Automobile part quality management method, device and equipment and storage medium

Abstract

The invention discloses an automobile part quality management method and device, equipment and a storage medium, and relates to the technical field of computers. The method comprises the following steps: receiving input data for performing quality management on a target automobile part based on an automobile part quality management system; calling a pre-deployed large model in the automobile part quality management system, enabling the large model to call a target analysis model based on the input data and the historical analysis data, and generating a response result based on a first analysis result of the target analysis model; the target analysis model is at least one of a plurality of analysis models pre-deployed in the automobile part quality management system; and outputting a second analysis result according with a preset format through the automobile part quality management system based on the response result of the large model. By adopting the method, the quality management efficiency of the automobile parts can be improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, equipment and storage medium for quality management of automotive parts. Background Technology

[0002] The manufacturing industry is undergoing a profound transformation centered on intelligentization. As a representative of high-end manufacturing, the quality of automotive parts directly impacts the safety and performance of the entire vehicle. Traditional automotive parts quality management methods heavily rely on human experience, resulting in inefficiency, inconsistent standards, and an inability to handle massive amounts of multimodal data (such as images, text, and data). Although artificial intelligence technology has made significant progress in areas such as image recognition and predictive analytics, current applications are mostly siloed, single-point solutions that struggle to deeply integrate with a company's core Quality Management System (QMS), leading to low processing efficiency when dealing with massive amounts of multimodal data.

[0003] Therefore, how to improve the efficiency of automotive parts quality management has become an urgent problem to be solved. Summary of the Invention

[0004] This application provides a method, apparatus, device, and storage medium for automotive parts quality management, which can improve the efficiency of automotive parts quality management.

[0005] In a first aspect, embodiments of this application provide a method for quality management of automotive parts, the method comprising:

[0006] The automotive parts quality management system receives input data for quality management of target automotive parts.

[0007] The system invokes a pre-deployed large model in the automotive parts quality management system. Based on the input data and historical analysis data, the large model invokes the target analysis model and generates a response result based on the first analysis result of the target analysis model. The target analysis model is at least one of multiple pre-deployed analysis models in the automotive parts quality management system.

[0008] Based on the response results of the large model, a second analysis result conforming to a preset format is output through the automotive parts quality management system.

[0009] In one embodiment, based on the response results of a large model, a second analysis result conforming to a preset format is output through the automotive parts quality management system. This includes: inputting the response results of the large model into a trained standardized transformation model to obtain a transformed response result; the transformed response result being a structured description of the response result; obtaining a verification dataset related to automotive parts; inputting the transformed response result and the verification dataset into a pre-built verification function to obtain a verification result; and based on the verification result, determining a second analysis result conforming to a preset format corresponding to the input data, and outputting the second analysis result through the automotive parts quality management system.

[0010] In one embodiment, based on the verification result, determining the second analysis result corresponding to the input data that conforms to a preset format includes: if the verification result indicates that the verification passed, using the converted response result as the second analysis result corresponding to the input data that conforms to a preset format; or, if the verification result indicates that the verification failed, re-calling the large model pre-deployed in the automotive parts quality management system to regenerate the response result until the verification result passes, thereby obtaining the second analysis result that conforms to the preset format.

[0011] In one embodiment, a pre-deployed large model in the automotive parts quality management system is invoked. This large model, based on input data and historical analysis data, invokes a target analysis model and generates a response result based on the first analysis result of the target analysis model. This includes: invoking the pre-deployed large model in the automotive parts quality management system, causing the large model to determine the target input type based on the input data and historical analysis data; if the target input type is non-dialogue type, the large model invokes a dedicated parameter extraction interface to extract parameters from the input data, obtaining key parameters corresponding to the input data; the large model invokes the target analysis model, generating an initial analysis result associated with the input data based on the key parameters; the initial analysis result is structured to obtain a first analysis result; and the large model is invoked to generate a response result based on the input data, historical analysis data, and the first analysis result.

[0012] In one embodiment, the method further includes generating a response result using a large model when the target input type is a dialogue type.

[0013] In one embodiment, the trained standardized conversion model is determined by: obtaining training sample pairs; the training sample pairs consist of sample analysis results corresponding to the sample input data and actual structured descriptions corresponding to the sample analysis results; inputting the sample analysis results into the conversion model to be trained to obtain the predicted structured descriptions corresponding to the sample analysis results; training the conversion model in a direction that reduces the difference between the predicted structured descriptions corresponding to the sample analysis results and the actual structured descriptions to obtain the trained standardized conversion model.

[0014] In one embodiment, the large model is determined by: constructing at least two databases: a first database for storing the real-time quality status of each automotive component, a second database for storing quantitative standards and visual samples for quality inspection, a third database for providing decision support for initializing the large model, a fourth database for providing semantic understanding for initializing the large model, and a fifth database for storing normative documents for the quality management system; based on the identification information corresponding to each automotive component, associating the multimodal data in the first, second, third, fourth, and fifth databases to obtain associated data related to the automotive components; and training the large model to be trained based on each automotive component and the associated data of each automotive component to obtain a large model for automotive component quality management.

[0015] Secondly, embodiments of this application provide an automotive parts quality management device, the device comprising:

[0016] The receiving module is used to receive input data for quality management of target automotive parts based on the automotive parts quality management system;

[0017] The processing module is used to call the large model pre-deployed in the automotive parts quality management system. The large model calls the target analysis model based on the input data and historical analysis data, and generates a response result based on the first analysis result of the target analysis model. The target analysis model is at least one of the multiple analysis models pre-deployed in the automotive parts quality management system.

[0018] The output module is used to output second analysis results in a preset format based on the response results of the large model through the automotive parts quality management system.

[0019] Thirdly, embodiments of this application provide a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the method provided in the first aspect.

[0020] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a computer device, implements the steps of the method provided in the first aspect above.

