A method, apparatus, equipment and medium for judging the surface quality of galvanized products

Abstract

This application discloses a method, apparatus, equipment, and medium for judging the surface quality of galvanized products. The method includes: acquiring product information and multiple defect detection data of the galvanized product to be inspected; wherein, the product information includes at least one of customer orders, steel grade specifications, and batch numbers; responding to the quality inspection task of the galvanized product, selecting quality inspection operation documents from a quality inspection knowledge base based on the product information; performing semantic parsing on the quality inspection operation documents based on a large language model to extract corresponding defect-redundant quality inspection rules; and performing quality inspection on the galvanized product based on the defect-redundant quality inspection rules and defect detection data to obtain the quality inspection result. This application achieves dynamic adaptation and accurate judgment for different quality inspection standards by constructing a quality inspection knowledge base and automatically extracting differentiated quality inspection rules using a large language model.

Description

Technical Field

[0001] This application relates to the field of industrial quality testing technology, and in particular to a method, apparatus, equipment and medium for determining the surface quality of galvanized products. Background Technology

[0002] Surface quality assessment of galvanized products is a crucial aspect of the steel manufacturing quality control system. The accuracy of the assessment standards and the efficiency of their implementation directly impact product delivery pass rates and customer satisfaction.

[0003] Currently, most steel companies in China still rely on manual review of paper-based work instructions or traditional judgment methods based on fixed thresholds in the quality inspection of galvanized products, which makes it difficult to quickly respond to complex and ever-changing quality inspection needs. Summary of the Invention

[0004] This application provides a method, apparatus, equipment, and medium for judging the surface quality of galvanized products, which solves the problem that existing technologies are unable to quickly respond to complex and ever-changing quality inspection needs.

[0005] The technical solution adopted in this application is as follows: Firstly, this application provides a method for determining the surface quality of galvanized products, including: Obtain product information and multiple defect detection data for the galvanized products to be inspected; the product information includes at least one of the following: customer order, steel grade and specification, and batch number. In response to the quality inspection tasks for galvanized products, quality inspection operation documents are selected from the quality inspection knowledge base based on product information; Semantic parsing of quality inspection work documents is performed based on a large language model to extract corresponding defect and redundant quality inspection rules. The quality inspection results are obtained by conducting quality inspection on galvanized products based on the defect redundancy quality inspection rules and defect detection data.

[0006] In one alternative approach of the first aspect, the galvanized products are inspected based on defect redundancy inspection rules and defect detection data to obtain inspection results, including: Based on the defect redundancy quality inspection rules and defect detection data, generate executable quality inspection instructions; Execute quality inspection instructions and generate quality inspection results.

[0007] In one alternative of the first aspect, based on defect redundancy quality inspection rules and defect detection data, executable quality inspection instructions are generated, including: The defect redundancy quality inspection rules are converted into data filtering instructions and pass / fail judgment instructions. The data filtering instructions are used to filter out target defect data from the defect detection data, and the pass / fail judgment instructions are used to determine whether the target defect data is qualified.

[0008] In one alternative approach of the first aspect, semantic parsing of the quality inspection work documents is performed based on a large language model to extract corresponding defect redundancy quality inspection rules, including: The large language model is called to perform semantic parsing on the quality inspection work documents to extract the corresponding quality judgment conditions, process constraint rules and exception judgment conditions. Among them, the quality judgment conditions are used to specify the criteria for judging the qualification of defects, the process constraint rules are used to limit the range of production process parameters, and the exception judgment conditions are used to specify special handling rules under specific circumstances. Defect redundancy quality inspection rules are generated based on quality judgment conditions, process constraint rules, and exception judgment conditions.

[0009] In one alternative to the first aspect, after obtaining the quality inspection results, it also includes: The quality inspection results are output to the target terminal in a preset format so that the target terminal can perform abnormal prompts, result verification, or defective product handling operations based on the quality inspection results.

[0010] In one alternative approach of the first aspect, the quality inspection results are output to the target terminal according to a preset format, including: Defect handling recommendations are generated based on quality inspection results and defect detection data; among them, defect handling recommendations include at least one of the following: rework solution, downgrade solution, or defect removal location guidance; The quality inspection results and defect handling suggestions are output to the target terminal in a preset format so that the target terminal can perform defective product handling operations based on the defect handling suggestions.

