Vehicle price evaluation method, device, equipment and medium

Abstract

This invention relates to the field of data analysis technology and discloses a vehicle price assessment method, apparatus, equipment, and medium, comprising: acquiring full lifecycle data and market data of a target vehicle; extracting features from the full lifecycle data and market data to obtain vehicle feature data and market trend feature data; generating a predicted price range and market pricing adjustment strategy for the target vehicle based on the vehicle feature data, market trend feature data, and a preset price prediction model; generating a risk assessment result for the target vehicle based on the vehicle feature data and a preset risk assessment model; and generating a price assessment report for the target vehicle based on the predicted price range, market pricing adjustment strategy, and risk assessment result. Through automated data processing and intelligent assessment models, the method achieves automatic and effective integration of multi-source heterogeneous data, and combined with risk identification capabilities, significantly improves the accuracy and efficiency of price assessment.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a method, apparatus, equipment and medium for vehicle price evaluation. Background Technology

[0002] Currently, used car valuation primarily relies on appraisers' experience and pricing based on basic vehicle information. However, with the continuous growth of the used car market and the increasing volume of related data, existing manual valuation methods are struggling to meet the demands of processing such massive amounts of data, leading to low valuation efficiency. Furthermore, manual pricing is susceptible to bias due to the appraiser's subjective judgment, resulting in poor accuracy of the valuation results. Summary of the Invention

[0003] This invention provides a vehicle price assessment method, apparatus, computer equipment, and medium to solve the technical problem that current used car assessments mainly rely on the subjective judgment of assessors, resulting in poor accuracy and low efficiency.

[0004] Firstly, a vehicle price assessment method is provided, including: Acquire the full lifecycle data and market data of the target vehicle. The full lifecycle data includes at least one of the following: basic vehicle information, vehicle exterior image data, historical operation data, and historical event data. The historical event data includes at least one of the following: historical maintenance event data, historical accident event data, historical transfer event data, and historical maintenance event data. Feature extraction is performed on the full lifecycle data and market data to obtain vehicle feature data and market trend feature data; Based on vehicle characteristic data, market trend characteristic data, and a preset price prediction model, the predicted price range for the target vehicle and the market pricing adjustment strategy are generated. Based on vehicle characteristic data and a pre-set risk assessment model, generate risk assessment results for the target vehicle; Based on the predicted price range, market pricing adjustment strategies, and risk assessment results, a price assessment report for the target vehicle is generated.

[0005] Secondly, a vehicle price evaluation device is provided, comprising: The acquisition module is used to acquire the full lifecycle data and market data of the target vehicle. The full lifecycle data includes at least one of the following: basic vehicle information, vehicle exterior image data, historical operation data, and historical event data. The historical event data includes at least one of the following: historical maintenance event data, historical accident event data, historical transfer event data, and historical maintenance event data. The feature extraction module is used to extract features from the full lifecycle data and market data respectively, to obtain vehicle feature data and market trend feature data; The first generation module is used to generate the predicted price range of the target vehicle and the market pricing adjustment strategy based on vehicle feature data, market trend feature data and a preset price prediction model. The second generation module is used to generate risk assessment results for the target vehicle based on vehicle feature data and a preset risk assessment model. The third generation module is used to generate a price assessment report for the target vehicle based on the predicted price range, market pricing adjustment strategies, and risk assessment results.

[0006] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described vehicle price assessment method.

[0007] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described vehicle price assessment method.

[0008] The aforementioned vehicle price assessment method, device, computer equipment, and storage medium, implemented in this scheme, firstly, collects full lifecycle data and market data of used cars, extracts features from each, and uses a pre-set price prediction model to uncover the non-linear relationship between features and vehicle value, outputting an objective basic valuation. Simultaneously, it simulates the market transaction environment to dynamically adjust pricing strategies, achieving intelligent pricing that coordinates vehicle and market dynamics. Secondly, it integrates multi-source feature data through a pre-set risk assessment model, identifying potential risks such as accident-damaged vehicles, flood-damaged vehicles, and vehicles with tampered odometers from multiple dimensions, providing comprehensive risk warnings to both parties in the transaction. Finally, it integrates the predicted price range, market pricing adjustment strategies, and risk assessment results to generate an intuitive and easy-to-understand price assessment report. Through automated data processing and intelligent assessment models, it achieves automatic and effective integration of multi-source heterogeneous data. Combined with risk identification capabilities, it significantly improves the accuracy and efficiency of price assessment, enhancing customer satisfaction and market competitiveness while effectively reducing labor costs. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1This is a schematic diagram of an application environment for a vehicle price evaluation method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating a vehicle price evaluation method according to an embodiment of the present invention; Figure 3 yes Figure 1 A schematic diagram of a specific implementation method for step S20; Figure 4 This is a schematic diagram of a vehicle price evaluation device according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention; Figure 6 This is another structural schematic diagram of a computer device according to one embodiment of the present invention. Detailed Implementation

[0011] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0012] The vehicle price evaluation method provided in this invention can be applied to, for example... Figure 1 In this application environment, the client communicates with the server via a network. The server can obtain the target vehicle's full lifecycle data and market data; it extracts features from the full lifecycle data and market data to obtain vehicle feature data and market trend feature data; based on the vehicle feature data, market trend feature data, and a preset price prediction model, it generates a predicted price range for the target vehicle and a market pricing adjustment strategy; based on the vehicle feature data and a preset risk assessment model, it generates a risk assessment result for the target vehicle; and based on the predicted price range, market pricing adjustment strategy, and risk assessment result, it generates a price assessment report for the target vehicle. Through automated data processing and intelligent assessment models, it achieves automatic and effective integration of multi-source heterogeneous data, combined with risk identification capabilities, significantly improving the accuracy and efficiency of price assessment, effectively reducing labor costs while enhancing customer satisfaction and market competitiveness. The client can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented using a standalone server or a server cluster consisting of multiple servers. The invention will be described in detail below through specific embodiments.

[0013] Please see Figure 2 As shown, Figure 2A flowchart illustrating the vehicle price assessment method provided in this embodiment of the invention includes the following steps: S10: Obtain full lifecycle data and market information for the target vehicle; The full lifecycle data includes at least one of the following: basic vehicle information, vehicle exterior image data, historical operation data, and historical event data. Historical event data includes historical maintenance event data, historical accident event data, historical transfer event data, and historical maintenance event data.

