CN-121980702-A - Welding fixture design method and system based on large language model

CN121980702ACN 121980702 ACN121980702 ACN 121980702ACN-121980702-A

Abstract

The invention provides a large language model-based welding jig design method and system, comprising the steps of clustering a plurality of historical welding jig assemblies to obtain a plurality of welding jig groups, establishing a parameterized template, traversing the initial welding jig assemblies to obtain initial welding jig assembly parameters, inputting the welding jig design requirements into a large language model to obtain a matched parameterized template, analyzing the initial welding jig assembly parameters through the large language model, combining the matched parameterized template to obtain a parameter modification instruction, and adjusting the initial welding jig assemblies until the initial welding jig assembly parameters meet the optimization requirements to obtain target welding jig assembly parameters so as to generate target welding jig assemblies, thereby completing the welding jig design. According to the invention, the design experience of the welding fixture is parameterized, and the parameter semantics are extracted through the large language model to correspondingly optimize and adjust the parameters of the assembly body of the welding fixture, so that the welding fixture design meeting engineering requirements is generated, and the design reliability of the welding fixture is improved.

Inventors

LV JUNCHENG
Liu changye
QI JIANDE
WANG HU
LI GUANGYAO
Lv Gaoguang
MENG DAQING
LAI WEILIN
NIU BEN

Assignees

上汽通用五菱汽车股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260122

Claims (10)