[0011] In one alternative to the first aspect, after obtaining the quality inspection results, it also includes: Obtain feedback data on the review of quality inspection results; Based on the review feedback data, adjust the semantic parsing parameters of the large language model or update the quality inspection knowledge base.

[0012] Secondly, this application provides a device for judging the surface quality of galvanized products, comprising: The data acquisition module is used to acquire product information and multiple defect detection data of the galvanized products to be inspected; among which, the product information includes at least one of customer orders, steel grade specifications and batch numbers; The document filtering module is used to respond to quality inspection tasks for galvanized products and filter quality inspection operation documents from the quality inspection knowledge base based on product information. The rule extraction module is used to perform semantic parsing on quality inspection work documents based on a large language model and extract the corresponding defective and redundant quality inspection rules. The quality assessment module is used to perform quality inspection on galvanized products based on defect redundancy inspection rules and defect detection data, and obtain the inspection results.

[0013] Thirdly, this application provides an electronic device including a memory and a processor. The memory is used to store computer programs or instructions that, when executed by the processor, implement the method described in the first aspect or any of the alternative solutions of the first aspect.

[0014] Fourthly, this application provides a computer-readable storage medium. The storage medium stores a computer program or instructions that, when executed by a processor, implement the method described in the first aspect or any of the alternative solutions to the first aspect.

[0015] The beneficial effects of the technical solutions provided in some embodiments of this specification include at least the following: This application, in response to the quality inspection tasks of galvanized products, accurately selects and matches quality inspection operation documents for specific customers, steel grades, and batches from a pre-built quality inspection knowledge base based on product information. Then, it uses a large language model to perform semantic parsing on the selected quality inspection operation documents to automatically extract the corresponding defect redundancy quality inspection rules. Based on the extracted rules, it performs defect redundancy judgment and complex logical operations on defect detection data, realizing dynamic adaptation and accurate judgment of quality inspection standards for different customer orders, steel grades, specifications, and batches. This meets the rapid response needs in complex, multi-variety order scenarios and significantly improves quality inspection efficiency and judgment accuracy. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating the surface quality determination method for galvanized products provided in this application embodiment; Figure 2 This is a schematic diagram of the surface quality determination device for galvanized products provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0018] Embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0019] The accompanying drawings illustrate various structural schematics according to embodiments of the present disclosure. These drawings are not to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.

[0020] In the context of this disclosure, when a layer / element is referred to as being "above" another layer / element, the layer / element may be directly above the other layer / element, or there may be an intermediate layer / element between them. Additionally, if a layer / element is "above" another layer / element in one orientation, then when the orientation is reversed, the layer / element may be "below" the other layer / element.

[0021] The following is a brief explanation of the terms used in this application: Surface quality: refers to the degree to which the surface of galvanized steel sheets and strips meets specific usage requirements. It is usually evaluated by visual inspection or instrument testing, taking into account the quantity, size, distribution density, and morphological characteristics of surface defects (such as zinc dross, air knife marks, passivation spots, scratches, color difference, zinc layer peeling, etc.). Based on the customer's technical agreement or national standards (such as EN 10143, JIS G3302, GB / T 2518, etc.), the product grade is determined (such as ordinary surface FB, higher grade surface FC, high grade surface FD). It is a key quality indicator that determines whether the product can be delivered and its pricing grade.

[0022] With the deepening of intelligent transformation and flexible production in the steel industry, the dynamic adaptation and intelligent extraction of quality inspection rules under multi-variety, small-batch order models has become an important technical direction for improving the response speed of quality control and reducing reliance on manual experience. Currently, the surface quality judgment of galvanized products mainly relies on manual review of paper work instructions or generalized automatic detection based on fixed thresholds. The inventors have found that in actual production, there are often differentiated defect acceptance rules (such as single-point defect tolerance limits, thresholds for merging similar defects, and exemption standards for specific areas) for different customer orders, steel grades, specifications, and batches. When faced with such complex quality inspection requirements involving defect tolerance judgment, multi-defect merging rules, and exemption conditions, existing solutions still require quality inspectors to manually screen, interpret, and match corresponding rules from a large number of unstructured documents based on their personal experience. This results in low rule extraction efficiency, standard execution being easily affected by subjective factors, and difficulty in ensuring the consistency and accuracy of complex rule judgments.