[0014] In this step, the target vehicle refers to the vehicle for which a value assessment is currently required. Traditional price assessment methods rely solely on basic vehicle information and cannot comprehensively reflect the vehicle's condition and market dynamics. To address these issues, this application proposes collecting full lifecycle data and market data for used vehicles. The full lifecycle data encompasses various data generated throughout the target vehicle's lifecycle, from production to its current usage. By collecting this data, the basic value range of the vehicle can be determined, the extent of exterior and interior wear and tear can be assessed, and the intensity of use and mechanical performance can be evaluated. The full lifecycle data includes at least one of the following: basic vehicle information, vehicle exterior image data, historical operating data, and historical event data. Historical event data is used to identify risk factors such as accident records, abnormal repairs, and frequent ownership transfers. This historical event data includes at least one of the following: historical repair event data, historical accident event data, historical ownership transfer event data, and historical maintenance event data.

[0015] Optionally, basic vehicle information includes static attribute information determined at the time of vehicle manufacture and configuration information added by the user later, such as the Vehicle Identification Number (VIN) and vehicle configuration information. This type of information can be obtained by the user through manual input. Vehicle exterior image data is uploaded by the user via a mobile terminal. Historical operating data includes data reflecting the vehicle's operating status during use, such as mileage, fuel consumption records, engine speed, fault code information, and driving trajectory. This type of data can be collected through the On-Board Diagnostics (OBD) system interface or obtained from the vehicle networking platform. Historical event data includes historical repair events, historical accident events, historical ownership transfer events, and historical maintenance events. Historical repair events record information such as parts replaced, repair type, and repair cost during repairs. Historical accident events record the time, type, claim amount, and whether structural damage was involved in the accident. Historical ownership transfer events record the number of ownership transfers, transfer time, and transfer location. Historical maintenance events record the maintenance time, maintenance items, and maintenance institution. This type of data can be obtained through third-party data interfaces (such as insurance companies, 4S store systems, and vehicle management offices) or extracted from user-uploaded repair and maintenance records using natural language processing technology. Market data includes historical transaction prices for the same vehicle model, current inventory for the same model, regional supply and demand indices, seasonal factors, new car price reduction information, and macroeconomic indicators. This type of data can be collected through used car trading platform interfaces, auction platform data interfaces, or web crawler technology.

[0016] S20: Extract features from the full lifecycle data and market data to obtain vehicle feature data and market trend feature data.

[0017] In this step, after acquiring the full lifecycle data and market data of the target vehicle, feature extraction is performed on the above data. Key information that reflects the vehicle's value, usage status and market environment is extracted from the raw data, and this information is converted into standardized feature vectors that can be processed by the model, ultimately obtaining vehicle feature data and market trend feature data.

[0018] Specifically, vehicle characteristic data includes at least one of the following: basic attribute characteristics, vehicle damage characteristics, mechanical performance characteristics, usage status characteristics, driving behavior characteristics, and historical event characteristics. Basic attribute characteristics include static information determined at the time of manufacture, such as vehicle age, brand, model, engine displacement, transmission type, and drive type; these characteristics are used to determine the vehicle's basic value range. Vehicle damage characteristics include the location, area, and severity of damage such as scratches, dents, and rust. Mechanical performance characteristics include information reflecting the health status of core vehicle components, such as engine condition, transmission condition, braking system condition, and chassis condition. Usage status characteristics include information reflecting the intensity of vehicle use, such as current mileage, average annual mileage, idling time percentage, and average load rate. Driving behavior characteristics include information reflecting the previous owner's driving habits, such as frequency of rapid acceleration, frequency of sudden braking, frequency of sharp turns, speeding rate, nighttime driving rate, and driving risk index. Historical event characteristics include information reflecting the vehicle's historical history, such as the number of repairs, accidents, ownership transfers, maintenance regularity, and major repair event markers.

[0019] Furthermore, market trend characteristic data includes information reflecting the external market environment of the target vehicle, such as historical transaction prices of the same model, inventory of the same model, regional supply and demand index, price volatility, seasonality factors, new car price reduction information, consumer confidence index, and policy impact factors.

[0020] By using the above methods, the original multi-source heterogeneous data is transformed into structured and quantifiable feature data, providing a high-quality data foundation for subsequent model evaluation.

[0021] In one embodiment of this application, the vehicle feature data includes at least one of the following: basic attribute features, vehicle damage features, mechanical performance features, usage status features, driving behavior features, and historical event features. For example... Figure 3 As shown, a specific feature extraction scheme is provided. In step S20, feature extraction is performed on the full lifecycle data and market data respectively to obtain vehicle feature data and market trend feature data, specifically including the following steps S21-S25: S21: Extract features from the basic information of the vehicle to obtain the basic attribute features of the target vehicle.

[0022] In this step, the Vehicle Identification Code (VIC) of the target vehicle is determined based on the vehicle's basic information. The VIC is then decoded and converted into structured, multi-dimensional basic attribute features, including at least one of the following: brand, model series, model year, engine displacement, transmission type, drive system, body style, and fuel type. Simultaneously, pre-defined multi-dimensional configuration features are extracted from the vehicle's basic information. For example, configuration features may include sunroof type, seat material, navigation system, audio brand, autonomous driving level, whether it has seat heating or ventilation functions, and whether it supports keyless entry, etc.

[0023] S22: Based on a preset instance segmentation model, feature extraction is performed on the vehicle appearance image data to obtain the vehicle damage features of the target vehicle.

[0024] In this step, a pre-defined instance segmentation model is used to segment the uploaded vehicle exterior image, accurately separating each damage instance (such as scratches, dents, rust, etc.) from the background. The vehicle exterior image data is input into the pre-defined instance segmentation model, which generates candidate damage regions through a region proposal network. For each candidate damage region, object classification, bounding box regression, and instance segmentation mask generation are performed in parallel. The final output includes the damage type, damage location bounding box, damage area, and damage severity score.

[0025] Optionally, the damage type includes scratches, dents, rust, and breakage; the damage location bounding box is used to locate the specific area of ​​damage on the vehicle body, such as the left front door, right rear fender, and front bumper; the area of ​​the damaged area is represented by pixel area or actual physical area; the damage severity score is calculated based on the ratio of the damaged area to the total area of ​​the vehicle body, the preset weight corresponding to the damage type, and the preset weight corresponding to the damage location, with a value range of 0 to 100 points, and the higher the score, the more severe the damage.

[0026] In one embodiment of this application, a specific damage feature extraction scheme is provided. In step S22, that is, based on a preset instance segmentation model, feature extraction is performed on the vehicle appearance image data to obtain the vehicle damage features of the target vehicle, specifically including the following steps S221-S226: S221: Acquire a set of images of vehicle exterior damage; The vehicle exterior damage image set includes images of vehicle exterior damage under different lighting conditions and shooting angles.

