Vehicle fault code to system component mapping method

Abstract

This invention proposes a method for mapping vehicle fault codes to system components, relating to the field of data mapping technology. It involves acquiring historical vehicle diagnostic records, analyzing the proportion, continuity, and range of garbled data to obtain anomaly judgment data, processing it, and then obtaining diagnostic anomaly processing data. This data is packaged to obtain processed data bar information. Fault code similarity analysis and combination are performed on multiple processed data bar information to obtain batch analysis data. A fault classification model is used to classify the batch analysis data for fault codes and component entity information, obtaining batch processing results. Entity normalization and granularity unification processing are performed on multiple component entity information to obtain standardized component classification information. A mapping table is established, associated numbers are generated, and the mapping table is updated to obtain an updated mapping table. This invention achieves accurate and intelligent mapping between fault codes and components.

Description

Technical Field

[0001] This invention proposes a method for mapping vehicle fault codes to system components, which relates to the field of data mapping technology, specifically the field of mapping vehicle fault codes to system components. Background Technology

[0002] In existing technologies, one type of method relies on personal experience to construct a rule-based classification system for fault code structures. This type of method not only requires the accumulation of sufficient data patterns but is also limited by the scope of personal experience. Its rules are often limited to data features specific to a particular brand and vehicle model, resulting in significant limitations and a high risk of misjudgment. When dealing with unfamiliar vehicle models, the accuracy further declines. Another type of method uses keywords, word vectors, or word embeddings combined with machine learning for clustering. However, due to the lack of industry-customized dictionaries and keywords, the clustering results are often insufficiently professional. While manually constructing a professional dictionary can improve professionalism, it requires a significant amount of manpower and time, resulting in low efficiency. Summary of the Invention

[0003] This invention provides a method for mapping vehicle fault codes to system components to solve the above-mentioned problems:

[0004] The present invention proposes a method for mapping vehicle fault codes to system components, the method comprising:

[0005] S1. Obtain historical vehicle diagnostic record information, perform analysis on the proportion, continuity and range of garbled data, obtain anomaly judgment data and process it, thereby obtaining diagnostic anomaly processing data, packaging it, and obtaining processing data bar information.

[0006] S2. Perform fault code similarity analysis and combination on multiple processed data bar information to obtain batch analysis data. Use the fault classification model to classify fault codes and component entity information in the batch analysis data to obtain batch processing results.

[0007] S3. Perform entity normalization and granularity unification processing on the entity information of multiple parts to obtain standardized part classification information;

[0008] S4. Establish a mapping table based on fault code classification data, system category and component granularity classification data, generate associated numbers, update the mapping table, and obtain an updated mapping table.

[0009] Further, S1 includes:

[0010] Historical vehicle diagnostic records are obtained through a big data platform, and the historical vehicle diagnostic records are preliminarily processed to obtain preliminary diagnostic processing data.

[0011] The preliminary diagnostic data is subjected to garbled character analysis to obtain anomaly judgment data. The anomaly judgment data is then removed to obtain diagnostic anomaly processing data.

[0012] The diagnostic anomaly processing data is packaged one by one to obtain the processing data item information.

[0013] Further, the step of performing garbled character analysis on the preliminary diagnostic data to obtain abnormal data, removing the abnormal data, and obtaining diagnostic anomaly processing data includes:

[0014] Perform ASCII character acceptability determination on the preliminary diagnostic data to obtain the character acceptability determination results;

[0015] Based on the character acceptance judgment result, determine the garbled character data of the preliminary diagnostic processing data;

[0016] The garbled character data is subjected to proportion analysis and continuous analysis, and then range analysis to obtain the garbled character range comparison results. Then, the garbled character data is judged as abnormal data to obtain abnormal judgment data.

[0017] The preliminary diagnostic data is processed to remove abnormal data, resulting in abnormal diagnostic data.

[0018] Furthermore, the step of performing proportional and continuous analysis on the garbled character data, followed by range analysis to obtain a comparison result of the garbled character range, and then determining abnormal data from the garbled character data to obtain abnormal determination data, includes:

[0019] Obtain the percentage of garbled character data in the initial diagnostic processing data, and obtain the garbled character quantity percentage coefficient.

[0020] Obtain the proportion of continuous length of garbled character data in the initial diagnostic processing data, and obtain the garbled character length proportion coefficient.

[0021] The ratio of the garbled text quantity percentage coefficient to the garbled text length percentage coefficient is used to obtain the garbled text range percentage coefficient.

[0022] The garbled character range proportion coefficient is compared with the preset garbled character range proportion threshold to obtain the garbled character range comparison result.

[0023] Based on the comparison results of the garbled character range, abnormal data is determined for the garbled character data, and abnormal determination data is obtained.

[0024] Further, S2 includes:

[0025] Perform fault code similarity analysis on multiple processed data bars to obtain similar fault code processing data;

[0026] Based on the similarity processing data of fault codes, similar combinations of diagnostic anomaly processing data are obtained to obtain similar combination data of diagnostic processing.

