Resource allocation method, device and equipment for automobile aftermarket and storage medium

Abstract

The application relates to a resource allocation method, device and equipment for an automobile aftermarket and a storage medium, wherein the method comprises the following steps: acquiring multi-source heterogeneous data and performing normalization processing to obtain normalized multi-source heterogeneous data; a vehicle maintenance event trajectory is constructed, and cumulative meteorological exposure is calculated; a post-market knowledge graph is constructed based on normalized diagnosis record data and space-time gridded meteorological data, and edge weights are dynamically updated; future meteorological forecast data and material sensitivity coefficients are acquired, and a meteorological comprehensive erosion index is calculated; meteorological conditions, associated weights of component failures, vehicle ages, driving distances and the meteorological comprehensive erosion index are taken as input characteristics, and a fault prediction model is used to predict the fault probability of a to-be-tested component; based on the fault probability and the distribution of vehicles in a region, the demand for accessories in the region is predicted, accessory stocking and allocation suggestions are generated based on the demand for accessories, and the suggestions are pushed to vehicle owners. The application realizes accurate accessory demand prediction and resource optimization allocation.

Description

Technical Field

[0001] This application relates to the field of automotive aftermarket technology, and in particular to a method, apparatus, equipment and storage medium for resource allocation in the automotive aftermarket. Background Technology

[0002] The automotive aftermarket has long faced the challenge of a lagging parts supply chain response. The underlying reason lies in the lack of technical means to cross-dimensionally correlate and quantify the macro-level meteorological environment, the meso-level regional distribution of vehicles, and the micro-level characteristics of vehicle components. Meteorological environment, as a key external variable, has a significant impact on the failure modes of vehicle components. For example, high temperature and humidity conditions accelerate the aging and cracking of rubber seals, while frequent freeze-thaw cycles easily lead to stress fatigue in metal structural components. However, existing fault prediction technologies are generally limited to static attribute analysis of vehicles, relying solely on historical data such as vehicle model, age, and mileage to build models, failing to transform real-time meteorological changes and their long-term cumulative effects into calculable characteristic parameters. This limitation prevents prediction systems from dynamically capturing sudden changes in fault risk caused by meteorological fluctuations, and also prevents the quantitative assessment of the hidden losses caused by continuous exposure to specific climatic conditions throughout the vehicle's lifespan, resulting in a significant temporal deviation between prediction results and actual fault occurrence.

[0003] In the parts demand forecasting stage, current methods rely excessively on linear extrapolation of historical sales data, failing to establish a dynamic transmission chain between meteorological conditions, failure probabilities, and regional demand. Due to the lack of a quantitative mapping mechanism from meteorological data to parts failure probabilities and then to geographically distributed demand, supply chain managers cannot optimize inventory layout in advance based on weather forecasts. For example, when a typhoon warning is issued, the system struggles to accurately identify the types of vehicles and their regional distribution that are vulnerable, resulting in critical parts such as windshield wipers and headlights not being promptly deployed to high-risk areas. Simultaneously, the spatial matching of repair shop service capacity with vehicle density is not incorporated into the forecasting framework, causing resource allocation imbalances. Some regions experience parts stockpiles while others face supply shortages, ultimately creating a dual dilemma of delayed service response and increased inventory costs. This passive response model severely restricts the timeliness and economy of aftermarket services. Summary of the Invention

[0004] The purpose of this application is to propose a resource allocation method, apparatus, device, and storage medium for the automotive aftermarket, which can dynamically quantify the impact of meteorological environment on vehicle component failures and achieve accurate parts demand forecasting and optimized resource allocation.

[0005] To address the aforementioned technical problems, embodiments of this application provide a resource allocation method for the automotive aftermarket, including: Acquire multi-source heterogeneous data and normalize the multi-source heterogeneous data to obtain normalized multi-source heterogeneous data, wherein the normalized multi-source heterogeneous data includes normalized vehicle data, normalized repair shop data, normalized diagnostic record data and spatiotemporal gridded meteorological data. The vehicle's maintenance event trajectory is constructed based on the normalized diagnostic record data, and the vehicle's cumulative meteorological exposure since its initial entry is calculated based on the spatiotemporal gridded meteorological data. Based on the normalized diagnostic record data and the spatiotemporal gridded meteorological data, an aftermarket knowledge graph is constructed, and the edge weights of the aftermarket knowledge graph are dynamically updated. The entities of the aftermarket knowledge graph include meteorological conditions, parts, vehicle models, and fault phenomena. The relationships of the aftermarket knowledge graph include meteorological conditions inducing part faults and fault-related vehicle models. Acquire future weather forecast data and material sensitivity coefficients of each component to different meteorological factors, and calculate the comprehensive meteorological erosion index based on the cumulative meteorological exposure, the material sensitivity coefficients, and the future weather forecast data; The correlation weight between meteorological conditions and component failures, vehicle age, mileage, and the comprehensive meteorological erosion index provided by the aftermarket knowledge graph are used as input features to predict the failure probability of the component under test in a future time window through a failure prediction model. Based on the normalized vehicle data, the distribution of vehicles within the region is obtained. Based on the failure probability and the vehicle distribution within the region, the demand for parts is predicted by region. Based on the demand for parts, suggestions for parts preparation and allocation are generated and pushed to vehicle owners.

[0006] To address the aforementioned technical problems, embodiments of this application provide a resource allocation device for the automotive aftermarket, comprising: The data normalization processing module is used to acquire multi-source heterogeneous data and normalize the multi-source heterogeneous data to obtain normalized multi-source heterogeneous data. The normalized multi-source heterogeneous data includes normalized vehicle data, normalized repair shop data, normalized diagnostic record data, and spatiotemporal gridded meteorological data. The cumulative meteorological exposure calculation module is used to construct the vehicle's maintenance event trajectory based on the normalized diagnostic record data, and to calculate the vehicle's cumulative meteorological exposure since its initial entry based on the spatiotemporal gridded meteorological data. The knowledge graph construction module is used to construct an aftermarket knowledge graph based on the normalized diagnostic record data and the spatiotemporal gridded meteorological data, and dynamically update the edge weights of the aftermarket knowledge graph. The entities of the aftermarket knowledge graph include meteorological conditions, parts, vehicle models and fault phenomena, and the relationships of the aftermarket knowledge graph include meteorological conditions inducing part faults and fault-related vehicle models. The erosion index calculation module is used to acquire future weather forecast data and the material sensitivity coefficients of each component to different meteorological factors, and to calculate the comprehensive meteorological erosion index based on the cumulative meteorological exposure, the material sensitivity coefficients and the future weather forecast data. The failure probability prediction module is used to use the correlation weight between meteorological conditions and component failures, vehicle age, mileage and the meteorological comprehensive erosion index provided by the aftermarket knowledge graph as input features, and predict the failure probability of the component under test in a future time window through the failure prediction model. The parts demand generation module is used to obtain the vehicle distribution within the region based on the normalized vehicle data, predict the parts demand by region based on the failure probability and the vehicle distribution within the region, generate parts preparation and allocation suggestions based on the parts demand, and push the parts preparation and allocation suggestions to vehicle owners.

[0007] To solve the above-mentioned technical problems, one technical solution adopted by the present invention is to provide a computer device, including one or more processors; and a memory for storing one or more programs, so that the one or more processors implement the resource allocation method for the automotive aftermarket as described in any one of the above-mentioned methods.

[0008] To solve the above-mentioned technical problems, one technical solution adopted by the present invention is: a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the resource allocation method for the automotive aftermarket as described in any one of the above-mentioned methods.