[0027] S222: Perform pixel-level annotation on the damaged areas in the vehicle exterior damage image set, and generate annotation data containing damage type labels and damage area contour masks.

[0028] S223: Use the labeled data as training data to train the Mask R-CNN model and obtain the preset instance segmentation model.

[0029] For steps S221-S223, a large number of vehicle exterior damage images are collected as sample data, and the damage areas in each image are pixel-level labeled. Closed polygonal contours are drawn for each type of damage instance, such as scratches, dents, rust, and breakage, using a labeling tool, and corresponding damage type labels are assigned, generating labeled data containing damage type labels and damage area contour masks. Subsequently, the labeled dataset is divided into training, validation, and test sets, and a transfer learning strategy is used to train the Mask R-CNN model. Specifically, the backbone network weights (such as ResNet) pre-trained on a general image dataset (such as ImageNet or COCO) are loaded as the initial feature extractor, and the Mask R-CNN model is fine-tuned using the training set. During training, the model generates candidate damage regions through a region proposal network, and for each candidate region, object classification, bounding box regression, and instance segmentation mask generation are performed in parallel. Simultaneously, classification loss, bounding box regression loss, and segmentation mask loss are calculated, and the three are weighted and summed to obtain the total loss value. The model parameters are then iteratively optimized using the backpropagation algorithm. Finally, the trained model is evaluated using a validation set, and metrics such as average precision and average intersection-over-union ratio are calculated. If the evaluation results meet the preset threshold requirements, the model is determined as the preset instance segmentation model that has been trained.

[0030] S224: Perform image enhancement processing on vehicle exterior image data.

[0031] In this step, in order to improve the accuracy of damage feature extraction, the uploaded appearance image is subjected to quality inspection to remove low-quality images with excessive blur, insufficient lighting, or improper angles. Images that meet the quality requirements are then enhanced, including histogram equalization and super-resolution reconstruction, to obtain the enhanced appearance image.

[0032] S225: Input the enhanced vehicle exterior image data into the preset instance segmentation model to detect and segment vehicle damage, and output the location coordinates, damage area, damage type, and damage severity score of the damaged area.

[0033] S226: Based on the location coordinates of the damaged area, the damaged area, the damaged type, and the damage severity score, the vehicle damage characteristics of the target vehicle are obtained.

[0034] For steps S225-S226, the vehicle exterior image data is input into a preset instance segmentation model. This model generates candidate damage regions through a region proposal network, and performs target classification, bounding box regression, and instance segmentation mask generation in parallel for each candidate damage region. Finally, it outputs the damage type, damage location bounding box, damage area, and damage severity score. Subsequently, the above information from each damage region output by the model is summarized to obtain the vehicle damage features of the target vehicle.

[0035] S23: Extract features from historical operating data to obtain the mechanical performance characteristics, usage status characteristics, and driving behavior characteristics of the target vehicle.

[0036] In this step, by analyzing the historical operating data of the target vehicle and based on pre-set status detection conditions for each core component, the health status and operating status of core components such as the engine, transmission, braking system, and emission system are identified. It is determined whether each component has a fault, and the severity of the faulty components is classified and rated to determine the fault type and severity. Finally, the mechanical performance characteristics are obtained based on the summarized status of each component.

[0037] Optionally, the mechanical performance characteristics include at least one of the following: engine status, transmission status, brake system status, chassis status, emission system status, battery status, and motor / electronic control system status. For example, the detection conditions for engine status include: the presence of engine fault codes, the type and severity of the fault codes, engine speed stability, whether the coolant temperature is abnormal, and whether the oil pressure is normal.

[0038] Furthermore, by using indicators such as mileage, time, and load from historical operating data, the current mileage, average annual mileage, idling time percentage, average driving speed, average load rate, cold start frequency, and cold start frequency of the target vehicle are calculated. Through the above calculation data, the usage status characteristics of the target vehicle are summarized to quantify the overall usage intensity of the vehicle and help determine the remaining service life of the vehicle.

[0039] Next, time-series analysis is performed on historical operational data such as acceleration, speed, and steering to obtain the target vehicle's frequency of rapid acceleration, sudden braking, sharp turning, speeding rate, nighttime driving rate, driving stability, following distance, and lane change frequency. Based on a weighted fusion of these multiple indicators, a comprehensive driving risk index is calculated as a driving characteristic to quantify the impact of previous owners' driving habits on vehicle wear and tear.

[0040] S24: Extract features from historical event data using a pre-defined semantic analysis model to obtain the historical event features of the target vehicle; Among them, the characteristics of historical events include at least one of the following: maintenance characteristics, accident characteristics, transfer characteristics, and upkeep characteristics.

[0041] In this step, a pre-set semantic analysis model is used to perform natural language processing on textual data such as repair records and insurance claim records in historical event data. This extracts structured historical event features, including key information such as event type, event time, involved components, operation type, repair amount, accident cause, and responsible party. Then, based on the key information of each event, repair features, accident features, and maintenance features are obtained. Simultaneously, by analyzing vehicle transfer data, the vehicle's transfer features are obtained.

[0042] In one embodiment of this application, a specific historical event feature extraction scheme is provided. In step S24, that is, feature extraction of historical event data is performed on the historical event data through a preset semantic analysis model to obtain the historical event features of the target vehicle, specifically including the following steps S241-S248: S241: Obtain the vehicle event dataset; The vehicle incident dataset includes repair records, insurance claim records, and maintenance records for multiple vehicles.

[0043] S242: Label the vehicle event dataset to generate labeled data containing event type, time information, involved parts, operation type, and amount involved.

[0044] S243: Use the labeled vehicle event dataset as training data to train the BERT model and obtain the preset semantic analysis model.

[0045] For steps S241-S243, a large amount of textual data, including vehicle repair records, insurance claim records, and maintenance records, is collected as training samples. Key information in the samples is manually labeled, including event type (repair, accident, maintenance, transfer of ownership), time information (event occurrence time, repair duration), involved components (engine, transmission, braking system, body panels, etc.), operation type (replacement, repair, painting, inspection), amount involved, accident type (rear-end collision, side impact, flooding, fire), responsible party, and other entities and their relationships. Based on the labeled dataset, transfer learning is performed on the BERT model, loading pre-trained BERT model weights from a general corpus. Fine-tuning is then performed using the labeled vehicle text dataset to adapt the model to the specialized terminology and expressions used in the automotive repair and accident fields. During fine-tuning, the model extracts the contextual semantic features of the text through a multi-layer Transformer structure, sets output layers for named entity recognition and relation extraction tasks respectively, optimizes them using the cross-entropy loss function, and iteratively updates the model parameters through the backpropagation algorithm until the model's recognition accuracy and recall on the validation set reach the preset threshold, thus obtaining the trained preset semantic analysis model.

