Process planning method and device based on large language model and storage medium
By using a process planning method based on a large language model, the problems of low recognition accuracy and long computation time of traditional systems in complex industrial scenarios and new parts are solved, achieving efficient and accurate process planning and rapid response.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN MUSICAL BOOK LANCE TECHNOLOGY CO LTD
- Filing Date
- 2026-01-04
- Publication Date
- 2026-05-01
AI Technical Summary
Traditional computer-aided process planning systems suffer from low recognition accuracy, long calculation time, and difficulty in meeting the needs of rapid quotation and production response when faced with complex industrial scenarios and new parts.
A process planning method based on a large language model is adopted. Through feature recognition, iterative reasoning and knowledge base matching, an optimal process list and process cost table are generated, and autonomous decision-making is carried out using the large language model.
It enables efficient and accurate process planning for complex industrial scenarios and new parts, generates transparent and precise process cost sheets, and adapts to rapid response to complex environments and innovative parts.
Smart Images
Figure CN121960145A_ABST
Abstract
Description
Process planning methods, equipment, and storage media based on large language models Technical Field
[0001] This invention relates to the field of intelligent manufacturing, and in particular to a process planning method, equipment, and storage medium based on a large language model. Background Technology
[0002] Computer-Aided Process Planning (CAPP) is a crucial bridge connecting computer-aided design and computer-aided manufacturing. Its core task is to transform the abstract geometric information in design drawings into detailed process flows and machining instructions required to guide production.
[0003] Traditional CAPP systems typically encompass core components such as feature recognition, process selection, process planning, process route generation, tool / fixture planning, and cost assessment. In the current field of computer-aided process planning, both traditional methods and emerging intelligent algorithms face significant technical bottlenecks and limitations in achieving full-process automation, optimization, and dynamic adaptability from part modeling to final process cost calculation.
[0004] Traditional geometric algorithms face significant challenges in developing the geometric modeling kernel for industrial software, resulting in high licensing costs. Traditional computer vision solutions rely on massive amounts of high-quality labeled data, and their accuracy drops significantly when faced with new part structures, different drawing standards, or damaged or occluded drawings, failing to meet industrial production requirements. Expert rule-based systems suffer from rigid reasoning capabilities, making them ill-suited for complex scenarios and difficult to maintain. Case-based reasoning heavily depends on the quality and breadth of the case library, rendering it completely ineffective for innovative parts with novel structures and no historical precedents. Optimization algorithms, for complex parts, present a combinatorial explosion problem in process planning, requiring numerous iterations and fitness function evaluations, leading to long computation times and hindering the need for rapid pricing and production response. Therefore, a new technology is needed to address the limitations of current computer-aided process planning technologies in handling complex industrial scenarios and new parts. Summary of the Invention
[0005] The main objective of this invention is to solve the technical problem that computer-aided process planning is difficult to apply to complex industrial scenarios and new parts.
[0006] The first aspect of this invention provides a process planning method based on a large language model, comprising the steps of: receiving engineering drawing files; performing feature recognition and extraction processing on the engineering drawing files according to a preset feature extraction algorithm to generate a structured feature list; performing memory block hit processing on the structured feature list based on a preset iterative reasoning algorithm to generate a verification feature list; extracting production requirements from the engineering drawing files; performing reasoning and matching from a preset knowledge base based on a preset large language model and preset processing rules, using the production requirements and the verification feature list as input parameters, to generate an optimal process list; and generating a process cost table based on the optimal process list.
[0007] Optionally, in a first implementation of the first aspect of the present invention, the step of performing feature recognition and extraction processing on the engineering drawing file according to a preset feature extraction algorithm to generate a structured feature list includes: performing geometric type recognition and extraction on the engineering drawing file to generate recognized geometric elements; extracting geometric relationships from the engineering drawing file based on the recognized geometric elements to obtain the dependency relationships corresponding to the recognized geometric elements; and generating a structured feature list based on the recognized geometric elements and the dependency relationships corresponding to the recognized geometric elements.
[0008] Optionally, in a second implementation of the first aspect of the present invention, the step of performing memory block hit processing on the structured feature list based on a preset loop inference algorithm to generate a verification feature list includes: extracting structured features from the structured feature list; determining whether the structured features exist in a preset memory cache; if they exist, marking the structured features as hits; if they do not exist, writing the structured features into a preset object framework to generate verification features, and writing the verification features into a preset memory cache; and integrating the data in the memory cache to generate a verification feature list.
