Automobile diagnosis data processing method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of intelligent automobile diagnosis and maintenance, and discloses an automobile diagnosis data processing method and device, computer equipment and a storage medium. The method comprises the following steps: firstly, carrying out frequency statistics on automobile fault codes in a historical record of diagnosis equipment, generating a to-be-processed fault code data set, inputting the to-be-processed fault code data set into a maintenance diagnosis model to generate a preliminary maintenance suggestion, and forming a preliminary maintenance suggestion data set; then, maintenance reference information is obtained according to fault codes in the data set, fusion preprocessing is carried out on the maintenance reference information and the initial maintenance suggestion data set, and a diagnosis fusion data set is generated; thirdly, performing quality correction on the fusion data set to obtain a corrected fusion diagnosis data set; and finally, integrating the three data sets according to a priority rule to generate a standard automobile diagnosis data set. According to the method, through multi-stage data processing and optimization, the accuracy and reliability of automobile diagnosis data are remarkably improved, high-quality data support is provided for intelligent automobile diagnosis, and the maintenance efficiency and precision are improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent automobile diagnosis and maintenance technology, and in particular to an automobile diagnostic data processing method, device, computer equipment and storage medium. Background Art

[0002] With the development of artificial intelligence (AI) technology, AI model training involves three core issues: data, computing power, and algorithms. In the automotive diagnostics field, the development of vertical AI applications and the pre-training of large industry models require a large amount of high-quality maintenance data as a corpus. This data is not only the foundation for model learning but also directly affects model performance and reliability.

[0003] However, the industry currently faces numerous challenges in obtaining automotive fault code repair data. First, the data provided by data portals is expensive, making it unaffordable for businesses. Furthermore, the data may not meet requirements after purchase, leading to increased future investments. Second, automotive brand data, due to commercial confidentiality, is not standardized in format and has different fault code naming conventions. For example, data from BBA manufacturers can only be accessed individually using specialized inspection tools and cannot be obtained in bulk. Third, while data from fragmented third-party providers is relatively inexpensive, it is fragmented and disorganized, requiring high labor costs for cleaning and organizing, and ultimately limited in usability. Fourth, the accuracy of publicly available data online cannot be guaranteed, manual review is time-consuming and labor-intensive, and its fragmented nature makes it difficult to systematically collect.

[0004] Therefore, how to effectively integrate multi-source automobile fault code maintenance data and build a high-quality and efficient data set has become an important issue that needs to be solved urgently. Summary of the Invention

[0005] In view of this, the embodiments of the present application provide a method, apparatus, computer equipment and storage medium for processing automobile diagnostic data, which can effectively solve the problems in the prior art such as high cost of obtaining automobile maintenance data, scattered data sources, uneven quality, confusing format and difficulty in ensuring correctness.

[0006] In a first aspect, an embodiment of the present application provides a method for processing automobile diagnostic data, comprising:

[0007] Perform frequency statistics on the vehicle fault codes in the diagnostic equipment history to obtain the vehicle fault code dataset to be processed;

[0008] Inputting the to-be-processed automobile fault code dataset into a pre-trained maintenance diagnosis model to obtain a preliminary automobile maintenance suggestion dataset;

[0009] Based on the vehicle fault code included in the preliminary vehicle repair suggestion dataset, obtaining repair reference information corresponding to the vehicle fault code, and performing fusion preprocessing on the repair reference information and the preliminary vehicle repair suggestion dataset to obtain a vehicle diagnosis fusion dataset;

[0010] Performing quality correction processing on the automobile diagnosis fusion dataset to obtain an automobile corrected fusion diagnosis dataset that meets a preset quality standard;

[0011] According to a preset priority rule, the automobile preliminary maintenance suggestion dataset, the automobile diagnosis fusion dataset, and the automobile correction fusion diagnosis dataset are integrated to form a standard automobile diagnosis dataset.

[0012] In some embodiments, the method further comprises:

[0013] When the standard automobile diagnostic dataset does not meet the preset quality standard, the steps of automobile fault code frequency statistical processing, automobile preliminary maintenance suggestion dataset generation, automobile diagnostic fusion dataset generation, quality correction processing and integration processing are repeatedly performed until the standard automobile diagnostic dataset meets the preset quality standard.

[0014] In some embodiments, performing frequency statistics on the vehicle fault codes in the diagnostic device history records to obtain a dataset of vehicle fault codes to be processed includes:

[0015] Extracting vehicle fault code information from the diagnostic device history records, and formatting the vehicle fault code information to obtain a standardized fault code list;

[0016] Grouping the standardized fault code list according to vehicle brand information to obtain a grouped fault code set classified by brand;

[0017] Counting the number of occurrences of each fault code in the grouped fault code set, and generating a descending sorted list based on the number of occurrences;

[0018] Based on the comparison between the number of occurrences of each fault code in the descending sorted list and a preset frequency threshold or cumulative frequency ratio, the fault codes that meet the frequency threshold or cumulative frequency ratio conditions are screened to form the to-be-processed automobile fault code data set.

[0019] In some embodiments, inputting the to-be-processed vehicle fault code dataset into a pre-trained maintenance diagnosis model to obtain a preliminary vehicle maintenance suggestion dataset includes:

[0020] Constructing a standardized prompt word text according to a preset prompt word template and the vehicle trouble codes and the corresponding vehicle brand information contained in the vehicle trouble code data set to be processed;

[0021] Inputting the prompt word text into the pre-trained maintenance diagnosis model to obtain initial maintenance suggestions including each vehicle fault code;

[0022] Extracting structured information of corresponding fields from the initial maintenance suggestion according to a preset field mapping rule to generate a structured maintenance suggestion entry;

[0023] The structured maintenance suggestion items are aggregated according to the order of the fault codes in the to-be-processed automobile fault code data set to form the automobile preliminary maintenance suggestion data set.