[0009] This invention provides a resource allocation method, apparatus, equipment, and storage medium for the automotive aftermarket. By integrating multi-source heterogeneous data to construct a dynamic knowledge graph, quantifying meteorological exposure and erosion index, and combining it with a fault prediction model to achieve regional parts demand forecasting, it effectively solves the problem of supply chain response lag caused by the lack of correlation between meteorological environment and vehicle faults in the prior art. It has significant advantages in improving fault prediction accuracy, optimizing inventory layout, and reducing service delays. Attached Figure Description

[0010] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0011] Figure 1 This is a flowchart of an implementation of the resource allocation method for the automotive aftermarket provided in this application embodiment; Figure 2This is a flowchart illustrating the implementation of the first sub-process in the resource allocation method for the automotive aftermarket provided in this application embodiment; Figure 3 This is a flowchart illustrating the implementation of the second sub-process in the resource allocation method for the automotive aftermarket provided in this application embodiment; Figure 4 This is a flowchart illustrating the implementation of the third sub-process in the resource allocation method for the automotive aftermarket provided in this application embodiment; Figure 5 This is a flowchart illustrating the implementation of the fourth sub-process in the resource allocation method for the automotive aftermarket provided in this application embodiment; Figure 6 This is a flowchart illustrating the implementation of the fifth sub-process in the resource allocation method for the automotive aftermarket provided in this application embodiment; Figure 7 This is a flowchart illustrating the implementation of the sixth sub-process in the resource allocation method for the automotive aftermarket provided in this application embodiment; Figure 8 This is a schematic diagram of a resource allocation device for the automotive aftermarket provided in an embodiment of this application; Figure 9 This is a schematic diagram of the computer device provided in the embodiments of this application. Detailed Implementation

[0012] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0013] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0014] The traditional automotive aftermarket lags behind in parts supply chain response, primarily due to a lack of technical means to cross-dimensionally correlate and quantify macro-level meteorological conditions, meso-level vehicle regional distribution, and micro-level vehicle component characteristics. Existing fault prediction models fail to incorporate meteorological data, thus failing to capture the dynamic fluctuations in fault risk caused by weather conditions. Furthermore, parts demand forecasting relies heavily on historical sales data, neglecting to correlate with fault prediction and regional vehicle distribution, resulting in passive resource allocation and a coexistence of inventory backlog and shortages.

[0015] This embodiment provides a resource allocation method for the automotive aftermarket, including: Multi-source heterogeneous data is acquired and normalized to obtain normalized multi-source heterogeneous data. This normalized multi-source heterogeneous data includes normalized vehicle data, normalized repair shop data, normalized diagnostic record data, and spatiotemporal gridded meteorological data. Based on the normalized diagnostic record data, vehicle repair event trajectories are constructed, and based on the spatiotemporal gridded meteorological data, the cumulative meteorological exposure of the vehicle since its initial entry is calculated. Based on the normalized diagnostic record data and the spatiotemporal gridded meteorological data, an aftermarket knowledge graph is constructed, and its edge weights are dynamically updated. The entities in the aftermarket knowledge graph include meteorological conditions, parts, vehicle models, and fault phenomena. The relationships in the aftermarket knowledge graph include meteorological conditions inducing parts... The system identifies component failures and associated vehicle models; it acquires future weather forecast data and material sensitivity coefficients for each component to different weather factors, and calculates a comprehensive weather erosion index based on cumulative weather exposure, material sensitivity coefficients, and future weather forecast data; it uses the association weights between weather conditions and component failures, vehicle age, mileage, and the comprehensive weather erosion index provided by the aftermarket knowledge graph as input features, and predicts the failure probability of the tested component within a future time window through a failure prediction model; it obtains vehicle distribution within a region based on normalized vehicle data, predicts parts demand by region based on failure probability and vehicle distribution within the region, and generates parts preparation and allocation suggestions based on parts demand, which are then pushed to vehicle owners.

[0016] For ease of understanding, the following explains some key terms in this embodiment: Multi-source heterogeneous data refers to data sets originating from different systems and possessing different formats and structures, such as vehicle static information, repair shop geographical locations, vehicle diagnostic records, and meteorological observation data. In their raw state, these data are typically difficult to directly correlate and analyze.

[0017] Normalization refers to converting data with different dimensions and ranges to a unified scale to eliminate the influence of different dimensions, thus facilitating subsequent data fusion and model training. For example, temperature and humidity data with different units can be uniformly converted into dimensionless values.

[0018] Spatiotemporal gridded meteorological data refers to a data format in which continuous meteorological observation data or forecast data is discretized and organized according to preset time intervals and geographic grids. In this way, it is convenient to query the meteorological conditions of a specific time and a specific area.

[0019] Cumulative weather exposure refers to the cumulative time and intensity of a vehicle's exposure to specific weather conditions since it was put into use. This indicator quantifies the long-term wear and tear of vehicle components under different environmental stresses, reflecting the impact of environmental factors on vehicle lifespan.

[0020] Aftermarket knowledge graphs are tools that represent knowledge using a graph structure, which includes entities (such as weather conditions, parts, vehicle models, and fault phenomena) and relationships (such as weather conditions causing part failures and faults being associated with vehicle models). Knowledge graphs can explicitly represent and infer the complex relationships between various elements in the automotive aftermarket.

[0021] The meteorological comprehensive erosion index is a comprehensive indicator used to quantify the degree of erosion of vehicle components by meteorological environment. This index combines the vehicle's historical cumulative meteorological exposure, the sensitivity of component materials to different meteorological factors, and future weather forecasts to assess the potential erosion risk that components may face in the future.

[0022] A fault prediction model is a model built based on machine learning or statistical methods. Its function is to output the probability of the tested component failing within a specific time window in the future, based on the input features (such as the correlation weight between meteorological conditions and component failure, vehicle age, mileage, meteorological comprehensive erosion index, etc.).

[0023] This application's embodiments integrate multi-source heterogeneous data to quantify the cumulative erosion and future risks of meteorological environments on vehicle components, constructing a knowledge graph linking meteorology and faults, thus achieving accurate prediction of component failure probabilities. Consequently, it can proactively predict parts demand based on regional vehicle distribution and generate inventory and allocation suggestions, effectively solving the problems of delayed supply chain response, passive resource allocation, and the coexistence of inventory backlog and shortages in the automotive aftermarket, thereby improving the scientific nature and responsiveness of resource allocation.

[0024] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0025] Please see Figure 1 , Figure 1 This illustrates a specific implementation of a resource allocation method for the automotive aftermarket.

[0026] It should be noted that if substantially the same result is obtained, the method of this invention is not based on... Figure 1 Limited to the order of the processes shown, this method includes the following steps: S1: Acquire multi-source heterogeneous data and normalize the multi-source heterogeneous data to obtain normalized multi-source heterogeneous data, wherein the normalized multi-source heterogeneous data includes normalized vehicle data, normalized repair shop data, normalized diagnostic record data, and spatiotemporal gridded meteorological data.

[0027] Specifically, multi-source heterogeneous data can be obtained from various channels, such as manual entry, export from existing databases, or crawling through public interfaces. Normalization can be performed using linear transformations, logarithmic transformations, etc., to unify data from different sources and with different dimensions onto a comparable scale. For example, meteorological data such as temperature, humidity, and pressure can be converted into values ​​between 0 and 1, or vehicle mileage can be standardized.

[0028] Please see Figure 2 , Figure 2 A specific implementation of step S1 is shown below: S11: Acquire the multi-source heterogeneous data, wherein the multi-source heterogeneous data includes original vehicle and component data, original repair shop distribution data, original diagnostic record data, and original meteorological data. S12: Using the vehicle identification number as the primary key, associate the original vehicle and component data and the original diagnostic record data to generate the normalized vehicle data and the normalized diagnostic record data. S13: Match the latitude and longitude of the repair shops in the original repair shop distribution data with the latitude and longitude of the repair shops in the original diagnostic record data to generate the normalized repair shop data. S14: Slice the original meteorological data according to timestamps and geographic grids to generate the spatiotemporal gridded meteorological data.