[0009] Optionally, in a third implementation of the first aspect of the present invention, the step of performing geometric type recognition and extraction on the engineering drawing file to generate recognized geometric elements includes: performing recognition and segmentation processing on the engineering drawing file according to a preset YOLO algorithm to obtain labeled image elements and labeled text elements; performing structure recognition on the image elements according to a preset RNN algorithm to obtain the geometric elements corresponding to the image elements; recognizing the text elements based on a preset semantic OCR algorithm to obtain text data; and matching the text data with the geometric elements according to the recognized and segmented label data to obtain the recognized geometric elements.
[0010] Optionally, in a fourth implementation of the first aspect of the present invention, the step of generating an optimal process list by reasoning and matching from a preset knowledge base based on a preset large language model and preset processing rules, using the production requirements and the verification feature list as input parameters, further includes: performing feature reasoning on the production requirements and the verification feature list based on the preset large language model and preset processing rules to generate constraints; selecting equipment and tools from the preset knowledge base based on the constraints to obtain target equipment and target tools; and generating an optimal process list based on the target equipment and the target tools.
[0011] Optionally, in a fifth implementation of the first aspect of the present invention, generating a process cost table based on the optimal process list includes: obtaining the equipment parameters of the target equipment; calculating the production hours based on the equipment parameters; generating a raw material cost table based on the materials and preset material costs in the optimal process list; and obtaining the process cost table based on the raw material cost table and the production hours.
[0012] Optionally, in a sixth implementation of the first aspect of the present invention, the step of selecting equipment and tools from a preset knowledge base based on the constraints to obtain target equipment and target tools includes: performing a process decision query from the preset knowledge base based on the constraints to generate a process combination decision graph; performing cost search processing on the process combination decision graph according to a preset search algorithm to obtain the lowest cost path; and selecting target tools and target equipment from the preset knowledge base based on the lowest cost path.
[0013] Optionally, in a seventh implementation of the first aspect of the present invention, the step of generating an optimal process list based on the target equipment and the target tool includes: combining the process of the target equipment and the target tool based on the lowest cost path to generate an optimal process list.
[0014] A second aspect of the present invention provides a process planning device based on a large language model, comprising: a memory and at least one processor, wherein the memory stores instructions, and the memory and the at least one processor are interconnected via a circuit; the at least one processor invokes the instructions in the memory to cause the process planning device based on the large language model to execute the aforementioned process planning method based on the large language model.
[0015] A third aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the above-described process planning method based on a large language model.
[0016] In this embodiment of the invention, feature recognition analysis is performed on engineering drawing files to extract relevant part information, including names, parameters, and other data. A loop reasoning algorithm is then used to deduplicate the part information, generating a validated feature list. By introducing a large language model and a knowledge base for massive industrial manufacturing data, the optimal process list is inferred from the feature list while meeting the production requirements of the engineering drawing files. Based on this optimal process list, a process cost table is generated. The large language model enables autonomous decision-making based on an understanding of process constraints, effectively adapting to complex industrial environments and innovative parts. It efficiently calculates the theoretically optimal solution, solving the technical problem that computer-aided process planning is difficult to apply to complex industrial scenarios and new parts. Attached Figure Description
[0017] Figure 1 is a schematic diagram of an embodiment of the process planning method based on a large language model in this invention; Figure 2 is a schematic diagram of a specific embodiment of step 102 of the process planning method based on a large language model in this invention; Figure 3 is a schematic diagram of a specific embodiment of step 1021 of the process planning method based on a large language model in this invention; Figure 4 is a schematic diagram of a specific embodiment of step 105 of the process planning method based on a large language model in this invention; Figure 5 is a schematic diagram of an embodiment of the process planning equipment based on a large language model in this invention. Detailed Implementation
[0018] This invention provides a process planning method, equipment, and storage medium based on a large language model.
[0019] The embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the accompanying drawings and embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.
[0020] In the description of the embodiments disclosed in this invention, the term "comprising" and similar terms should be understood as open-ended inclusion, i.e., "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc., may refer to different or the same objects. Other explicit and implicit definitions may also be included below.
