Technical drawing generation system and technical drawing generation method

Abstract

This invention provides a design drawing generation system and a design drawing generation method. The design drawing generation system includes a storage device and a processor. The processor is coupled to the storage device and is used to execute a parameter set generation module, a macro code generation module, and a drawing rendering module. The parameter set generation module parses and processes a requirement file to obtain design requirement information. The parameter set generation module performs correlation analysis on the design requirement information and sample data to obtain analysis results, and matches the corresponding drawing type according to the analysis results to obtain the parameter set of the design drawing. The macro code generation module generates drawing macro code according to the design logic of the corresponding drawing type and the parameter set of the design drawing. The drawing rendering module renders the drawing according to the drawing macro code to generate the design drawing.

Claims

1. A design drawing generation system, comprising: Storage device for storing parameter set generation module, macro code generation module, and drawing rendering module; The system includes a processor coupled to the storage device, which executes the parameter set generation module, the macro code generation module, and the drawing rendering module, and receives a requirement file. The parameter set generation module parses the requirement file to obtain design requirement information. It performs correlation analysis on the design requirement information and sample data from a design experience base to obtain analysis results, and matches the corresponding drawing type based on the analysis results to obtain the parameter set for the design drawing. The macro code generation module generates drawing macro code based on the design logic corresponding to the drawing type and the parameter set of the design drawing. The drawing rendering module renders the drawing based on the drawing macro code to generate the design drawing. The parameter set generation module includes a correlation analysis module and a parameter completion module. The correlation analysis module compares the analysis results with a threshold; if the analysis result is greater than the threshold, the correlation analysis module matches the drawing type based on the drawing parameter set in the design experience base to obtain the design logic. In response to the analysis result being less than the threshold, the correlation analysis module performs slicing processing on the design requirement information to obtain a minimum subset of the design requirement information, and performs the correlation analysis again on the minimum subset to obtain a new analysis result, until the analysis result is greater than the threshold. The parameter completion module completes the relevant parameters of the design requirement information based on the matched drawing type using a nearest neighbor algorithm to obtain the parameter set.

2. The design drawing generation system as described in claim 1, wherein the parameter set generation module includes a parsing module, wherein the parsing module parses the requirement document to determine the type of the requirement document, wherein in response to the type being a file type, the parsing module parses the requirement document to obtain file information, and summarizes the file information through a large language model to obtain the requirement information, wherein in response to the type being a text type, the parsing module organizes and analyzes the requirement document through the large language model to obtain the requirement information.

3. The design drawing generation system as described in claim 1, wherein the macro code generation module includes a rule comparison module and a code generation module, wherein the rule comparison module searches for corresponding design rules from the design experience base according to the design logic, wherein the rule comparison module compares the parameter set and the design rules to determine whether the parameter set conforms to the design rules, and in response to the parameter set not conforming to the design rules, the rule comparison module generates an error message, and in response to the parameter set conforming to the design rules, the code generation module generates the drawing macro code according to the design logic and the parameter set through a code generation model.

4. The design drawing generation system as claimed in claim 1, wherein the storage device further stores a verification module, wherein the processor executes the verification module to perform verification processing on the drawing macro code, and in response to the drawing macro code passing the verification processing, the verification module inputs the drawing macro code to the drawing rendering module, and in response to the drawing macro code failing the verification processing, the verification module outputs a verification prompt to the display according to the failed verification rules.

5. The design drawing generation system as described in claim 1, wherein the design drawings include three-dimensional design drawings and two-dimensional cross-sectional drawings, wherein the parameter set generation module receives drawing adjustment instructions and adjusts the design requirement information based on the drawing adjustment instructions.

6. The design drawing generation system as described in claim 1, wherein the correlation analysis module performs document normalization on the design requirement information, thereby adjusting the design requirement information into a format that conforms to the design experience base.

7. The design drawing generation system as described in claim 1, wherein the storage device further stores a model training module, wherein the processor executes the model training module to collect and organize design rules and design documents to generate a dataset, wherein the model training module slices the dataset to input the sliced ​​dataset into a large model for iterative training of the large model, wherein the model training module inputs fine data into the large model for reinforcement learning to generate a code generation model, wherein the model training module establishes a design experience base based on the dataset and the fine data.

8. The design drawing generation system as described in claim 8, wherein the model training module inputs the data into the large model at a coarse-to-fine ratio of 9:1 for N iterations of training, where N is a hyperparameter; wherein after X iterations, the model training module performs random checks on the code generated by the large model to generate a check result; wherein if the check result is a failure, the model training module changes the coarse-to-fine ratio of the data to 8:2 and inputs it into the large model, repeating this process N times.

9. The design drawing generation system as claimed in claim 9, wherein the sampling result is passed, and wherein the model training module then performs the reinforcement learning on the large model; The steps of the sampling process performed by the model training module include: The model training module performs random checks on the code based on the test set in the data to obtain code quality; In response to the code quality being greater than a threshold, the model training module generates a passing sampling result; in response to the code quality not being greater than the threshold, the model training module generates a failing sampling result.