[0029] The acquisition of this multi-source heterogeneous data aims to comprehensively collect various types of raw information related to the automotive aftermarket, laying the foundation for subsequent data processing and analysis. This multi-source heterogeneous data covers vehicle static attributes (such as model and configuration), component information, repair history, the geographical distribution of repair service networks, and real-time environmental and meteorological conditions. Specifically, a unified data acquisition platform can be built, capable of periodically acquiring and aggregating this raw data from vehicle manufacturer databases, third-party diagnostic systems, repair shop management software, and meteorological service providers via API interfaces or data scraping tools. Another approach is to employ a batch processing mechanism, periodically extracting data from different data sources (such as relational databases, non-relational databases, and file storage systems) and temporarily storing it in a data lake to maintain the integrity of the raw data.

[0030] Using the vehicle identification number (VIN) as the primary key, the system links the original vehicle and component data with the original diagnostic record data to generate normalized vehicle data and normalized diagnostic record data. This aims to effectively integrate scattered static vehicle information and dynamic maintenance events. The VIN, as a globally unique vehicle identifier, is crucial for connecting different data sources. Specifically, in a data warehouse, a SQL JOIN operation can be used to join the original vehicle and component data table containing the VIN with the original diagnostic record data table containing the VIN, thereby generating a normalized dataset containing detailed vehicle attributes and all its diagnostic history. Alternatively, in a distributed computing framework (such as Apache Spark), the VIN can be used as the key to aggregate and merge records from different data sources, forming structured normalized vehicle data and normalized diagnostic record data.

[0031] The normalized repair shop data is generated by matching the latitude and longitude of repair shops in the original data of the repair shop distribution with the latitude and longitude of repair shops in the original data of the diagnostic records. This aims to accurately bind repair activities to their actual geographical locations. The latitude and longitude of the repair shops provide their precise geographical coordinates. Specifically, a standardized repair shop geographic information database can be constructed first, containing a unique identifier for each repair shop and its precise latitude and longitude information. Then, for each diagnostic record, a spatial query or fuzzy matching is performed in this database based on the latitude and longitude of the recorded repair shop to determine the corresponding standardized repair shop entity and associate it with the diagnostic record. Alternatively, if there are slight discrepancies in the latitude and longitude of the diagnostic records, geocoding services or distance-based clustering algorithms can be used to group similar latitude and longitude values ​​to the same repair shop entity, thereby generating accurate normalized repair shop data.

[0032] The raw meteorological data is sliced ​​according to timestamps and geographic grids to generate spatiotemporally gridded meteorological data. This aims to transform continuous, large-scale meteorological information into discrete data units that can be precisely matched with vehicle events. The raw meteorological data is typically continuous and covers a wide area. In practice, the target area can be divided into geographic grids of a preset size (e.g., one grid per square kilometer), and the meteorological data (such as temperature, humidity, precipitation, etc.) within each grid can be aggregated or interpolated. Simultaneously, time slices are performed at fixed time intervals (e.g., hourly or daily). In this way, each data slice represents the meteorological conditions of a specific geographic grid within a specific time period. Furthermore, multidimensional arrays or spatial database technologies can be used to store meteorological data in a structure with three dimensions: time, longitude, and latitude, facilitating efficient querying and matching based on the time and location of vehicle events.

[0033] This application's embodiments effectively solve the logical discontinuity problem in the integration process of multi-source heterogeneous data through primary key association, spatial matching, and spatiotemporal slicing techniques. This ensures that subsequent fault prediction models can be calculated based on high-quality, highly correlated data, thereby improving the scientificity and accuracy of resource allocation recommendations.

[0034] S2: Construct the vehicle's maintenance event trajectory based on the normalized diagnostic record data, and calculate the vehicle's cumulative meteorological exposure since its initial entry based on the spatiotemporal gridded meteorological data.

[0035] Specifically, the vehicle's maintenance event trajectory can be obtained by sorting the vehicle's diagnostic records by time; for example, arranging the maintenance records of the same vehicle in chronological order. Cumulative weather exposure can be calculated by statistically analyzing the total duration or intensity of specific weather conditions (such as high temperature, high humidity, heavy rainfall, etc.) experienced by the vehicle in different geographical locations and time periods.

[0036] Please see Figure 3 , Figure 3 A specific implementation of step S2 is shown below: S21: The normalized diagnostic records corresponding to the same vehicle identification code are sorted according to the diagnostic start timestamp to form a trajectory point sequence.

[0037] This application aims to reconstruct the maintenance history sequence of vehicles at different time points, laying the foundation for subsequent spatiotemporal analysis. Specifically, a database management system can be used to query the stored normalized diagnostic records, using the vehicle identification number as the grouping key and the diagnostic start time stamp as the sorting key to sort the diagnostic records of each vehicle in ascending or descending order, thereby obtaining a sequence of trajectory points arranged in chronological order. Alternatively, a data processing program can load the normalized diagnostic record data, store the diagnostic records of each vehicle using a data structure (such as a list or array), and sort these records according to the diagnostic start time stamp using a sorting algorithm (such as quicksort or mergesort).

[0038] S22: Convert the latitude and longitude of the repair shop for each trajectory point in the trajectory point sequence into a geographic grid code.

[0039] The purpose of this application is to map discrete maintenance locations onto a unified geospatial framework for efficient matching with spatiotemporally gridded meteorological data. For example, the Geohash algorithm can be used to encode the latitude and longitude coordinates of the maintenance facility into a string-based geogrid code. The length of this code can be adjusted according to the required precision; for instance, a longer Geohash code represents a smaller geographic area and achieves higher precision. Alternatively, a predefined geogrid system (such as an H3 grid system or a quadtree grid system) can be used to map the latitude and longitude of the maintenance facility to a specific grid cell ID within that system, thereby achieving geographic standardization and gridding.

[0040] S23: Match the meteorological snapshot at the time of diagnosis from the spatiotemporal gridded meteorological data based on the diagnosis time and the geographic grid code.

[0041] The key to this application's embodiments lies in achieving a precise correlation between maintenance activities and the local meteorological environment at that time, ensuring the spatiotemporal consistency of meteorological data. Specifically, a multidimensional index can be constructed, using a timestamp and geographic grid code as a composite key to store spatiotemporally gridded meteorological data. When matching is needed, querying this index allows for the rapid retrieval of meteorological snapshots that precisely match a specific diagnostic time and geographic grid code. Furthermore, spatial databases (such as PostGIS) or big data processing frameworks (such as Apache Spark) can be used to preprocess and partition the spatiotemporally gridded meteorological data, enabling efficient spatiotemporal proximity queries or precise matching operations given a diagnostic time and geographic grid code.

[0042] S24: Determine the activity area of ​​the vehicle, and based on the historical meteorological data of the activity area, calculate the cumulative exposure of the vehicle under preset extreme weather conditions to obtain the cumulative meteorological exposure.

[0043] This application aims to quantify the potential environmental erosion suffered by a vehicle during its historical operation. When determining the vehicle's activity area, the minimum bounding box or convex hull covered by the geogrid codes in all trajectory point sequences corresponding to the same vehicle identification number can be calculated as the vehicle's primary activity area. Another method is to analyze the vehicle's dwell time or maintenance frequency under different geogrid codes, defining the geogrid codes with the longest dwell time or highest maintenance frequency, along with their adjacent areas, as the vehicle's core activity area. When calculating cumulative exposure, a series of extreme weather conditions (e.g., high temperature, low temperature, high humidity, heavy rainfall, sandstorms, etc.) can be predefined, and corresponding thresholds can be set for each condition. Then, historical meteorological data within the vehicle's activity area is traversed, and the total duration or intensity integral experienced by the vehicle under these extreme weather conditions is calculated to obtain the cumulative exposure. Alternatively, machine learning models can be used to perform pattern recognition on historical meteorological data, identifying specific meteorological event sequences that significantly affect vehicle components, and calculating the cumulative frequency or duration of the vehicle's exposure to these event sequences.

[0044] Based on the acquisition of normalized diagnostic record data and spatiotemporal gridded meteorological data, this application further constructs a refined vehicle maintenance event trajectory and achieves spatiotemporal alignment between maintenance behavior and meteorological data by converting the latitude and longitude of the repair shop into geographic grid codes.