[0027] Batch analysis data were obtained based on similar combinations of diagnostic treatments.

[0028] By setting keywords and system categories in the big data model, a fault classification model can be obtained;

[0029] The batch analysis data is input into the fault classification model. The fault classification model is used to classify the batch analysis data into fault codes and extract the component entity information associated with the faults to obtain fault code classification data.

[0030] Multiple batches of data can be processed asynchronously to obtain batch processing results.

[0031] Further, S3 includes:

[0032] Obtain information on multiple component entities and input the information on these multiple component entities into the fault classification model;

[0033] The fault classification model is used to perform entity normalization and granularity unification processing on the entity information of multiple components to obtain entity normalization processing information and granularity unification processing information.

[0034] Standardized component classification information is obtained based on entity normalization processing information and granularity unified processing information.

[0035] Furthermore, the step of performing entity normalization and granularity unification processing on the entity information of multiple components through the fault classification model to obtain entity normalization processing information and granularity unification processing information includes:

[0036] The fault classification model is used to extract entity feature information from the entity information of multiple components to obtain entity feature extraction information.

[0037] The entity feature extraction information is mapped to the feature standard information to obtain entity feature correspondence information;

[0038] The entity feature correspondence information is used to perform entity normalization annotation on the entity information of multiple parts to obtain entity normalization annotation information of multiple parts.

[0039] The granularity analysis of the normalized annotation information of multiple component entities is performed by the fault classification model to obtain component entity granularity analysis data.

[0040] Based on the granularity analysis data of the component entities, granularity analysis is performed on the normalized annotation information of multiple component entities to obtain component granularity classification data.

[0041] Further, the step of performing granularity analysis on the normalized annotation information of multiple component entities based on the component entity granularity analysis data to obtain component granularity classification data includes:

[0042] Obtain the normalized annotation information of each component entity and its corresponding entity feature extraction information;

[0043] Obtain multiple preset element difference data of the normalized annotation information and entity feature extraction information of the component entities;

[0044] Obtain normalized data of multiple preset element difference data to obtain element difference normalized data;

[0045] Based on the element normalization data, obtain the difference distance coefficient between the normalized annotation information of the component entity and the entity feature extraction information;

[0046] Based on the difference distance coefficient, the granularity of the normalized annotation information of the component entities is divided to obtain feature normalized granularity information;

[0047] Based on the feature normalization granularity information, the normalization annotation information of each component entity is classified to obtain component granularity classification information.

[0048] Further, S4 includes:

[0049] Obtain the system category based on the fault code classification data;

[0050] Obtain component granularity classification data based on system category;

[0051] Establish a mapping table for fault code classification data, system category, and component granularity classification data;

[0052] Generate corresponding associated numbers for fault code classification data based on component granularity classification data;

[0053] The mapping table is updated according to the corresponding association number to obtain the updated mapping table.

[0054] Furthermore, the system includes:

[0055] The data anomaly processing module is used to obtain historical vehicle diagnostic record information, perform analysis on the proportion, continuity and range of garbled data, obtain anomaly judgment data and process it, thereby obtaining diagnostic anomaly processing data, packaging it and obtaining processing data bar information.

[0056] The model classification module is used to perform fault code similarity analysis and combination on multiple processed data bars to obtain batch analysis data. The fault classification model is used to classify fault codes and component entity information in the batch analysis data to obtain batch processing results.

[0057] The component classification module is used to perform entity normalization and granularity unification processing on the entity information of multiple components to obtain standardized component classification information.

[0058] The mapping analysis module is used to establish a mapping table based on fault code classification data, system category and component granularity classification data, generate associated numbers, update the mapping table, and obtain an updated mapping table.

[0059] Beneficial effects of this invention:

[0060] This method achieves end-to-end transformation from unstructured raw diagnostic data to structured fault codes, system, and component mapping relationships. It eliminates the need for extensive manual annotation, significantly reducing labor costs and solving the problems of time-consuming and inefficient manual annotation in traditional methods. Breaking brand and vehicle model limitations, through fault classification models and standardized component processing, the mapping results are adaptable to different brands and vehicle models, providing broader coverage and addressing the pain point of poor brand and vehicle compatibility in traditional methods. A standardized mapping table is established to provide unified data support for scenarios such as fault diagnosis, repair dispatch, and parts inventory management. For example, repair personnel can quickly locate associated parts using fault codes, improving repair efficiency; the supply chain can accurately prepare inventory based on mapping relationships, reducing inventory redundancy.

[0061] The entire process is driven by data and assisted by models, avoiding the bias of subjective human judgment, making the association between fault codes and components more professional and consistent, and improving the reliability of data use in application scenarios.