[0021] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to Figure 1, which is a schematic diagram of an embodiment of the process planning method based on a large language model in the present invention, including the following steps: 101, receiving engineering drawing files; in this embodiment, the engineering drawing files can be common files such as CAD two-dimensional files, STEP three-dimensional files, STP three-dimensional files, and PDF files.
[0022] 102. According to the preset feature extraction algorithm, the engineering drawing file is subjected to feature recognition and extraction processing to generate a structured feature list; in this embodiment, according to the preset feature extraction algorithm, the description geometric information and geometric relationship EXPRESS text of the engineering drawing file are extracted, and a structured feature list similar to JSON, XML and other structured documents is generated.
[0023] Specifically, please refer to Figure 2. Figure 2 is a schematic diagram of a specific embodiment of step 102 of the process planning method based on a large language model in this invention. Step 102 includes the following specific implementation: 1021. Geometric type identification and extraction are performed on the engineering drawing file to generate identification geometric elements; 1022. Based on the identification geometric elements, geometric relationships are extracted from the engineering drawing file to obtain the dependency relationships corresponding to the identification geometric elements; 1023. Based on the identification geometric elements and the dependency relationships corresponding to the identification geometric elements, a structured feature list is generated.
[0024] In steps 1021-1023, assume that you are inputting a STEP file and PDF drawing of a flange part with a diameter of 100mm, a thickness of 20mm, a Ø30mm H7 through hole in the center, and four Ø10mm through holes evenly distributed around the circumference.
[0025] The engineering drawing file is subjected to geometric type identification and extraction, which extracts geometric elements and their relational text from the STEP file in EXPRESS format. For example: "#34=ADVANCED_FACE(...)", "#52=EDGE_CURVE(...)", "#78=AXIS2_PLACEMENT_3D(...)", etc.
[0026] Based on the identified geometric elements, geometric relationships are extracted from the engineering drawing file to obtain the dependency relationships corresponding to the identified geometric elements.
[0027] Finally, based on the dependencies corresponding to the identified geometric elements, the identified geometric elements are combined and organized to generate a structured feature list, for example: {"name":"upper face","type":"Plane","roughness":"Ra1.6"}{"name":"lower face","type":"Plane","roughness":"Ra3.2"}{"name":"outer cylindrical face","type":"Cylinder","diameter":100}{"name":"center through hole","type":"ThroughHole","diameter":30,"tolerance":"H7"}{"name":"mounting through hole","type":"ThroughHole","diameter":10,"quantity":4}.
[0028] Further, please refer to Figure 3. Figure 3 is a schematic diagram of a specific embodiment of step 1021 of the process planning method based on a large language model in this invention. Step 1021 includes the following specific implementation: 10211. According to a preset YOLO algorithm, the engineering drawing file is segmented to obtain labeled image elements and labeled text elements; 10212. According to a preset RNN algorithm, the image elements are structurally identified to obtain the corresponding geometric elements; 10213. According to a preset semantic OCR algorithm, the text elements are identified to obtain text data; 10214. According to the identified and segmented label data, the text data is matched with the geometric elements to obtain the identified geometric elements.
[0029] In steps 10211-10214, the YOLO algorithm is first used to distinguish between image elements and text elements in the engineering drawing file. Text elements and image elements whose adjacent distance meets the set threshold are marked with the same label, resulting in labeled image elements and labeled text elements.
[0030] Then, based on the RNN algorithm, structural recognition is performed on the image elements to obtain the corresponding geometric elements, such as cylinders, rhombuses, and circular holes.
[0031] Then, a semantic OCR algorithm is introduced to recognize text elements and obtain text data. Finally, based on the initially identified and segmented label data, the text data is matched with geometric elements to obtain the recognized geometric elements.
[0032] 103. Based on a preset loop inference algorithm, the structured feature list is subjected to memory block hit processing to generate a verification feature list; in this example, the loop inference algorithm performs hit and deduplication processing on the structured feature list in the memory cache to generate a verification feature list.
[0033] Specifically, step 103 includes the following implementation methods: 1031. Extracting structured features from the structured feature list; 1032. Determining whether the structured features exist in a preset memory cache; 1033. If they exist, marking the structured features as hits; 1034. If they do not exist, writing the structured features into a preset object framework to generate verification features, and writing the verification features into a preset memory cache; 1035. Integrating the data in the memory cache to generate a verification feature list.