[0045] S3: Based on the normalized diagnostic record data and the spatiotemporal gridded meteorological data, construct an aftermarket knowledge graph and dynamically update the edge weights of the aftermarket knowledge graph. The entities in the aftermarket knowledge graph include meteorological conditions, parts, vehicle models, and fault phenomena. The relationships in the aftermarket knowledge graph include meteorological conditions inducing part faults and fault-related vehicle models.

[0046] Specifically, the construction of an aftermarket knowledge graph can begin by manually defining entities (e.g., weather conditions, parts, vehicle models, fault phenomena) and their initial relationships (e.g., weather conditions inducing part failures, faults associated with vehicle models) or extracting them from structured documents. The edge weights of the knowledge graph can be adjusted based on the frequency of actual repair events. For example, when the frequency of a certain weather condition and a specific part failure occurring simultaneously increases, the weight of the corresponding edge can be increased.

[0047] Please see Figure 4 , Figure 4 A specific implementation of step S3 is shown below: S31: Obtain the vehicle repair manual and extract static association knowledge of vehicle model, parts and standard fault codes from the vehicle repair manual to establish an initial map skeleton.

[0048] The implementation methods of this application embodiment may include: manually having domain experts read, analyze, and summarize the vehicle repair manual, and structurally organize the static association knowledge such as the compatibility relationship between vehicle models and parts, and the correspondence between parts and standard fault codes, and input it into a graph database to form initial graph nodes and edges; or, using natural language processing (NLP) technology to automatically parse the digital repair manual text, identify entities such as vehicle models, parts, and fault codes through named entity recognition (NER) technology, and identify the static association relationships between these entities through relation extraction (RE) technology, thereby automatically constructing the initial graph skeleton.

[0049] S32: Determine the meteorological condition type corresponding to each normalized diagnostic record based on the meteorological snapshot at the diagnostic time, and establish an inducing edge from the meteorological condition entity to the component entity in the aftermarket knowledge graph.

[0050] This application embodiment determines the meteorological condition type corresponding to each normalized diagnostic record based on the meteorological snapshot at the diagnostic time, and establishes induced edges from meteorological condition entities to component entities in the aftermarket knowledge graph. Its function is to associate the vehicle's actual diagnostic records with the specific meteorological environment at the time of the malfunction, thereby revealing the potential induced mechanism of meteorological conditions on component malfunctions. This step can be implemented by: predefining a series of meteorological condition types, such as high temperature, low temperature, high humidity, heavy rainfall, and sandstorms, and setting corresponding meteorological parameter thresholds or rule sets for each type; after obtaining the meteorological snapshot at the diagnostic time, matching the meteorological parameters such as temperature, humidity, precipitation, and wind speed in the snapshot with the preset rules to determine the corresponding meteorological condition type, and establishing induced edges from the meteorological condition entity to the malfunctioning component entity in the knowledge graph; or, using a machine learning classification model, using multi-dimensional meteorological data from the meteorological snapshot at the diagnostic time as input features, training the model to automatically identify and output the most likely meteorological condition type, and then establishing the corresponding induced edges.

[0051] S33: Establish a fault-related vehicle model edge based on the fault phenomena and vehicle models in the normalized diagnostic records.

[0052] The implementation methods of this application may include: directly extracting the text description of the fault phenomenon and the vehicle model information from the normalized diagnostic records, treating the fault phenomenon as a new entity or attribute, and establishing an association edge between it and the corresponding vehicle model entity, such as the association between "engine abnormal noise" and "a certain brand A model"; or, the fault phenomenon may be standardized, for example, by using keyword matching, text clustering or ontology mapping and other technologies to map the fault phenomenon in free text form to predefined fault type entities, and then establishing an association edge between these standardized fault type entities and vehicle model entities.

[0053] S34: Dynamically update the weight of the triggering edge based on the frequency and geographical distribution of fault codes in the normalized diagnostic record.

[0054] The implementation methods of this application may include: directly extracting the text description of the fault phenomenon and the vehicle model information from the normalized diagnostic records, treating the fault phenomenon as a new entity or attribute, and establishing an association edge between it and the corresponding vehicle model entity, such as the association between "engine abnormal noise" and "a certain brand A model"; or, the fault phenomenon may be standardized, for example, by using keyword matching, text clustering or ontology mapping and other technologies to map the fault phenomenon in free text form to predefined fault type entities, and then establishing an association edge between these standardized fault type entities and vehicle model entities.

[0055] The construction method of this application, from static skeleton to dynamic weight update, provides high-confidence knowledge support for subsequent fault prediction, effectively improving the targeting of resource allocation. This solves the problem that relying solely on static maintenance manual knowledge cannot reflect the dynamic induction mechanism of meteorological environment on vehicle faults, and lacks the logic to associate meteorological snapshots with specific fault phenomena, resulting in the knowledge graph being too general in describing the causal relationship between meteorology and faults, making it difficult to support subsequent accurate fault prediction.

[0056] Please see Figure 5 , Figure 5 A specific implementation of step S34 is shown below: S341: Slice the post-market knowledge graph according to a preset time period and record the temporal changes in the weights of each side.

[0057] This application aims to capture the dynamic evolution of internal association weights over time by periodically taking snapshots of the state of the aftermarket knowledge graph. One implementation involves configuring the system to automatically archive and store the edge weight data of the knowledge graph daily, weekly, or monthly, forming a series of timestamped historical versions. Another implementation involves recording only the edge weights that change within a preset time period and their magnitude of change, combined with timestamps, to construct a lightweight time-series change log. This approach provides historical references for subsequent weight updates and supports the analysis of trends in the correlation between weather conditions and component failures.

[0058] S342: Based on the frequency of occurrence of regional codes and fault codes in the normalized diagnostic records, the weights of the edge of component failure induced by meteorological conditions are updated using a weighted evolution formula.

[0059] This application embodiment updates the weights of the edges related to meteorological conditions inducing component failures based on the frequency of occurrence of regional codes and fault codes in the normalized diagnostic records, using a weighted evolution formula. This technical feature utilizes a mathematical model to quantitatively adjust the correlation strength between meteorological conditions and component failures based on the latest diagnostic record data. For example, a Bayesian update-based formula can be used, taking the frequency of occurrence of new regional codes and fault codes as observational evidence and combining it with prior weights to calculate the updated posterior weights. Alternatively, a weighted average formula can be designed to weight and fuse historical weights with the failure incidence rate of a specific region within the current period, where the weighting factors can be adjusted according to the data's age or confidence level. This allows the edge weights of the knowledge graph to dynamically reflect the actual impact of different regional meteorological conditions on component failures, avoiding the limitations of relying solely on static statistics.

[0060] S343: Track the next normalized diagnostic record of the same vehicle after maintenance using the vehicle identification code. If the fault code disappears, strengthen the weight of the triggering edge; otherwise, weaken the weight of the triggering edge.

[0061] This application establishes a closed-loop feedback mechanism based on actual repair results to verify and correct the correlation strength between meteorological conditions and component failures in the knowledge graph. One implementation is that when the system tracks a vehicle through its vehicle identification code and detects a specific fault that has been repaired, and the corresponding fault code no longer appears in subsequent diagnostic records, the repair is considered effective, and the weight of the inducing edge between the meteorological conditions causing the fault and the component is slightly increased (strengthened). Conversely, if the fault code persists or reappears, the weight of the inducing edge is slightly decreased (attenuated). Another implementation is that the magnitude of strengthening or attenuation can be dynamically adjusted based on factors such as the severity of the fault, the timeliness of the repair, and the time interval between the disappearance or persistence of the fault code. For example, the longer the fault code disappearance time, the greater the strengthening; the shorter the fault code recurrence time, the greater the attenuation. This mechanism can utilize real repair feedback to calibrate the accuracy of the knowledge graph, effectively filtering out the influence of accidental faults or misdiagnosis, and making the weights more accurately reflect causal relationships.

