New energy automobile product full life cycle problem management method based on knowledge graph

Through a multimodal knowledge mining algorithm based on knowledge graph, a knowledge graph for life cycle problem management of new energy vehicle products is constructed, and the problem of manual dependence in the management of the entire life cycle problem of new energy vehicle products is solved, and efficient and systematic problem management is achieved. , improves the quality management level of product life cycle.

CN118966549AActive Publication Date: 2024-11-15CHONGQING UNIV
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
CN202411100105.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-15
Estimated Expiration
2044-08-12

AI Technical Summary

Technical Problem

In the management of the entire life cycle of new energy vehicle products, the existing technology relies on manual discovery, confirmation, analysis and resolution of problems, resulting in low efficiency, poor quality, and lack of systematic management, which is easy to miss important problems and inconvenient data management. Information leakage risk.

Method used

Using a knowledge graph-based method, the problem data of different modalities is converted into triple clusters through multimodal knowledge mining algorithms to build a problem management knowledge graph for the life cycle of new energy vehicle products, and realize the automated discovery, classification and solution of problems. Extraction and verification, combined with dynamic cross-modal weighted fusion, form an intelligent problem management system.

Benefits of technology

It improves the efficiency and quality of problem discovery and resolution, ensures systematic management of problems, reduces manual intervention, enhances data accuracy and consistency, reduces information leakage risks, and improves the quality management level of product life cycle.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118966549A_ABST
    Figure CN118966549A_ABST
Patent Text Reader

Abstract

The invention discloses a new energy automobile product full life cycle problem management method based on a knowledge graph, and the method comprises the following steps: 1, inputting a problem, searching whether the problem is a repeated problem from a historical problem library, and if yes, executing a step 7; if not, executing the step 2; 2, classifying the questions based on professional knowledge and experience, and determining the types and properties of the questions; step 3, finding out measures for solving the problem in the historical problem library and proposing a problem solution; 4, the feasibility and effectiveness of the proposed problem solution are evaluated, and if the evaluation is passed, the step 5 is executed; if the evaluation is not passed, executing the step 3; 5, the problem solution is actually verified, and if verification is passed, the step 6 is executed; if the verification is not passed, executing the step 3; step 6, uploading the problem data to a data center to update a historical problem library; and 7, completing problem solving.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of product information management, and specifically provides a method for managing problems in the entire life cycle of new energy vehicle products based on a knowledge graph. Background Art

[0002] In the current market environment, enterprises are facing increasingly fierce competition. Discovering and solving problems and ensuring the quality of products and services have become key elements of enterprise competitiveness. The importance of discovering and solving problems has risen to the strategic height of enterprise survival and development. Enterprises should attach great importance to the ability to discover and solve problems, and establish perfect mechanisms and measures to continuously improve the ability to discover and solve problems, so as to achieve the strategic goal of first-class work efficiency and competitiveness, and constantly move closer to the ultimate goal of "eliminating recurring problems and solving high-value prevention and pursuit problems".

[0003] Knowledge graph is a knowledge management carrier. By associating knowledge of different modalities and different cycles in the form of triples, it can efficiently organize and manage complex information. In the full-cycle problem management of new energy vehicles, building a knowledge graph helps to systematically manage various types of problems and their solutions. The knowledge graph for the full-cycle problem management of new energy vehicles is constructed so that the knowledge graph contains a large amount of vertical domain knowledge of various problems in the life cycle of new energy vehicles. This knowledge includes not only technical problems in the R&D stage, process problems in the production stage, and management problems in the supply chain, but also customer complaints and feedback problems in the use stage. However, the amount of data involved in the life cycle of new energy vehicles is huge and the sources are diverse. How to efficiently collect, organize and integrate these data is the primary challenge in building a knowledge graph. The accuracy, completeness and consistency of the data need to be considered; in addition, the knowledge graph needs to have strong semantic parsing and reasoning capabilities to understand and infer the relationship between different problems. This requires deep technical accumulation in natural language processing, machine learning and knowledge representation; new energy vehicle technology and market demand are constantly changing, and the knowledge graph needs to be able to be dynamically updated and maintained to reflect the latest technological progress and market feedback in a timely manner. This involves automated data collection and update mechanisms, as well as effective management of old knowledge.