10. The design drawing generation system as described in claim 1, wherein the drawing rendering module integrates the design drawing into third-party 3D computer-aided design software to present the design drawing, wherein the drawing rendering module receives editing instructions and adjusts the design drawing according to the editing instructions.

11. A method for generating design drawings, comprising: The parameter set generation module is executed by the processor to parse and process the requirement file, so that the parameter set generation module can obtain design requirement information. The processor executes a parameter set generation module to perform correlation analysis on the design requirement information and sample data from the design experience base, so that the parameter set generation module obtains analysis results; the parameter set generation module matches the corresponding drawing type according to the analysis results to obtain the parameter set of the design drawing; the processor executes a macro code generation module to generate drawing macro code according to the design logic of the drawing type and the parameter set of the design drawing; the processor executes a drawing rendering module to render the drawing according to the drawing macro code, thereby generating the design drawing, wherein the step of matching the corresponding drawing type according to the analysis results to obtain the parameter set of the design drawing includes: comparing the analysis results and a threshold through a correlation analysis module; In response to the analysis result being greater than the threshold, the correlation analysis module performs drawing type matching based on the drawing parameter set in the design experience base to obtain the design logic; in response to the analysis result being less than the threshold, the correlation analysis module performs slicing processing on the design requirement information to obtain the smallest subset of the design requirement information, and performs the correlation analysis again on the smallest subset to obtain a new analysis result, until the analysis result is greater than the threshold; and the parameter completion module completes the relevant parameters of the design requirement information based on the matched drawing type using a nearest neighbor algorithm to obtain the parameter set.

12. The method as described in request item 11, wherein the step of the parameter set generation module performing the parsing process on the requirement file includes: The parsing module parses the requirement document to determine its type. If the type is a file, the parsing module parses the requirement document to obtain file information and then parses the file information using a large language model to obtain requirement information. If the type is a text, the parsing module analyzes the requirement document using the code generation model to obtain requirement information.

13. The method as described in claim 11, wherein the step of generating the drawing macro set code includes: The rule comparison module retrieves the corresponding design rules from the design experience base according to the design logic. The rule comparison module then compares the parameter set with the design rules to determine whether the parameter set conforms to the design rules. If the parameter set does not conform to the design rules, the rule comparison module generates an error message. And in response to the parameter set conforming to the design rules, the code generation module generates the drawing macro code through the code generation model based on the design logic and the parameter set.

14. The method of claim 11, wherein prior to the step of rendering the drawing based on the drawing macro code, the method comprises the following steps: executing a verification module through the processor to perform verification processing on the drawing macro code; in response to the drawing macro code passing the verification processing, inputting the drawing macro code into the drawing rendering module through the verification module; and in response to the drawing macro code failing the verification processing, outputting a verification prompt to the display through the verification module according to the failed verification rules.

15. The method as described in claim 11, wherein the design drawings include three-dimensional design drawings and two-dimensional cross-sectional drawings, wherein the method further comprises the steps of: receiving drawing adjustment instructions through the parameter set generation module, and adjusting the design requirement information based on the drawing adjustment instructions through the parameter set generation module.

16. The method of claim 11, wherein prior to performing the correlation analysis, it further comprises: The correlation analysis module performs document standardization on the design requirement information, thereby adjusting the design requirement information into a format that conforms to the design experience base.

17. The method as described in claim 11, further comprising: The processor executes the model training module to generate a dataset according to the design rules and design documents; The dataset is sliced ​​using the model training module, and the sliced ​​data is then input into a large model for iterative training. The model training module inputs fine-grained data into the large model for reinforcement learning to generate a code generation model; and the model training module establishes a design experience base based on the dataset and the fine-grained data.

18. The method of claim 17, wherein the step of iteratively training the large model comprises: The data is input into the large model at a coarse-to-fine ratio of 9:1 through the model training module for N iterations of training, where N is a hyperparameter. After X iterations, the code generated by the large model is sampled and inspected through the model training module to generate inspection results. In response to the sampling result being a failure, the coarse-to-fine ratio of the data is changed to 8:2 through the model training module and input into the large model, and this process is repeated N times.

19. The method of claim 18, wherein in response to the sampling result being passed, the reinforcement learning is then performed on the large model via the model training module; The steps of performing the sampling inspection include: sampling the code based on the test set in the data using the model training module to obtain code quality; generating a passing sampling result using the model training module in response to the code quality being greater than a threshold; and generating a failing sampling result using the model training module in response to the code quality not being greater than the threshold.

20. The method as described in claim 11, further comprising: The processor executes the drawing rendering module to integrate the design drawings into third-party software to present the design drawings. The processor executes the drawing rendering module to receive editing instructions and adjusts the design drawings according to the editing instructions.

Images