[0062] This application achieves refined dynamic management of induced edge weights in the aftermarket knowledge graph by introducing time-series slice records, weight evolution formulas, and a feedback mechanism based on maintenance closed loops. This significantly improves the accuracy and reliability of characterizing the correlation between meteorological conditions and component failures, thereby providing more accurate input features for fault prediction models and ultimately optimizing the resource allocation efficiency of the automotive aftermarket.

[0063] In a specific example, the formula for updating the weights of the aftermarket knowledge graph is: ; Wherein, the initial weight: W initial =1 (based on static knowledge from the maintenance manual), α is a dynamic correction factor. This represents the number of times fault code X occurs in region Y. W represents the total number of diagnoses in region Y. t−1 The weights before the update.

[0064] S4: Obtain future weather forecast data and material sensitivity coefficients of each component to different weather factors, and calculate the comprehensive meteorological erosion index based on the cumulative weather exposure, the material sensitivity coefficients and the future weather forecast data.

[0065] Specifically, future weather forecast data can be obtained from meteorological service agencies. Material sensitivity coefficients can be determined by consulting component material property manuals or conducting laboratory simulation tests; for example, the sensitivity of a certain rubber material to ultraviolet radiation or ozone. The comprehensive meteorological erosion index can be obtained by performing a simple weighted sum or product of cumulative meteorological exposure, material sensitivity coefficients, and future weather forecast data.

[0066] Please see Figure 6 , Figure 6 A specific implementation of step S4 is shown below: S41: Obtain the material sensitivity coefficients of each component to different meteorological factors.

[0067] Specifically, the material sensitivity coefficient of each component to different meteorological factors refers to the quantitative index of the changes in the material properties (such as aging rate, corrosion degree, embrittlement tendency, etc.) of different automotive components (e.g., rubber seals, plastic parts, metal parts, electronic components, etc.) when faced with specific meteorological factors (such as high temperature, low temperature, high humidity, ultraviolet radiation, acid rain, salt spray, etc.). These coefficients can be obtained through accelerated aging tests in laboratories, materials science analysis, technical specifications or industry standards provided by component suppliers, etc. For example, by fitting the performance degradation curves of different materials under simulated extreme environments, their sensitivity to specific meteorological factors can be determined.

[0068] S42: The cumulative meteorological exposure is classified and normalized according to meteorological factors, and the historical cumulative erosion index is calculated with the sensitivity coefficient of each material as the weight. The historical cumulative erosion index is the sum of the products of the sensitivity coefficient of each material and the corresponding normalized exposure factor.

[0069] Specifically, cumulative weather exposure is the accumulated amount of various weather conditions experienced by a vehicle since its initial registration. Categorizing it by weather factor means decomposing the total exposure into cumulative amounts for specific weather factors (such as duration of high-temperature exposure, duration of high-humidity exposure, and intensity of ultraviolet radiation exposure). Normalization transforms these cumulative amounts with different dimensions and ranges to a unified scale, for example, through maximum-minimum normalization or Z-score normalization, ensuring the values ​​are between 0 and 1 for subsequent weighted calculations. This step aims to eliminate dimensional differences between different weather factors and highlight their relative impact. The historical cumulative erosion index aims to quantify the cumulative damage to vehicle components caused by weather conditions over a long period. In calculation, the normalized exposure to each weather factor is multiplied by the corresponding component's material sensitivity coefficient, and then the products of all weather factors are summed. This weighted summation reflects the actual destructive contribution of different weather factors to the material of specific components, thus more accurately assessing historical cumulative damage.

[0070] S43: Obtain the future weather forecast data of the area where the vehicle is currently located, and extract the future weather stress factor based on the future weather forecast data.

[0071] Specifically, obtaining future weather forecast data for the vehicle's current location refers to acquiring weather forecast information for a future period based on the vehicle's current geographical location or its planned driving route. This can be achieved by using the vehicle's GPS positioning system to obtain the current location and then sending a request to a weather service interface to obtain weather forecast data for the area for the next few hours, days, or weeks. This step ensures the regional accuracy of the future weather assessment. Extracting future weather stress factors based on future weather forecast data refers to extracting key meteorological parameters from the raw weather forecast data that may negatively impact component performance. For example, if heavy rainfall is forecast, "rainfall intensity" can be used as a stress factor; if high temperatures are forecast, "maximum temperature" or "duration of sustained high temperatures" can be used as a stress factor. Extraction methods can include threshold judgment (e.g., temperature exceeding 35°C), duration calculation (e.g., humidity exceeding 90% for X hours), or combining multiple meteorological parameters into a comprehensive stress index using specific algorithms.

[0072] S44: Calculate the future erosion index using the sensitivity coefficients of each material as weights, wherein the future erosion index is the sum of the products of the sensitivity coefficients of each material and the aggregated value of the corresponding future meteorological stress factor.

[0073] Specifically, the Future Erosion Index aims to predict the short-term impacts or accelerated wear that components may suffer due to changes in meteorological conditions over a future period. The calculation method is similar to the Historical Cumulative Erosion Index, but it uses future meteorological stress factors. Specifically, the extracted future meteorological stress factors (which may have undergone aggregation processing) are multiplied by the corresponding component's material sensitivity coefficient, and then the products of all meteorological factors are summed. For example, the Future Erosion Index can be expressed as: `Σ(Material Sensitivity Coefficient_j * Aggregated Future Meteorological Stress Factor Value_j)`, where `j` represents different future meteorological stress factors. This calculation method can proactively assess the potential impact of future meteorological conditions on components, providing a basis for preventative maintenance.

[0074] This application constructs a multi-dimensional meteorological erosion quantification model, realizing full-cycle risk assessment from historical cumulative damage to future environmental stress. First, by acquiring future weather forecast data and material sensitivity coefficients of various components to different meteorological factors, a foundational data support is provided for subsequent quantitative calculations. When calculating the historical cumulative erosion index, the cumulative meteorological exposure is classified and normalized, and the material sensitivity coefficient is introduced as a weight. This approach effectively distinguishes the different degrees of damage caused by different meteorological factors to specific materials, thus transforming the abstract exposure duration into a concrete material loss index, accurately assessing the long-term cumulative damage to components. When calculating the future erosion index, by extracting future meteorological stress factors and combining them with the material sensitivity coefficient for weighted aggregation, the potential stress impact of future environmental changes on components can be accurately captured, enabling the prediction of short-term risks. Finally, the historical cumulative erosion index and the future erosion index are combined to form the meteorological comprehensive erosion index. This index not only covers the "environmental history" of the vehicle's past experiences, but also predicts the "pathogenic risk" of the future environment. It provides more in-depth and comprehensive feature input for the fault prediction model, effectively solves the problem of insufficient assessment of the impact of meteorological environment under a single prediction dimension, significantly improves the accuracy of component fault prediction, and thus provides a more reliable decision-making basis for resource allocation in the automotive aftermarket.

[0075] S45: Combine the historical cumulative erosion index and the future erosion index into the meteorological comprehensive erosion index.

[0076] Specifically, the historical cumulative erosion index and the future erosion index are combined into a comprehensive meteorological erosion index. This combination involves fusing the historical cumulative erosion index and the future erosion index to form a single indicator that comprehensively reflects the meteorological-related risks of components. There are various methods for this combination, such as simple weighted summation or fusion using more complex nonlinear functions.

[0077] In one specific embodiment, the formula for calculating the meteorological comprehensive erosion index is: ; in, I weather The meteorological comprehensive erosion index, Meteorological stress factor, This represents the material sensitivity weight. Meteorological stress factors include humidity factor, temperature variability factor, and precipitation / salt spray factor.

[0078] S5: Using the correlation weight between meteorological conditions and component failures, vehicle age, mileage, and the meteorological comprehensive erosion index provided by the aftermarket knowledge graph as input features, the failure probability of the component under test in the future time window is predicted through the failure prediction model.

[0079] Specifically, fault prediction models can employ various machine learning algorithms, such as support vector machines, decision trees, or simple neural network models. These input features are directly fed into the model, which outputs a value between 0 and 1, representing the probability of a component failing within a specific future timeframe.