[0004] However, most companies still use text editing software, emails, and even human memory to promote problem discovery and resolution, which has the following disadvantages:

[0005] 1. The discovery, confirmation, analysis, and resolution of problems completely rely on people’s spontaneous promotion and memory, which can easily lead to omissions or forgetfulness. On the one hand, this reduces the efficiency of problem discovery and resolution, and on the other hand, it affects the quality of problem resolution.

[0006] 2. There are many and complicated issues related to business operations and management. Which issue or issues to be resolved first depends entirely on people's cognition and preferences. It is easy to overlook important and high-priority issues, which may bring hidden dangers or losses to business operations and management.

[0007] 3. The correctness of problem analysis depends entirely on human cognition and experience. The lack of a universal analysis method reduces the efficiency and quality of problem analysis.

[0008] 4. Problem discovery, confirmation, analysis, resolution, statistics and analysis (including process and results) require a lot of manpower and are prone to errors.

[0009] 5. Problem discovery and resolution are not fully aligned with the corporate organizational structure. On the one hand, manual promotion affects efficiency, and there is a risk of corporate management information leakage due to improper authority management. On the other hand, if the performance of employees is evaluated based on problem discovery and resolution, it will take a huge amount of manpower (or even impossible).

[0010] Based on the above problems, using software technology to support enterprises in efficiently discovering and solving problems, replacing people's use of word editing software, emails, and even human brain memory to promote problem discovery and solving, is an issue that needs to be addressed urgently. Summary of the invention

[0011] In view of this, in order to solve the shortcomings of the prior art in product life cycle problem management, the purpose of the present invention is to provide a new energy vehicle product life cycle problem management method based on knowledge graph.

[0012] In order to achieve the above object, the present invention provides the following technical solutions:

[0013] A method for managing problems in the entire life cycle of new energy vehicle products based on a knowledge graph includes the following steps:

[0014] Step 1: Enter the question and search the historical question database to see if the question is a repeated question. If so, proceed to step 2. If not, the question is a new question and proceed to step 3.

[0015] Step 2: Classify the problems based on professional knowledge and experience to determine the type and nature of the problems;

[0016] Step 3: For repeated problems, according to the type of problem, find solutions to the problem in the historical problem database and propose solutions to the problem; for new problems, input the new problem into the new energy vehicle product life cycle problem management knowledge map built in the historical problem database, make judgments and inferences on the new problem, and propose solutions to solve the new problem;

[0017] Step 2: Classify the problems based on professional knowledge and experience to determine the type and nature of the problems;

[0018] Step 3: According to the type of problem, find solutions to the problem in the historical problem database and propose solutions to the problem;

[0019] Step 4: Evaluate the feasibility and effectiveness of the proposed solution: If the solution passes the evaluation, proceed to step 5; if the solution fails the evaluation, proceed to step 3;

[0020] Step 5: Perform actual verification of the problem solution: If the problem solution is verified, proceed to step 6; if the problem solution is not verified, proceed to step 3;

[0021] Step 6: Upload the problem data including problems, problem solutions and implementation experience to the data center to update the historical problem database;

[0022] Step 7: Problem solving is complete.

[0023] Furthermore, a knowledge graph of new energy vehicle product life cycle problem management is constructed in the historical problem database based on the problem data. The construction method of the knowledge graph of new energy vehicle product life cycle problem management is as follows:

[0024] S1: Obtain relevant data information from the entire life cycle of new energy vehicles, including: product design information in the R&D stage, product production information in the manufacturing stage, and user feedback information in the service stage;

[0025] S2: Based on the product history maintenance database, combined with the new energy vehicle full life cycle problem database and case database, a new energy vehicle problem management database is formed;

[0026] S3: For the new energy vehicle problem management database, a multimodal knowledge mining algorithm is used to present problem data of different modes in text form to form triple clusters of problem data of each mode;

[0027] S4: Globally encode the triplet clusters of each modal problem data, and perform modal fusion and cross-modal alignment to achieve cross-modal problem data fusion and form a knowledge graph for new energy vehicle product life cycle problem management.

[0028] Furthermore, in step S3, the method steps of converting question data of different modalities into triples using a multimodal knowledge mining algorithm are as follows:

[0029] S31: Explore the natural language description method of multimodal industrial data and establish an industrial knowledge mining framework for modal feature-natural language description consistency generation;

[0030] S32: An industrial knowledge mining framework generated by modal feature-natural language description consistency is used to convert non-text modal data into text or graph form, and multimodal knowledge mining is achieved through text extraction methods, including:

[0031] For text modality: First, the PL-Marker strategy is used to encode text sentences or paragraphs, and then Albert is used for embedding representation. Finally, a hypergraph is used to model the embedded representation, and a hypergraph neural network is constructed to extract text modality knowledge and form a text modality triple group.