[0062] By constructing a standardized mapping relationship between fault codes, systems, and components, the format of fault codes, descriptions, and fault systems is unified, providing a high-quality data foundation for computational analysis. This mapping relationship exists in the form of a structured form, with a response speed far superior to real-time large-scale model computation. It can be applied with only a query operation, avoiding the high resource consumption of large-scale model deployment. The standardized mapping supports multilingual scenarios, eliminating the need for additional translation functions for different regional languages, and has good international applicability. Relying on large-scale model inference and continuous fine-tuning optimization, it effectively alleviates the problem of misjudgment caused by incomplete training data. It covers a wide range of fields and a large data scale, and a small amount of noisy data can be corrected through error correction rules to ensure the accuracy of the output results. Attached Figure Description

[0063] Figure 1 This is a schematic diagram illustrating the method of mapping vehicle fault codes to system components. Detailed Implementation

[0064] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0065] In one embodiment of the present invention, a method for mapping vehicle fault codes to system components is proposed, the method comprising:

[0066] S1. Obtain historical vehicle diagnostic record information, perform analysis on the proportion, continuity and range of garbled data, obtain anomaly judgment data and process it, thereby obtaining diagnostic anomaly processing data, packaging it, and obtaining processing data bar information.

[0067] S2. Perform fault code similarity analysis and combination on multiple processed data bar information to obtain batch analysis data. Use the fault classification model to classify fault codes and component entity information in the batch analysis data to obtain batch processing results.

[0068] S3. Perform entity normalization and granularity unification processing on the entity information of multiple parts to obtain standardized part classification information;

[0069] S4. Establish a mapping table based on fault code classification data, system category, and component granularity classification data; generate associated numbers; update the mapping table to obtain an updated mapping table, such as... Figure 1 As shown.

[0070] The working principle and technical effects of the above technical solution are as follows: data is extracted from historical vehicle diagnostic records, abnormal data is removed through garbled code analysis and packaged into standardized data bars; fault code similarity analysis and combination are performed on the data bars to form batch data, which is then input into the fault classification model to complete fault code classification and component entity extraction; component entities are normalized and granularity is unified to obtain standardized classification information; a mapping table is established based on fault code classification, system category and component granularity classification, associated numbers are generated and the mapping table is updated, forming a closed loop of the entire process of data cleaning, classification extraction, standardization and mapping establishment.

[0071] This method achieves end-to-end transformation from unstructured raw diagnostic data to structured fault codes, system, and component mapping relationships. It eliminates the need for extensive manual annotation, significantly reducing labor costs and solving the problems of time-consuming and inefficient manual annotation in traditional methods. Breaking brand and vehicle model limitations, through fault classification models and standardized component processing, the mapping results are adaptable to different brands and vehicle models, providing broader coverage and addressing the pain point of poor brand and vehicle compatibility in traditional methods. A standardized mapping table is established to provide unified data support for scenarios such as fault diagnosis, repair dispatch, and parts inventory management. For example, repair personnel can quickly locate associated parts using fault codes, improving repair efficiency; the supply chain can accurately prepare inventory based on mapping relationships, reducing inventory redundancy.

[0072] The entire process is driven by data and assisted by models, avoiding the bias of subjective human judgment, making the association between fault codes and components more professional and consistent, and improving the reliability of data use in application scenarios.

[0073] In one embodiment of the present invention, S1 includes:

[0074] Historical vehicle diagnostic records are obtained through a big data platform, and these records are preliminarily processed to obtain preliminary diagnostic data. The preliminary processing includes duplicate data removal and missing value filling.

[0075] The preliminary diagnostic data is subjected to garbled character analysis to obtain anomaly judgment data. The anomaly judgment data is then removed to obtain diagnostic anomaly processing data.

[0076] The diagnostic anomaly processing data is packaged one by one to obtain processing data bar information. A processing data bar information can be 2024-xx-xx, brand: XX, vehicle model: XXX, fault description: brake noise, fault code P0504, etc.

[0077] The working principle and technical effect of the above technical solution are as follows: Historical vehicle diagnostic records are obtained through a big data platform. First, preliminary processing such as duplicate data removal (deleting redundant records) and missing value supplementation (such as using common data of the same brand and model to fill in key missing fields) is performed to obtain preliminary diagnostic processing data. Then, the data is analyzed for garbled characters to identify and remove abnormal data, resulting in abnormal diagnostic processing data. Finally, the abnormal diagnostic processing data is packaged one by one according to a fixed format of time, brand, model, fault description, and fault code to generate processing data bar information.

[0078] Data duplication removal and missing value supplementation reduce data redundancy, complete key information, avoid wasting computing resources due to data duplication, or causing analytical bias due to data missing, and ensure the integrity and effectiveness of the initial diagnostic data processing.