[0023] Based on the above-mentioned technical problems, the inventive concept of this application is to: construct a quality inspection knowledge base covering multi-dimensional product information, and automatically extract defective and redundant quality inspection rules by semantic parsing of differentiated quality inspection operation documents based on a large language model, thereby realizing dynamic adaptation and accurate judgment of quality inspection standards for different customer orders, steel specifications and batches, thereby replacing the traditional manual interpretation mode and improving the quality inspection response speed and judgment accuracy in complex multi-variety order scenarios.

[0024] The technical solutions of this application and how they solve the aforementioned technical problems will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0025] It should be noted that the executor of this application may be, but is not limited to, edge computing gateways, industrial quality inspection servers, or quality inspection workstations, which are quality inspection devices with the ability to manage quality inspection knowledge bases and extract intelligent rules.

[0026] refer to Figure 1 , Figure 1 A flowchart illustrating the surface quality assessment method for galvanized products provided in this application embodiment. Figure 1 As shown, the method for judging the surface quality of galvanized products includes at least the following steps: S101: Obtain product information and multiple defect detection data for galvanized products to be inspected.

[0027] The product information includes at least one of the following: customer order, steel grade and specification, and batch.

[0028] Specifically, the quality inspection equipment obtains product information of the galvanized products to be inspected from the Manufacturing Execution System (MES) or Enterprise Resource Planning (ERP) through a data interface, and at the same time obtains multiple defect detection data from surface defect detection equipment (such as machine vision inspection system and laser scanning equipment).

[0029] Product information is the key attribute information used to uniquely identify and classify the galvanized product. It includes at least one of the following: customer order (identifying customer needs and corresponding special quality requirements), steel grade and specifications (such as DC51D+Z, DX53D+Z, etc., determining the characteristics of the base material and the corresponding defect acceptance standards), and production batch (used for production traceability and matching specific process parameters for that batch). Defect detection data includes quantitative indicators such as defect type (such as zinc dross, air knife marks, passivation spots, scratches, etc.), defect location coordinates, defect size parameters, and defect severity level.

[0030] S103: In response to the quality inspection task of galvanized products, the quality inspection operation documents are obtained by filtering from the quality inspection knowledge base based on product information.

[0031] Specifically, in response to the quality inspection task trigger command for the galvanized product, the quality inspection equipment, based on key attributes such as customer orders, steel grades and / or batches from the product information obtained in step S101, precisely filters from a pre-built quality inspection knowledge base to obtain quality inspection operation documents that match the attributes of the galvanized product. This quality inspection knowledge base stores unstructured text data such as historical quality inspection standard documents, technical agreements, and process specifications for different customers, steel grades, and batches.

[0032] S105: Based on a large language model, perform semantic parsing on quality inspection work documents to extract corresponding defect and redundant quality inspection rules.

[0033] Specifically, the quality inspection equipment uses a large language model to perform deep semantic analysis on the quality inspection work documents selected in step S103. This large language model uses natural language processing technology to automatically identify and extract structured defect redundancy quality inspection rules from unstructured document text, including but not limited to complex judgment logic such as single-point defect tolerance limits, thresholds for merging similar defects, and exemption standards for specific areas.

[0034] Furthermore, the aforementioned large language model can be a pre-trained language model based on the Transformer architecture (including but not limited to the GPT series, BERT series, Wenxin Yiyan, Tongyi Qianwen, or domain-specific models fine-tuned from corpora in the steel quality inspection field); natural language processing technologies include Named Entity Recognition (NER) to extract key parameters such as defect type and size threshold, relation extraction to identify logical relationships between conditions (such as "and" and "or" relationships), and semantic role labeling to parse the decision subject and standard value in the subject-verb-object structure, thereby transforming unstructured natural language text into structured rule triples (condition-operation-threshold) or machine-executable logic in the form of decision trees.

[0035] S107: Conduct quality inspection on galvanized products based on defect redundancy inspection rules and defect detection data to obtain inspection results.