[0024] In some embodiments, the obtaining of maintenance reference information corresponding to the vehicle fault code based on the vehicle fault code included in the preliminary vehicle maintenance suggestion dataset, and performing fusion preprocessing on the maintenance reference information and the preliminary vehicle maintenance suggestion dataset to obtain a vehicle diagnosis fusion dataset includes:

[0025] According to a preset search template, the vehicle fault code and the corresponding vehicle brand information are combined as search keywords, a maintenance reference information retrieval module is called to send a search request to a target maintenance knowledge base, and the original maintenance reference information is received back;

[0026] Performing field extraction, standardization, and deduplication processing on the original maintenance reference information to obtain structured maintenance reference information that is consistent with the fields of the automobile preliminary maintenance suggestion dataset;

[0027] According to the unique identifier of the vehicle fault code, the structured maintenance reference information is matched at the field level with the corresponding structured maintenance suggestion entry in the vehicle preliminary maintenance suggestion dataset and merged to generate a preliminary fusion record;

[0028] Semantic consistency correction and redundant data elimination are performed on the preliminary fusion records to form the automobile diagnosis fusion dataset.

[0029] In some embodiments, performing quality correction processing on the vehicle diagnostic fusion dataset to obtain a vehicle corrected fusion diagnostic dataset that meets a preset quality standard includes:

[0030] Receiving calibration reference data, and performing field standardization processing on the calibration reference data according to the preset field mapping rules to generate a structured calibration data set;

[0031] Based on the unique identifier of the vehicle fault code, each record in the structured correction data set is compared with the corresponding record in the vehicle diagnosis fusion data set at the field level to identify the field that needs to be corrected;

[0032] According to a preset correction priority rule, replacing or supplementing the field value corresponding to the fault code in the structured correction data set with the corresponding field to be corrected in the vehicle diagnostic fusion data set to obtain a corrected vehicle diagnostic fusion data set;

[0033] Quality verification is performed on the corrected automobile diagnosis fusion dataset to obtain the automobile corrected fusion diagnosis dataset.

[0034] In some embodiments, the integration of the preliminary vehicle maintenance suggestion dataset, the vehicle diagnosis fusion dataset, and the vehicle correction fusion diagnosis dataset according to a preset priority rule to form a standard vehicle diagnosis dataset includes:

[0035] For each of the vehicle fault codes, extracting corresponding structured records from the vehicle preliminary repair suggestion dataset, the vehicle diagnosis fusion dataset, and the vehicle correction fusion diagnosis dataset to generate a multi-source record set to be integrated;

[0036] According to the priority rule, performing replacement or supplementation processing on the same field in the multi-source record set to be integrated to generate an integrated structured record;

[0037] Performing field detection and verification on the integrated structured records to generate a priority integration result set;

[0038] The priority integration result sets are aggregated in sequence according to the fault code sequence to form the standard automobile diagnostic data set.

[0039] In a second aspect, an embodiment of the present application provides an automobile diagnostic data processing device, comprising:

[0040] The fault code processing module is used to perform frequency statistics on the vehicle fault codes in the diagnostic equipment history records to obtain the vehicle fault code data set to be processed;

[0041] a suggestion processing module, configured to input the to-be-processed vehicle fault code dataset into a pre-trained maintenance diagnosis model, and generate preliminary maintenance suggestions corresponding to the vehicle fault codes, so as to form a preliminary vehicle maintenance suggestion dataset;

[0042] a reference information acquisition module, configured to acquire, based on the vehicle fault code included in the preliminary vehicle repair suggestion dataset, maintenance reference information corresponding to the vehicle fault code, and perform fusion preprocessing on the maintenance reference information and the preliminary vehicle repair suggestion dataset to obtain a vehicle diagnosis fusion dataset;

[0043] a correction module, configured to perform quality correction processing on the automobile diagnosis fusion dataset to obtain an automobile corrected fusion diagnosis dataset that meets a preset quality standard;

[0044] The standard data processing module is used to integrate the automobile preliminary maintenance suggestion dataset, the automobile diagnosis fusion dataset and the automobile correction fusion diagnosis dataset according to a preset priority rule to form a standard automobile diagnosis dataset.

[0045] In a third aspect, an embodiment of the present application provides a computer device, comprising a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the computer program to implement the automobile diagnostic data processing method of the first aspect.

[0046] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, and when the computer program is executed on a processor, the automobile diagnostic data processing method of the first aspect mentioned above is implemented.

[0047] The embodiments of the present application have the following beneficial effects:

[0048] The method in this application performs frequency statistics on vehicle trouble codes in the diagnostic equipment's historical records, filtering out high-frequency, practically valuable datasets for processing. This process not only reduces the processing burden of low-frequency or invalid trouble codes but also improves the relevance and efficiency of data processing, laying a solid foundation for subsequent steps.

[0049] Secondly, the pre-trained repair diagnosis model is fed into the vehicle fault code dataset to generate preliminary repair recommendations for the corresponding fault codes, forming a preliminary repair recommendation dataset. Leveraging the advantages of the pre-trained model, a large-scale, basic repair recommendation dataset can be quickly generated, effectively reducing the time and cost of manual intervention and expanding the coverage and diversity of preliminary repair recommendations.

[0050] Furthermore, based on the vehicle fault codes contained in the preliminary repair recommendation dataset, corresponding repair reference information is obtained and pre-processed with the preliminary repair recommendation dataset to generate a fused vehicle diagnostic dataset. By incorporating multi-source repair reference information, the richness and comprehensiveness of the data are significantly enhanced, while also addressing the issues of inconsistent formatting and limited content associated with single-source data, thereby significantly improving the reliability and accuracy of repair recommendations.