[0046] S244: Extract historical maintenance data, historical insurance claim data, historical transfer data, and historical maintenance data of the target vehicle from historical event data.

[0047] In this step, historical event data includes various types of record-keeping text data generated during the use of the target vehicle. Historical maintenance data, historical insurance claims data, historical transfer data, and historical maintenance data of the target vehicle are extracted from this text data.

[0048] In practical applications, historical maintenance data includes records generated when vehicles are repaired at 4S dealerships, comprehensive repair shops, and other similar institutions. This data covers information such as repair time, repair items, involved parts, repair cost, type of replaced parts (original or aftermarket), and the name of the repair shop. This type of data can be extracted by connecting with the repair shop's system or from repair work orders and electronic repair records provided by the user. Historical insurance claim data includes claim records after accidents that occurred while the vehicle is insured with an insurance company. This data covers the time of the accident, accident type, claim amount, repair items, whether personal injury was involved, and liability allocation. This type of data can be extracted by connecting with the insurance company's system or from insurance claim records provided by the user. Historical ownership transfer data includes records generated when vehicle ownership is transferred at the vehicle management office. This data covers the number of transfers, transfer time, transfer location, and information about the buyer and seller. This type of data can be extracted by connecting with the vehicle management office's system or from the vehicle registration certificate. Historical maintenance data includes records generated during routine vehicle maintenance, covering information such as maintenance time, maintenance mileage, maintenance items (such as changing engine oil, oil filter, air filter, etc.), and maintenance institution. This type of data can be extracted by connecting with the maintenance institution's system or from the maintenance manual or electronic maintenance records provided by the user.

[0049] S245: Input historical maintenance data into a preset semantic analysis model to obtain the maintenance characteristics of the target vehicle.

[0050] In this step, historical maintenance data is input into a preset semantic analysis model. The preset semantic analysis model performs entity recognition and relation extraction on the maintenance record text, extracting key information such as the time of the maintenance event, maintenance type, involved parts, maintenance amount, whether original parts were replaced, and type of maintenance organization, and summarizing the maintenance features.

[0051] S246: Input historical insurance claim data into a preset semantic analysis model to obtain the accident characteristics of the target vehicle.

[0052] In this step, historical insurance claim data is input into a preset semantic analysis model. The preset semantic analysis model extracts key information such as accident time, accident type, accident severity, claim amount, whether structural damage is involved, whether airbags deploy, and responsible party from insurance claim records and accident repair records. Based on the above information, statistical features such as the number of accidents, accident frequency, major accident identification, flooded vehicle identification, and fire vehicle identification are calculated. Finally, by combining the key information and statistical features of each accident, the accident features are summarized.

[0053] S247: Extract features from historical transfer data to obtain the transfer features of the target vehicle.

[0054] In this step, the number of times the target vehicle has been transferred, the time of transfer, the region of transfer, the interval between transfers, and other information are statistically analyzed by using the transfer records in the historical event data, and the transfer characteristics are summarized.

[0055] S248: Input historical maintenance data into the preset semantic analysis model to obtain the maintenance characteristics of the target vehicle.

[0056] In this step, historical maintenance data is input into a preset semantic analysis model. The preset semantic analysis model extracts information such as maintenance time, maintenance items, maintenance mileage, and maintenance institutions from the maintenance record text. Based on the above information, key indicators such as maintenance frequency, maintenance regularity, maintenance interval rationality, and whether maintenance is performed on time are calculated, and then the maintenance characteristics are summarized.

[0057] S25: Based on basic vehicle information, feature extraction is performed on market data to obtain market trend features.

[0058] In this step, market data includes real-time transaction data, listing data, and user behavior data collected through used car trading platform interfaces, auction platform interfaces, or web crawler technology, as well as macroeconomic data and policy information obtained through third-party data sources. Based on the target vehicle's basic information, relevant data of the same brand, model, and year as the target vehicle are extracted from the market data. Specifically, this includes the vehicle's historical transaction price, transaction speed, negotiation space, inventory for sale, new supply speed, listing price distribution, search popularity, click-through rate, inquiry conversion rate, new car price reduction information, policy influencing factors, and seasonal factors. Based on this, a time-series analysis is conducted using historical transaction prices of the same vehicle model to calculate indicators such as price moving average, price trend slope, and price volatility, revealing the price trend characteristics of the same vehicle model. The supply-demand ratio is calculated based on the ratio of current inventory to historical transaction volume of the same vehicle model, revealing supply-demand balance characteristics. Search volume, click volume, and inquiry volume for the same vehicle model are statistically analyzed to calculate a demand heat index, revealing demand heat characteristics. Periodic analysis is performed based on historical transaction price differences of the same vehicle model in different seasons, generating a seasonal adjustment factor, revealing seasonal fluctuation characteristics. The transmission coefficient of new car prices to used car prices is calculated based on changes in the official guide price of new cars and terminal discounts for the same vehicle model, revealing the new car impact characteristics. Regional economic indicators and policy information are quantified into macroeconomic impact factors, revealing macroeconomic environment characteristics. Finally, the above characteristics are summarized to obtain market trend characteristics, used to reflect the changing trends of the external market environment in which the target vehicle operates.

[0059] S30: Based on vehicle characteristic data, market trend characteristic data, and a preset price prediction model, generate the predicted price range for the target vehicle and the market pricing adjustment strategy.

[0060] In this step, a pre-set price prediction model is used to comprehensively assess the vehicle's condition and the external market environment, outputting a pricing result that reflects the vehicle's true value while adapting to market fluctuations. Vehicle characteristic data and market trend characteristic data are input into the pre-set price prediction model. The model uses an attention mechanism to fuse vehicle damage characteristics, mechanical performance characteristics, usage status characteristics, driving behavior characteristics, historical event characteristics, and market trend characteristics, extracting the non-linear relationship between these characteristics and vehicle value, and outputting a predicted price range for the target vehicle. This price range reflects a comprehensive assessment result based on the vehicle's condition and historical market patterns. Building on this, the model further simulates the market transaction environment based on the predicted price range, vehicle characteristic data, and current market dynamic data. With the goal of maximizing the probability of actual transactions, it learns dynamic pricing strategies under different market conditions, ultimately obtaining a dynamically adjusted pricing strategy. This pricing strategy includes suggested pricing, price adjustment range, recommended promotional labels, and suggested negotiation space.