[0079] Garbled data analysis and anomaly removal eliminate invalid and dirty data, ensuring the purity of diagnostic anomaly processing data and preventing garbled or invalid information from interfering with subsequent fault code classification and component extraction, thus providing high-quality data input for subsequent steps. Standardized packaging creates a unified format for processed data, eliminating format differences between data sources (such as different output formats from different diagnostic devices), making similarity analysis, batch processing, and other operations easier to perform, and reducing the complexity of data preprocessing.

[0080] In one embodiment of the present invention, the step of performing garbled character analysis on the preliminary diagnostic processing data to obtain abnormal data, and then removing the abnormal data to obtain diagnostic anomaly processing data includes:

[0081] Perform ASCII character acceptability determination on the preliminary diagnostic data to obtain the character acceptability determination results;

[0082] Based on the character acceptance judgment result, determine the garbled character data of the preliminary diagnostic processing data;

[0083] The garbled character data is subjected to proportion analysis and continuous analysis, and then range analysis to obtain the garbled character range comparison results. Then, the garbled character data is judged as abnormal data to obtain abnormal judgment data.

[0084] The preliminary diagnostic data is processed to remove abnormal data, resulting in abnormal diagnostic data.

[0085] The working principle and technical effect of the above technical solution are as follows: For each character in the preliminary diagnostic data, ASCII acceptability is determined to distinguish between valid ASCII characters and non-ASCII garbled characters; after filtering out garbled character data based on the determination results, the data is analyzed from two dimensions: garbled character proportion (the proportion of garbled characters to the total number of characters) and continuous garbled character length (the number of consecutively appearing garbled characters), and then combined with range analysis (the overall influence range of the combined proportion and continuous features) to calculate the garbled character range proportion coefficient; this coefficient is compared with a preset threshold to determine whether the data is abnormal; finally, abnormal data is removed to obtain diagnostic abnormal processing data.

[0086] Multi-dimensional (proportion, continuity, range) analysis avoids the limitations of single-dimensional judgment. For example, looking only at the proportion may miss short, continuous strings of garbled characters, and looking only at continuity may miss scattered high-proportion garbled characters. Combining multiple dimensions can more accurately identify abnormal data. It accurately retains data containing a small number of valid non-ASCII characters (such as ×- in vehicle models) to avoid accidentally deleting valid information, while thoroughly removing severely garbled data (such as large areas of unrecognizable characters), balancing data purity and valid data retention rate. A standardized garbled character judgment process is established, allowing diagnostic data from different batches and sources to be processed according to a unified standard, avoiding subjective bias from manual judgment and improving the consistency and reproducibility of garbled character processing results.

[0087] In one embodiment of the present invention, the step of performing proportion analysis and continuous analysis on the garbled character data, and then performing range analysis to obtain a comparison result of the garbled character range, and then determining abnormal data on the garbled character data to obtain abnormal determination data, includes:

[0088] Obtain the percentage of garbled character data in the initial diagnostic processing data, and obtain the garbled character quantity percentage coefficient.

[0089] Obtain the proportion of continuous length of garbled character data in the initial diagnostic processing data, and obtain the garbled character length proportion coefficient.

[0090] The ratio of the garbled text quantity percentage coefficient to the garbled text length percentage coefficient is used to obtain the garbled text range percentage coefficient.

[0091] The garbled character range proportion coefficient is compared with the preset garbled character range proportion threshold to obtain the garbled character range comparison result.

[0092] Based on the comparison results of the garbled character range, abnormal data is determined for the garbled character data, and abnormal determination data is obtained.

[0093] The working principle and technical effect of the above technical solution are as follows: The proportion of garbled characters in the initial diagnostic processing data is statistically analyzed to obtain the garbled character quantity proportion coefficient; the ratio of the longest consecutive garbled character length to the total data length is calculated to obtain the garbled character length proportion coefficient; the ratio of the two (e.g., quantity proportion coefficient ÷ length proportion coefficient) is converted into a quantifiable garbled character range proportion coefficient; this coefficient is compared with a preset garbled character range proportion threshold (e.g., 0.3). If the coefficient exceeds the threshold, it is determined to be abnormal data; otherwise, it is retained.

[0094] By fusing the quantitative and continuous characteristics of garbled data into a single range proportion coefficient, complex multi-dimensional judgments are transformed into simple threshold comparisons, reducing the complexity of the judgment logic and facilitating engineering implementation. Quantitative coefficient and threshold comparisons avoid biases from subjective experience-based judgments. For example, different personnel may have different definitions of a small number of garbled data, but the coefficient threshold can be standardized, making the results of abnormal data judgments more objective and verifiable. The preset thresholds can be adjusted to adapt to different scenarios (e.g., loosening the threshold for data from older diagnostic equipment and tightening the threshold for data from newer equipment), improving the flexibility and adaptability of garbled data judgment and avoiding misjudgments or omissions caused by a one-size-fits-all approach.