[0051] Subsequently, the fused vehicle diagnostic dataset was quality-corrected to remove erroneous or low-quality data, generating a corrected fused vehicle diagnostic dataset that met pre-defined quality standards. This step effectively addressed potential noise issues that may exist during the multi-source data integration process, significantly enhancing the quality of the final dataset.

[0052] Finally, according to preset priority rules, the initial vehicle repair recommendation dataset, the vehicle diagnosis fusion dataset, and the vehicle correction fusion diagnosis dataset are integrated to form a standard vehicle diagnosis dataset. By setting priority rules, high-quality data is ensured to dominate the integration process, preventing the impact of low-quality data on overall performance, thereby further optimizing the overall effectiveness of the dataset.

[0053] In summary, the method of this application realizes the full process optimization from raw fault code data to high-quality standard automobile diagnostic data set, greatly improving the efficiency, accuracy and reliability of data processing, providing high-quality data support for the field of intelligent automobile diagnosis, and laying a solid technical foundation for the precise and automated development of future intelligent diagnostic systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0055] Figure 1 A flow chart of a method for processing automobile diagnostic data according to an embodiment of the present application is shown;

[0056] Figure 2 Another flow chart of the automobile diagnostic data processing method according to an embodiment of the present application is shown;

[0057] Figure 3 Another flow chart of the automobile diagnostic data processing method according to an embodiment of the present application is shown;

[0058] Figure 4 A structural diagram of the automobile diagnostic data processing method according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0059] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments.

[0060] The components of the embodiments of the present application generally described and illustrated in the drawings herein may be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed application, but rather merely represents selected embodiments of the present application. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative effort are within the scope of protection of the present application.

[0061] Hereinafter, the terms "including", "having" and their cognates used in various embodiments of the present application are intended only to indicate specific features, numbers, steps, operations, elements, components or combinations of the aforementioned items, and should not be understood as excluding the existence of one or more other features, numbers, steps, operations, elements, components or combinations of the aforementioned items or adding the possibility of one or more features, numbers, steps, operations, elements, components or combinations of the aforementioned items. In addition, the terms "first", "second", "third" and the like are only used to distinguish descriptions and should not be understood as indicating or implying relative importance.

[0062] Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by those skilled in the art to which the various embodiments of the present application belong. The terms (such as those defined in generally used dictionaries) will be interpreted as having the same meaning as in the context of the relevant technical field and will not be interpreted as having an idealized meaning or an overly formal meaning unless clearly defined in the various embodiments of the present application.

[0063] The following describes some embodiments of the present application in detail with reference to the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments may be combined with each other.

[0064] Considering the high data acquisition costs, fragmented data sources, uneven data quality, chaotic formats, and difficulty ensuring accuracy in existing technologies, a method for processing automotive diagnostic data is proposed. This method statistically analyzes the frequency of automotive fault codes in the diagnostic equipment's historical records to select high-frequency and practically valuable fault code datasets, effectively reducing the processing burden of invalid data. This filtered fault code dataset is then fed into a pre-trained repair diagnosis model to generate preliminary repair recommendations. This dataset is then fused with multi-source repair reference information for pre-processing, significantly improving the richness and accuracy of the data. Furthermore, quality correction is performed on the fused dataset to eliminate low-quality or erroneous data, ensuring that the final data meets high standards. Finally, the datasets from each stage are integrated according to priority rules to form a high-quality, standardized automotive diagnostic dataset, providing reliable support for intelligent vehicle diagnosis.

[0065] The automobile diagnostic data processing method is described below with reference to some specific embodiments.

[0066] Figure 1 A flow chart of a method for processing automobile diagnostic data according to an embodiment of the present application is shown. Exemplarily, the method for processing automobile diagnostic data includes the following steps:

[0067] Step S100 , performing frequency statistics processing on the vehicle fault codes in the history records of the diagnostic equipment to obtain a data set of vehicle fault codes to be processed.

[0068] In this embodiment, fault code frequency statistics processing involves centralized, structured statistical analysis of the large amount of fault code information recorded in vehicle diagnostic equipment history records to identify the most representative fault codes. Diagnostic equipment history records typically consist of raw logs containing fault codes and brand information generated by various brands of diagnostic tools during maintenance. These records can reach approximately 50 million, and after deduplication, approximately 1 million brand-fault code combinations are generated.

[0069] By performing frequency statistics, we can quickly locate the most common fault codes in maintenance practice, avoiding ineffective processing of rarely occurring or obsolete fault codes. This step reduces the size of the data to be processed, improving operational efficiency, and ensures that the selected fault codes are sufficiently representative to meet the quality requirements of subsequent repair recommendation generation and data fusion. The final output is a dataset of vehicle fault codes to be processed, sorted by frequency of occurrence and accounting for over 80% of the total.

[0070] In an optional embodiment, step S100 includes:

[0071] Extract vehicle fault code information from the diagnostic equipment history records, format the vehicle fault code information, and obtain a standardized fault code list.

[0072] For example, when pre-processing diagnostic device history records, a script or ETL tool can be used to parse the raw logs and extract the fault code field information. The extracted fault code strings are then formatted (e.g., capitalized, with extra spaces and prefixes removed) to generate a standardized fault code list.

[0073] The standardized fault code list is grouped according to vehicle brand information to obtain a grouped fault code set classified by brand.

[0074] Exemplarily, based on the vehicle brand information field in the record, the standardized fault code list is grouped by brand to generate a grouped fault code set (for example, BMW, Mercedes-Benz, and Audi each correspond to an independent subset).

[0075] The number of occurrences of each fault code in the grouped fault code set is counted, and a descending sorted list is generated based on the number of occurrences. The number of occurrences of each fault code in the descending sorted list is compared with the preset frequency threshold or cumulative frequency ratio, and the fault codes that meet the frequency threshold or cumulative frequency ratio conditions are screened to form a data set of automobile fault codes to be processed.