[0061] Through the above methods, an objective basic valuation based on the vehicle's own condition is provided, while the pricing strategy is dynamically adjusted according to real-time changes in the market environment. This enables intelligent pricing that is coordinated between the vehicle and the market, effectively improving the accuracy of pricing and transaction efficiency, and meeting the needs of large-scale transactions.

[0062] In one embodiment of this application, a specific price evaluation scheme is provided. In step S30, based on vehicle characteristic data, market trend characteristic data, and a preset price prediction model, a predicted price range for the target vehicle and a market pricing adjustment strategy are generated, specifically including the following steps S31-S32: S31: Input vehicle feature data and market trend feature data into the deep neural network sub-model, extract the mapping relationship between feature data and vehicle value through the deep neural network sub-model, and output the predicted price range of the target vehicle.

[0063] In this step, the pre-set price prediction model includes a deep neural network sub-model and a reinforcement learning sub-model, which work together to complete the entire process from basic valuation to dynamic pricing. The deep neural network sub-model is used to extract the non-linear relationship between vehicle feature data and market trend feature data. By learning complex patterns from a large amount of historical transaction data, it outputs the basic price range for the target vehicle. Specifically, a multimodal input structure is used to input vehicle feature data and market trend feature data into the trained deep neural network sub-model. The model integrates vehicle damage features, mechanical performance features, usage status features, driving behavior features, historical event features, and market trend features through an attention mechanism. It automatically learns the complex mapping relationship between these features and vehicle value, capturing the comprehensive impact of multiple factors such as mileage, vehicle age, accident records, maintenance history, driving behavior, market supply and demand, seasonal fluctuations, and new car price reductions on the price. This fully reflects the value formation pattern under the interaction of the vehicle's own condition and the external market environment, ultimately outputting a predicted price range that reflects the combined effect of the vehicle's objective value and market conditions. Compared to traditional linear regression or simple weighted methods, deep neural network sub-models can handle non-linear interaction effects between features. For example, the price reduction for the same scratch may differ between low-mileage and high-mileage vehicles, or the discount rate for accident vehicles may be further amplified during the off-season. This clarifies the value of different vehicles in different market environments and provides a relatively objective basic valuation.

[0064] S32: Input the predicted price range and market data into the reinforcement learning sub-model. Simulate the current market trading environment through the reinforcement learning sub-model. With the goal of maximizing the actual transaction probability, output the market pricing adjustment strategy for the target vehicle.

[0065] In this step, the reinforcement learning sub-model, based on the base price range output by the deep neural network sub-model, simulates the current market trading environment. With the goal of maximizing the probability of actual transactions, it learns dynamic pricing strategies under different market conditions. Specifically, the predicted price range output by the deep neural network sub-model and market data are input into the reinforcement learning sub-model. Through multiple rounds of trial and error learning in the simulated market trading environment, the model can dynamically adjust its pricing strategy according to the current market environment. For example, when the market is oversupplied, it suggests a moderate price reduction and adding an "urgent sale" label to expedite transactions; when the market is undersupplied, it suggests maintaining or slightly increasing the price to obtain higher profits. Finally, the reinforcement learning sub-model outputs market pricing adjustment strategies, including suggested pricing, price adjustment range, recommended promotional labels, and suggestions for negotiation space, providing both parties with pricing guidance that aligns with the vehicle's objective value and adapts to real-time market fluctuations.

[0066] By utilizing the synergistic effect of the two sub-models, we can provide an accurate basic valuation based on the objective value of the vehicle, and flexibly adjust the trading strategy according to real-time market changes, thereby achieving intelligent pricing that coordinates the vehicle and the market.

[0067] In one embodiment of this application, a model training scheme is provided, which, before generating the predicted price range for the target vehicle and the market pricing adjustment strategy based on vehicle feature data, market trend feature data, and a preset price prediction model, further includes the following steps: Obtain full lifecycle data, historical transaction prices, vehicle risk levels, and market conditions for multiple traded vehicles; Feature extraction is performed on full lifecycle data and market data to obtain a feature dataset for each traded vehicle. Based on the feature dataset and historical transaction prices, the initial deep neural network sub-model is trained with the goal of minimizing the difference between the predicted price and the historical transaction price. The model parameters are iteratively updated to obtain the trained deep neural network sub-model. Based on the basic price range, feature dataset and simulated market trading environment output by the trained deep neural network sub-model, the initial reinforcement learning sub-model is trained. With the goal of maximizing the actual transaction probability, the model parameters are iteratively updated to obtain the trained reinforcement learning sub-model. Based on vehicle feature data and vehicle risk level in the feature dataset, an initial deep neural network is trained with the goal of minimizing the difference between the predicted risk level and the vehicle risk level. The model parameters are iteratively updated to obtain a trained preset risk assessment model.

[0068] In this embodiment, firstly, a large amount of full lifecycle data, historical transaction prices, vehicle risk levels, and market data of traded used cars are collected. Then, feature extraction is performed on the full lifecycle data and market data to obtain a feature dataset for each traded vehicle. Further, the vehicle feature data and market trend feature data from the feature dataset are used as input, and historical transaction prices are used as supervision labels. With the goal of minimizing the difference between the predicted price and the historical transaction price, the model parameters are iteratively updated using the backpropagation algorithm to obtain a trained deep neural network sub-model. Subsequently, a simulated market trading environment is constructed. The base price range output by the deep neural network sub-model, along with the vehicle features and market trend features from the feature dataset, are used as input to the state space. With the goal of maximizing the actual transaction probability, a deep Q-network algorithm is used to perform multiple rounds of trial and error learning in the simulated environment, iteratively updating the model parameters to obtain a trained reinforcement learning sub-model.

[0069] Furthermore, using vehicle feature data from the feature dataset as input and vehicle risk level as supervision label, with the goal of minimizing the difference between the predicted risk level and the vehicle risk level, the model parameters are iteratively updated through the backpropagation algorithm to obtain the trained preset risk assessment model.

[0070] By applying reinforcement learning to used car pricing in this way, the model achieves adaptability to market fluctuations, enabling it to flexibly respond to the impact of external factors such as economic cycle changes and policy adjustments. Simultaneously, through transfer learning techniques, the evaluation experience of high-residual-value vehicles can be effectively transferred to less popular models, significantly improving evaluation accuracy in small sample scenarios.