[0021] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a computer device, implements the steps of the method provided in the first aspect above.

[0022] The aforementioned automotive parts quality management method, apparatus, equipment, and storage medium receive input data for quality management of target automotive parts from an automotive parts quality management system; invoke a pre-deployed large model within the automotive parts quality management system, enabling the large model to invoke a target analysis model based on the input data and historical analysis data, and generate a response result based on the first analysis result of the target analysis model; the target analysis model is at least one of multiple pre-deployed analysis models in the automotive parts quality management system; based on the response result of the large model, the automotive parts quality management system outputs a second analysis result conforming to a preset format. This method integrates a large model into the automotive parts quality management system, using the system as an intelligent hub. By dynamically scheduling analysis models through the large model to analyze user-input data for quality management of target automotive parts, the system obtains the response result of the large model, and based on the response result, outputs an analysis result conforming to a preset format (i.e., the second analysis result) through the automotive parts quality management system. This improves the efficiency of automotive parts quality management when dealing with massive amounts of multimodal data and different types of input data. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating a method for quality management of automotive parts provided in an embodiment of this application;

[0025] Figure 2 This is a flowchart illustrating another method for quality management of automotive parts provided in this application embodiment;

[0026] Figure 3 This is a schematic diagram of the structure of an automotive parts quality management device provided in an embodiment of this application;

[0027] Figure 4 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0029] The global manufacturing industry is undergoing a profound transformation centered on intelligentization. As a representative of high-end manufacturing, the quality of automotive parts directly impacts the safety and performance of the entire vehicle. Traditional automotive parts quality management methods heavily rely on human experience, resulting in inefficiency, inconsistent standards, and an inability to handle massive amounts of multimodal data (such as images, text, and data). Although artificial intelligence technology has made significant progress in areas such as image recognition and predictive analytics, current applications are mostly siloed, single-point solutions that struggle to deeply integrate with a company's core Quality Management System (QMS), leading to low processing efficiency when dealing with massive amounts of multimodal data.

[0030] To address the aforementioned problems, embodiments of this application provide a method, apparatus, device, and storage medium for automotive parts quality management. The method includes: a computer device receiving input data for quality management of a target automotive part based on an automotive parts quality management system; invoking a pre-deployed large model within the automotive parts quality management system, causing the large model to invoke a target analysis model based on the input data and historical analysis data, and generating a response result based on a first analysis result of the target analysis model; the target analysis model is at least one of multiple pre-deployed analysis models within the automotive parts quality management system; and outputting a second analysis result conforming to a preset format through the automotive parts quality management system based on the response result of the large model. This method integrates a large model into the automotive parts quality management system, using the system as an intelligent hub. By dynamically scheduling analysis models through the large model to analyze user-inputted input data for quality management of the target automotive parts, the system obtains the response result of the large model, and outputs an analysis result (i.e., the second analysis result) conforming to a preset format through the automotive parts quality management system based on the response result. This improves the efficiency of automotive parts quality management when dealing with massive amounts of multimodal data and different types of input data.

[0031] Optionally, the computer equipment mentioned above can be a terminal, which may include, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices may include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection equipment, etc. Portable wearable devices may include smartwatches, smart bracelets, etc.

[0032] The following describes the automotive parts quality management method provided in the embodiments of this application.

[0033] Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for quality management of automotive parts provided in an embodiment of this application. The method can be executed by a computer device in the vehicle. Figure 1 As shown, this automotive parts quality management method may include, but is not limited to, the following steps:

[0034] S101. Receive input data for quality management of target automotive parts based on the automotive parts quality management system.

[0035] The input data refers to image and / or text data entered by the user for quality management of the target automotive parts. For example, assuming the target automotive part is component xx, the input data may include an image of component xx and the text "Analyze whether there are scratches on component xx".

[0036] In one alternative implementation, the computer device receives input data for quality management of a target automotive component based on the automotive component quality management system. This may involve: detecting a startup operation of the automotive component quality management system; responding to the startup operation by displaying the homepage of the automotive component instruction management system on the user interface; and receiving input data for quality management of the target automotive component entered into the input boxes on the homepage of the automotive component instruction management system.

[0037] Optionally, the activation operation may include, but is not limited to, click operation, voice operation, button operation, etc., and there is no limitation here.

[0038] For example, when a user needs to manage the quality of a target automotive part, they can click the icon corresponding to the automotive parts quality management system displayed in the computer device's user interface. The computer device will then respond to this click by displaying the automotive parts quality management system's homepage. As another example, when a user needs to manage the quality of a target automotive part, they can output the voice command "Open automotive parts management system." The computer device will then respond to this voice command by displaying the automotive parts quality management system's homepage. Yet another example is that when a user needs to manage the quality of a target automotive part, they can press the button corresponding to the automotive parts quality management system. The computer device will then respond to this button press by displaying the automotive parts quality management system's homepage.

[0039] S102. Call the pre-deployed large model in the automotive parts quality management system, so that the large model calls the target analysis model based on the input data and historical analysis data, and generates response results based on the first analysis result of the target analysis model.

[0040] The target analysis model is at least one of several pre-deployed analysis models in the automotive parts quality management system.

[0041] Among them, the large model can also be called the Large Language Model (LLM).

[0042] In one alternative implementation, the multiple analysis models pre-deployed in the automotive parts quality management system may include at least two of the following: quality prediction model, image recognition model, image classification model, defect detection model, etc.

[0043] In one optional implementation, the target analysis model is related to the analysis type corresponding to the input data. For example, assuming the analysis type corresponding to the input data is image recognition, the target analysis model is an image recognition model; assuming the analysis type corresponding to the input data is quality prediction, the target analysis model is a quality prediction model.

[0044] In one alternative implementation, prior to step S102, the computer device may pre-train multiple analysis models and deploy the trained analysis models into the automotive parts quality management system.