[0032] For image modality: select several multi-image description models, perform image description tasks on the same image, obtain multiple image description texts, embed and represent the description texts and images, perform feature recognition, and sort the description text quality. Input the top-k description texts into a combined prompt and input it into the large language model to generate a fused image description. Use the text modality knowledge extraction method to extract image modality industrial knowledge and form an image modality triple group.

[0033] For the graphic modality: first, all 3D models are converted into a standard format, and then the PCL library is introduced to process the 3D models into point clouds to obtain the 3D point clouds of the models; then the 3D object detector is used to obtain the relevant features of the point cloud, and the point cloud description text is generated based on multi-head cross attention; finally, the text modality knowledge extraction method is used to realize the graphic modality industrial knowledge extraction and form a graphic modality triple group;

[0034] For signal modality: first, the signal data is stored in a standard database, then the signal is processed and the typical time-frequency domain features of the signal are extracted; under the guidance of prior knowledge such as sensors, signal-text association rules are constructed; finally, the text modality knowledge extraction method is used to extract industrial knowledge of signal modality and form a signal modality triple group;

[0035] Regarding expression modality: according to the different forms of expressions, it is divided into unstructured expression data and semi-structured expression data; for semi-structured expression data, the discretized information is directly extracted and a graph structure is formed; for unstructured expression data, the expression in the image is first recognized through OCR, and then it is discretized to realize the construction of the scene graph of discrete elements, and finally form an expression modality triple group.

[0036] Furthermore, in step S4, the method steps for forming a knowledge graph of new energy vehicle product life cycle problem management are as follows:

[0037] According to the correlation relationship of modal problem data triple clusters, the modal original data features are introduced, and the modal triple group graph features of each modal problem data triple cluster are combined for global encoding. The weights of each part of the features are assigned through dynamic cross-modal weighting, and then modal fusion and cross-modal alignment operations are performed to achieve cross-modal knowledge fusion and form a knowledge graph for new energy vehicle product life cycle problem management.

[0038] The new energy vehicle product life cycle problem management knowledge graph uses the URI number as the core node and is loaded with text attribute descriptions, modal attribute descriptions, standardized data attributes and feature attributes.

[0039] Furthermore, in step three, when the problem type is a production quality problem, dynamically adjust IQC sampling, failure mode and effect analysis methods and control plan updates to find measures to solve the problem in the historical problem library and propose solutions to the problem.

[0040] Further, the failure mode and effects analysis method includes the following steps:

[0041] 31) Identify failure modes: determine the failure modes that may cause problems in product production quality;

[0042] 32) Failure impact assessment: assess the impact of failure modes of product production quality problems on system or product performance;

[0043] 33) Determine the cause of failure: Analyze the potential causes that lead to the failure mode;

[0044] 34) Calculation of risk priority number: Calculate the risk priority number based on the severity, frequency and detection difficulty of the failure mode:

[0045] RPN = severity × frequency × difficulty of detection

[0046] 35) Develop improvement measures: When the risk priority number is greater than the set threshold, develop and implement improvement measures;

[0047] 36) Verify the effectiveness of improvement measures and regularly update FMEA documents to reflect the latest situation.

[0048] The beneficial effects of the present invention are:

[0049] The present invention provides a new energy vehicle product life cycle problem management method based on knowledge graph, which has the following technical effects:

[0050] (1) The method of the present invention can not only effectively solve many problems in the traditional engineering production management system PMS and quality management system QMS, but also improve the quality management level of the entire product life cycle through intelligent and data-based management. In high-tech industries such as new energy vehicles, the application of product life cycle problem management methods based on knowledge graphs will bring significant efficiency improvements and competitive advantages, and can solve the deficiencies in the discovery and solution of problems in the existing technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to make the purpose, technical solution and beneficial effects of the present invention clearer, the present invention provides the following drawings for illustration:

[0052] Figure 1 It is a flow chart of the problem management method of the whole life cycle of new energy vehicle products based on the knowledge graph of the present invention;

[0053] Figure 2 This is a framework diagram of the new energy vehicle product life cycle problem management method based on the knowledge graph in this embodiment;

[0054] Figure 3 A flowchart of the method for constructing a knowledge graph for managing product life cycle issues of new energy vehicles;

[0055] Figure 4 Part of the knowledge graph for managing new energy vehicle product life cycle issues. DETAILED DESCRIPTION

[0056] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0057] like Figure 1 As shown, the method for managing problems in the entire life cycle of new energy vehicle products based on the knowledge graph in this embodiment includes the following steps.