[0076] Demonstratively, for each grouped fault code set, the number of record entries for each fault code is counted, and a descending sorted list is generated from high to low according to the number of occurrences. Then, the number of occurrences of each fault code in the descending sorted list is compared with the preset frequency threshold or the cumulative frequency proportion (such as 80% of all records). Only fault codes that meet the single code frequency ≥ threshold or the cumulative proportion reaches the set proportion are retained, and extremely low frequency or marginal fault codes are eliminated to form the final vehicle fault code data set to be processed. By screening high-frequency fault codes, the data set not only reduces the data scale for subsequent processing and improves operational efficiency, but also ensures that the selected fault codes are sufficiently representative, laying the foundation for the subsequent model to generate high-quality preliminary maintenance recommendations.

[0077] Step S200: Input the to-be-processed vehicle fault code dataset into a pre-trained maintenance diagnosis model to obtain a preliminary vehicle maintenance suggestion dataset.

[0078] The pre-trained repair diagnosis model is a deep learning model pre-trained on a large-scale automotive fault repair corpus. It possesses semantic understanding and knowledge inference capabilities for fault codes and brand information. Based on the input dataset of pending automotive fault codes, it automatically generates preliminary repair recommendations corresponding to each fault code. The preliminary automotive repair recommendation dataset: A collection of repair recommendations for specific fault codes output by the pre-trained repair diagnosis model. While these recommendations have not been manually verified, they already include basic fault analysis, possible causes, and repair steps.

[0079] For example, a dataset of vehicle fault codes to be processed is fed into a pre-trained repair diagnosis model. This model, which has been extensively trained on a wealth of repair documentation and fault code data, is able to generate comprehensive repair recommendations based on the input fault code. For example, when a BMW fault code, B3820, is input, the model, based on its internal knowledge base, outputs preliminary repair recommendations with detailed information, including [Fault Code Definition], [Fault Code Symptoms], [Circuit System Verification], [Component Testing], and [Repair Instructions]. Furthermore, multiple preliminary repair recommendations are aggregated to form a preliminary repair recommendation dataset.

[0080] In an optional embodiment, as Figure 2 As shown, step S200 includes:

[0081] Step S210 , constructing a standardized prompt word text according to a preset prompt word template and the vehicle trouble codes and corresponding vehicle brand information contained in the vehicle trouble code data set to be processed.

[0082] The prompt word template defines multiple placeholders such as fault code, brand, and output field structure, and generates professional maintenance prompt words for each fault code in a standardized format. For example, based on the preset prompt word template, each fault code and its corresponding vehicle brand information in the to-be-processed vehicle fault code dataset is filled into the template placeholder. For example, for the input "Brand is BMW, Fault Code is B3820" and the language requirement is Chinese, the prompt word template can be set as follows:

[0083] Role definition: You are a professional car breakdown repair expert.

[0084] Skill requirements: Use professional and comprehensive answers to help users solve problems from {{input}}, and output as [answer].

[0085] Field mapping requirements: Based on maintenance theory, sort out and filter the following content from the answers:

[0086]

Fault code definition

[0087]

Fault code symptoms

[0088]

Circuit system inspection

[0089] [Circuit System Test] = guide_sys_test (involves using tools to perform system electronic function testing);

[0090] Component test = guide_comp_test (including possible component damage);

[0091] Repair Guide = guide (detailed explanation of solutions and steps);

[0092] [Answer] = guide_detail (complete output content);

[0093] Language requirements: All content must be translated into {{language}} (e.g. Chinese);

[0094] Rich text format requirements: Organize the output content into a rich text format that meets the customer's needs and wrap it with tags;

[0095] Restriction: Only {{language}} content can be output; when there is no relevant content, "NA" is returned.

[0096] The prompt word template replaces placeholders with specific fault codes (such as B3820) and brand information (such as BMW) to generate a standardized prompt word text sequence, which is then stored in a prompt word queue in order of fault codes. This design ensures a highly automated and consistent prompt word generation process while meeting the diverse needs of different fault codes and brands.

[0097] In step S220 , the prompt word text is input into the pre-trained maintenance diagnosis model to obtain an initial maintenance suggestion including each vehicle fault code.

[0098] As an example, standardized prompt text is sequentially retrieved from the prompt word queue and fed into a pre-trained maintenance diagnosis model via a model call interface. During this process, the model generates a text output containing repair suggestions for each fault code based on the prompt word content and its internal knowledge base. Generation parameters (such as temperature control randomness and maximum generated text length) can be dynamically configured to ensure that the generated repair suggestions meet practical needs and are of high quality. Finally, the suggested text returned by the maintenance diagnosis model is automatically associated with the corresponding fault code and stored as the initial repair suggestion.

[0099] Step S230 : extracting structured information of corresponding fields from the initial maintenance suggestion according to a preset field mapping rule, and generating a structured maintenance suggestion entry.

[0100] Exemplarily, field extraction is performed on each output in the initial maintenance suggestion list, and structured fields such as code_def, code_case, and guide_sys_ex are parsed according to preset field mapping rules (regular expressions or JSON parsing), and then a structured maintenance suggestion entry is generated.

[0101] In step S240 , the structured maintenance suggestion items are aggregated according to the order of the fault codes in the vehicle fault code data set to be processed to form a preliminary vehicle maintenance suggestion data set.

[0102] For example, all structured entries are summarized sequentially according to the fault code sequence in the vehicle fault code dataset to be processed, and missing or incomplete entries are removed to ensure the quality and integrity of the dataset and avoid errors or invalid data in subsequent processing. Ultimately, a preliminary vehicle repair recommendation dataset is formed.