[0045] Historical analysis data refers to all interaction data prior to the user's input data and after the user's interaction with the large model. Historical analysis data may include, but is not limited to, all historical input data after the previous user-large model interaction and the historical output data corresponding to each historical input data. For example, suppose the input data for quality management of a target automotive component is the data input by the user in the nth instance (denoted as x). n Then, the historical analysis data consists of all historical interaction data after the (n-1)th interaction between the user and the large model (denoted as H). n-1 ), where H n-1 ={(x1,o1),(x2,o2),...,(x n-1 ,o n-1 )},x i This represents the data input by the user for the i-th time, o i This represents the response of the large model to the data input by the user for the i-th time.

[0046] S103. Based on the response results of the large model, the second analysis results conforming to the preset format are output through the automotive parts quality management system.

[0047] In one optional implementation, the preset format can be as specified by the company in quality improvement or as specified in automotive parts quality management standards, etc., and is not limited here. Optionally, the preset format can be a one-page report or an 8-dimensional report, etc.

[0048] In this embodiment, by integrating a large model into the automotive parts quality management system, and using the automotive parts quality management system as the intelligent hub, the large model dynamically schedules and analyzes the input data of the user for quality management of the target automotive parts, and obtains the response results of the large model. Based on the response results, the automotive parts quality management system outputs analysis results (i.e., the second analysis results) that conform to a preset format. In this way, the efficiency of automotive parts quality management can be improved when facing massive multimodal data and different types of input data.

[0049] In one alternative implementation, Figure 1Step S103 in the automotive parts quality management method shown, namely, the method by which the computer equipment outputs a second analysis result conforming to a preset format based on the response result of the large model through the automotive parts quality management system, can be as follows: inputting the response result of the large model into the trained standardized transformation model to obtain the transformed response result; the transformed response result is a structured description of the response result; obtaining a verification dataset related to automotive parts; inputting the transformed response result and the verification dataset into a pre-built verification function to obtain the verification result; based on the verification result, determining the second analysis result conforming to the preset format corresponding to the input data, and outputting the second analysis result through the automotive parts quality management system.

[0050] In some embodiments, when a computer device inputs the response result of a large model into a trained standardized transformation model to obtain the transformed response result, the following formula (1) can be used.

[0051] (1)

[0052] In formula (1), T output T represents the transformed response; T() represents the trained, standardized transformation model; L output This represents the response results of the large model.

[0053] Because the output structures of various analytical models are inconsistent—for example, the XGBoost model outputs a risk value of 0.92, while the convolutional neural network-based image recognition model outputs a bounding box and classification probability [X1, X2, Y1, Y2, "scratches", 0.98]—these data lack business context and have poor readability and operability. Large models can generate fluent text based on instructions, but their style, format, terminology, and completeness may not conform to enterprise standards. Different employees asking questions may produce reports with vastly different formats, hindering rapid information extraction and standardized management. Therefore, computer equipment can address these issues by introducing a trained standardized transformation model. In other words, the trained standardized transformation model primarily addresses the problem of professional models (analytical models) lacking business context and having inconsistent styles with the output of large models. Through the trained standardized transformation model, the non-standardized response results of large models can be forcibly converted into structured, executable standardized content that meets the enterprise's quality system requirements. For example, through fine-tuning training of question-and-answer pairs, the output of each analysis model can be described using the Five Ws and Two Hs Method (5W2H) to describe the problem and to find the root cause of the problem using the Five Whys method.

[0054] In some embodiments, the trained standardized transformation model can be determined by a computer device through the following methods: acquiring training sample pairs; the training sample pairs consist of sample analysis results corresponding to the sample input data and actual structured descriptions corresponding to the sample analysis results; inputting the sample analysis results into the transformation model to be trained to obtain the predicted structured descriptions corresponding to the sample analysis results; training the transformation model in a direction that reduces the difference between the predicted structured descriptions corresponding to the sample analysis results and the actual structured descriptions to obtain the trained standardized transformation model. This approach helps to address the problem of professional models (analysis models) lacking business context in their output and having inconsistent styles with the output of larger models, thereby ensuring that the final output analysis results are structured, executable, and standardized content that meets the requirements of the enterprise's quality system.

[0055] The training sample pairs may include sample input data and sample analysis results.

[0056] For example, the sample analysis results included in the training sample pair are as follows: the detection result of the convolutional neural network is a scratch probability of 0.92, and the actual structured description corresponding to this sample analysis result is as follows: the scratch depth is 0.15 mm, which exceeds the benchmark threshold of 0.1 mm.

[0057] In some embodiments, the verification dataset related to automotive parts may be a dataset stored locally on a computer device, or a dataset stored in a database that is accessible to the computer device, etc., without limitation here.

[0058] In some embodiments, the computer device inputs the converted response result and the verification dataset into a pre-built verification function to obtain the verification result, which can be obtained using the following formula (2).

[0059] (2)

[0060] In formula (2), R represents the verification result; C() represents the pre-constructed verification function, whose output is a Boolean value, True indicates that the verification passed, and False indicates that the verification failed; T output represents the converted response result; D represents the verification dataset related to automotive parts.

[0061] In some embodiments, the computer device determines a second analysis result corresponding to the input data that conforms to a preset format based on the verification result. This can be achieved by: if the verification result indicates that the verification passed, using the converted response result as the second analysis result corresponding to the input data that conforms to the preset format; or, if the verification result indicates that the verification failed, re-invoking the pre-deployed large model in the automotive parts quality management system to regenerate the response result until the verification result passes, thus obtaining a second analysis result conforming to the preset format. In this way, by embedding constraint standards into the automotive parts quality management process, the compliance of the final output result can be ensured.

[0062] In other words, computer equipment can verify results as follows: In this case, determine the final output (i.e., the second analysis result that conforms to the preset format) (denoted as O). n )for ; when the verification result is In such cases, the large model is instructed to regenerate the response based on the validation results until the validation passes.