[0095] In one embodiment of the present invention, S2 includes:

[0096] Perform fault code similarity analysis on multiple processed data bars to obtain similar fault code processing data;

[0097] Based on the similarity processing data of fault codes, similar combinations of diagnostic anomaly processing data are obtained to obtain similar combination data of diagnostic processing; cluster analysis is then performed based on the similarity.

[0098] Batch analysis data is obtained based on similar combinations of diagnostic processing data; each batch analysis data includes processing data entries for multiple diagnostic anomaly processing data.

[0099] By setting keywords and system categories in the big data model, a fault classification model can be obtained;

[0100] The batch analysis data is input into the fault classification model. The fault classification model is used to classify the batch analysis data into fault codes and extract the component entity information associated with the faults to obtain fault code classification data.

[0101] Multiple batches of data can be processed asynchronously to obtain batch processing results.

[0102] The working principle and technical effect of the above technical solution are as follows: Similarity analysis is performed on fault codes in multiple processed data entries (e.g., based on fault code coding rules and the similarity of associated fault types) to filter out similar fault code processing data; similar data are combined and grouped using clustering algorithms (e.g., K-Means) to form diagnostic processing similar combination data; the combination data is divided into batch analysis data according to a fixed number (e.g., 100-1000 entries / batch); industry keywords (e.g., brake discs and spark plugs) and preset system categories (e.g., brake system and engine system) are set for the big data model (e.g., LLM) to construct a fault classification model; batch data is input into the model, and the model outputs the system category to which the fault code belongs and the associated component entity; batch asynchronous processing (multi-batch parallel request model) is adopted to avoid single-batch blocking and obtain batch processing results.

[0103] Similarity analysis and clustering group similar fault data into the same batch, ensuring the model processes similar data within the same batch, reducing the interference of data differences on model classification results, and improving the accuracy of fault code classification and component extraction. Batch asynchronous processing significantly improves the processing efficiency of massive amounts of data, avoiding the long waiting time of traditional serial processing, and shortening the full data processing cycle, especially suitable for processing millions or even tens of millions of diagnostic records. The model is set with industry keywords and system categories, making the model output more closely aligned with the professional needs of the automotive diagnostic field, avoiding the generalization problem of generalized model output, and ensuring the professionalism of fault code classification (such as accurately classifying P0504 as the braking system) and the accuracy of component extraction.

[0104] In one embodiment of the present invention, S3 includes:

[0105] Obtain information on multiple component entities and input the information on these multiple component entities into the fault classification model;

[0106] The fault classification model is used to perform entity normalization and granularity unification processing on the entity information of multiple components to obtain entity normalization processing information and granularity unification processing information.

[0107] Standardized component classification information is obtained based on entity normalization processing information and granularity unified processing information.

[0108] The working principle and technical effect of the above technical solution are as follows: Collect the entity information of multiple components (such as brake disc, brake disc, left front brake disc) output by the fault classification model and input it into the model; the model first extracts the core features of each entity (such as function, installation location, and compatible system) to obtain entity feature extraction information; then match the feature information with the preset component feature standard library (such as industry standard terminology features), and label entities with different names but consistent features as the same standard name (such as brake disc, left front brake disc is unified as brake disc), thus completing entity unification; perform granular analysis on the unified entities according to the hierarchy from system to component category to specific component, and unify the granularity (such as avoiding mixing brake disc and left front brake disc); integrate the unification results and the granularity unification results to generate standardized component classification information.

[0109] Entity unification addresses the issue of inconsistent component naming (e.g., different brands using different names for the same part), eliminating ambiguity and ensuring data compatibility across different sources, preventing subsequent mapping confusion caused by naming differences. Unified granularity establishes a standardized component hierarchy (e.g., from brake system to brake disc to left front brake disc), avoiding inconsistent granularity (e.g., the simultaneous appearance of "brake disc" and "brake disc screw"), making component classification more organized and providing a clear hierarchical basis for fault code mapping. Standardized component classification information can be directly integrated with downstream systems such as automotive repair and parts management. For example, repair systems can quickly retrieve parts by category, improving repair efficiency; parts inventory management can statistically analyze inventory by category, optimizing inventory structure.

[0110] In one embodiment of the present invention, the step of performing entity normalization processing and granularity unification processing on the entity information of multiple component parts through the fault classification model to obtain entity normalization processing information and granularity unification processing information includes:

[0111] The fault classification model is used to extract entity feature information from the entity information of multiple components to obtain entity feature extraction information.

[0112] The entity feature extraction information is mapped to the feature standard information to obtain entity feature correspondence information;

[0113] The entity feature correspondence information is used to perform entity normalization annotation on the entity information of multiple parts to obtain entity normalization annotation information of multiple parts.

[0114] The granularity analysis of the normalized annotation information of multiple component entities is performed by the fault classification model to obtain component entity granularity analysis data.

[0115] Based on the granularity analysis data of the component entities, granularity analysis is performed on the normalized annotation information of multiple component entities to obtain component granularity classification data.