[0034] In steps 1031-1035, the list of structured features is as follows: {"name":"Upper end face","type":"Plane","roughness":"Ra1.6"}{"name":"Lower end face","type":"Plane","roughness":"Ra3.2"}{"name":"Outer cylindrical surface","type":"Cylinder","diameter":100}{"name":"Center through hole","type":"ThroughHole","diameter":30,"tolerance":"H7"}{"name":"Mounting through hole","type":"ThroughHole","diameter":10,"quantity":4} First, extract a structured feature {"name":"Upper end face","type":"Plane","roughness":"Ra1.6"}, and determine whether this structured feature exists in the preset memory cache.
[0035] If it exists, it means that the structured feature has already been written to the memory cache, and there is no need to write it again.
[0036] If the structured feature does not exist, it is written into a preset object framework to generate a verification feature, and the verification feature is written into a preset memory cache. An example of the verification feature structure is as follows: {"part_id": "top face","geometric_elements": [{"id": "#34","type":"Plane","roughness":"Ra1.6"},],"identified_features": [],"reasoning_log": [],"is_complete": false} Finally, all the data written into the memory cache is combined and merged to generate a verification feature list.
[0037] 104. Extract the production requirements from the engineering drawing file; in this embodiment, semantic analysis is used to extract the production requirements from the engineering drawing file.
[0038] 105. Based on the preset large language model and preset processing rules, and using the production requirements and the verification feature list as input parameters, reasoning and matching are performed from the preset knowledge base to generate the optimal process list; in this embodiment, the identified features and related constraints are sent as context information to the large model to use its knowledge for reasoning. For example, if a cylindrical surface (#45) intersects with two external planes (#34, #60) at the same time, it is likely to be a "through hole".
[0039] A knowledge base is a pre-built semantic network containing massive amounts of manufacturing knowledge, covering processes, equipment, tools / fixtures, quality requirements, etc. Knowledge base implementations can include graph databases (such as Neo4j) for storing knowledge graphs, vector databases, and relational or document databases (such as PostgreSQL and MongoDB) for storing business data.
[0040] Based on production demand, the knowledge graph is queried to identify all "process" nodes that can process the input features and meet production requirements. These processes constitute the initial node set of the decision graph. Connections (edges) are established between these process nodes. Each edge is assigned one or more weights, representing the "cost" of moving from one process to the next. This cost is a comprehensive function (calculated from the attributes associated with equipment and tools in the knowledge graph).
[0041] The process planning problem is transformed into finding the shortest path from the "raw material" state to the "finished product" state on a dynamic decision graph. The A* algorithm is used for solving this problem. When external information is received (such as a "DMG-CTX-510 lathe malfunction" report from the MES system), the state attributes of the corresponding node in the knowledge graph are updated in real time. Then, the algorithm is rerun. This time, the algorithm automatically avoids any paths that would require using the malfunctioning equipment, thereby generating a new, executable optimal process route and an optimal process list.
[0042] Specifically, please refer to Figure 4. Figure 4 is a schematic diagram of a specific embodiment of step 105 of the process planning method based on a large language model in this invention. Step 105 includes the following specific implementation: 1051. Based on a preset large language model and preset processing rules, perform feature reasoning on the production requirements and the verification feature list to generate constraints; 1052. Based on the constraints, screen equipment and tools from a preset knowledge base to obtain target equipment and target tools; 1053. Based on the target equipment and the target tools, generate an optimal process list.
[0043] In steps 1051-1053, based on the preset large language model and preset machining rules, the H7 tolerance was identified as a strong constraint. Consulting the knowledge graph revealed that a simple drilling operation cannot satisfy H7; a combination of drilling, reaming, or drilling and boring must be used.
[0044] Based on the constraints, a decision graph was constructed that includes multiple possible combinations of operations such as turning the outer circle, turning the end face, drilling the center hole, reaming the center hole, and drilling the mounting hole. The target equipment and target tool were selected through the decision graph.
[0045] Analyze the process paths that the target equipment and tooling can execute, and generate an optimal process list.
[0046] Furthermore, step 1052 includes the following specific implementation: 10521. Based on the constraints, perform process decision query from a preset knowledge base to generate a process combination decision graph; 10522. According to a preset search algorithm, perform cost search processing on the process combination decision graph to obtain the lowest cost path; 10523. Based on the lowest cost path, select target tools and target equipment from the preset knowledge base.