[0063] By adopting this implementation method, the response results of the large model are standardized and then verified. On the one hand, this ensures that the final output analysis results are standardized, structured, and executable, in accordance with the requirements of the enterprise's quality system. On the other hand, it strictly constrains the final output results within the scope of industry standards and enterprise specifications, thereby solving the problem of compliance of the generated content.

[0064] In one alternative implementation, Figure 1 In step S102 of the automotive parts quality management method shown, the computer device calls a pre-deployed large model in the automotive parts quality management system, enabling the large model to call a target analysis model based on input data and historical analysis data, and generate a response result based on the first analysis result of the target analysis model. This can be achieved by: calling a pre-deployed large model in the automotive parts quality management system, enabling the large model to determine the target input type based on input data and historical analysis data; if the target input type is non-dialogue type, using the large model to call a dedicated parameter extraction interface to extract parameters from the input data, obtaining the key parameters corresponding to the input data; using the large model to call the target analysis model, generating an initial analysis result associated with the input data based on the key parameters; performing structured processing on the initial analysis result to obtain the first analysis result; and calling the large model to generate a response result based on the input data, historical analysis data, and the first analysis result.

[0065] In some embodiments, the computer device invokes a pre-deployed large model in the automotive parts quality management system, enabling the large model to determine the target input type based on input data and historical analysis data. This can be achieved by invoking a pre-deployed large model in the automotive parts quality management system, enabling the large model to determine the target input type based on input data, historical analysis data, a set of input types, and user requirements.

[0066] Historical analysis data refers to the data that was previously input into the large model, and can be denoted as H. n-1 .

[0067] Optionally, the set of input types can be denoted as S = {S0, S1, S2}. 2, ...,S k}, where k > 2; S0 can represent dialogue type, S1 can represent quality prediction type, S2 can represent image recognition type, S k It can represent extended types, etc.

[0068] Optionally, when determining the target input type based on input data, historical analysis data, input type set and user needs, the large model can use the following formula (3).

[0069] (3)

[0070] In formula (3), x n This represents the input data; H n-1 This represents historical analysis data; This indicates the target input type, where, S = {S0, S1, S2, ..., S} k}, S0 represents the dialogue type, S1 represents the quality prediction type, S2 represents the image recognition type, S k L() represents an extended type; L() represents a large model.

[0071] In some embodiments, when a computer device calls a dedicated parameter extraction interface through a large model to extract parameters from the input data and obtain the key parameters corresponding to the input data, the following formula (4) can be used. For example, the key parameter is, for instance, the "raw material batch" in quality prediction.

[0072] (4)

[0073] In formula (4), x n This represents the input data; H n-1 This refers to historical analysis data; L ext () indicates a dedicated interface for parameter extraction, where L ext()=L(“Extract key parameters of the calling tool model:”,x n H n-1 ); param represents the key parameters corresponding to the input data.

[0074] The target analysis model can be any one of the analysis models in the set, and the target analysis model is associated with the target input type. The set of analysis models can be denoted as M = {M1, M2, ..., M}. k}, k>2, where M1 represents the quality prediction model, M2 represents the image recognition model, M k This represents the extended model.

[0075] For example, assuming the target input type is quality prediction, the target analysis model is M; assuming the target input type is image recognition, the target analysis model is M2; assuming the target input type is an extended type, the target analysis model is M... k .

[0076] In some embodiments, when a computer device calls a target analysis model through a large model and generates initial analysis results associated with the input data based on key parameters, the following formula (5) may be used.

[0077] (5)

[0078] In formula (5), M i_result This represents the initial analysis results associated with the input data, for example, M. i_result q represents defect detection results, quality risk values, etc.; param represents the key parameters corresponding to the input data; M i () represents the target analysis model.

[0079] In some embodiments, the computer device performs structured processing on the initial analysis results to obtain a first analysis result. This can be achieved by inputting the initial analysis results into a trained standardized transformation model. In this way, the raw output of the target analysis model (i.e., the initial analysis result) can be transformed into quantifiable or interpretable content. For example, assuming the initial analysis result is a risk value of 0.85, the transformed analysis result (i.e., the first analysis result) could be high risk, with the key factor being welding temperature (weight 0.35).

[0080] The process by which the computer equipment performs structured processing on the initial analysis results to obtain the first analysis result can be expressed as the following formula (6).

[0081] (6)

[0082] In formula (6), T() represents the standardized transformation model after training; This represents the first analysis result; the physical meaning of the other parameters can be found in the previous explanation of the physical meaning of the parameters in formula (5), and will not be repeated here.

[0083] In some embodiments, the computer device invokes a large model to generate a response result based on the input data, historical analysis data, and the first analysis result. This can be achieved by invoking the large model to integrate the input data, historical analysis data, and the first analysis result to obtain the response result. The process of generating the response result described above can be represented by the following formula (7).

[0084] (7)

[0085] In formula (7), L output This represents the response result; L() represents the large model; x n This represents the input data; H n-1 This represents historical analysis data; This indicates the results of the first analysis.

[0086] In some embodiments, the computer device may also generate response results from a large model when the target input type is a dialogue type.

[0087] For example, suppose Dialogue type, i.e. Then the computer device can call the large model pre-deployed in the automotive parts quality management system, so that the large model can generate the original dialogue response based on the user's input data and historical analysis data, and use the original dialogue response as the response result.

[0088] By adopting this implementation method, a large model pre-deployed in the automotive parts quality management system is invoked. Based on the input data and historical analysis data, the large model determines the target input type, which is beneficial for subsequent invocation of the target analysis model based on the target input type, thereby improving the efficiency of automotive parts quality management.