[0116] The working principle and technical effect of the above technical solution are as follows: The fault classification model first extracts features from the input component entity information (such as brake discs), obtaining the functional features (such as those used for braking), structural features (such as disc shape), and adaptation system features (such as brake system) of each entity, thus obtaining entity feature extraction information; these features are compared with a preset feature standard information library (including feature descriptions of industry standard components) to find the standard component name that matches the features, and normalize the label for each entity (such as labeling a brake disc as a brake disk); then, the normalized entities are analyzed according to a preset granularity level from system level to category level to specific component level to specification level to determine the granularity level to which each entity belongs (such as brake discs belonging to the category level, and the left front brake disc belonging to the specific component level); finally, based on the level judgment result, the entities are uniformly adjusted to the target granularity level (such as unified to the specific component level), thus completing the granularity unification.

[0117] Feature extraction and standard matching ensure the accuracy of entity normalization, avoid errors caused by normalization based solely on name similarity (e.g., brake pumps and brake discs have similar names but different functions; feature matching can accurately distinguish them), and improve the reliability of normalization results.

[0118] Granularity analysis and unification are performed according to a preset hierarchy, ensuring consistency in the hierarchy of component entities and avoiding gaps caused by granularity mismatch during mapping (e.g., fault codes can only be located at the category level, while component classification is at the specification level), thus guaranteeing smooth mapping. The entire process is automated by the model, eliminating the need for manual annotation, significantly reducing the labor costs of component standardization, and improving processing efficiency. It is suitable for the standardization of large-scale component entities.

[0119] In one embodiment of the present invention, the step of performing granularity analysis on the normalized annotation information of multiple component entities based on the component entity granularity analysis data to obtain component granularity classification data includes:

[0120] Obtain the normalized annotation information of each component entity and its corresponding entity feature extraction information;

[0121] Obtain multiple preset element difference data of the normalized annotation information and entity feature extraction information of the component entities;

[0122] Obtain normalized data of multiple preset element difference data to obtain element difference normalized data;

[0123] The difference distance coefficient between the normalized annotation information of the component entity and the extracted entity feature information is obtained based on the element normalization data; the difference distance coefficient is the average value of the element difference normalization data of multiple preset element difference data;

[0124] Based on the difference distance coefficient, the granularity of the normalized annotation information of the component entities is divided to obtain feature normalized granularity information;

[0125] Based on the feature normalization granularity information, the normalization annotation information of each component entity is classified to obtain component granularity classification information.

[0126] The working principle and technical effect of the above technical solution are as follows: Obtain the normalized annotation information of each component entity (e.g., brake disc) and its corresponding entity feature extraction information (e.g., function: brake, location: wheel, adaptation system: brake system); preset multiple key elements (e.g., functional differences, location differences, adaptation system differences), calculate the difference data between the normalized annotation information and the feature extraction information on each element (e.g., if the functions are completely identical, the difference is 0; if partially identical, the difference is 0.5); normalize the difference data of each element (mapping the difference value to the 0-1 range) to obtain element difference normalized data; average the normalized data of all elements to obtain the difference distance coefficient (the closer the coefficient is to 0, the better the entity matches the standard feature); set the granularity division threshold according to the difference distance coefficient, divide the component entity normalized annotation information into granularity levels, and obtain feature normalized granularity information; classify the entities according to the granularity level to generate component granularity classification data.

[0127] Multi-element difference analysis and normalization transform abstract granular differences into quantifiable difference distance coefficients, avoiding biases caused by subjective judgments of granularity and making granularity classification more objective and reproducible. Granularity levels are defined based on coefficient thresholds, ensuring consistent differences between component entities and standard features at the same level, and clear distinctions between different levels, improving the consistency and accuracy of granularity classification. By adjusting preset elements and classification thresholds, the granularity requirements of different business scenarios can be adapted (e.g., repair dispatch requires specific component level, inventory statistics require category level), improving the scenario adaptability of granularity classification.

[0128] In one embodiment of the present invention, S4 includes:

[0129] Obtain the system category based on the fault code classification data;

[0130] Obtain component granularity classification data based on system category;

[0131] Establish a mapping table for fault code classification data, system category, and component granularity classification data;

[0132] Generate corresponding associated numbers for fault code classification data based on component granularity classification data;

[0133] The mapping table is updated according to the corresponding association number to obtain the updated mapping table.

[0134] The working principle and technical effect of the above technical solution are as follows: From the fault code classification data output by S2, extract the system category corresponding to each fault code (e.g., fault code P0504 corresponds to the braking system); based on the system category, associate the component granularity classification data under that system output by S3 (e.g., the braking system is associated with granularity classification data such as brake discs and brake pads); construct an initial mapping table with the fault code classification data as rows and the system category and component granularity classification data as columns; generate a unique association number for each fault code and component correspondence (e.g., P0504 corresponds to number 001 for brake disc) to identify the unique mapping relationship; when new fault code classification data or component granularity classification data is added, locate the corresponding position in the mapping table through the association number, update or supplement the mapping relationship, and obtain an updated mapping table.