[0103] Step S300 , based on the vehicle fault codes included in the preliminary vehicle repair suggestion dataset, obtain the repair reference information corresponding to the vehicle fault codes, and perform fusion preprocessing on the repair reference information and the preliminary vehicle repair suggestion dataset to obtain a vehicle diagnosis fusion dataset.

[0104] Main sources for obtaining repair reference information include publicly available auto repair materials from both domestic and international sources, as well as local data held within companies. To efficiently collect this information, appropriate crawler tools can be selected based on specific needs. For example, Scrapy is a Python framework suitable for large-scale crawling tasks, featuring powerful middleware and pipeline systems; BeautifulSoup + Requests is a lightweight combination suitable for simple crawling tasks; and Selenium / Playwright is ideal for dynamic web pages that require JavaScript rendering. These tools enable the acquisition of high-quality repair reference information from both public channels and internal resources.

[0105] As an example, based on the vehicle fault codes contained in the preliminary repair recommendation dataset and combined with external repair reference information, new prompt words were constructed to fuse the repair reference information and the preliminary repair recommendation dataset. Both were then input into the repair diagnosis model, resulting in a more comprehensive and consistent fused vehicle diagnosis dataset. This process not only improves the accuracy of data integration but also provides reliable support for subsequent quality correction and priority integration.

[0106] In an optional embodiment, as Figure 3 As shown, step S300 includes:

[0107] Step S310 , based on a preset search template, the vehicle fault code and the corresponding vehicle brand information are combined as search keywords, the maintenance reference information search module is called to send a search request to the target maintenance knowledge base, and the returned original maintenance reference information is received.

[0108] For example, based on a preset search template, each fault code in the preliminary vehicle repair recommendation dataset and its corresponding vehicle brand information are combined into search keywords. The template defines the placeholders {{Brand}} and {{Fault Code}}, generating, for example, "BMW B3820." To improve the accuracy and structure of the search, the search classification method can be further optimized, such as using brand classification as the first level and the second level classification based on the system to which the fault code belongs (such as the engine system, air conditioning system, etc.). This classification method helps narrow the search scope and improve search efficiency.

[0109] During the retrieval process, these keywords are read in sequence, and a retrieval request is sent to the target maintenance knowledge base. After receiving the retrieval request, the retrieval module extracts the original maintenance reference information that matches the keywords from the target maintenance knowledge base. This information is usually stored in JSON or table format, and contains fields such as fault definition, fault cause, detection steps, and repair operation suggestions. In this way, not only can maintenance reference information related to specific brands and fault codes be quickly located, but it can also provide more accurate and comprehensive data support for subsequent fusion preprocessing. This combination of classification and retrieval significantly improves the ability to extract effective information from massive maintenance knowledge bases, laying the foundation for building a high-quality automotive diagnostic fusion dataset.

[0110] Step S320 , performing field extraction, standardization, and deduplication processing on the original maintenance reference information to obtain structured maintenance reference information that is consistent with the fields of the automobile preliminary maintenance suggestion dataset.

[0111] Specifically, field extraction is performed on the retrieved raw repair reference information, using regular expressions or data mapping rules to extract fields such as the fault definition (code_def) and test steps (guide_sys_test) from the JSON. The extracted fields are then standardized, including unifying field names, formatting text, removing HTML tags, and deduplicating based on fault codes and brands (to prevent duplicate information from interfering with subsequent fusion and correction processes, ensuring data accuracy and consistency). This generates a structured repair reference information set that is consistent with the fields in the preliminary vehicle repair recommendation dataset.

[0112] In step S330 , according to the unique identifier of the vehicle fault code, the structured maintenance reference information is matched at the field level with the corresponding structured maintenance suggestion item in the vehicle preliminary maintenance suggestion dataset and merged to generate a preliminary fusion record.

[0113] For example, based on the vehicle fault code identifier, the externally acquired structured maintenance reference information is matched at the field level with the corresponding entries in the vehicle preliminary maintenance suggestion dataset, and the contents of the same fields (such as fault definition, fault cause, maintenance steps, etc.) are aligned one by one, and finally a complete preliminary fusion record is generated. In this process, by constructing new prompt words for the fusion of structured maintenance reference information and the vehicle preliminary maintenance suggestion dataset, the maintenance diagnosis model is guided to understand the task requirements more accurately, thereby improving the quality and efficiency of data integration. For example, when encountering field conflicts (such as differences in the definition of the same fault between two sources), the maintenance diagnosis model can select the optimal expression and supplement additional maintenance suggestions based on the contextual information provided by the prompt words.

[0114] It's important to note that the main function of the prompt here is to assist in information merging. For example, when there are two inputs (input 1 and input 2), both of which are solutions to a specific brand's fault code (input 3), the prompt can guide the model to weight and logically reorganize the information from different sources, generating a logically coherent, concise, and well-organized summary.

[0115] Ultimately, the generated preliminary fusion record not only contains the advantages of information from both data sources, but also further improves the accuracy and consistency of the data with the assistance of the maintenance diagnosis model and new prompt words.

[0116] Step S340 : performing semantic consistency correction and redundant data elimination on the preliminary fusion records to form a vehicle diagnosis fusion dataset.

[0117] As an example, semantic consistency correction is performed on the preliminary fused records. By calculating the textual similarity of the query field values, fields with significant expression differences are identified and corrected. Furthermore, duplicate or highly similar records are removed to ensure the uniqueness and high quality of the fusion results, ultimately forming a standardized fusion dataset for automotive diagnostics.

[0118] Step S400 : performing quality correction processing on the automobile diagnosis fusion dataset to obtain an automobile corrected fusion diagnosis dataset that meets a preset quality standard.

[0119] Quality correction involves repairing and optimizing the fused automotive diagnostic dataset using pre-verified correction reference data. This correction reference data is derived from high-quality repair information annotated by industry experts and cleaned and mapped to internal system fields (such as code_def, code_case, and guide_sys_ex).