[0089] In one alternative implementation, the aforementioned large model can be determined by a computer device in the following manner: Constructing at least two databases: a first database for storing the real-time quality status of each automotive component, a second database for storing quantitative standards and visual samples for quality inspection, a third database for providing decision support for initializing the large model, a fourth database for providing semantic understanding for initializing the large model, and a fifth database for storing normative documents of the quality management system; associating the multimodal data in the first, second, third, fourth, and fifth databases based on the identification information corresponding to each automotive component to obtain associated data related to the automotive components; and training the large model to be trained based on each automotive component and the associated data of each automotive component to obtain a large model for automotive component quality management.

[0090] In some embodiments, the first database may also be referred to as the quality state database. The quality state database can be used to provide data support for supervised learning and real-time decision-making of quality prediction models (e.g., XGBoost models), and the data included is also an important basis for describing the variable weight parameters in the transformation of quality prediction results. For example, the content structure of the quality state database can be shown in Table 1 below.

[0091] Table 1 Content Structure of the Quality Status Database

[0092]

[0093] In some embodiments, the second database may also be referred to as a quality benchmark database. The quality benchmark database can be used to provide training data for image recognition models (e.g., Convolutional Neural Networks (CNN) models) while simultaneously constraining the output compliance of quality prediction models and large models. The data in the quality benchmark database serves as a benchmark source for transforming image recognition results into quantifiable descriptions of non-compliance. For example, the content structure of the quality benchmark database may be shown in Table 2 below.

[0094] Table 2 Content Structure of the Quality Benchmark Database

[0095]

[0096] In some embodiments, the third database may also be referred to as a quality improvement database. The quality improvement database can be used to record the closed-loop improvement process of quality issues throughout the entire chain from R&D to after-sales service. It serves as a knowledge base providing root cause analysis and decision support for large-scale models, and is also an important reference for large-scale models to output relevant improvement suggestion templates. For example, the content structure of the quality improvement database can be shown in Table 3 below.

[0097] Table 3 Content Structure of the Quality Improvement Database

[0098]

[0099] The 5 Whys analysis report is a formal document that systematically records and demonstrates the entire process and results of using the "5 Whys analysis method" to solve problems. Its core concept is that to thoroughly solve a problem, one cannot just deal with the surface symptoms, but needs to dig deeper until the "root cause" of the problem is found. A professional and complete 5 Whys analysis report usually includes the following parts: 1. Problem description; 2. Team formation; 3. Problem chain analysis (the core of the report); 4. Root cause identification; 5. Development and implementation of corrective actions; 6. Effect verification and standardization; 7. Horizontal expansion and recurrence prevention.

[0100] Fishbone diagrams, also known as cause-and-effect diagrams, are graphical tools used to systematically and structurally identify and display the potential root causes of problems.

[0101] In some embodiments, the fourth database may also be referred to as a quality knowledge database. The quality knowledge database can be used to archive quality-related technical documents and experiential knowledge, enriching the domain semantic understanding capabilities of large models and helping them to more accurately explain technical terms and English abbreviations. For example, the content structure of the quality knowledge database can be shown in Table 4 below.

[0102] Table 4 Content Structure of the Quality Knowledge Database

[0103]

[0104] Design Failure Mode and Effects Analysis (DFMEA) refers to the analysis of potential failure modes during the design phase. It is a means of preventing product quality issues from the design stage and a control tool for ensuring product quality meets standards during the formal production and delivery process to customers. Due to the similarity of similar products, the DFMEA stage often draws on the advantages and disadvantages of previously mass-produced or currently in-production products to evaluate and improve upon the new product.

[0105] Retrieval-Augmented Generation (RAG) is a framework that combines information retrieval techniques with large model generation techniques. Its core idea can be summarized as: retrieve first, then generate, or use facts to support generation. Simply put, RAG allows large models to answer questions not only by relying on their internal memory (parameterized knowledge), but also by first searching for relevant, up-to-date, and specific information from external knowledge bases, and then organizing the answer based on these retrieved results.

[0106] In some embodiments, the fifth database may also be referred to as the quality system database. The quality system database can be used to align the output of the large model with industry compliance requirements and serves as an important basis for binding verification functions. For example, the content structure of the quality system database can be shown in Table 5 below.

[0107] Table 5 Content Structure of the Quality System Database

[0108]

[0109] In some embodiments, the multimodal data in the quality status database, quality benchmark database, quality improvement database, quality knowledge base, and quality system database may include, but is not limited to, structured data, text data, and image data, etc., without limitation here.

[0110] For example, assuming the identification information for an automotive component is a left headlight assembly, the computer equipment can correlate multimodal data in the quality status database, quality benchmark database, quality improvement database, quality knowledge base, and quality system database associated with the left headlight assembly to obtain the associated data corresponding to the left headlight assembly. This not only achieves alignment of the inspection data, process text, and images for the same automotive component, but also avoids the problem of low semantic understanding accuracy and inability to perform knowledge association in automotive component quality management scenarios due to inconsistent data terminology (e.g., "product quality planning" may exist in different forms such as the full Chinese name, English abbreviation, or the full English name in different databases such as R&D, production, and suppliers).

[0111] In some embodiments, during the process of constructing training sample pairs based on various automotive components and their associated data, the constructed training sample pairs can be represented in the form of a terminology mapping table. This mapping table serves as the basis for subsequent model fine-tuning and the execution of the T() function. Based on this mapping table, it can be fundamentally ensured that both the internal cognition and the external output conform to strict industry terminology standards when processing multi-source information.

[0112] The process of constructing a terminology mapping table using computer equipment may include: obtaining standard documents and internal documents related to each automotive component; initially screening the first terminology corresponding to the associated data of each automotive component from the standard documents and internal documents; and / or obtaining the second terminology corresponding to the associated data of each automotive component collected in historical work; outputting the first terminology and / or the second terminology so that the quality expert team can review the first terminology and / or the second terminology; and, if the review is passed, constructing a terminology mapping table based on each automotive component and the first terminology and / or the second terminology corresponding to the associated data of each automotive component.