[0058] Step 1: Enter the question and search the historical question database to see if the question is a repeated question. If so, proceed to step 2. If not, the question is a new question and proceed to step 3.

[0059] Specifically, repetitive problems will appear in each stage of the new energy vehicle product life cycle, and new problems that have not been included in the historical database will also appear. These problems include: R&D problems, supplier problems, manufacturing problems and customer complaints. Use the historical problem database, product production information, product design information and material information to judge whether the problems are repetitive problems.

[0060] Step 2: Classify the problems based on professional knowledge and experience to determine the type and nature of the problems.

[0061] For repeated problems, engineers rely on their rich experience and expertise to judge the type of problem and classify it into categories, such as material problems in the automobile manufacturing process, automobile quality problems, and reliability problems.

[0062] Step 3: For repeated problems, according to the type of problem, find solutions to the problem in the historical problem library and propose solutions. Specifically, when the problem type is a production quality problem, dynamically adjust IQC sampling, failure mode and effect analysis methods and control plan updates to find solutions to the problem in the historical problem library and propose solutions. For new problems, after entering the new problem into the new energy vehicle product life cycle problem management knowledge map built in the historical problem library, judge and reason about the new problem, and propose a solution to the new problem.

[0063] Specifically, IQC sampling is dynamically adjusted based on product production quality data and historical problem records, and the sampling plan for product inspection is dynamically adjusted to improve the efficiency and effectiveness of inspection. This includes: data collection and analysis, risk assessment, sampling plan adjustment, implementation, and monitoring.

[0064] Failure Mode and Effects Analysis (FMEA) is a systematic method used to identify potential failure modes and evaluate their impact on the system or product so that measures can be taken to prevent or reduce failures. Specifically, the failure mode and effects analysis method includes the following steps:

[0065] 31) Identify failure modes: determine the failure modes that may cause problems in product production quality;

[0066] 32) Failure impact assessment: assess the impact of failure modes of product production quality problems on system or product performance;

[0067] 33) Determine the cause of failure: Analyze the potential causes that lead to the failure mode;

[0068] 34) Risk priority number calculation: Calculate the risk priority number based on the severity, frequency and detection difficulty of the failure mode:

[0069] RPN = severity × frequency × difficulty of detection

[0070] 35) Develop improvement measures: When the risk priority number is greater than the set threshold, develop and implement improvement measures;

[0071] 36) Verify the effectiveness of improvement measures and regularly update FMEA documents to reflect the latest situation.

[0072] The control plan is a document that describes in detail the control measures and inspection methods in the product production process to ensure that the product quality meets the requirements. It includes: identifying change requirements, revising control measures, updating inspection methods, implementation and monitoring.

[0073] Step 4: Evaluate the feasibility and effectiveness of the proposed solution: If the solution passes the evaluation, proceed to step 5; if the solution fails the evaluation, proceed to step 3.

[0074] Specifically, the feasibility and effectiveness of the proposed solution can be evaluated using the cost-effectiveness method. By collecting and analyzing solution-related data, the solution's return on investment, cost-effectiveness ratio and other indicators can be calculated. If the problem-solving solution passes the evaluation indicators, step five is executed to actually verify the solution; if the problem-solving solution fails the evaluation, step three is executed to re-propose the problem-solving solution.

[0075] Step 5: Perform actual verification of the problem solution: If the problem solution is verified, proceed to step 6; if the problem solution is not verified, proceed to step 3.

[0076] Step 6: Upload the problem data including problems, problem solutions and implementation experience to the data center to update the historical problem library, and then update the new energy vehicle product life cycle problem management knowledge graph.

[0077] Step 7: Problem solving is complete.