[0071] S40: Generate risk assessment results for the target vehicle based on vehicle feature data and a preset risk assessment model.

[0072] In this process, there are numerous hidden risks in used car transactions, such as accident-damaged vehicles, flood-damaged vehicles, fire-damaged vehicles, and vehicles with tampered odometers. These vehicles often have their exteriors repaired, making them difficult to detect based on price alone, yet they pose serious safety hazards and future usage problems. To effectively identify these risks, prevent buyers from purchasing problematic vehicles, and help sellers avoid subsequent disputes, this application proposes a pre-trained risk assessment model. By integrating multi-source data such as repair records, insurance claim records, and image detection results, this model identifies potential risks of the target vehicle from multiple dimensions, providing comprehensive risk warnings for both parties in the transaction.

[0073] S50: Generates a price assessment report for the target vehicle based on the predicted price range, market pricing adjustment strategies, and risk assessment results.

[0074] In this step, after obtaining the predicted price range for the target vehicle, market pricing adjustment strategies, and risk assessment results, a price assessment report is generated based on the above information, providing users with comprehensive decision support. This price assessment report helps users understand the reasonable price range for the vehicle and the optimal pricing strategy under the current market conditions, and also helps users fully grasp the potential problems of the vehicle, thereby effectively improving the transparency and efficiency of used car transactions.

[0075] In one embodiment of this application, a specific price assessment report generation scheme is provided. In step S50, that is, based on the predicted price range, market pricing adjustment strategy, and risk assessment results, a price assessment report for the target vehicle is generated, specifically including the following steps S51-S53: S51: Input vehicle feature data, market trend feature data, and predicted price range into the SHAP algorithm to calculate the feature contribution value of each feature to the predicted price.

[0076] In this step, the predicted price range output by the deep neural network sub-model is analyzed using the SHAP (Shapley Additive Explanations) algorithm. Vehicle feature data, market trend feature data, and the predicted price range are input into the SHAP algorithm. Based on the Shapley value in cooperative game theory, the SHAP algorithm decomposes the predicted price into the sum of the contributions of each input feature, outputting the effect of each feature on pushing up or lowering the predicted price. For example, lower mileage increases the price by 5000 yuan, while an accident record decreases the price by 8000 yuan. This visually demonstrates to users the key factors affecting vehicle prices and their degree of influence, providing traceability of the evaluation results and allowing users to view the data source and model decision-making logic at each step.

[0077] S52: Based on natural language generation technology, text conversion is performed on various features and their contribution values ​​to generate pricing text data.

[0078] In this step, the feature contribution values ​​and their corresponding feature names output by the SHAP algorithm are input into the natural language generation model. The model converts the numerical contribution values ​​into natural language descriptions that are easy for users to understand through preset templates and semantic rules. For example, the car has only 30,000 kilometers on the odometer, which is lower than the average level of cars of the same age, thus increasing the price by about 5,000 yuan. The car has a front bumper replacement record, thus reducing the price by about 2,000 yuan.

[0079] S53: Integrate the predicted price range, pricing text data, market pricing adjustment strategies, and risk assessment results to obtain a price assessment report.

[0080] In this step, the predicted price range, the pricing text data output by the natural language generation technology, the market pricing adjustment strategy output by the reinforcement learning sub-model, and the risk assessment results output by the preset risk assessment model are integrated to form a complete price assessment report.

[0081] Optionally, the price assessment report also includes visual content, such as a feature influence waterfall chart generated based on the SHAP algorithm, a vehicle body damage annotation chart generated based on the instance segmentation model, and a market trend prediction chart generated based on the long short-term memory network model. By combining text and graphics, the report provides users with comprehensive, intuitive, and understandable assessment information, effectively improving the transparency and decision-making efficiency of used car transactions.

[0082] As can be seen, the above solution first collects full lifecycle data and market data of used cars, extracts features from each, and uses a pre-set price prediction model to uncover the non-linear relationship between features and vehicle value, outputting an objective basic valuation. Simultaneously, it simulates the market transaction environment to dynamically adjust pricing strategies, achieving intelligent pricing that coordinates vehicle and market dynamics. Next, a pre-set risk assessment model integrates multi-source feature data to identify potential risks such as accident-damaged vehicles, flood-damaged vehicles, and vehicles with tampered odometers from multiple dimensions, providing comprehensive risk warnings for both parties in the transaction. Finally, the predicted price range, market pricing adjustment strategies, and risk assessment results are integrated to generate an intuitive and easy-to-understand price assessment report. Through automated data processing and intelligent assessment models, the automatic and effective integration of multi-source heterogeneous data is achieved. Combined with risk identification capabilities, this significantly improves the accuracy and efficiency of price assessment, enhancing customer satisfaction and market competitiveness while effectively reducing labor costs.

[0083] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0084] In one embodiment, a vehicle price assessment device is provided, which corresponds one-to-one with the vehicle price assessment method described in the above embodiments. For example... Figure 4 As shown, the vehicle price evaluation device includes an acquisition module 101, a feature extraction module 102, a first generation module 103, a second generation module 104, and a third generation module 105. Detailed descriptions of each functional module are as follows: The acquisition module 101 is used to acquire the full life cycle data and market data of the target vehicle. The full life cycle data includes at least one of the following: basic vehicle information, vehicle exterior image data, historical operation data and historical event data. The historical event data includes at least one of the following: historical maintenance event data, historical accident event data, historical transfer event data and historical maintenance event data. Feature extraction module 102 is used to extract features from full lifecycle data and market data respectively to obtain vehicle feature data and market trend feature data; The first generation module 103 is used to generate the predicted price range of the target vehicle and the market pricing adjustment strategy based on vehicle feature data, market trend feature data and a preset price prediction model. The second generation module 104 is used to generate risk assessment results for the target vehicle based on vehicle feature data and a preset risk assessment model. The third generation module 105 is used to generate a price assessment report for the target vehicle based on the predicted price range, market pricing adjustment strategy, and risk assessment results.

[0085] In one embodiment, the vehicle feature data includes at least one of the following: basic attribute features, vehicle damage features, mechanical performance features, usage status features, driving behavior features, and historical event features. The feature extraction module 102 is specifically used for: Feature extraction is performed on the basic information of the vehicle to obtain the basic attribute features of the target vehicle; Based on a pre-defined instance segmentation model, feature extraction is performed on vehicle exterior image data to obtain vehicle damage features of the target vehicle. Feature extraction is performed on historical operating data to obtain the mechanical performance characteristics, usage status characteristics, and driving behavior characteristics of the target vehicle; By extracting features from historical event data using a pre-defined semantic analysis model, the historical event features of the target vehicle are obtained. These historical event features include at least one of the following: maintenance features, accident features, ownership transfer features, and upkeep features. Based on basic vehicle information, features are extracted from market data to obtain market trend characteristics.