[0036] Specifically, the quality inspection equipment performs logical matching and calculation between the defect redundancy quality inspection rules extracted in step S105 and the defect detection data obtained in step S101, automatically executes quality inspection logic such as defect tolerance judgment, multi-defect merging calculation and regional exemption verification, and finally generates the quality inspection result (qualified / unqualified / conditionally acceptable, etc.) of the galvanized product.

[0037] Through the coordinated execution of the above four steps, the quality inspection equipment achieves intelligent and precise full-process judgment of the surface quality of galvanized products. First, because step S101 acquires multi-dimensional product information including customer orders, steel grades, specifications, and batches, step S103 can accurately select specific quality inspection operation documents corresponding to the current product from the quality inspection knowledge base based on these key attributes, thus replacing the inefficient operation of manually sifting through massive amounts of documents. Furthermore, because step S103 provides precisely matched documents, step S105 can perform deep semantic analysis on this specific document based on a large language model, automatically extracting defect-redundant quality inspection rules specific to the product, such as customer-specific requirements, steel grade-specific acceptance standards, and batch process parameters, reducing the subjective bias of manually interpreting non-standardized text. Subsequently, by In step S105, structured and precise rules are extracted. In step S107, these rules are logically matched and calculated with the actual defect detection data obtained in step S101. Complex quality inspection logics such as defect tolerance determination, multi-defect merging calculation, and exemption condition verification are automatically executed. Finally, through the closed-loop processing of product information-document screening-rule extraction-quality inspection determination, the dynamic adaptation and accurate determination of differentiated quality inspection standards for different customer orders, steel grades, specifications, and batches are realized. This significantly improves the quality inspection response speed and determination accuracy in complex multi-variety order scenarios and effectively solves the technical problems of low rule extraction efficiency and easy deviation in standard execution under the traditional manual mode.

[0038] In some embodiments, galvanized products are inspected based on defect redundancy inspection rules and defect detection data to obtain inspection results, including: Based on the defect redundancy quality inspection rules and defect detection data, generate executable quality inspection instructions; Execute quality inspection instructions and generate quality inspection results.

[0039] Specifically, the quality inspection equipment integrates the extracted defect redundancy inspection rules with the acquired defect detection data, and converts the quality requirements described in these rules (such as how large a defect can be tolerated, which defects can be combined for calculation, which areas can be ignored, etc.) into specific calculation conditions and judgment formulas, generating quality inspection instructions that the machine can directly execute. This is equivalent to turning the written standards on paper into automatically calculable digital logic.

[0040] Based on this, the quality inspection equipment automatically performs calculations and judgments according to the generated quality inspection instructions, compares the actual detected defect data with the standards set in the instructions, automatically completes calculations such as whether the defects exceed the standards and whether points need to be deducted, and finally draws a quality inspection conclusion on whether the product is qualified.

[0041] Through the coordinated execution of the above two steps, the quality inspection equipment has achieved a technological improvement from "manual document review and judgment" to "automatic rule conversion and machine calculation." First, by converting the textually described rules into standardized calculation instructions, errors in reading or understanding that might occur during human reading and comprehension are avoided. Second, because the calculation and judgment are performed automatically by the machine, manual errors and omissions in the process of manually looking up tables and performing arithmetic calculations are reduced. Finally, by automatically calculating according to unified logic, the consistency and stability of standard execution when different personnel inspect the same batch of products at different times are ensured, significantly improving the accuracy and processing speed of the quality inspection results.

[0042] In some embodiments, based on defect redundancy inspection rules and defect detection data, executable inspection instructions are generated, including: The defect redundancy quality inspection rules are converted into data filtering instructions and pass / fail judgment instructions. The data filtering instructions are used to filter out target defect data from the defect detection data, and the pass / fail judgment instructions are used to determine whether the target defect data is qualified.