[0135] Establish a structured mapping table for fault codes, systems, and components to make the relationships between the three clearly visible, avoiding the problems of scattered and difficult-to-query relationships in traditional methods, and improving the ease of use of the relationships.

[0136] The unique association number provides an identity for the mapping relationship, which facilitates operations such as querying, modifying and deleting the mapping relationship (e.g., the mapping relationship of P0504 and brake disc can be quickly located through the number 001), thus improving the management efficiency of the mapping table.

[0137] It supports dynamic updates of the mapping table. When new fault codes or new part types appear, the mapping relationship can be quickly supplemented through the associated number without rebuilding the entire mapping table, ensuring the timeliness and scalability of the mapping table and adapting to the dynamic changes in vehicle faults and parts.

[0138] According to one embodiment of the present invention, the system includes:

[0139] The data anomaly processing module is used to obtain historical vehicle diagnostic record information, perform analysis on the proportion, continuity and range of garbled data, obtain anomaly judgment data and process it, thereby obtaining diagnostic anomaly processing data, packaging it and obtaining processing data bar information.

[0140] The model classification module is used to perform fault code similarity analysis and combination on multiple processed data bars to obtain batch analysis data. The fault classification model is used to classify fault codes and component entity information in the batch analysis data to obtain batch processing results.

[0141] The component classification module is used to perform entity normalization and granularity unification processing on the entity information of multiple components to obtain standardized component classification information.

[0142] The mapping analysis module is used to establish a mapping table based on fault code classification data, system category and component granularity classification data, generate associated numbers, update the mapping table, and obtain an updated mapping table.

[0143] The working principle and technical effects of the above technical solution are as follows: The data anomaly processing module collects historical vehicle diagnostic records through a big data platform, performs duplicate data removal, missing value supplementation, garbled code analysis, and anomaly removal, and packages the processed data into processed data bar information; the model classification module receives the processed data bar information, performs fault code similarity analysis, combined clustering, and batch partitioning, constructs a fault classification model, and performs batch asynchronous processing on the batch data, outputting fault code classification data and component entity information; the component classification module receives component entity information, completes entity normalization and granularity unification through the model, and generates standardized component classification information; the mapping analysis module receives fault code classification data, system category, and standardized component classification information, establishes a mapping table and generates associated numbers, dynamically updates the mapping table, and outputs the updated mapping table.

[0144] Modular design breaks down the entire process into independent modules, each focusing on a single function (e.g., data anomaly handling only handles data cleaning). This reduces coupling between modules and facilitates subsequent optimization and upgrades of individual modules (e.g., optimizing the garbled character detection logic only requires modifying the data anomaly handling module), improving system maintainability. The modules work collaboratively to form a complete closed loop. Data flows from the data anomaly handling module to the mapping analysis module without manual intervention, achieving end-to-end automated processing, significantly reducing manual operation costs and improving overall processing efficiency. Data transfer between modules is based on standardized formats (e.g., processing data bar information, standardized component classification information), avoiding flow obstacles caused by data format incompatibility and ensuring system stability and smooth operation. Simultaneously, the output data of each module can be reused independently (e.g., the output of the component classification module can be used in the parts management system), improving system resource utilization.

[0145] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for mapping vehicle trouble codes to system components, the method comprising: The method comprises: S1, obtaining historical vehicle diagnosis record information, analyzing the proportion, continuity and range of random code data, obtaining abnormal judgment data and processing, and then obtaining diagnosis abnormal processing data, packaging the data, obtaining processing data information; S2, performing fault code similarity analysis and combination on the plurality of processing data information, obtaining batch analysis data, classifying the fault codes of the batch analysis data through a fault classification model and extracting fault-related component entity information, obtaining fault code classification data, performing batch asynchronous processing on the plurality of batch analysis data, and obtaining batch processing results; S3, performing entity normalization processing and granularity uniformity processing on the plurality of component entity information, and obtaining standardized component classification information; S4, establishing a mapping table according to the fault code classification data, system category and component granularity classification data, generating an association number, updating the mapping table, and obtaining an updated mapping table.

2. The method of claim 1, wherein, The S1 comprises: Obtain historical vehicle diagnosis record information through a big data platform, and preliminarily process the historical vehicle diagnosis record information to obtain diagnosis preliminary processing data; Perform random code analysis on the diagnosis preliminary processing data to obtain abnormal judgment data, and perform abnormal judgment data elimination to obtain diagnosis abnormal processing data; Packaging the diagnosis abnormal processing data piece by piece to obtain processing data information.