1. The method for designing the welding fixture based on the large language model is characterized by comprising the following steps of: Clustering and grouping the plurality of history welding jig assemblies based on the plurality of history welding jig assemblies obtained in advance to obtain a plurality of welding jig groups and establishing a parameterized template based on the plurality of welding jig groups; traversing the pre-obtained initial welding fixture assembly to obtain initial welding fixture assembly parameters; Based on the pre-acquired welding fixture design requirement, inputting the welding fixture design requirement into a preset large language model to match a corresponding parameterized template, and obtaining a matched parameterized template; Inputting the initial welding fixture assembly parameters and the matching parameterized templates into the preset large language model, analyzing the initial welding fixture assembly parameters through the preset large language model, and obtaining parameter modification instructions based on the matching parameterized templates; adjusting the initial welding jig assembly based on the parameter modification instruction until the initial welding jig assembly parameters meet preset optimization requirements, so as to obtain target welding jig assembly parameters; And generating a target welding jig assembly based on the target welding jig assembly parameters to complete the welding jig design.
2. The method for designing a welding jig based on a large language model according to claim 1, wherein the clustering grouping of the plurality of history welding jig assemblies based on the plurality of pre-acquired history welding jig assemblies to obtain a plurality of welding jig groups and establishing a parameterized template based on the plurality of welding jig groups comprises: acquiring a plurality of history welding fixture assemblies from a preset history design library; extracting historical semantic information corresponding to a plurality of historical welding fixture assemblies; Carrying out semantic clustering on the historical semantic information corresponding to each historical welding jig assembly according to a preset clustering rule, and taking each historical welding jig assembly with the same type of the historical semantic information as a corresponding welding jig group to obtain a plurality of welding jig groups; And selecting a plurality of target parameters meeting preset parameter dimensions from each welding fixture group, and setting constraint relations for the target parameters to obtain parameterized templates corresponding to the welding fixture groups.
3. The method for designing a welding jig based on a large language model according to claim 1, wherein said traversing the pre-acquired initial welding jig assembly to obtain initial welding jig assembly parameters comprises: Traversing the pre-acquired initial welding fixture assemblies through a preset CAD script interface, and extracting the corresponding spatial relation parameters, functional semantic parameters and dependency relation parameters of each initial welding fixture assembly; And converting the spatial relation parameters, the functional semantic parameters and the dependency relation parameters into a preset structured text format based on a preset parameter structuring standard to obtain initial welding fixture assembly parameters.
4. The welding jig design method based on a large language model as claimed in claim 3, wherein based on the pre-acquired welding jig design requirement, and inputting the welding jig design requirement into a parameterized template corresponding to the pre-set large language model for matching, obtaining a matched parameterized template, comprises: Based on the pre-acquired geometric feature information, welding process information and clamp installation environment information, converting the geometric feature information, the welding process information and the clamp installation environment information into a target text format, and obtaining welding clamp design requirements; inputting the design requirement of the welding fixture into a preset large language model, and matching the parameterized templates meeting the preset design requirement to obtain matched parameterized templates.
5. The method of claim 4, wherein inputting the initial welding jig assembly parameters and the matching parameterized templates into the pre-set large language model, parsing the initial welding jig assembly parameters through the pre-set large language model, and obtaining parameter modification instructions based on the matching parameterized templates, comprises: Inputting the parameters of the initial welding fixture assembly and the matching parameterized template into a preset large language model, and analyzing the functional relation of the initial welding fixture assembly through the preset large language model to obtain a parameter dependence spectrum; Generating a parameter modification suggestion based on the matching parameterized template and the parameter dependent graph; and converting the parameter modification suggestion into an instruction script format to obtain a parameter modification instruction.
6. The large language model based welding jig design method of claim 1, wherein adjusting the initial welding jig assembly based on the parameter modification instructions until the initial welding jig assembly parameters meet a preset optimization requirement, obtaining target welding jig assembly parameters, comprises: adjusting the initial welding jig assembly based on the parameter modification instruction to obtain an optimized welding jig assembly; sequentially performing interference test and stability test on the optimized welding fixture assembly to obtain a collision report and a stability report; Inputting the collision report and the stability report into a preset large model, and evaluating the initial welding fixture assembly parameters based on a pre-constructed loss function to generate an evaluation feedback result; and if the evaluation feedback result does not meet the preset optimization requirement, adjusting the initial welding fixture assembly until the evaluation feedback result meets the preset optimization requirement, and obtaining the parameters of the target welding fixture assembly.
7. The method of designing a welding jig based on a large language model as claimed in claim 6, wherein in the step of inputting the collision report and the stability report into a preset large model, evaluating the initial welding jig assembly parameters based on a pre-constructed loss function, and generating an evaluation feedback result, the pre-construction process of the loss function includes: Acquiring an initial point location based on the initial welding fixture assembly; Acquiring a target point location based on the matching parameterized template; calculating the distance error of the initial point position and the target point position to obtain a clamping point error item; and setting corresponding weight coefficients for a preset collision punishment item, a preset clamping stability index and the clamping point error item to obtain a loss function.
8. A large language model based welding jig design system, characterized by performing a large language model based welding jig design method as claimed in any one of claims 1 to 7, comprising: the device comprises a parameterized template construction module, a parameter acquisition module, a matching parameterized template acquisition module, a parameter modification instruction generation module, a parameter optimization module and a welding fixture design module; The parameterized template construction module is used for clustering and grouping a plurality of history welding jig assemblies based on the plurality of history welding jig assemblies obtained in advance to obtain a plurality of welding jig groups and establishing a parameterized template based on the plurality of welding jig groups; the parameter acquisition module is used for traversing the pre-acquired initial welding fixture assembly to obtain initial welding fixture assembly parameters; The matching parameterized template acquisition module is used for inputting the welding fixture design requirement into a corresponding parameterized template matched with a preset large language model based on the pre-acquired welding fixture design requirement to obtain a matching parameterized template; the parameter modification instruction generation module is used for inputting the initial welding fixture assembly parameters and the matching parameterized templates into the preset large language model, analyzing the initial welding fixture assembly parameters through the preset large language model, and obtaining parameter modification instructions based on the matching parameterized templates; the parameter optimization module is used for adjusting the initial welding jig assembly based on the parameter modification instruction until the initial welding jig assembly parameters meet preset optimization requirements, so as to obtain target welding jig assembly parameters; The welding jig design module is used for generating a target welding jig assembly based on the target welding jig assembly parameters so as to complete the welding jig design.
9. The large language model based welding jig design system of claim 8, wherein the parameterized template construction module is configured to cluster a plurality of historical welding jig assemblies based on the pre-acquired plurality of historical welding jig assemblies to obtain a plurality of welding jig groups and to establish a parameterized template based on the plurality of welding jig groups, comprising: the device comprises a history assembly body acquisition unit, a history semantic information acquisition unit, a welding fixture group construction unit and a template construction unit; The history assembly acquisition unit is used for acquiring a plurality of history welding fixture assemblies from a preset history design library; the history semantic information acquisition unit is used for extracting history semantic information corresponding to a plurality of history welding jig assemblies; The welding jig set construction unit is used for carrying out semantic clustering on the historical semantic information corresponding to each historical welding jig assembly according to a preset clustering rule, and taking each historical welding jig assembly with the same type of the historical semantic information as a corresponding welding jig set to obtain a plurality of welding jig sets; The template construction unit is used for selecting a plurality of target parameters meeting the preset parameter dimension in each welding fixture group, and setting constraint relations for the target parameters to obtain parameterized templates corresponding to the welding fixture groups.
10. The large language model based welding jig design system of claim 8, wherein the parameter acquisition module is configured to traverse the pre-acquired initial welding jig assembly to obtain initial welding jig assembly parameters, comprising: a parameter extraction unit and a text format conversion unit; The parameter extraction unit is used for traversing the pre-acquired initial welding fixture assemblies through a preset CAD script interface and extracting the spatial relation parameters, the functional semantic parameters and the dependency relation parameters corresponding to each initial welding fixture assembly; The text format conversion unit is used for converting the spatial relation parameters, the functional semantic parameters and the dependency relation parameters into a preset structured text format based on a preset parameter structuring standard to obtain initial welding fixture assembly parameters.