[0043] Specifically, when the quality inspection equipment converts the defect redundancy inspection rules into executable instructions, it generates a data filtering instruction based on the defect characteristic conditions (such as defect type, size threshold, location range, etc.) defined in the rules. This instruction is used to quickly filter out the target defect data that needs to be focused on from the original defect detection data (such as defects that exceed the normal tolerance or defects located in critical areas). At the same time, it generates a conformity judgment instruction based on the acceptance criteria, merging calculation logic and exemption clauses defined in the rules. This instruction is used to perform a refined conformity judgment on the selected target defect data to determine whether it meets the quality requirements.

[0044] Through the above embodiments, the quality inspection equipment realizes a "screen-then-judgment" layered processing mechanism for defect quality inspection. First, due to the pre-execution of data screening instructions, the target defects of concern are quickly extracted from massive defect detection data, reducing the resource consumption of ineffective calculations on irrelevant defect data. Second, since the conformity judgment instructions only perform complex judgment logic (such as multi-defect merging calculation, regional exemption verification, etc.) on the screened target defects, the computational load is significantly reduced and the judgment accuracy is improved. Finally, through the separation design of screening and judgment instructions, it is ensured that key defects are not missed, and the modularity and maintainability of the quality inspection process are realized, maintaining good processing efficiency and judgment accuracy even when facing large-scale defect data.

[0045] In some embodiments, semantic parsing of quality inspection work documents is performed based on a large language model to extract corresponding defect redundancy quality inspection rules, including: The large language model is called to perform semantic parsing on the quality inspection work documents to extract the corresponding quality judgment conditions, process constraint rules and exception judgment conditions. Among them, the quality judgment conditions are used to specify the criteria for judging the qualification of defects, the process constraint rules are used to limit the range of production process parameters, and the exception judgment conditions are used to specify special handling rules under specific circumstances. Defect redundancy quality inspection rules are generated based on quality judgment conditions, process constraint rules, and exception judgment conditions.

[0046] Specifically, the quality inspection equipment utilizes a large language model to perform deep semantic analysis on the quality inspection work documents. This model, through natural language understanding technology, automatically identifies and extracts three types of key information from the document text: first, quality judgment conditions (such as upper limits for the size and quantity thresholds of various defects, and other acceptance standards); second, process constraint rules (such as production parameter limitations such as coating thickness range and zinc bath temperature); and third, exception judgment conditions (such as special acceptance standards for specific customers and exemption clauses for specific regions, and other special case handling rules). Through this step, non-standardized textual descriptions scattered throughout the document are transformed into structured digital conditions.

[0047] Furthermore, the quality inspection equipment logically correlates and integrates the three types of conditions extracted, constructing a complete defect redundancy quality inspection rule system according to the priority order of "basic standards - process modification - exception handling". Specifically, it uses general defect compliance standards as the basic judgment basis, process parameter constraints as the judgment weight adjustment factor, and exception conditions as the judgment coverage logic for special cases, ultimately generating unified quality inspection rules covering both routine and special scenarios.

[0048] Thus, through the coordinated execution of the above two steps, the quality inspection equipment achieves intelligent conversion from unstructured documents to structured quality inspection rules. First, because the large language model can simultaneously identify explicit judgment criteria (quality conditions) and implicit constraints (process rules, exception clauses) in the document, it avoids the problem of overlooking implicit rules or special clauses during manual reading. Second, by systematically integrating the three types of conditions to generate unified rules, it ensures the completeness and consistency of the judgment logic in complex production scenarios (such as when process fluctuations and special customer requirements coexist). Finally, through automatic extraction and rule generation, it significantly improves the efficiency of rule construction for differentiated documents of different steel grades and batches, reduces the subjective bias of manually reviewing documents line by line, and ensures the accuracy and completeness of the digital conversion of quality inspection standards.

[0049] In some embodiments, after obtaining the quality inspection results, the method further includes: The quality inspection results are output to the target terminal in a preset format so that the target terminal can perform abnormal prompts, result verification, or defective product handling operations based on the quality inspection results.

[0050] Specifically, after generating the quality inspection results, the quality inspection equipment outputs the results to a designated target terminal according to a preset data exchange format (such as JSON, XML, or standardized report format). This target terminal can be a Manufacturing Execution System (MES) terminal, a quality inspector's workstation, or a production line control device. Upon receiving the quality inspection results, the target terminal triggers corresponding subsequent operations based on the result type. For example, when a non-conforming item is detected, an automatic audio-visual anomaly alert is issued to notify relevant personnel; when the result is at a critical value or requires manual confirmation, a result review process is initiated for the quality inspector to verify; when a product is determined to be defective, the production line control system is linked to automatically perform defective product marking, sorting, or isolation.