[0120] As an example, using a fused automotive diagnostic dataset and calibration reference data as input, the system generates a fused automotive calibration dataset that meets pre-defined quality standards through a series of steps, including field standardization, field comparison, field replacement / supplementation, and quality verification. This improves the integrity and accuracy of the fused data, provides a solid data foundation for subsequent priority integration, and achieves seamless integration and closed-loop security for multi-source data.

[0121] In an optional embodiment, step S400 includes:

[0122] Receive calibration reference data and perform field standardization on the calibration reference data according to preset field mapping rules to generate a structured calibration dataset.

[0123] Specifically, the system first receives calibration reference data and maps its repair information to internal structured fields according to pre-set field mapping rules. Secondly, during the mapping process, field names, text formats, and encoding styles are unified. Numeric fields are also converted, and HTML tags and extra spaces are removed. Finally, a structured calibration dataset is generated, ensuring its field structure is consistent with the vehicle diagnostic fusion dataset.

[0124] Based on the unique identifier of the vehicle fault code, each record in the structured correction dataset is compared with the corresponding record in the vehicle diagnosis fusion dataset at the field level to identify the fields that need to be corrected.

[0125] As an example, a structured correction dataset is compared field by field with the corresponding record in the vehicle diagnostic fusion dataset, using the vehicle fault code as a unique identifier. During this process, preset thresholds (such as the allowable error range for field values) and verification rules (such as data type and format requirements) are compared to identify fields requiring correction due to missing values, inconsistent data formats, or semantic deviations. These fields are then marked, providing clear target positioning for subsequent correction operations.

[0126] According to the preset correction priority rules, the field values ​​corresponding to the fault codes in the structured correction dataset are replaced or supplemented to the corresponding fields to be corrected in the vehicle diagnosis fusion dataset to obtain the corrected vehicle diagnosis fusion dataset.

[0127] For example, for identified fields requiring correction, the corrected data set replaces the corresponding fields in the original fused data with their values ​​based on correction priority (corrected reference data takes precedence over the original fused data). For fields missing from the original fused data, the corrected reference values ​​are added. This replacement and addition process generates a corrected automotive diagnostic fused data set, while preserving the original values ​​of unchanged fields to ensure the integrity of the entire data record.

[0128] Quality verification is performed on the corrected automobile diagnosis fusion dataset to obtain an automobile corrected fusion diagnosis dataset.

[0129] As an example, the corrected vehicle diagnostic fusion dataset undergoes quality verification again, including field integrity testing, data type validation, and semantic consistency assessment. Records that fail verification are automatically removed, and the reasons for the anomaly are recorded. Once verification is complete, the remaining records constitute the final corrected vehicle fusion dataset.

[0130] Step S500 : integrating the preliminary vehicle maintenance suggestion dataset, the vehicle diagnosis fusion dataset, and the vehicle correction fusion diagnosis dataset according to a preset priority rule to form a standard vehicle diagnosis dataset.

[0131] Priority rules refer to the pre-defined order for merging multi-source data, determining the overlap and supplementary order of data from different sources during merging. Priority rules are divided into three tiers based on data credibility and source quality: the vehicle calibration fusion dataset has the highest credibility due to its rigorous calibration, followed by the vehicle diagnosis fusion dataset, and the vehicle preliminary repair recommendation dataset serves as a supplementary source. This hierarchical mechanism ensures that the system prioritizes the most reliable information when integrating multi-source data.

[0132] For example, the prioritized integration process first extracts records from various data sources, then performs field overwriting and supplementation operations based on rules. The integrated data is then checked for completeness and consistency to ensure data quality. Finally, all records are aggregated in order of priority to generate a standard automotive diagnostic dataset. This mechanism automatically selects the best information based on the quality and credibility of the data sources, and combines field overwriting and supplementation rules to ensure data integrity and accuracy.

[0133] In an optional embodiment, step S500 includes:

[0134] For each vehicle fault code, the corresponding structured records in the vehicle preliminary repair suggestion dataset, vehicle diagnosis fusion dataset, and vehicle correction fusion diagnosis dataset are extracted to generate a multi-source record set to be integrated.

[0135] Specifically, for each pending vehicle fault code, the corresponding structured records are retrieved from the three datasets in sequence, forming a set of records to be integrated, including preliminary suggested entries, fused entries, and corrected entries. This set is linked by the unique identifier of the fault code and maintains the same sorting index as the original pending list, ensuring sequential consistency in subsequent processing.

[0136] According to the priority rules, the same field in the multi-source record set to be integrated is replaced or supplemented to generate the integrated structured record.

[0137] The pre-set priority rules prioritize the vehicle calibration fusion dataset over the vehicle diagnosis fusion dataset, which in turn takes precedence over the vehicle preliminary repair recommendation dataset. The specific logic is: if a high-priority data entry contains a valid value in that field, the medium- and low-priority data are overwritten. If a high-priority field is missing, the next highest level is filled in, step by step, until the field is complete, resulting in the most complete and trustworthy integrated record.

[0138] Perform field detection and verification on the integrated structured records to generate a prioritized integration result set.

[0139] For example, each consolidated record undergoes field integrity checks and consistency verification. Integrity checks ensure that key fields (such as fault code definitions and repair procedures) are not empty, while consistency verification includes data type matching and logical proofreading based on semantic similarity. Record systems that fail verification are marked and removed, retaining only qualified entries that meet all pre-defined quality criteria to form a prioritized consolidation result set.

[0140] According to the order of fault codes, the priority integration result sets are summarized in sequence to form a standard automobile diagnostic data set.