[0080] Please see Figure 7 , Figure 7 A specific implementation of step S5 is shown below: S51: Determine the target-induced edge weights from the aftermarket knowledge graph based on the meteorological conditions of the vehicle's location and the component under test. S52: Use the vehicle age and mileage provided by the aftermarket knowledge graph, the target-induced edge weights, and the comprehensive meteorological erosion index as input features. S53: Predict the failure probability of the component under test within a future time window using the fault prediction model based on the input features.

[0081] Specifically, several methods can be used to determine the weight of the target inducing edge. One approach is for the system to first determine the current or predicted weather condition type (e.g., high temperature, high humidity, acid rain, snow, etc.) based on the vehicle's current geographical location and real-time or predicted meteorological data. Then, in the constructed aftermarket knowledge graph, using the entity representing this weather condition type and the entity representing the component under test as the starting and ending points, the system searches for a relationship edge between them: "weather condition induces component failure." If such an edge exists, its current weight is directly extracted as the target inducing edge weight. Another approach is to construct a query using a knowledge graph query language, inputting the weather condition type of the vehicle's location and the information of the component under test. The knowledge graph inference engine automatically identifies and returns the inducing edges related to the specific weather condition and component based on predefined rules and graph structure, and extracts their weights. For example, if an edge is found where "high temperature" induces "engine cooling system" failure, the weight of that edge is returned.

[0082] When using the vehicle age and mileage provided by the aftermarket knowledge graph, the target-induced edge weights, and the meteorological comprehensive erosion index as input features, a feature vector concatenation method can be adopted. That is, the vehicle age and mileage data obtained from the aftermarket knowledge graph, the target-induced edge weights determined in the previous step, and the pre-calculated meteorological comprehensive erosion index are concatenated in a preset order to form a multi-dimensional numerical feature vector, for example, [vehicle age, mileage, target-induced edge weights, meteorological comprehensive erosion index]. Alternatively, a structured data table can be constructed, using these features as different columns to create a row of data for each component instance to be predicted, containing fields such as "vehicle ID," "component type," "vehicle age," "mileage," "target-induced edge weights," and "meteorological comprehensive erosion index," for the fault prediction model to train and infer.

[0083] When predicting the failure probability of the tested component within a future time window based on the input features using the aforementioned failure prediction model, various failure prediction models can be employed. One approach is to use supervised learning algorithms as the failure prediction model, such as Support Vector Machines (SVM), Random Forests, Gradient Boosting Trees (e.g., XGBoost or LightGBM), or neural networks. These models are trained on historical failure data, learn the mapping relationship between input features and component failures, and then predict new input features to output the failure probability. Another approach is to use recurrent neural network models such as Long Short-Term Memory Networks (LSTM), Gated Recurrent Units (GRU), or Transformer models combined with attention mechanisms, if the failure probability prediction needs to consider time-series characteristics. These models can capture patterns of feature changes over time, thereby more accurately predicting the failure probability within the future time window.

[0084] This application's embodiments achieve refined and dynamic prediction of vehicle component failure risks. Specifically, by extracting target-induced edge weights directly related to the atmospheric conditions of the vehicle's location and the tested component from the aftermarket knowledge graph, the specific impact of the macro-environment on micro-components is quantified, enabling the prediction model to capture the differential influence of environmental factors. Simultaneously, by combining the vehicle's inherent static attributes such as age and mileage with dynamically changing target-induced edge weights and a cumulative atmospheric erosion index, a multi-dimensional, highly correlated input feature vector is constructed. This feature fusion method not only considers the vehicle's own wear state but also transforms the originally discrete external environmental influences into input variables directly related to component failures through dynamic association of the knowledge graph and quantification of the atmospheric index. This allows the failure prediction model to more accurately assess the failure probability of the tested component within a future time window based on these comprehensive and highly interpretable features, effectively solving the prediction bias problem caused by the single feature in traditional prediction methods and significantly improving the accuracy and reliability of failure prediction. Ultimately, this accurate prediction of failure probability provides a solid foundation for subsequent parts demand forecasting and resource allocation, helping to achieve intelligent and proactive response in the automotive aftermarket supply chain and avoid the dilemma of coexisting inventory backlog and shortage.

[0085] Furthermore, a fault prediction model is provided: ; in, The baseline failure rate is the base probability calculated based on vehicle mileage and model years. Iweather(t) is the weather-weighted increment, which is the weather-weighted increment from the last forecast; β is the environmental acceleration factor, which is an adjustment parameter used to control the amplification factor of weather factors on the failure rate. β values ​​are larger in harsh environments and smaller in dry environments.

[0086] S6: Based on the normalized vehicle data, obtain the vehicle distribution within the region, predict the demand for parts by region based on the failure probability and the vehicle distribution within the region, generate parts preparation and allocation suggestions based on the parts demand, and push the parts preparation and allocation suggestions to the vehicle owners.

[0087] Specifically, vehicle distribution within a region can be obtained by counting the number of registered vehicles in that specific area. Parts demand can be estimated by simply multiplying the predicted failure probability by the number of relevant vehicle models in the region. Parts replenishment and allocation recommendations can be generated by comparing the predicted parts demand with the current inventory, producing simple replenishment or transfer instructions. These recommendations can be sent to vehicle owners via SMS, app notifications, etc.

[0088] Please refer to Figure 8 As a response to the above Figure 1 The implementation of the method shown in this application provides an embodiment of a resource allocation device for the automotive aftermarket, which is similar to... Figure 1 Corresponding to the method embodiments shown, the device can be specifically applied to various computer devices.

[0089] like Figure 8 As shown, the resource allocation device for the automotive aftermarket in this embodiment includes: a data normalization processing module 71, a cumulative meteorological exposure calculation module 72, a knowledge graph construction module 73, an erosion index calculation module 74, a fault probability prediction module 75, and a parts demand generation module 76, wherein: The data normalization processing module 71 is used to acquire multi-source heterogeneous data and perform normalization processing on the multi-source heterogeneous data to obtain normalized multi-source heterogeneous data, wherein the normalized multi-source heterogeneous data includes normalized vehicle data, normalized repair shop data, normalized diagnostic record data and spatiotemporal gridded meteorological data. The cumulative meteorological exposure calculation module 72 is used to construct the vehicle's maintenance event trajectory based on the normalized diagnostic record data, and to calculate the vehicle's cumulative meteorological exposure since its initial entry based on the spatiotemporal gridded meteorological data. The knowledge graph construction module 73 is used to construct an aftermarket knowledge graph based on the normalized diagnostic record data and the spatiotemporal gridded meteorological data, and dynamically update the edge weights of the aftermarket knowledge graph. The entities of the aftermarket knowledge graph include meteorological conditions, parts, vehicle models and fault phenomena, and the relationships of the aftermarket knowledge graph include meteorological conditions inducing part faults and fault-related vehicle models. The erosion index calculation module 74 is used to acquire future weather forecast data and the material sensitivity coefficients of each component to different meteorological factors, and to calculate the comprehensive meteorological erosion index based on the cumulative meteorological exposure, the material sensitivity coefficients and the future weather forecast data. The fault probability prediction module 75 is used to use the correlation weight between meteorological conditions and component faults, vehicle age, mileage and meteorological comprehensive erosion index provided by the aftermarket knowledge graph as input features, and predict the fault probability of the component under test in a future time window through the fault prediction model. The parts demand generation module 76 is used to obtain the vehicle distribution within the region based on the normalized vehicle data, predict the parts demand by region based on the failure probability and the vehicle distribution within the region, generate parts preparation and allocation suggestions based on the parts demand, and push the parts preparation and allocation suggestions to the vehicle owners.