[0078] Specifically, a knowledge graph of new energy vehicle product life cycle problem management is constructed in the historical problem database based on the problem data. The construction method of the knowledge graph of new energy vehicle product life cycle problem management is as follows:

[0079] S1: Obtain relevant data information from the entire life cycle of new energy vehicles, including: product design information in the R&D stage, product production information in the manufacturing stage, and user feedback information in the service stage, to build a problem library and case library for the entire life cycle of new energy vehicles.

[0080] S2: Based on the product historical maintenance database, combined with the new energy vehicle full life cycle problem database and case library, a new energy vehicle problem management database is formed.

[0081] S3: For the new energy vehicle problem management database, a multimodal knowledge mining algorithm is used to present problem data of different modalities in text form to form triple clusters of problem data of each modality.

[0082] In this embodiment, the method steps of converting question data of different modalities into triples using a multimodal knowledge mining algorithm are as follows.

[0083] S31: Explore the natural language description methods of multimodal industrial data and establish an industrial knowledge mining framework for consistent generation of modal features and natural language descriptions.

[0084] S32: An industrial knowledge mining framework generated by modal feature-natural language description consistency is used to convert non-text modal data into text or graph form, and multimodal knowledge mining is achieved through text extraction methods, including:

[0085] For text modality: First, the PL-Marker strategy is used to encode text sentences or paragraphs, and then Albert is used for embedding representation. Finally, a hypergraph is used to model the embedded representation, and a hypergraph neural network is constructed to extract text modality knowledge and form a text modality triple group.

[0086] For image modality: select several multi-image description models, perform image description tasks on the same image, obtain multiple image description texts, embed and represent the description texts and images, perform feature recognition, and sort the description text quality. Input the top-k description texts into a combined prompt and input it into the large language model to generate a fused image description. Use the text modality knowledge extraction method to extract image modality industrial knowledge and form an image modality triple group.

[0087] For the graphic modality: first, all 3D models are converted into a standard format, and then the PCL library is introduced to process the 3D models into point clouds to obtain the 3D point clouds of the models; then the 3D object detector is used to obtain the relevant features of the point cloud, and the point cloud description text is generated based on multi-head cross attention; finally, the text modality knowledge extraction method is used to realize the graphic modality industrial knowledge extraction and form a graphic modality triple group;

[0088] For signal modality: first, the signal data is stored in a standard database, then the signal is processed and the typical time-frequency domain features of the signal are extracted; under the guidance of prior knowledge such as sensors, signal-text association rules are constructed; finally, the text modality knowledge extraction method is used to extract industrial knowledge of signal modality and form a signal modality triple group;

[0089] Regarding expression modality: according to the different forms of expressions, it is divided into unstructured expression data and semi-structured expression data; for semi-structured expression data, the discretized information is directly extracted and a graph structure is formed; for unstructured expression data, the expression in the image is first recognized through OCR, and then it is discretized to realize the construction of the scene graph of discrete elements, and finally form an expression modality triple group.

[0090] S4: Globally encode the triplet clusters of each modal problem data, and perform modal fusion and cross-modal alignment to achieve cross-modal problem data fusion and form a knowledge graph for new energy vehicle product life cycle problem management.

[0091] In this embodiment, the method steps for forming a new energy vehicle product life cycle problem management knowledge graph are as follows:

[0092] According to the correlation between the modal problem data triplet clusters, the modal original data features are introduced, and the modal triplet group graph features of each modal problem data triplet cluster are combined for global encoding. The weights of each part of the features are assigned through dynamic cross-modal weighting, and then modal fusion and cross-modal alignment operations are performed to achieve cross-modal knowledge fusion and form a knowledge graph for new energy vehicle product life cycle problem management.

[0093] The new energy vehicle product life cycle problem management knowledge graph uses the URI number as the core node and is loaded with text attribute descriptions, modal attribute descriptions, standardized data attributes and feature attributes.

[0094] The specific implementation of this embodiment is described below with reference to specific examples.

[0095] Case example: Implementation of a new energy vehicle manufacturer

[0096] When receiving input of a research and development problem, such as an overheating problem with a certain model of engine;

[0097] Based on the historical question database, product production information, product design information, material information and other information, determine whether the problem is a repeated problem.

[0098] If it is judged to be a repetitive problem, record the problem details for subsequent analysis;

[0099] For repetitive problems, experienced engineers analyzed and determined that the problem was a design defect in the engine cooling system and classified it as a design problem. For new problems not included in the historical database, the new problems were input into the knowledge graph of new energy vehicle product life cycle problem management, and the new problems were judged and reasoned using knowledge reasoning and other methods to propose solutions to solve the new problems.