[0113] In some embodiments, training sample pairs can also be referred to as compliance training sample pairs. During the process of training an initial large-scale model for automotive parts quality management based on these training sample pairs, the computer device can inject core constraint principles into the model parameters. For example, it can modify the vague "stable process, good quality" to a quantifiable "Process Capability Index (CPK) of 1.67, better than the acceptance criterion of 1.33," to train the initial large-scale model to proactively avoid non-quantitative expressions. Furthermore, the computer device can extract explicit verification rules (such as "root cause analysis must include 5 Whys logic") from the quality system database and automotive parts-related verification datasets, and pre-set them in the verification function. Finally, a quality closed loop is formed through real-time verification and correction: during inference, the verification function performs compliance matching and judgment on the initial output of the large-scale model; if the verification fails, a feedback and correction mechanism is triggered, feeding back the violation content and specific reasons to the large-scale model, so that the large-scale model regenerates the response results until the output is fully compliant.

[0114] This implementation method ensures the compliance and accuracy of the output results of the large model used for automotive parts quality management.

[0115] The following is combined Figure 2 This paper provides an overall description of the automotive parts quality management method provided in the embodiments of this application. Please refer to [link / reference]. Figure 2 , Figure 2 This is a flowchart illustrating another method for quality management of automotive parts provided in an embodiment of this application. Figure 2 As shown, this automotive parts quality management method may include, but is not limited to, the following steps:

[0116] S201. Receive input data for quality management of target automotive parts based on the automotive parts quality management system.

[0117] In an optional implementation, the relevant description of step S201 can be found in the description of step S101 above, and will not be repeated here.

[0118] S202. Call the pre-deployed large model in the automotive parts quality management system, so that the large model can determine the target input type based on the input data and historical analysis data.

[0119] In some embodiments, when a computer device calls a pre-deployed large model in the automotive parts quality management system, and the large model determines the target input type based on input data and historical analysis data, the aforementioned formula (3) can be used.

[0120] S203. Determine whether the target input type is a dialog type. If yes, proceed to step S204; otherwise, proceed to steps S205 to S207.

[0121] S204. Generate the response results corresponding to the input data through the large model.

[0122] S205. By calling the dedicated parameter extraction interface through the large model, parameters are extracted from the input data to obtain the key parameters corresponding to the input data.

[0123] In some embodiments, when a computer device calls a dedicated parameter extraction interface through a large model to extract parameters from the input data and obtain the key parameters corresponding to the input data, the aforementioned formula (4) can be used.

[0124] S206. Using the large model, call the target analysis model and generate initial analysis results associated with the input data based on key parameters.

[0125] In some embodiments, when a computer device calls a target analysis model through a large model and generates initial analysis results associated with the input data based on key parameters, the aforementioned formula (5) may be used.

[0126] S207. The initial analysis results are structured to obtain the first analysis result. The large model is then invoked to generate the response result based on the input data, historical analysis data, and the first analysis result.

[0127] In some embodiments, when a computer device performs structured processing on the initial analysis results to obtain the first analysis result, the aforementioned formula (6) may be used.

[0128] In some embodiments, when a computer device invokes a large model to generate a response result based on input data, historical analysis data, and the first analysis result, the aforementioned formula (7) may be used.

[0129] S208. Input the response results of the large model into the trained standardized transformation model to obtain the transformed response results.

[0130] In some embodiments, when a computer device inputs the response result of a large model into a trained standardized transformation model to obtain the transformed response result, the aforementioned formula (1) can be used.

[0131] S209. Obtain the verification dataset related to automotive parts.

[0132] S210. Input the converted response result and the verification dataset into the pre-built verification function to obtain the verification result.

[0133] In some embodiments, the computer device inputs the converted response result and the verification dataset into a pre-built verification function to obtain the verification result, and the aforementioned formula (2) can be used.

[0134] S211. Based on the verification results, determine the second analysis result that conforms to the preset format corresponding to the input data, and output the second analysis result through the automotive parts quality management system.

[0135] In some embodiments, the computer device determines a second analysis result corresponding to the input data that conforms to a preset format based on the verification result. This can be achieved by: if the verification result indicates that the verification passed, using the converted response result as the second analysis result corresponding to the input data that conforms to the preset format; or, if the verification result indicates that the verification failed, re-executing steps S202 to S211 until the verification result passed, and using the last generated response result as the second analysis result corresponding to the input data that conforms to the preset format.

[0136] The following is an overall description of the automotive parts quality management process provided in the embodiments of this application.

[0137] Input: x n (nth user input), H n-1 (The history of the (n-1)th interaction, including input, output and type);

[0138] Routing decision (Type calculation):

[0139] / / Call the routing decision function

[0140] The calculation process for Type is as follows:

[0141]

[0142] (The range of values ​​for Type:) )

[0143] Response generation:

[0144] Depending on the Type, the corresponding processing flow is executed, and the final output needs to undergo transformation training and constraint verification:

[0145] (1) If Type=S0 (dialogue type), then

[0146] Step 1: Generate initial dialogue content from the large model:

[0147] / / Generate the original dialogue response based on user input and history.

[0148] Step 2: Specialized training for transformation (standardized dialogue expression):

[0149] / / Transform the original dialogue into content that conforms to industry terminology and formatting standards.

[0150] Step 3: Constraint verification (compliance with system documents and calibration data):

[0151] like The final output will be... ;

[0152] like If the error persists, return to step 1 and have L regenerate the dialogue content until the verification passes.

[0153]

[0154] Repeat steps 2-3 until... ,at this time .