[0047] In steps 10521-10523, based on constraints, a decision graph is constructed containing various possible combinations of operations, including turning the outer diameter, turning the end face, drilling the center hole, reaming the center hole, and drilling the mounting hole. Using the A* search algorithm, a cost search is performed on the operation combination decision graph to obtain the lowest-cost path. Assuming the knowledge graph shows that the hourly cost of the "machining center" is much higher than that of the "drilling machine," and that the transfer and clamping costs are similar, the A* algorithm will ultimately select path B as the cost-optimal solution. Based on the lowest-cost path, target tools and target equipment are selected from a pre-defined knowledge base.
[0048] Specifically, step 1053 includes the following specific implementation: 10531. Based on the lowest cost path, perform process processing combination on the target equipment and the target tool to generate an optimal process list.
[0049] In step 10531, based on the lowest cost path, the target equipment and the target tool are combined to generate an optimal process list, which is shown below: 1. Process 10: Turning (equipment: lathe CL-20, tool: ...), machining the upper and lower end faces and outer circle.
[0050] 2. Operation 20: Drilling and reaming (equipment: lathe CL-20, cutting tools: ...), machining Ø30mm H7 center hole.
[0051] 3. Operation 30: Drilling (Equipment: Drilling machine ZD-05, Fixture: ...), machining 4 x Ø10mm mounting holes.
[0052] 106. Based on the optimal process list, generate a process cost table.
[0053] In this embodiment, based on the optimal process route, all main, auxiliary, and preparation time is automatically accumulated, and cost elements such as materials, equipment depreciation, and labor are associated, ultimately outputting a highly transparent, clearly structured cost estimate with accurate results.
[0054] Furthermore, step 106 includes the following specific implementation methods: 1061. Obtain the equipment parameters of the target equipment; 1062. Calculate the production time based on the equipment parameters; 1063. Generate a raw material cost table according to the materials and preset material costs in the optimal process list; 1064. Obtain a process cost table based on the raw material cost table and the production time.
[0055] In steps 1061-1064, the equipment parameters of the target equipment are obtained, and then the production time is calculated based on the equipment parameters. The raw material cost table is generated by multiplying the materials in the optimal process list and the preset material costs. Finally, the raw material cost tables are summed to obtain the material cost data. The production time is combined with the unit price of the time fluctuation to generate the processing cost data. The material cost data and the processing cost data are added together to obtain the total cost data. All the above cost data are combined to obtain the process cost table.
[0056] In this embodiment of the invention, feature recognition analysis is performed on engineering drawing files to extract relevant part information, including names, parameters, and other data. A loop reasoning algorithm is then used to deduplicate the part information, generating a validated feature list. By introducing a large language model and a knowledge base for massive industrial manufacturing data, the optimal process list is inferred from the feature list while meeting the production requirements of the engineering drawing files. Based on this optimal process list, a process cost table is generated. The large language model enables autonomous decision-making based on an understanding of process constraints, effectively adapting to complex industrial environments and innovative parts. It efficiently calculates the theoretically optimal solution, solving the technical problem that computer-aided process planning is difficult to apply to complex industrial scenarios and new parts.
[0057] Figure 5 is a schematic diagram of a process planning device based on a large language model according to an embodiment of the present invention. This process planning device 500 can vary considerably depending on its configuration or performance, and may include one or more central processing units (CPUs) 510 and memory 520, and one or more storage media 530 storing application programs 533 or data 532. The memory 520 and storage media 530 can be temporary or persistent storage. The program stored in the storage media 530 may include one or more modules (not shown in the figure), each module including a series of instruction operations on the process planning device 500 based on the large language model. Furthermore, the processor 510 may be configured to communicate with the storage media 530 and execute the series of instruction operations in the storage media 530 on the process planning device 500 based on the large language model.
[0058] The large language model-based process planning device 500 may also include one or more power supplies 540, one or more wired or wireless network interfaces 550, one or more input / output interfaces 560, and / or one or more operating systems 531, such as Windows Server, Mac OS X, Unix, Linux, Free BSD, etc. Those skilled in the art will understand that the structure of the large language model-based process planning device shown in Figure 5 does not constitute a limitation on the large language model-based process planning device, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.
[0059] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the process planning method based on a large language model.