[0086] In one embodiment, the feature extraction module 102 is further configured to: Acquire a set of vehicle exterior damage images, which includes vehicle exterior damage images under different lighting conditions and shooting angles; Pixel-level annotations are performed on the damaged areas in the vehicle exterior damage image set to generate annotation data containing damage type labels and damage area contour masks. The labeled data is used as training data to train the Mask R-CNN model, resulting in a preset instance segmentation model; Image enhancement processing is performed on vehicle exterior image data; The enhanced vehicle exterior image data is input into a preset instance segmentation model to detect and segment vehicle damage, and outputs the location coordinates, damage area, damage type and damage severity score of the damaged area. Based on the location coordinates of the damaged area, the damaged area, the damaged type, and the damage severity score, the vehicle damage characteristics of the target vehicle are obtained.

[0087] In one embodiment, the feature extraction module 102 is further configured to: Obtain the vehicle event dataset, which includes repair records, insurance claim records, and maintenance records for multiple vehicles; Label the vehicle event dataset to generate labeled data containing event type, time information, involved parts, operation type, and amount involved; The labeled vehicle event dataset is used as training data to train the BERT model, resulting in a pre-defined semantic analysis model. Extract historical maintenance data, historical insurance claim data, historical ownership transfer data, and historical maintenance data of the target vehicle from historical event data; Historical maintenance data is input into a preset semantic analysis model to obtain the maintenance characteristics of the target vehicle; Historical insurance claims data are input into a pre-defined semantic analysis model to obtain the accident characteristics of the target vehicle; Feature extraction is performed on historical transfer data to obtain the transfer characteristics of the target vehicle; Historical maintenance data is input into a preset semantic analysis model to obtain the maintenance characteristics of the target vehicle.

[0088] In one embodiment, the preset price prediction model includes a deep neural network sub-model and a reinforcement learning sub-model. The first generation module 103 is specifically used for: Vehicle feature data and market trend feature data are input into a deep neural network sub-model. The deep neural network sub-model extracts the mapping relationship between feature data and vehicle value, and outputs the predicted price range of the target vehicle. The predicted price range and market data are input into the reinforcement learning sub-model. The reinforcement learning sub-model simulates the current market trading environment and outputs the market pricing adjustment strategy for the target vehicle with the goal of maximizing the actual transaction probability.

[0089] In one embodiment, the acquisition module 101 is further configured to acquire full lifecycle data, historical transaction prices, vehicle risk levels, and market data of multiple traded vehicles; The feature extraction module 102 is also used to extract features from the full lifecycle data and market data to obtain a feature dataset for each traded vehicle.

[0090] In one embodiment, the device further includes: The model training module is used to train the initial deep neural network sub-model based on the feature dataset and historical transaction prices. The goal is to minimize the difference between the predicted price and the historical transaction price. The model parameters are iteratively updated to obtain the trained deep neural network sub-model. The model training module is also used to train the initial reinforcement learning sub-model based on the basic price range, feature dataset and simulated market trading environment output by the trained deep neural network sub-model, and iteratively update the model parameters with the goal of maximizing the actual transaction probability, so as to obtain the trained reinforcement learning sub-model. The model training module is also used to train the initial deep neural network based on the vehicle feature data and vehicle risk level in the feature dataset, with the goal of minimizing the difference between the predicted risk level and the vehicle risk level, and iteratively update the model parameters to obtain the trained preset risk assessment model.

[0091] In one embodiment, the third generation module 105 is specifically used for: Input vehicle feature data, market trend feature data, and predicted price range into the SHAP algorithm to calculate the feature contribution value of each feature to the predicted price. Based on natural language generation technology, text conversion is performed on various features and their contribution values ​​to generate pricing text data; The predicted price range, pricing text data, market pricing adjustment strategies, and risk assessment results are integrated to obtain a price assessment report.

[0092] This invention provides a vehicle price assessment device. First, it collects full lifecycle data and market data of used cars, extracting features from each. A pre-set price prediction model is then used to uncover the non-linear relationship between these features and vehicle value, outputting an objective basic valuation. Simultaneously, it simulates the market transaction environment to dynamically adjust pricing strategies, achieving intelligent pricing that coordinates vehicle and market dynamics. Next, a pre-set risk assessment model integrates multi-source feature data to identify potential risks such as accident-damaged vehicles, flood-damaged vehicles, and vehicles with tampered odometers from multiple dimensions, providing comprehensive risk warnings to both parties in the transaction. Finally, the predicted price range, market pricing adjustment strategies, and risk assessment results are integrated to generate an intuitive and easy-to-understand price assessment report. Through automated data processing and intelligent assessment models, the device achieves automatic and effective integration of multi-source heterogeneous data. Combined with risk identification capabilities, it significantly improves the accuracy and efficiency of price assessments, enhancing customer satisfaction and market competitiveness while effectively reducing labor costs.

[0093] Specific limitations regarding the vehicle price assessment device can be found in the limitations of the vehicle price assessment method described above, and will not be repeated here. Each module in the aforementioned vehicle price assessment device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0094] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 5As shown. The computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external clients via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a vehicle price assessment method on the server side.

[0095] In one embodiment, a computer device is provided, which may be a client, and its internal structure diagram may be as follows: Figure 6 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with an external server via a network connection. When executed by the processor, the computer program implements the functions or steps of a vehicle price assessment method on the client side.

[0096] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: Acquire the full lifecycle data and market data of the target vehicle. The full lifecycle data includes at least one of the following: basic vehicle information, vehicle exterior image data, historical operation data, and historical event data. The historical event data includes at least one of the following: historical maintenance event data, historical accident event data, historical transfer event data, and historical maintenance event data. Feature extraction is performed on the full lifecycle data and market data to obtain vehicle feature data and market trend feature data; Based on vehicle characteristic data, market trend characteristic data, and a preset price prediction model, the predicted price range for the target vehicle and the market pricing adjustment strategy are generated. Based on vehicle characteristic data and a pre-set risk assessment model, generate risk assessment results for the target vehicle; Based on the predicted price range, market pricing adjustment strategies, and risk assessment results, a price assessment report for the target vehicle is generated.