[0155] (2) If Type=S i If 1≤i≤k (non-dialogue type), then

[0156] Step 1: Extract key parameters:

[0157] / / Extract M i Required parameters (such as "raw material batch" for quality prediction)

[0158] Step 2: Call the target analysis model:

[0159] / / Output the raw results (such as defect detection probability, quality risk value)

[0160] Step 3: Transform the results of the target analysis model (structured processing):

[0161] / / Transform the raw output of the model into quantifiable or interpretable content (e.g., "Risk value 0.85 → High risk, key factor: welding temperature (weight 0.35)")

[0162] Step 4: Integrating the large language model to generate preliminary content:

[0163] / / Generate response results by combining user input, historical data, and the converted model output.

[0164] Step 5: Convert the large model output (template processing):

[0165] / / Convert the response results into content that conforms to a preset format (such as a fixed report format specified by the company in quality improvement).

[0166] Step 6: Constraint verification (compliance with system documents and calibration data):

[0167] like The final output will be... ;

[0168] like If the error persists, return to step 4 and have L() regenerate the response until the verification passes.

[0169]

[0170] Repeat steps 5-6 until... ,at this time .

[0171] In some embodiments, the computer device may also include the inputs, outputs, and types of the current interaction in a history for use in the next interaction:

[0172] / / Add the current interaction type to the history.

[0173] In some embodiments, the above-mentioned automotive parts quality management process can be expressed as the following formula (8).

[0174] (8)

[0175] The physical meaning of each parameter in formula (8) can be found in the descriptions of the other formulas mentioned above, and will not be repeated here.

[0176] In this embodiment, by integrating a large model into the automotive parts quality management system, and using the automotive parts quality management system as the intelligent hub, the large model dynamically schedules and analyzes the input data of the user for quality management of the target automotive parts, and obtains the response results of the large model. Based on the response results, the automotive parts quality management system outputs analysis results (i.e., the second analysis results) that conform to a preset format. In this way, the efficiency of automotive parts quality management can be improved when facing massive multimodal data and different types of input data.

[0177] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0178] In some embodiments, the main application scenarios and integration schemes of the large model mentioned in this application in the automotive parts quality management system are shown in Table 6 below.

[0179] Table 6. Main application scenarios and integration solutions of the large model in QMS.

[0180]

[0181]

[0182]

[0183]

[0184] Based on the same inventive concept, this application also provides an automotive parts quality management device for implementing the aforementioned automotive parts quality management method. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the automotive parts quality management device provided below can be found in the limitations of the automotive parts quality management method described above, and will not be repeated here.

[0185] Please see Figure 3 , Figure 3This is a schematic diagram of the structure of an automotive parts quality management device provided in an embodiment of this application. Figure 3 As shown, the automotive parts quality management device may include, but is not limited to:

[0186] The receiving module 301 is used to receive input data for quality management of target automotive parts based on the automotive parts quality management system;

[0187] The processing module 302 is used to call a pre-deployed large model in the automotive parts quality management system, so that the large model calls the target analysis model based on the input data and historical analysis data, and generates a response result based on the first analysis result of the target analysis model; the target analysis model is at least one of multiple analysis models pre-deployed in the automotive parts quality management system.

[0188] Output module 303 is used to output a second analysis result conforming to a preset format through the automotive parts quality management system based on the response results of the large model.

[0189] In some embodiments, when the output module 303 outputs a second analysis result conforming to a preset format through the automotive parts quality management system based on the response result of the large model, the processing module 302 is used to input the response result of the large model into the trained standardized conversion model to obtain the converted response result; the converted response result is a structured description of the response result; obtain a verification dataset related to automotive parts; input the converted response result and the verification dataset into a pre-built verification function to obtain the verification result; based on the verification result, determine the second analysis result conforming to the preset format corresponding to the input data, and the output module 303 is used to output the second analysis result through the automotive parts quality management system.

[0190] In some embodiments, when the processing module 302 determines the second analysis result corresponding to the input data that conforms to a preset format based on the verification result, it is specifically used to: if the verification result indicates that the verification is passed, use the converted response result as the second analysis result corresponding to the input data that conforms to a preset format; or, if the verification result indicates that the verification is failed, re-call the large model pre-deployed in the automotive parts quality management system to regenerate the response result until the verification result passes, and obtain the second analysis result that conforms to the preset format.

[0191] In some embodiments, when the processing module 302 invokes a pre-deployed large model in the automotive parts quality management system, causing the large model to invoke a target analysis model based on input data and historical analysis data, and generate a response result based on the first analysis result of the target analysis model, the specific steps are as follows: Invoking a pre-deployed large model in the automotive parts quality management system, causing the large model to determine the target input type based on input data and historical analysis data; when the target input type is a non-dialogue type, using the large model to invoke a dedicated parameter extraction interface to extract parameters from the input data, obtaining the key parameters corresponding to the input data; using the large model to invoke the target analysis model, generating an initial analysis result associated with the input data based on the key parameters; performing structured processing on the initial analysis result to obtain a first analysis result; and invoking the large model to generate a response result based on the input data, historical analysis data, and the first analysis result.

[0192] In some embodiments, the processing module 302 is further configured to: generate a response result using a large model when the target input type is a dialogue type.

[0193] In some embodiments, the apparatus may further include a training module. The training module is used to acquire training sample pairs; the training sample pairs consist of sample analysis results corresponding to sample input data and actual structured descriptions corresponding to the sample analysis results; inputting the sample analysis results into the conversion model to be trained to obtain the predicted structured descriptions corresponding to the sample analysis results; training the conversion model in a direction that reduces the difference between the predicted structured descriptions corresponding to the sample analysis results and the actual structured descriptions to obtain the trained standardized conversion model.