[0060] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0061] Furthermore, although the operations are described in a specific order, this should be understood as requiring that such operations be performed in the specific order shown or in sequential order, or requiring that all illustrated operations be performed to achieve the desired result. In certain environments, multitasking and parallel processing may be advantageous. Similarly, although several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single implementation. Conversely, various features described in the context of a single implementation may also be implemented individually or in any suitable sub-combination in multiple implementations.
[0062] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.
Claims
1. A process planning method based on a large language model, characterized in that, The steps include: receiving engineering drawing files; performing feature recognition and extraction processing on the engineering drawing files according to a preset feature extraction algorithm to generate a structured feature list; Based on a preset loop reasoning algorithm, the structured feature list is processed by memory block hits to generate a verification feature list; production requirements are extracted from the engineering drawing file; based on a preset large language model and preset processing rules, the production requirements and the verification feature list are used as input parameters to perform reasoning and matching from a preset knowledge base to generate an optimal process list; based on the optimal process list, a process cost table is generated.
2. The process planning method based on a large language model according to claim 1, characterized in that, The step of performing feature recognition and extraction processing on the engineering drawing file according to a preset feature extraction algorithm to generate a structured feature list includes: performing geometric type recognition and extraction on the engineering drawing file to generate recognized geometric elements; extracting geometric relationships from the engineering drawing file based on the recognized geometric elements to obtain the dependency relationships corresponding to the recognized geometric elements; and generating a structured feature list based on the recognized geometric elements and the dependency relationships corresponding to the recognized geometric elements.
3. The process planning method based on a large language model according to claim 2, characterized in that, The steps of generating a verification feature list by performing memory block hit processing on the structured feature list based on a preset loop inference algorithm include: extracting structured features from the structured feature list; determining whether the structured features exist in a preset memory cache; marking the structured features as hits when they exist; writing the structured features into a preset object framework to generate verification features when they do not exist, and writing the verification features into the preset memory cache; and integrating the data in the memory cache to generate a verification feature list.
4. The process planning method based on a large language model according to any one of claims 2 or 3, characterized in that, The steps of extracting geometric types from the engineering drawing file and generating identifiable geometric elements include: performing recognition and segmentation processing on the engineering drawing file according to a preset YOLO algorithm to obtain labeled image elements and labeled text elements; performing structure recognition on the image elements according to a preset RNN algorithm to obtain the corresponding geometric elements; recognizing the text elements based on a preset semantic OCR algorithm to obtain text data; and matching the text data with the geometric elements according to the identified and segmented label data to obtain the identifiable geometric elements.
5. The process planning method based on a large language model according to claim 1, characterized in that, The step of generating an optimal process list by inferring and matching from a preset knowledge base based on a preset large language model and preset processing rules, using the production requirements and the verification feature list as input parameters, further includes: performing feature inference on the production requirements and the verification feature list based on the preset large language model and preset processing rules to generate constraints; filtering equipment and tools from the preset knowledge base based on the constraints to obtain target equipment and target tools; and generating an optimal process list based on the target equipment and the target tools.
6. The process planning method based on a large language model according to claim 5, characterized in that, The step of generating a process cost table based on the optimal process list includes: obtaining the equipment parameters of the target equipment; calculating the production hours based on the equipment parameters; generating a raw material cost table based on the materials and preset material costs in the optimal process list; and obtaining the process cost table based on the raw material cost table and the production hours.
7. The process planning method based on a large language model according to claim 5, characterized in that, The step of selecting target equipment and target tools from a preset knowledge base based on the constraints includes: performing a process decision query from the preset knowledge base based on the constraints to generate a process combination decision graph; performing a cost search on the process combination decision graph according to a preset search algorithm to obtain the lowest cost path; and selecting target tools and target equipment from the preset knowledge base based on the lowest cost path.
8. The process planning method based on a large language model according to claim 7, characterized in that, The step of generating an optimal process list based on the target equipment and the target cutting tool includes: combining the process of the target equipment and the target cutting tool based on the lowest cost path to generate an optimal process list.
9. A process planning device based on a large language model, characterized in that, The large language model-based process planning device includes: a memory and at least one processor, wherein the memory stores instructions, and the memory and the at least one processor are interconnected via a circuit; the at least one processor invokes the instructions in the memory to cause the large language model-based process planning device to execute the large language model-based process planning method as described in any one of claims 1-8.
10. A computer-readable storage medium storing a computer program thereon, characterized in that, When the computer program is executed by the processor, it implements the process planning method based on a large language model as described in any one of claims 1-8.