[0090] Furthermore, the data normalization processing module 71 includes: The data acquisition unit is used to acquire the multi-source heterogeneous data, wherein the multi-source heterogeneous data includes original data of vehicles and parts, original data of repair shop distribution, original data of diagnostic records, and original meteorological data; The normalized vehicle data generation unit is used to generate the normalized vehicle data and the normalized diagnostic record data by associating the vehicle identification code as the primary key with the original data of the vehicle and components and the original data of the diagnostic record. The normalized repair shop data generation unit is used to match the latitude and longitude of the repair shops in the original data of the repair shop distribution with the latitude and longitude of the repair shops in the original data of the diagnostic records to generate the normalized repair shop data. The spatiotemporal gridded meteorological data generation unit is used to slice the original meteorological data according to timestamps and geographic grids to generate the spatiotemporal gridded meteorological data.

[0091] Furthermore, the cumulative meteorological exposure calculation module 72 includes: The trajectory point sequence travel unit is used to sort the normalized diagnostic records corresponding to the same vehicle identification code according to the diagnostic start timestamp to form a trajectory point sequence. A geographic grid coding conversion unit is used to convert the latitude and longitude of the repair shop for each trajectory point in the trajectory point sequence into a geographic grid code. A meteorological snapshot matching unit is used to match a meteorological snapshot at the time of diagnosis from the spatiotemporal gridded meteorological data based on the diagnosis time and the geographic grid code. The cumulative exposure statistics unit is used to determine the activity area of ​​the vehicle, and based on the historical meteorological data of the activity area, to calculate the cumulative exposure of the vehicle under preset extreme weather conditions, and obtain the cumulative meteorological exposure level.

[0092] Furthermore, the knowledge graph construction module 73 includes: An initial atlas skeleton suggestion unit is used to obtain the vehicle repair manual and extract static association knowledge of vehicle model, parts and standard fault codes from the vehicle repair manual to establish an initial atlas skeleton; An edge-establishing unit is used to determine the meteorological condition type corresponding to each normalized diagnostic record based on the meteorological snapshot at the diagnostic time, and to establish an edge from the meteorological condition entity to the component entity in the aftermarket knowledge graph. The fault-associated vehicle model edge establishment unit is used to establish fault-associated vehicle model edges based on the fault phenomena and vehicle models in the normalized diagnostic records. The weight update unit is used to dynamically update the weight of the triggering edge based on the frequency and geographical distribution of fault codes in the normalized diagnostic record.

[0093] Furthermore, the weight update unit includes: The slicing subunit is used to slice the aftermarket knowledge graph according to a preset time period and record the temporal changes in the weights of each side. The weight update subunit is used to update the weight of the meteorological condition-induced component failure edge based on the frequency of occurrence of regional codes and fault codes in the normalized diagnostic records, using a weight evolution formula. The weight enhancement subunit is used to track the next normalized diagnostic record of the same vehicle after maintenance through the vehicle identification code. If the fault code disappears, the weight of the triggering edge is enhanced; otherwise, the weight of the triggering edge is weakened.

[0094] Furthermore, the erosion index calculation module 74 includes: The material sensitivity coefficient acquisition unit is used to acquire the material sensitivity coefficient of each component to different meteorological factors; The cumulative meteorological exposure classification unit is used to classify and normalize the cumulative meteorological exposure according to meteorological factors, and calculate the historical cumulative erosion index with the sensitivity coefficient of each material as the weight. The historical cumulative erosion index is the sum of the products of the sensitivity coefficient of each material and the corresponding normalized exposure factor. The future meteorological stress factor extraction unit is used to acquire the future meteorological forecast data of the area where the vehicle is currently located, and extract the future meteorological stress factor based on the future meteorological forecast data. The future erosion index calculation unit is used to calculate the future erosion index by weighting each of the material sensitivity coefficients, wherein the future erosion index is the sum of the products of each of the material sensitivity coefficients and the corresponding aggregated value of the future meteorological stress factor; The meteorological comprehensive erosion index generation unit is used to combine the historical cumulative erosion index and the future erosion index into the meteorological comprehensive erosion index.

[0095] Furthermore, the fault probability prediction module 75 includes: The target-induced edge weight determination unit is used to determine the target-induced edge weight from the aftermarket knowledge graph based on the meteorological conditions of the area where the vehicle is located and the component to be tested. The input feature determination unit is used to take the vehicle age and mileage provided by the aftermarket knowledge graph, the target-induced edge weight, and the meteorological comprehensive erosion index as the input features; The fault probability prediction unit is used to predict the fault probability of the component under test within a future time window based on the input features using the fault prediction model.

[0096] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 9 , Figure 9 This is a basic structural block diagram of the computer device in this embodiment.

[0097] Computer device 8 includes storage 81, processor 82, and network interface 83, which are interconnected via a system bus. It should be noted that... Figure 9 The computer device 8 shown only has three components: storage 81, processor 82, and network interface 83. However, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0098] Storage 81 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, storage 81 may be an internal storage unit of computer device 8, such as the hard disk or memory of computer device 8. In other embodiments, storage 81 may also be an external storage device of computer device 8, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on computer device 8. Of course, storage 81 may include both internal storage units and external storage devices of computer device 8. In this embodiment, storage 81 is typically used to store the operating system and various application software installed on computer device 8, such as program code for resource configuration methods in the automotive aftermarket. In addition, storage 81 may also be used to temporarily store various types of data that have been output or will be output.

[0099] In some embodiments, processor 82 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. This processor 82 is typically used to control the overall operation of computer device 8. In this embodiment, processor 82 is used to run program code stored in storage 81 or process data, for example, to run the program code of the above-described resource allocation method for the automotive aftermarket, to implement various embodiments of the resource allocation method for the automotive aftermarket.

[0100] The network interface 83 may include a wireless network interface or a wired network interface, which is typically used to establish a communication connection between the computer device 8 and other electronic devices.

[0101] This application also provides another embodiment, namely, a computer-readable storage medium storing a computer program that can be executed by at least one processor to cause the at least one processor to perform the steps of the resource allocation method for the automotive aftermarket as described above.

[0102] Obviously, the embodiments described above are merely some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the scope of this application. This application can be implemented in many different forms; rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of protection of this application.

Claims

1. A resource allocation method for the automotive aftermarket, characterized in that, include: Acquire multi-source heterogeneous data and normalize the multi-source heterogeneous data to obtain normalized multi-source heterogeneous data, wherein the normalized multi-source heterogeneous data includes normalized vehicle data, normalized repair shop data, normalized diagnostic record data and spatiotemporal gridded meteorological data. The vehicle's maintenance event trajectory is constructed based on the normalized diagnostic record data, and the vehicle's cumulative meteorological exposure since its initial entry is calculated based on the spatiotemporal gridded meteorological data. Based on the normalized diagnostic record data and the spatiotemporal gridded meteorological data, an aftermarket knowledge graph is constructed, and the edge weights of the aftermarket knowledge graph are dynamically updated. The entities of the aftermarket knowledge graph include meteorological conditions, parts, vehicle models, and fault phenomena. The relationships of the aftermarket knowledge graph include meteorological conditions inducing part faults and fault-related vehicle models. Acquire future weather forecast data and material sensitivity coefficients of each component to different meteorological factors, and calculate the comprehensive meteorological erosion index based on the cumulative meteorological exposure, the material sensitivity coefficients, and the future weather forecast data; The correlation weight between meteorological conditions and component failures, vehicle age, mileage, and the comprehensive meteorological erosion index provided by the aftermarket knowledge graph are used as input features to predict the failure probability of the component under test in a future time window through a failure prediction model. Based on the normalized vehicle data, the vehicle distribution within the region is obtained. Based on the failure probability and the vehicle distribution within the region, the demand for parts is predicted by region. Based on the demand for parts, a part preparation and allocation suggestion is generated and pushed to the vehicle owner. The process of acquiring future weather forecast data and the material sensitivity coefficients of each component to different meteorological factors, and calculating the comprehensive meteorological erosion index based on the cumulative meteorological exposure, the material sensitivity coefficients, and the future weather forecast data, includes: Obtain the material sensitivity coefficients of each component to different meteorological factors; The cumulative meteorological exposure is classified and normalized according to meteorological factors. The historical cumulative erosion index is calculated using the sensitivity coefficient of each material as the weight. The historical cumulative erosion index is the sum of the products of the sensitivity coefficient of each material and the corresponding normalized exposure factor. Acquire the future weather forecast data for the area where the vehicle is currently located, and extract the future weather stress factor based on the future weather forecast data; Using the sensitivity coefficients of each material as weights, a future erosion index is calculated, wherein the future erosion index is the sum of the products of the sensitivity coefficients of each material and the aggregated value of the corresponding future meteorological stress factor; The historical cumulative erosion index and the future erosion index are combined to form the meteorological comprehensive erosion index; The process involves using the association weights between meteorological conditions and component failures provided by the aftermarket knowledge graph, vehicle age, mileage, and the comprehensive meteorological erosion index as input features, and predicting the failure probability of the tested component within a future time window using a failure prediction model. This includes: From the aftermarket knowledge graph, the target-induced edge weights are determined based on the meteorological conditions of the vehicle's location and the components to be tested. The vehicle age and mileage provided by the aftermarket knowledge graph, the target-induced edge weights, and the meteorological comprehensive erosion index are used as the input features. The fault prediction model uses the input features to predict the probability of failure of the component under test within a future time window.