[0051] In this way, by outputting the inspection results in a standardized format and linking the target terminal to automatically execute anomaly prompts, result verification, or defective product handling, the quality inspection equipment achieves seamless closed-loop management of the quality inspection and production execution links. This not only ensures the compatibility and real-time transmission of data between different systems, but also reduces the time delay of manual information transmission and manual operation, significantly improves the response speed and handling efficiency of quality problems, and effectively reduces the risk of defective products flowing into subsequent processes.

[0052] In some embodiments, outputting the quality inspection results to the target terminal according to a preset format includes: Defect handling recommendations are generated based on quality inspection results and defect detection data; among them, defect handling recommendations include at least one of the following: rework solution, downgrade solution, or defect removal location guidance; The quality inspection results and defect handling suggestions are output to the target terminal in a preset format so that the target terminal can perform defective product handling operations based on the defect handling suggestions.

[0053] Specifically, after acquiring quality inspection results and defect detection data (including defect type, location coordinates, size parameters, etc.), the quality inspection equipment automatically generates targeted defect handling suggestions based on the severity and distribution characteristics of the defects, combined with a pre-set processing strategy knowledge base. These suggestions include rework solutions (such as recommending a replating process for severe zinc slag defects), downgrade solutions (such as reclassifying products with high surface grade requirements as ordinary grades), or defect removal location guidance (providing precise start and end coordinates or area ranges to remove unqualified parts), providing clear execution basis for subsequent handling operations.

[0054] Next, the quality inspection equipment encapsulates the generated defect handling suggestions and corresponding quality inspection results into data and outputs them to the target terminal (such as a production line PLC (Programmable Logic Controller), robot control terminal, or MES terminal) according to a preset communication format (such as JSON, XML, or data frames of a specific protocol). After parsing the data, the target terminal can directly drive the automated equipment to perform the corresponding defective product handling operations based on the handling suggestions, such as triggering rework, executing downgrade marking, or guiding the cutting equipment to perform precise removal.

[0055] In light of this, by outputting abstract quality inspection results along with concrete defect handling suggestions to the target terminal, the quality inspection equipment achieves deep coupling and automated closed-loop of quality judgment and production handling. Since the handling suggestions clearly define specific implementation plans for rework, downgrading, or removal, the target terminal can directly drive the equipment to perform the corresponding operations without human intervention. This reduces the time delays and decision-making biases associated with manually interpreting reports, formulating handling plans, and manually operating equipment in the traditional model, significantly shortening the defective product handling cycle, improving the automation level and processing accuracy of the production line, and reducing resource waste caused by human error.

[0056] In some embodiments, after obtaining the quality inspection results, the method further includes: Obtain feedback data on the review of quality inspection results; Based on the review feedback data, adjust the semantic parsing parameters of the large language model or update the quality inspection knowledge base.

[0057] Specifically, after generating and outputting the quality inspection results, the quality inspection equipment receives review feedback data from quality inspectors or verification terminals. This feedback data includes confirmation of the accuracy of the quality inspection results, correction annotations for misjudged cases, or evaluation information on the rationality of the rules extracted by the large language model. Based on this feedback data, the quality inspection equipment achieves system self-optimization and knowledge iteration by fine-tuning the semantic parsing parameters of the large language model (such as adjusting the weights of specific terms and optimizing the confidence threshold for rule extraction) or updating relevant work documents and rule templates in the quality inspection knowledge base.

[0058] This demonstrates that by introducing a review and feedback mechanism, the quality inspection equipment achieves self-learning and continuous optimization capabilities for the quality inspection system. Because the equipment can collect real-time confirmations and corrections from quality inspectors regarding the judgment results and feed these deviation cases from practical applications back to the large language model, the model can adjust semantic parsing parameters accordingly (such as optimizing the understanding weight of specific steel grade terms or correcting the logic of rule extraction). Simultaneously, the quality inspection knowledge base can be promptly supplemented with new standard documents or corrected outdated rules. This allows the system to continuously absorb professional human experience and automatically correct comprehension biases when faced with complex quality inspection documents from different customers and for different steel grades. It avoids the rule extraction errors or standard adaptation lag problems caused by the "unchanging" nature of traditional static systems, significantly improving the accuracy of quality inspection rule extraction and the system's adaptability to diverse scenarios, ensuring that quality inspection judgment standards remain accurate and complete over the long term.