[0141] As an example, the prioritized integration result set is accumulated sequentially according to the original order of the fault codes in the pending vehicle fault code dataset to form the final standard vehicle diagnostic dataset. This dataset, which aggregates optimal field information from multiple sources, possesses high integrity and consistency and can be directly used for downstream AI model training or intelligent diagnostic system calls, ensuring the accuracy and reliability of the entire closed-loop data processing chain.

[0142] In an optional embodiment, the automobile diagnostic data processing method further includes:

[0143] When the standard automobile diagnostic dataset does not meet the preset quality standards, the steps of automobile fault code frequency statistical processing, automobile preliminary repair suggestion dataset generation, automobile diagnostic fusion dataset generation, quality correction processing and integration processing are repeated until the standard automobile diagnostic dataset meets the preset quality standards.

[0144] Specifically, after generating a standard automotive diagnostic dataset, the dataset is evaluated based on pre-set quality standards, including checking fault code coverage, field integrity ratio, and verification pass rate. If any of these indicators falls below the threshold, the update condition is considered true and the update process begins:

[0145] Fault code frequency statistics processing: re-count the frequency of fault codes in the diagnostic equipment history records and update the fault code data set to be processed;

[0146] Preliminary repair suggestion generation: The newly generated fault code dataset is input into the pre-trained repair diagnosis model to obtain a new preliminary repair suggestion dataset;

[0147] Fusion preprocessing: Perform data fusion processing based on preliminary recommendations and maintenance reference information to generate a new vehicle diagnosis fusion dataset;

[0148] Quality correction processing: Use the correction reference data to perform field standardization, comparison, and replacement to obtain a new automobile correction fusion dataset;

[0149] Priority integration processing: hierarchically cover and supplement multi-source data based on priority rules to generate a new standard automobile diagnostic dataset.

[0150] After each cycle, a new standard automotive diagnostic dataset is re-evaluated until all quality indicators meet the preset standards.

[0151] Figure 4 A schematic diagram of the structure of an automobile diagnostic data processing device according to an embodiment of the present application is shown. Exemplarily, the automobile diagnostic data processing device 100 includes:

[0152] The fault code processing module 110 is used to perform frequency statistics processing on the vehicle fault codes in the history records of the diagnostic equipment to obtain a data set of vehicle fault codes to be processed;

[0153] The suggestion processing module 120 is used to input the to-be-processed vehicle fault code dataset into a pre-trained maintenance diagnosis model to generate preliminary maintenance suggestions corresponding to the vehicle fault codes to form a preliminary vehicle maintenance suggestion dataset;

[0154] a reference information acquisition module 130 for acquiring, based on the vehicle fault code included in the preliminary vehicle repair suggestion dataset, maintenance reference information corresponding to the vehicle fault code, and performing fusion preprocessing on the maintenance reference information and the preliminary vehicle repair suggestion dataset to obtain a vehicle diagnosis fusion dataset;

[0155] A correction module 140 is configured to perform quality correction processing on the automobile diagnosis fusion dataset to obtain an automobile corrected fusion diagnosis dataset that meets a preset quality standard;

[0156] The standard data processing module 150 is used to integrate the preliminary automobile maintenance suggestion dataset, the automobile diagnosis fusion dataset, and the automobile correction fusion diagnosis dataset according to a preset priority rule to form a standard automobile diagnosis dataset.

[0157] It can be understood that the apparatus of this embodiment corresponds to the method of the above embodiment, and the options in the above embodiment are also applicable to this embodiment, so they will not be described again here.

[0158] The present application also provides a computer device. Exemplarily, the computer device includes a processor and a memory, wherein the memory stores a computer program, and the processor runs the computer program to enable the computer device to execute the functions of each module in the above method or the above apparatus.

[0159] Among them, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, including a central processing unit (CPU), a graphics processing unit (GPU) and a network processor (NP), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or at least one of other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc., which can implement or execute the disclosed methods, steps and logic block diagrams in the embodiments of the present application.

[0160] The memory may be, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc. The memory is used to store a computer program, and the processor may execute the computer program accordingly after receiving an execution instruction.

[0161] The present application also provides a computer-readable storage medium for storing the computer program used in the above-mentioned computer device. For example, the computer-readable storage medium may include, but is not limited to, various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0162] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and structure diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to the multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in an alternative implementation, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the structure diagram and / or flowchart, and the combination of boxes in the structure diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0163] In addition, the functional modules or units in the various embodiments of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0164] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a smart phone, personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application.

[0165] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A method for processing automobile diagnostic data, characterized in that: The method comprises: Perform frequency statistics on the vehicle fault codes in the diagnostic equipment history records to obtain the vehicle fault code dataset to be processed; Inputting the to-be-processed automobile fault code dataset into a pre-trained maintenance diagnosis model to obtain a preliminary automobile maintenance suggestion dataset; Based on the vehicle fault code included in the preliminary vehicle repair suggestion dataset, obtaining repair reference information corresponding to the vehicle fault code, and performing fusion preprocessing on the repair reference information and the preliminary vehicle repair suggestion dataset to obtain a vehicle diagnosis fusion dataset; Performing quality correction processing on the automobile diagnosis fusion dataset to obtain an automobile corrected fusion diagnosis dataset that meets a preset quality standard; According to a preset priority rule, the automobile preliminary maintenance suggestion dataset, the automobile diagnosis fusion dataset, and the automobile correction fusion diagnosis dataset are integrated to form a standard automobile diagnosis dataset.

2. The automobile diagnostic data processing method according to claim 1, characterized in that: The method further comprises: When the standard automobile diagnostic dataset does not meet the preset quality standard, the steps of automobile fault code frequency statistical processing, automobile preliminary maintenance suggestion dataset generation, automobile diagnostic fusion dataset generation, quality correction processing and integration processing are repeatedly performed until the standard automobile diagnostic dataset meets the preset quality standard.