[0100] For this design problem, engineers used methods such as semantic parsing, graph embedding, and graph reasoning to search for the problem in the existing knowledge graph and found previous solutions to similar problems, including optimizing the heat sink design and adjusting the coolant flow.

[0101] Semantic parsing: Perform deep semantic parsing on the problem description to extract key elements (such as engine location, supplier name, problem type, etc.).

[0102] Graph embedding: embed the parsed elements into the pre-built knowledge graph to perform similarity calculation and matching.

[0103] Graph reasoning: Use historical experience and associated information in the graph to infer the most relevant solutions and preventive measures for the current problem.

[0104] Based on the graph, a solution to the problem was obtained. It was concluded that the cause of the problem was unreasonable design of the heat sink. The solution proposed was to optimize the shape of the heat sink and increase the coolant flow rate. At the same time, preventive measures were formulated, such as performing thermal simulation analysis in future designs.

[0105] Evaluate the solutions for optimizing heat sink design and increasing coolant flow to determine whether the solutions are feasible and effective. The evaluation generally adopts the expert review method, organizing experts in related fields or experienced engineers to discuss and review the proposed solutions. The expert review method relies on the knowledge and experience of experts to evaluate the solutions to the problem from multiple angles.

[0106] The optimized cooling system is tested on a real vehicle to verify the effectiveness of the solution. The test results and solution data are stored in the data center.

[0107] Update relevant information based on the adjusted measures, including design standards, production information, and material properties. For example, update the design standards for heat sinks and production information for cooling systems.

[0108] The updated relevant information and the categorized questions are used as input to update the knowledge graph. The updated knowledge graph includes new heat sink design standards and coolant flow adjustment solutions, which will serve as guidance to support future graph retrieval and problem solving.

[0109] The above-described embodiments are only preferred embodiments for fully illustrating the present invention, and the protection scope of the present invention is not limited thereto. Equivalent substitutions or changes made by those skilled in the art based on the present invention are within the protection scope of the present invention. The protection scope of the present invention shall be subject to the claims.

Claims

1. A new energy vehicle product life cycle problem management method based on knowledge graph, characterized by: The steps include: Step 1: Enter the question and search the historical question database to see if the question is a repeated question. If so, proceed to step 2. If not, the question is a new question and proceed to step 3. Step 2: Classify the problems based on professional knowledge and experience to determine the type and nature of the problems; Step 3: For repeated problems, according to the type of problem, find solutions to the problem in the historical problem database and propose solutions to the problem; for new problems, input the new problem into the new energy vehicle product life cycle problem management knowledge map built in the historical problem database, make judgments and inferences on the new problem, and propose solutions to solve the new problem; Step 4: Evaluate the feasibility and effectiveness of the proposed solution: If the solution passes the evaluation, proceed to step 5; if the solution fails the evaluation, proceed to step 3; Step 5: Perform actual verification of the problem solution: If the problem solution is verified, proceed to step 6; if the problem solution is not verified, proceed to step 3; Step 6: Upload the problem data including problems, problem solutions and implementation experience to the data center to update the historical problem database; Step 7: Problem solving is complete.

2. The method for managing problems in the entire life cycle of new energy vehicle products based on knowledge graph according to claim 1 is characterized in that: The historical question library has a new energy vehicle product life cycle problem management knowledge graph based on the problem data. The method for constructing the new energy vehicle product life cycle problem management knowledge graph is as follows: S1: Obtain relevant data information from the entire life cycle of new energy vehicles, including: product design information in the R&D stage, product production information in the manufacturing stage, and user feedback information in the service stage, to build a problem library and case library for the entire life cycle of new energy vehicles; S2: Based on the product history maintenance database, combined with the new energy vehicle full life cycle problem database and case database, a new energy vehicle problem management database is formed; S3: For the new energy vehicle problem management database, a multimodal knowledge mining algorithm is used to present problem data of different modes in text form to form triple clusters of problem data of each mode; S4: Globally encode the triplet clusters of each modal problem data, and perform modal fusion and cross-modal alignment to achieve cross-modal problem data fusion and form a knowledge graph for new energy vehicle product life cycle problem management.