[0059] Based on the same technical concept, this application also provides a device for judging the surface quality of galvanized products, see reference. Figure 2 , Figure 2 This is a schematic diagram of the surface quality assessment device for galvanized products provided in an embodiment of this application. Figure 2 As shown, the galvanized product surface quality assessment device includes at least a data acquisition module 201, a document filtering module 202, a rule extraction module 203, and a quality assessment module 204, wherein: The data acquisition module 201 is used to acquire product information and multiple defect detection data of the galvanized products to be inspected; among which, the product information includes at least one of customer order, steel grade specification and batch; The document filtering module 202 is used to respond to the quality inspection task of galvanized products and filter the quality inspection operation documents from the quality inspection knowledge base based on product information. Rule extraction module 203 is used to perform semantic parsing on quality inspection work documents based on a large language model and extract corresponding defect and redundant quality inspection rules. The quality judgment module 204 is used to perform quality inspection on galvanized products based on defect redundancy quality inspection rules and defect detection data, and obtain the quality inspection results.

[0060] It should be noted that this galvanized product surface quality assessment device can be used to implement any of the above method embodiments.

[0061] Based on the same technical concept, this application also provides an electronic device, see reference. Figure 3 , Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 3 As shown, the electronic device includes a memory 301 and a processor 302. The memory 301 is used to store computer instructions; when the processor 302 executes the computer instructions, it implements any of the above-described method embodiments.

[0062] The specific entity of the electronic device can be a smartphone, wearable device, tablet computer, personal computer, in-vehicle terminal, game console, virtual device, workbench, digital assistant, set-top box, robot, or other terminal device. In other embodiments, it can be a rack server, blade server, tower server, or cabinet server (including a standalone server or a server cluster composed of multiple servers).

[0063] The memory 301 includes at least one type of computer-readable storage medium, including flash memory, hard disk, multimedia card, random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), magnetic disk, optical disk, etc. In some embodiments, the computer-readable storage medium may be an internal storage unit of an electronic device, such as the hard disk or memory of the electronic device. In other embodiments, the computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk, secure digital card (SD card), flash memory card, etc., equipped on the electronic device. Of course, the computer-readable storage medium may include both internal storage units and external storage devices of the electronic device. In this embodiment, the computer-readable storage medium is typically used to store the operating system and various application software installed on the electronic device, such as the program code of the method for judging the surface quality of galvanized products in the embodiment. In addition, the computer-readable storage medium may also be used to temporarily store various types of data that have been output or will be output.

[0064] In some embodiments, processor 302 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other chip. Processor 302 is typically used to control the overall operation of the processing device, such as performing control and processing related to data interaction or communication with other entities. In this embodiment, processor 302 is used to run program code stored in memory 301 or process data.

[0065] Based on the same technical concept, this application also provides a computer-readable storage medium, which includes a computer program or instructions stored in the storage medium. When the computer program or instructions are executed by a processing device, they implement any of the above-described method embodiments. Further details can be found in the method embodiments, which will not be repeated here. In this embodiment, the computer-readable storage medium includes flash memory, hard disk, multimedia card, random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), magnetic disk, optical disk, etc. In some embodiments, the computer-readable storage medium can be an internal storage unit of an electronic device, such as the hard disk or memory of the electronic device. In other embodiments, the computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, a secure digital card (SD card), a flash memory card, etc., equipped on the electronic device. Of course, the computer-readable storage medium can also include both internal storage units and external storage devices of the electronic device. In this embodiment, the computer-readable storage medium is typically used to store the operating system and various application software installed on the electronic device, such as the program code of the method for judging the surface quality of galvanized products in the embodiment. Furthermore, the computer-readable storage medium can also be used to temporarily store various types of data that have been output or will be output.