3. The method of claim 2, wherein, The random code analysis on the diagnosis preliminary processing data, the abnormal data elimination, and the diagnosis abnormal processing data comprise: Perform ASCII character acceptability judgment on the diagnosis preliminary processing data to obtain character acceptance judgment results; Determine the random code character data of the diagnosis preliminary processing data according to the character acceptance judgment results; Perform proportion analysis and continuity analysis on the random code character data, and then perform range analysis to obtain a random code range comparison result, and then perform abnormal data judgment on the random code character data to obtain abnormal judgment data; Perform abnormal judgment data elimination on the diagnosis preliminary processing data to obtain diagnosis abnormal processing data.

4. The method of claim 3, wherein, The random code analysis on the diagnosis preliminary processing data, the abnormal data elimination, and the diagnosis abnormal processing data comprise: Obtain the proportion of the number of random code character data in the diagnosis preliminary processing data to obtain a random code quantity proportion coefficient; Obtain the proportion of the length of random code character data in the diagnosis preliminary processing data to obtain a random code length proportion coefficient; Obtain the ratio of the random code quantity proportion coefficient to the random code length proportion coefficient to obtain a random code range proportion coefficient; Compare the random code range proportion coefficient with a preset random code range proportion threshold to obtain a random code range comparison result; Perform abnormal data judgment on the random code character data according to the random code range comparison result to obtain abnormal judgment data.

5. The method of claim 1, wherein, The S2 comprises: Perform fault code similarity analysis on the plurality of processing data information to obtain fault code similarity processing data; Perform similar combination of diagnosis abnormal processing data according to the fault code similarity processing data to obtain diagnosis processing similar combination data; Batch analysis data is obtained according to the diagnostic processing similar combination data; Keywords and system categories are set for the big data model to obtain a fault classification model; The batch analysis data is input into the fault classification model.

6. The method of claim 1, wherein, The S3 comprises: Obtaining a plurality of component entity information, and inputting the plurality of component entity information into the fault classification model; Performing entity normalization processing and granularity uniformity processing on the plurality of component entity information through the fault classification model to obtain entity normalization processing information and granularity uniformity processing information; Obtaining standardized component classification information according to the entity normalization processing information and the granularity uniformity processing information.

7. The method of claim 6, wherein, The entity normalization processing and the granularity uniformity processing on the plurality of component entity information through the fault classification model to obtain the entity normalization processing information and the granularity uniformity processing information comprises: Performing entity feature information extraction on the plurality of component entity information through the fault classification model to obtain entity feature extraction information; Corresponding the entity feature extraction information with feature standard information to obtain entity feature corresponding information; Performing entity normalization labeling on the plurality of component entity information through the entity feature corresponding information to obtain a plurality of component entity normalization labeling information; Performing granularity analysis on the plurality of component entity normalization labeling information through the fault classification model to obtain component entity granularity analysis data; Performing granularity analysis on the plurality of component entity normalization labeling information according to the component entity granularity analysis data to obtain component granularity classification data.

8. The method of claim 1, wherein, The granularity analysis on the plurality of component entity normalization labeling information according to the component entity granularity analysis data to obtain the component granularity classification data comprises: Obtaining each component entity normalization labeling information and corresponding entity feature extraction information thereof; Obtaining a plurality of preset element difference data of the component entity normalization labeling information and the entity feature extraction information; Obtaining normalization data of the plurality of preset element difference data to obtain element difference normalization data; Obtaining a difference distance coefficient of the component entity normalization labeling information and the entity feature extraction information according to the element normalization data; Performing granularity division on the component entity normalization labeling information according to the difference distance coefficient to obtain feature normalization granularity information; Classifying each component entity normalization labeling information according to the feature normalization granularity information to obtain component granularity classification information.

9. The method of claim 1, wherein, The S4 comprises: Obtaining a system category according to the fault code classification data; Obtaining component granularity classification data according to the system category; Establishing a mapping table for the fault code classification data, the system category and the component granularity classification data; Generating a corresponding association number for the fault code classification data according to the component granularity classification data; Updating the mapping table according to the corresponding association number to obtain an updated mapping table.

10. A system for implementing the method of mapping vehicle trouble codes to system components as recited in claim 1, wherein, The system comprises: A data exception processing module is configured to obtain historical vehicle diagnostic record information, analyze the proportion, continuity and range of messy data, obtain exception determination data and process the same, further obtain diagnostic exception processing data, package the same, and obtain processing data item information; The model classification module is configured to perform fault code similarity analysis and combination on the plurality of processing data information, obtain batch analysis data, perform fault code classification on the batch analysis data by using a fault classification model, extract fault-related component entity information, obtain fault code classification data, perform batch asynchronous processing on the plurality of batch analysis data, and obtain batch processing results. The component classification module is configured to perform entity normalization processing and granularity unification processing on the plurality of component entity information, and obtain standardized component classification information. The mapping analysis module is configured to establish a mapping table according to the fault code classification data, system categories, and component granularity classification data, generate an association number, update the mapping table, and obtain an updated mapping table.

Images