Description

Welding fixture design method and system based on large language model Technical Field The invention relates to the field of automatic design of vehicle body clamps, in particular to a welding clamp design method and system based on a large language model. Background The welding fixture is used for positioning, restraining and supporting welding workpieces in a welding process on an automobile production assembly line, has various structural types, is generally a customized non-standard part, has a large number of reusable design resources along with accumulation of enterprise historical CAD data module data, and is a technical problem to be studied at present, and how to reliably adapt the design resources to various complex engineering requirements. At present, the prior art relies on engineer experience and manual modeling, and is repeatedly constructed and adjusted in a parameter or sketch mode in three-dimensional modeling software according to the same welding workpiece model under different engineering requirements, and interference check and process verification are carried out, but the method is complicated in flow and is easily influenced by human subjective factors, the design reliability of the welding fixture is poor, moreover, the prior art adopts an automatic path such as a rule script and a template drive, when facing various engineering requirements, the method generally needs a large amount of manual maintenance rules and templates, relies on manual experience, and is easily influenced by the human subjective factors under complex working conditions, so that the design reliability of the welding fixture is poor. Therefore, in the prior art, when facing to complex working conditions, the problem of poor design reliability of the welding fixture is caused by difficulty in meeting complex engineering requirements. Disclosure of Invention In order to solve the problems, the invention provides a welding jig design method and a system based on a large language model, which are used for parameterizing the welding jig design experience and extracting parameter semantics through the large language model to correspondingly optimize and adjust welding jig assembly parameters so as to generate a welding jig design meeting engineering requirements and improve the design reliability of the welding jig. The embodiment of the invention provides a welding jig design method based on a large language model, which comprises the steps of clustering and grouping a plurality of history welding jig assemblies based on a plurality of pre-acquired history welding jig assemblies to obtain a plurality of welding jig groups and establishing a parameterized template based on the plurality of welding jig groups, traversing the pre-acquired initial welding jig assemblies to obtain initial welding jig assembly parameters, inputting the welding jig design requirements into a corresponding parameterized template matched with the preset large language model based on the pre-acquired welding jig design requirements to obtain a matched parameterized template, inputting the initial welding jig assembly parameters and the matched parameterized template into the preset large language model, analyzing the initial welding jig assembly parameters through the preset large language model, obtaining a parameter modification instruction based on the matched parameterized template, adjusting the initial welding jig assemblies based on the parameter modification instruction until the initial welding jig assembly parameters meet preset optimization requirements to obtain target welding jig assembly parameters, and generating the target welding jig assemblies based on the target welding jig assembly parameters to complete the welding jig design. The embodiment of the invention provides a welding fixture design method based on a large language model, which comprises the steps of constructing a parameterized template, solidifying changeable design experience, extracting initial welding fixture assembly parameters which are understandable by a preset large language model from an initial welding fixture assembly, matching according to the design requirement of the welding fixture to obtain a matched parameterized template, inputting the initial welding fixture assembly parameters and the matched parameterized template into the preset large language model to replace manual reasoning, generating a parameter modification instruction to adjust the initial welding fixture assembly, acquiring target welding fixture assembly parameters by adopting an iterative optimization mode to generate a target welding fixture assembly, and finally completing the welding fixture design. Therefore, by constructing a parameterized template library and extracting structural parameters of the assembly, then carrying out semantic analysis and parameter reasoning by utilizing a preset large language model, and generating a welding fixture design meeting engineering requ