[0066] The above description does not provide detailed technical specifications regarding the structure of each layer. However, those skilled in the art should understand that layers and regions of desired shapes can be formed using various technical means. Furthermore, to form the same structure, those skilled in the art can also design methods that are not entirely identical to those described above. Additionally, although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be advantageously combined.

[0067] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0068] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for judging the surface quality of galvanized products, characterized in that, include: Obtain product information and multiple defect detection data of the galvanized products to be inspected; wherein, the product information includes at least one of customer order, steel grade specification and batch number; In response to the quality inspection task of the galvanized product, the quality inspection operation document is obtained by filtering from the quality inspection knowledge base based on the product information; Semantic parsing of the quality inspection work document is performed based on a large language model to extract the corresponding defect and redundancy quality inspection rules. The galvanized product is inspected based on the defect redundancy inspection rules and the defect detection data to obtain the inspection results.

2. The method according to claim 1, characterized in that, The quality inspection of the galvanized product based on the defect redundancy inspection rules and the defect detection data, and the resulting quality inspection results, include: Based on the defect redundancy quality inspection rules and the defect detection data, an executable quality inspection instruction is generated. Execute the quality inspection command to generate the quality inspection result.

3. The method according to claim 2, characterized in that, The step of generating executable quality inspection instructions based on the defect redundancy quality inspection rules and the defect detection data includes: The defect redundancy quality inspection rules are converted into data filtering instructions and pass / fail judgment instructions; wherein, the data filtering instructions are used to filter out target defect data from the defect detection data, and the pass / fail judgment instructions are used to determine the pass / fail status of the target defect data to determine whether the target defect data is pass / fail.

4. The method according to claim 1, characterized in that, The semantic parsing of the quality inspection work document based on a large language model to extract corresponding defect and redundancy quality inspection rules includes: The large language model is invoked to perform semantic parsing on the quality inspection work document, and the corresponding quality judgment conditions, process constraint rules and exception judgment conditions are extracted; wherein, the quality judgment conditions are used to specify the defect qualification judgment standard, the process constraint rules are used to limit the range of production process parameters, and the exception judgment conditions are used to specify special handling rules under specific circumstances. Defect redundancy quality inspection rules are generated based on the quality judgment conditions, the process constraint rules, and the exception judgment conditions.

5. The method according to claim 1, characterized in that, After obtaining the quality inspection results, the process also includes: The quality inspection results are output to the target terminal in a preset format so that the target terminal can perform anomaly alerts, result verification, or defective product handling operations based on the quality inspection results.

6. The method according to claim 5, characterized in that, The step of outputting the quality inspection results to the target terminal according to a preset format includes: Based on the quality inspection results and the defect detection data, a defect handling suggestion is generated; wherein, the defect handling suggestion includes at least one of the following: a rework solution, a downgrade solution, or a defect removal location guide; The quality inspection results and the defect handling suggestions are output to the target terminal in a preset format so that the target terminal can perform defective product handling operations based on the defect handling suggestions.

7. The method according to claim 5, characterized in that, After obtaining the quality inspection results, the process also includes: Obtain the review feedback data for the quality inspection results; Based on the review feedback data, adjust the semantic parsing parameters of the large language model or update the quality inspection knowledge base.

8. A device for judging the surface quality of galvanized products, characterized in that, include: The data acquisition module is used to acquire product information and multiple defect detection data of the galvanized products to be inspected; wherein, the product information includes at least one of customer order, steel grade specification and batch number; The document filtering module is used to respond to the quality inspection task of the galvanized product and filter the quality inspection operation documents from the quality inspection knowledge base based on the product information. The rule extraction module is used to perform semantic parsing on the quality inspection work document based on a large language model and extract the corresponding defective and redundant quality inspection rules. The quality assessment module is used to perform quality inspection on the galvanized product based on the defect redundancy quality inspection rules and the defect detection data, and obtain the quality inspection results.

9. An electronic device, characterized in that, It includes a memory and a processor, the memory being used to store computer programs or instructions; when the computer programs or instructions are executed by the processor, the method of any one of claims 1-7 is implemented.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program or instructions, which, when executed by a processor, implement the method of any one of claims 1-7.

Images