3. The method for managing problems in the entire life cycle of new energy vehicle products based on knowledge graph according to claim 2 is characterized in that: In step S3, the method steps of converting question data of different modes into triples using a multimodal knowledge mining algorithm are as follows: S31: Explore the natural language description method of multimodal industrial data and establish an industrial knowledge mining framework for modal feature-natural language description consistency generation; S32: An industrial knowledge mining framework generated by modal feature-natural language description consistency is used to convert non-text modal data into text or graph form, and multimodal knowledge mining is achieved through text extraction methods, including: For text modality: First, the PL-Marker strategy is used to encode text sentences or paragraphs, and then Albert is used for embedding representation. Finally, a hypergraph is used to model the embedded representation, and a hypergraph neural network is constructed to extract text modality knowledge and form a text modality triple group. For image modality: select several multi-image description models, perform image description tasks on the same image, obtain multiple image description texts, embed and represent the description texts and images, perform feature recognition, and sort the description text quality. Input the top-k description texts into a combined prompt and input it into the large language model to generate a fused image description. Use the text modality knowledge extraction method to extract image modality industrial knowledge and form an image modality triple group. For the graphic modality: first, all 3D models are converted into a standard format, and then the PCL library is introduced to process the 3D models into point clouds to obtain the 3D point clouds of the models; then the 3D object detector is used to obtain the relevant features of the point cloud, and the point cloud description text is generated based on multi-head cross attention; finally, the text modality knowledge extraction method is used to realize the graphic modality industrial knowledge extraction and form a graphic modality triple group; For signal modality: first, the signal data is stored in a standard database, then the signal is processed and the typical time-frequency domain features of the signal are extracted; under the guidance of prior knowledge such as sensors, signal-text association rules are constructed; finally, the text modality knowledge extraction method is used to extract industrial knowledge of signal modality and form a signal modality triple group; Regarding expression modality: according to the different forms of expressions, it is divided into unstructured expression data and semi-structured expression data; for semi-structured expression data, the discretized information is directly extracted and a graph structure is formed; for unstructured expression data, the expression in the image is first recognized through OCR, and then it is discretized to realize the construction of the scene graph of discrete elements, and finally form an expression modality triple group.

4. The method for managing problems in the entire life cycle of new energy vehicle products based on knowledge graph according to claim 2 is characterized in that: In step S4, the method steps for forming a knowledge graph of new energy vehicle product life cycle problem management are as follows: According to the correlation relationship of modal problem data triple clusters, the modal original data features are introduced, and the modal triple group graph features of each modal problem data triple cluster are combined for global encoding. The weights of each part of the features are assigned through dynamic cross-modal weighting, and then modal fusion and cross-modal alignment operations are performed to achieve cross-modal knowledge fusion and form a knowledge graph for new energy vehicle product life cycle problem management. The new energy vehicle product life cycle problem management knowledge graph uses the URI number as the core node and is loaded with text attribute descriptions, modal attribute descriptions, standardized data attributes and feature attributes.

5. The method for managing problems in the entire life cycle of new energy vehicle products based on knowledge graph according to claim 2 is characterized in that: In the step three, when the problem type is a production quality problem, dynamically adjust the IQC sampling, failure mode and effect analysis method and control plan update, find measures to solve the problem in the historical problem library and propose a solution to the problem.

6. The method for managing problems in the entire life cycle of new energy vehicle products based on knowledge graph according to claim 5 is characterized by: The FMEA method includes the following steps: 31) Identify failure modes: determine the failure modes that may cause problems in product production quality; 32) Failure impact assessment: assess the impact of failure modes of product production quality problems on system or product performance; 33) Determine the cause of failure: Analyze the potential causes that lead to the failure mode; 34) Risk priority number calculation: Calculate the risk priority number based on the severity, frequency and detection difficulty of the failure mode: RPN = severity × frequency × difficulty of detection 35) Develop improvement measures: When the risk priority number is greater than the set threshold, develop and implement improvement measures; 36) Verify the effectiveness of improvement measures and regularly update FMEA documents to reflect the latest situation.

Citation Information

Patent Citations

  • A dialogue service method and device for a vehicle

    CN109543019A

  • Construction method and system for automobile field knowledge graph

    CN110245239A

  • Problem library analysis method and device

    CN112115234A

  • Knowledge graph completion method and device based on semantic alignment and symmetric structure coding

    CN114117075A

  • Hybrid electric vehicle transmission system knowledge graph construction method, reasoning method and rapid design system

    CN116821355A