[0194] In some embodiments, the training module is further configured to construct at least two databases: a first database for storing the real-time quality status of each automotive component, a second database for storing quantitative standards and visual samples for quality inspection, a third database for providing decision support for the initial large model, a fourth database for providing semantic understanding for the initial large model, and a fifth database for storing normative documents of the quality management system; based on the identification information corresponding to each automotive component, the multimodal data in the first, second, third, fourth, and fifth databases are associated to obtain associated data related to the automotive components; based on each automotive component and the associated data of each automotive component, the large model to be trained is trained to obtain a large model for automotive component quality management.

[0195] It is understood that the specific implementation of each module in the automotive parts quality management device provided in this application embodiment and the beneficial effects that can be achieved can be referred to the description of the aforementioned automotive parts quality management method embodiment, and will not be repeated here.

[0196] Each module in the aforementioned automotive parts quality management device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored as software in the memory of the automotive parts quality management device, so that the processor can call and execute the corresponding operations of each module.

[0197] In one exemplary embodiment, a computer device is provided, the internal structure of which can be as shown in the figure. Figure 4 As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for quality management of automotive parts. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads installed inside the computer device.

[0198] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0199] In one exemplary embodiment, this application provides a computer device including a memory and a processor, wherein the memory stores a computer program; when the processor executes the computer program, it implements the steps in the above-described automotive parts quality management methods.

[0200] In one exemplary embodiment, this application provides a computer-readable storage medium having a computer program stored thereon. When executed by a processor, the computer program implements the steps in the above-described automotive parts quality management methods.

[0201] In one exemplary embodiment, this application provides a computer program product, including a computer program. When executed by a processor, the computer program implements the steps in the aforementioned automotive parts quality management methods.

[0202] It should be noted that the data involved in this application (including but not limited to data used for analysis, data stored, data displayed, etc.) are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0203] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0204] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0205] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for quality management of automotive parts, characterized in that, The method includes: The automotive parts quality management system receives input data for quality management of target automotive parts. The system invokes a pre-deployed large model in the automotive parts quality management system, which, based on the input data and historical analysis data, invokes a target analysis model and generates a response result based on the first analysis result of the target analysis model; the target analysis model is at least one of multiple pre-deployed analysis models in the automotive parts quality management system. Based on the response results of the large model, the automotive parts quality management system outputs a second analysis result that conforms to a preset format.

2. The method according to claim 1, characterized in that, The response results based on the large model are used by the automotive parts quality management system to output a second analysis result conforming to a preset format, including: The response result of the large model is input into the trained standardized transformation model to obtain the transformed response result; the transformed response result is a structured description of the response result. Obtain the verification dataset related to automotive parts; The transformed response and the verification dataset are input into a pre-built verification function to obtain the verification result; Based on the verification result, a second analysis result corresponding to the input data that conforms to a preset format is determined, and the second analysis result is output through the automotive parts quality management system.

3. The method according to claim 2, characterized in that, The step of determining the second analysis result corresponding to the input data in a preset format based on the verification result includes: If the verification result indicates that the verification passed, the converted response result will be used as the second analysis result corresponding to the input data and conforming to the preset format; or, If the verification result indicates that the verification fails, the pre-deployed large model in the automotive parts quality management system is called again to regenerate the response result until the verification result passes, and a second analysis result conforming to the preset format is obtained.

4. The method according to claim 1, characterized in that, The step of calling a pre-deployed large model in the automotive parts quality management system, enabling the large model to call a target analysis model based on the input data and historical analysis data, and generate a response result based on the first analysis result of the target analysis model, includes: The system invokes a pre-deployed large model in the automotive parts quality management system, enabling the large model to determine the target input type based on the input data and historical analysis data. When the target input type is non-dialogue type, the large model calls the dedicated parameter extraction interface to extract parameters from the input data and obtain the key parameters corresponding to the input data. Using the large model, the target analysis model is invoked, and based on the key parameters, initial analysis results associated with the input data are generated; The initial analysis results are structured to obtain the first analysis result; The large model generates a response result based on the input data, the historical analysis data, and the first analysis result.

5. The method according to claim 4, characterized in that, The method further includes: When the target input type is a dialogue type, the response result is generated through the large model.

6. The method according to claim 2, characterized in that, The trained standardized transformation model was determined in the following way: Obtain training sample pairs; each training sample pair consists of the sample analysis results corresponding to the sample input data and the actual structured description corresponding to the sample analysis results; The sample analysis results are input into the conversion model to be trained to obtain the predicted structured description corresponding to the sample analysis results; The conversion model is trained in the direction of reducing the difference between the predicted structured description and the actual structured description corresponding to the sample analysis results, so as to obtain the trained standardized conversion model.

7. The method according to any one of claims 1 to 6, characterized in that, The large model was determined in the following way: Construct at least two databases: a first database for storing the real-time quality status of each automotive component, a second database for storing quantitative standards and visual samples for quality inspection, a third database for providing decision support for the initial large model, a fourth database for providing semantic understanding of the initial large model, and a fifth database for storing normative documents of the quality management system. Based on the identification information corresponding to each of the aforementioned automotive parts, the multimodal data in the first database, the second database, the third database, the fourth database, and the fifth database are associated to obtain associated data related to the automotive parts; Based on each of the aforementioned automotive parts and the associated data associated with each of the aforementioned automotive parts, a large model to be trained is obtained to obtain a large model for automotive parts quality management.

8. A quality management device for automotive parts, characterized in that, The device includes: The receiving module is used to receive input data for quality management of target automotive parts based on the automotive parts quality management system; The processing module is used to call a pre-deployed large model in the automotive parts quality management system, so that the large model calls a target analysis model based on the input data and historical analysis data, and generates a response result based on the first analysis result of the target analysis model; the target analysis model is at least one of a plurality of analysis models pre-deployed in the automotive parts quality management system. The output module is used to output a second analysis result conforming to a preset format through the automotive parts quality management system based on the response results of the large model.

9. A computer device, comprising a memory and a processor, characterized in that, The memory stores a computer program; when the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images