deviation, process deviation, structural deviation, and annotation deviation is non-zero, a correction instruction for the corresponding dimension is generated based on the non-zero deviation. The correction instruction is fed back to the AI ​​drawing model, which then corrects the mold drawing data according to the correction instruction, generating a corrected version of the mold drawing data. The process then jumps to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all the geometric deviation, process deviation, structural deviation, and annotation deviation are reduced to zero.

7. The high-precision AI mapping generation method based on associated datasets according to claim 1, characterized in that, After all parameters meet the preset convergence condition, the target mold drawing data is output, and a drawing data traceability list is generated simultaneously, specifically including: After all parameters meet the preset convergence condition, the current version of the mold drawing data is determined as the target mold drawing data, and the target mold drawing data is output. Obtain the unique association ID of all related data records called during this mapping process, the version number of the AI ​​mapping model, the deviation value generated in each round of comparison, and the correction record of each round of correction. Combine the obtained unique association ID, version number, deviation value, and correction record into a mapping data traceability list, and store the mapping data traceability list in association with the target mold mapping data.

8. A high-precision AI mapping generation system based on associated datasets, characterized in that, The AI ​​mapping high-precision generation system based on associated datasets is used to implement the AI ​​mapping high-precision generation method based on associated datasets as described in any one of claims 1-7, wherein the AI ​​mapping high-precision generation system based on associated datasets includes: The dataset construction module is used to construct a multi-level associated dataset, wherein the associated datasets at each level in the multi-level associated dataset establish a bidirectional mapping relationship through an associated index engine; The retrieval and matching module is used to obtain product demand data, perform multi-level matching retrieval in the multi-layered associated dataset based on the product demand data, recall the matched associated data records, and combine all the recalled associated data records into a constraint parameter package. The constraint generation module is used to input the constraint parameter package into the AI ​​drawing model, and generate mold drawing data through the AI ​​drawing model based on the product requirement data and the constraint parameter package; The multidimensional comparison module is used to extract multidimensional parameters from the mold drawing data, compare the multidimensional parameters with the standard values ​​in the multi-layer association dataset item by item, and calculate the deviations of each item. The iterative correction module is used to generate correction instructions based on the deviation when the deviation exists, feed the correction instructions back to the AI ​​drawing model, control the AI ​​drawing model to iteratively correct the mold drawing data according to the correction instructions, and jump to the step of extracting multi-dimensional parameters from the mold drawing data for a new round of comparison until all parameters meet the preset convergence conditions. The output archiving module is used to output the target mold drawing data and generate a drawing data traceability list simultaneously after all parameters meet the preset convergence conditions.

9. A terminal, characterized in that, The terminal includes: a memory, a processor, and an AI mapping high-precision generation program based on an associated dataset stored in the memory and executable on the processor. When the AI ​​mapping high-precision generation program based on an associated dataset is executed by the processor, it implements the steps of the AI ​​mapping high-precision generation method based on an associated dataset as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an AI mapping high-precision generation program based on an associated dataset, which, when executed by a processor, implements the steps of the AI ​​mapping high-precision generation method based on an associated dataset as described in any one of claims 1-7.

Images