3. The automobile diagnostic data processing method according to claim 1, characterized in that: The method of performing frequency statistics processing on the automobile fault codes in the history records of the diagnostic equipment to obtain the automobile fault code dataset to be processed includes: Extracting vehicle fault code information from the diagnostic device history records and formatting the vehicle fault code information to obtain a standardized fault code list; Grouping the standardized fault code list according to vehicle brand information to obtain a grouped fault code set classified by brand; Counting the number of occurrences of each fault code in the grouped fault code set, and generating a descending sorted list based on the number of occurrences; Based on the comparison between the number of occurrences of each fault code in the descending sorted list and a preset frequency threshold or cumulative frequency ratio, the fault codes that meet the frequency threshold or cumulative frequency ratio conditions are screened to form the to-be-processed automobile fault code data set.

4. The automobile diagnostic data processing method according to claim 3, characterized in that: The step of inputting the to-be-processed automobile fault code dataset into a pre-trained maintenance diagnosis model to obtain a preliminary automobile maintenance suggestion dataset includes: Constructing a standardized prompt word text according to a preset prompt word template and the vehicle trouble codes and the corresponding vehicle brand information contained in the vehicle trouble code data set to be processed; Inputting the prompt word text into the pre-trained maintenance diagnosis model to obtain initial maintenance suggestions including each vehicle fault code; Extracting structured information of corresponding fields from the initial maintenance suggestion according to a preset field mapping rule to generate a structured maintenance suggestion entry; The structured maintenance suggestion items are aggregated according to the order of the fault codes in the to-be-processed automobile fault code data set to form the automobile preliminary maintenance suggestion data set.

5. The automobile diagnostic data processing method according to claim 4, characterized in that: The method of obtaining maintenance reference information corresponding to the vehicle fault code based on the vehicle fault code included in the vehicle preliminary maintenance suggestion dataset, and fusing the maintenance reference information with the vehicle preliminary maintenance suggestion dataset to obtain a vehicle diagnosis fusion dataset includes: According to a preset search template, the vehicle fault code and the corresponding vehicle brand information are combined as search keywords, a maintenance reference information retrieval module is called to send a search request to a target maintenance knowledge base, and the original maintenance reference information is received back; Performing field extraction, standardization, and deduplication processing on the original maintenance reference information to obtain structured maintenance reference information that is consistent with the fields of the automobile preliminary maintenance suggestion dataset; According to the unique identifier of the vehicle fault code, the structured maintenance reference information is matched at the field level with the corresponding structured maintenance suggestion entry in the vehicle preliminary maintenance suggestion dataset and merged to generate a preliminary fusion record; Semantic consistency correction and redundant data elimination are performed on the preliminary fusion records to form the automobile diagnosis fusion dataset.

6. The automobile diagnostic data processing method according to claim 1, characterized in that: The performing quality correction processing on the automobile diagnosis fusion dataset to obtain an automobile corrected fusion diagnosis dataset that meets a preset quality standard includes: Receiving calibration reference data, and performing field standardization processing on the calibration reference data according to the preset field mapping rules to generate a structured calibration data set; Based on the unique identifier of the vehicle fault code, each record in the structured correction data set is compared with the corresponding record in the vehicle diagnosis fusion data set at the field level to identify the field that needs to be corrected; According to a preset correction priority rule, replacing or supplementing the field value corresponding to the fault code in the structured correction data set with the corresponding field to be corrected in the vehicle diagnostic fusion data set to obtain a corrected vehicle diagnostic fusion data set; Quality verification is performed on the corrected automobile diagnosis fusion dataset to obtain the automobile corrected fusion diagnosis dataset.

7. The automobile diagnostic data processing method according to claim 1, characterized in that: The method integrates the preliminary vehicle maintenance suggestion dataset, the vehicle diagnosis fusion dataset, and the vehicle correction fusion diagnosis dataset according to a preset priority rule to form a standard vehicle diagnosis dataset, including: For each of the vehicle fault codes, extracting corresponding structured records from the vehicle preliminary repair suggestion dataset, the vehicle diagnosis fusion dataset, and the vehicle correction fusion diagnosis dataset to generate a multi-source record set to be integrated; According to the priority rule, performing replacement or supplementation processing on the same field in the multi-source record set to be integrated to generate an integrated structured record; Performing field detection and verification on the integrated structured records to generate a priority integration result set; The priority integration result sets are aggregated in sequence according to the fault code sequence to form the standard automobile diagnostic data set.

8. An automobile diagnostic data processing device, characterized in that: include: The fault code processing module is used to perform frequency statistics on the vehicle fault codes in the diagnostic equipment history records to obtain the vehicle fault code data set to be processed; a suggestion processing module, configured to input the to-be-processed vehicle fault code dataset into a pre-trained maintenance diagnosis model, and generate preliminary maintenance suggestions corresponding to the vehicle fault codes, so as to form a preliminary vehicle maintenance suggestion dataset; a reference information acquisition module, configured to acquire, based on the vehicle fault code included in the preliminary vehicle repair suggestion dataset, maintenance reference information corresponding to the vehicle fault code, and perform fusion preprocessing on the maintenance reference information and the preliminary vehicle repair suggestion dataset to obtain a vehicle diagnosis fusion dataset; a correction module, configured to perform quality correction processing on the automobile diagnosis fusion dataset to obtain an automobile corrected fusion diagnosis dataset that meets a preset quality standard; The standard data processing module is used to integrate the automobile preliminary maintenance suggestion dataset, the automobile diagnosis fusion dataset and the automobile correction fusion diagnosis dataset according to a preset priority rule to form a standard automobile diagnosis dataset.

9. A computer device, characterized in that: The computer device includes a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the computer program to implement the automobile diagnostic data processing method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The device stores a computer program, which, when executed on a processor, implements the vehicle diagnostic data processing method according to any one of claims 1 to 7.

Images