2. The resource allocation method for the automotive aftermarket according to claim 1, characterized in that, The process of acquiring multi-source heterogeneous data and normalizing the multi-source heterogeneous data to obtain normalized multi-source heterogeneous data includes: The multi-source heterogeneous data is acquired, including raw data on vehicles and parts, raw data on the distribution of repair shops, raw data on diagnostic records, and raw meteorological data. Using the vehicle identification number as the primary key, the original data of the vehicle and its components, as well as the original data of the diagnostic records, are associated to generate the normalized vehicle data and the normalized diagnostic record data. The latitude and longitude of the repair shops in the original data of the repair shop distribution are matched with the latitude and longitude of the repair shops in the original data of the diagnostic records to generate the normalized repair shop data. The original meteorological data is sliced ​​according to timestamps and geographic grids to generate the spatiotemporal gridded meteorological data.

3. The resource allocation method for the automotive aftermarket according to claim 2, characterized in that, The process of constructing the vehicle's maintenance event trajectory based on the normalized diagnostic record data and calculating the vehicle's cumulative meteorological exposure since its initial entry based on the spatiotemporal gridded meteorological data includes: The normalized diagnostic records corresponding to the same vehicle identification code are sorted according to the diagnostic start timestamp to form a trajectory point sequence; Convert the latitude and longitude of the repair shop for each trajectory point in the trajectory point sequence into a geographic grid code; Based on the diagnosis time and the geographic grid code, match the meteorological snapshot at the diagnosis time from the spatiotemporal gridded meteorological data; The activity area of ​​the vehicle is determined, and based on the historical meteorological data of the activity area, the cumulative exposure of the vehicle under preset extreme weather conditions is calculated to obtain the cumulative meteorological exposure.

4. The resource allocation method for the automotive aftermarket according to claim 3, characterized in that, The process of constructing a post-market knowledge graph based on the normalized diagnostic record data and the spatiotemporal gridded meteorological data, and dynamically updating the edge weights of the post-market knowledge graph, includes: Obtain the vehicle repair manual and extract static association knowledge of vehicle model, parts and standard fault codes from the vehicle repair manual to establish an initial map skeleton; Based on the meteorological snapshot at the time of diagnosis, determine the meteorological condition type corresponding to each normalized diagnostic record, and establish an inducing edge from meteorological condition entity to component entity in the aftermarket knowledge graph; Establish a fault-related vehicle model edge based on the fault phenomena and vehicle models in the normalized diagnostic records; The weights of the triggering edges are dynamically updated based on the frequency and geographical distribution of fault codes in the normalized diagnostic records.

5. The resource allocation method for the automotive aftermarket according to claim 4, characterized in that, The step of dynamically updating the weights of the triggering edges based on the frequency and geographical distribution of fault codes in the normalized diagnostic records includes: The post-market knowledge graph is sliced ​​according to a preset time period, and the temporal changes in the weights of each side are recorded. Based on the frequency of occurrence of regional codes and fault codes in the normalized diagnostic records, the weights of the edge of component failure induced by meteorological conditions are updated using a weighted evolution formula. By tracking the next normalized diagnostic record of the same vehicle after maintenance using the vehicle identification code, if the fault code disappears, the weight of the triggering edge is increased; otherwise, the weight of the triggering edge is decreased.

6. A resource allocation device for the automotive aftermarket, characterized in that, include: The data normalization processing module is used to acquire multi-source heterogeneous data and normalize the multi-source heterogeneous data to obtain normalized multi-source heterogeneous data. The normalized multi-source heterogeneous data includes normalized vehicle data, normalized repair shop data, normalized diagnostic record data, and spatiotemporal gridded meteorological data. The cumulative meteorological exposure calculation module is used to construct the vehicle's maintenance event trajectory based on the normalized diagnostic record data, and to calculate the vehicle's cumulative meteorological exposure since its initial entry based on the spatiotemporal gridded meteorological data. The knowledge graph construction module is used to construct an aftermarket knowledge graph based on the normalized diagnostic record data and the spatiotemporal gridded meteorological data, and dynamically update the edge weights of the aftermarket knowledge graph. The entities of the aftermarket knowledge graph include meteorological conditions, parts, vehicle models and fault phenomena, and the relationships of the aftermarket knowledge graph include meteorological conditions inducing part faults and fault-related vehicle models. The erosion index calculation module is used to acquire future weather forecast data and the material sensitivity coefficients of each component to different meteorological factors, and to calculate the comprehensive meteorological erosion index based on the cumulative meteorological exposure, the material sensitivity coefficients and the future weather forecast data. The failure probability prediction module is used to use the correlation weight between meteorological conditions and component failures, vehicle age, mileage and the meteorological comprehensive erosion index provided by the aftermarket knowledge graph as input features, and predict the failure probability of the component under test in a future time window through the failure prediction model. The parts demand generation module is used to obtain the vehicle distribution in the region based on the normalized vehicle data, predict the parts demand by region based on the failure probability and the vehicle distribution in the region, generate parts preparation and allocation suggestions based on the parts demand, and push the parts preparation and allocation suggestions to the vehicle owners. The erosion index calculation module includes: The material sensitivity coefficient acquisition unit is used to acquire the material sensitivity coefficient of each component to different meteorological factors; The cumulative meteorological exposure classification unit is used to classify and normalize the cumulative meteorological exposure according to meteorological factors, and calculate the historical cumulative erosion index with the sensitivity coefficient of each material as the weight. The historical cumulative erosion index is the sum of the products of the sensitivity coefficient of each material and the corresponding normalized exposure factor. The future meteorological stress factor extraction unit is used to acquire the future meteorological forecast data of the area where the vehicle is currently located, and extract the future meteorological stress factor based on the future meteorological forecast data. The future erosion index calculation unit is used to calculate the future erosion index by weighting each of the material sensitivity coefficients, wherein the future erosion index is the sum of the products of each of the material sensitivity coefficients and the corresponding aggregated value of the future meteorological stress factor; The meteorological comprehensive erosion index generation unit is used to combine the historical cumulative erosion index and the future erosion index into the meteorological comprehensive erosion index. The fault probability prediction module includes: The target-induced edge weight determination unit is used to determine the target-induced edge weight from the aftermarket knowledge graph based on the meteorological conditions of the area where the vehicle is located and the component to be tested. The input feature determination unit is used to take the vehicle age and mileage provided by the aftermarket knowledge graph, the target-induced edge weight, and the meteorological comprehensive erosion index as the input features; The fault probability prediction unit is used to predict the fault probability of the component under test within a future time window based on the input features using the fault prediction model.

7. A computer device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the resource allocation method for the automotive aftermarket as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the resource allocation method for the automotive aftermarket as described in any one of claims 1 to 5.

Images