[0097] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: Acquire the full lifecycle data and market data of the target vehicle. The full lifecycle data includes at least one of the following: basic vehicle information, vehicle exterior image data, historical operation data, and historical event data. The historical event data includes at least one of the following: historical maintenance event data, historical accident event data, historical transfer event data, and historical maintenance event data. Feature extraction is performed on the full lifecycle data and market data to obtain vehicle feature data and market trend feature data; Based on vehicle characteristic data, market trend characteristic data, and a preset price prediction model, the predicted price range for the target vehicle and the market pricing adjustment strategy are generated. Based on vehicle characteristic data and a pre-set risk assessment model, generate risk assessment results for the target vehicle; Based on the predicted price range, market pricing adjustment strategies, and risk assessment results, a price assessment report for the target vehicle is generated.

[0098] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0099] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0100] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0101] It should be noted that any neural network models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.

[0102] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for evaluating vehicle prices, characterized in that, include: Acquire the full lifecycle data and market data of the target vehicle. The full lifecycle data includes at least one of the following: basic vehicle information, vehicle exterior image data, historical operation data, and historical event data. The historical event data includes at least one of the following: historical maintenance event data, historical accident event data, historical transfer event data, and historical maintenance event data. Feature extraction is performed on the full lifecycle data and the market data to obtain vehicle feature data and market trend feature data; Based on the vehicle characteristic data, the market trend characteristic data, and the preset price prediction model, a predicted price range for the target vehicle and a market pricing adjustment strategy are generated. Based on the vehicle feature data and the preset risk assessment model, a risk assessment result for the target vehicle is generated; Based on the predicted price range, the market pricing adjustment strategy, and the risk assessment results, a price assessment report for the target vehicle is generated.

2. The vehicle price assessment method as described in claim 1, characterized in that, The vehicle feature data includes at least one of the following: basic attribute features, vehicle damage features, mechanical performance features, usage status features, driving behavior features, and historical event features. Feature extraction is performed on the full lifecycle data and the market trend data to obtain vehicle feature data and market trend feature data, including: Feature extraction is performed on the basic information of the vehicle to obtain the basic attribute features of the target vehicle; Based on a preset instance segmentation model, feature extraction is performed on the vehicle exterior image data to obtain the vehicle damage features of the target vehicle. Feature extraction is performed on the historical operating data to obtain the mechanical performance characteristics, usage status characteristics, and driving behavior characteristics of the target vehicle; By extracting features from the historical event data using a preset semantic analysis model, the historical event features of the target vehicle are obtained. The historical event features include at least one of the following: maintenance features, accident features, transfer features, and upkeep features. Based on the vehicle's basic information, feature extraction is performed on the market data to obtain the market trend features.

3. The vehicle price assessment method as described in claim 2, characterized in that, The step of extracting features from the vehicle exterior image data based on a preset instance segmentation model to obtain the vehicle damage features of the target vehicle includes: Acquire a set of vehicle exterior damage images, wherein the set of vehicle exterior damage images includes vehicle exterior damage images under different lighting conditions and different shooting angles; The damaged areas in the vehicle exterior damage image set are annotated at the pixel level to generate annotation data containing damage type labels and damage area contour masks. The labeled data is used as training data to train the Mask R-CNN model, thereby obtaining the preset instance segmentation model; Image enhancement processing is performed on vehicle exterior image data; The enhanced vehicle exterior image data is input into the preset instance segmentation model to detect and segment vehicle damage, and outputs the location coordinates, damage area, damage type and damage severity score of the damaged area. The vehicle damage characteristics of the target vehicle are obtained based on the location coordinates of the damaged area, the damaged area, the damaged type, and the damage severity score.

4. The vehicle price assessment method as described in claim 2, characterized in that, The step of extracting features from the historical event data using a preset semantic analysis model to obtain the historical event features of the target vehicle includes: Obtain a vehicle event dataset, which includes repair records, insurance claim records, and maintenance records for multiple vehicles; The vehicle event dataset is labeled to generate labeled data containing event type, time information, involved components, operation type, and amount involved; The labeled vehicle event dataset is used as training data to train the BERT model, thereby obtaining the preset semantic analysis model; Extract historical maintenance data, historical insurance claim data, historical ownership transfer data, and historical maintenance data of the target vehicle from historical event data; The historical maintenance data is input into the preset semantic analysis model to obtain the maintenance characteristics of the target vehicle; The historical insurance claims data is input into the preset semantic analysis model to obtain the accident characteristics of the target vehicle; Feature extraction is performed on the historical transfer data to obtain the transfer features of the target vehicle; The historical maintenance data is input into the preset semantic analysis model to obtain the maintenance characteristics of the target vehicle.

5. The vehicle price assessment method as described in claim 1, characterized in that, The preset price prediction model includes a deep neural network sub-model and a reinforcement learning sub-model. The step of generating a predicted price range for the target vehicle and a market pricing adjustment strategy based on the vehicle feature data, the market trend feature data, and the preset price prediction model includes: The vehicle feature data and the market trend feature data are input into the deep neural network sub-model. The deep neural network sub-model extracts the mapping relationship between the feature data and the vehicle value, and outputs the predicted price range of the target vehicle. The predicted price range and the market data are input into the reinforcement learning sub-model. The reinforcement learning sub-model simulates the current market trading environment and outputs the market pricing adjustment strategy for the target vehicle with the goal of maximizing the actual transaction probability.

6. The vehicle price assessment method as described in claim 1, characterized in that, Before generating the predicted price range and market pricing adjustment strategy for the target vehicle based on the vehicle characteristic data, the market trend characteristic data, and the preset price prediction model, the method further includes: Obtain full lifecycle data, historical transaction prices, vehicle risk levels, and market conditions for multiple traded vehicles; Feature extraction is performed on full lifecycle data and market data to obtain a feature dataset for each traded vehicle. Based on the feature dataset and the historical transaction prices, the initial deep neural network sub-model is trained with the goal of minimizing the difference between the predicted price and the historical transaction price. The model parameters are iteratively updated to obtain the trained deep neural network sub-model. Based on the base price range output by the trained deep neural network sub-model, the feature dataset, and the simulated market trading environment, the initial reinforcement learning sub-model is trained. With the goal of maximizing the actual transaction probability, the model parameters are iteratively updated to obtain the trained reinforcement learning sub-model. Based on the vehicle feature data and the vehicle risk level in the feature dataset, an initial deep neural network is trained to minimize the difference between the predicted risk level and the vehicle risk level. The model parameters are iteratively updated to obtain a trained preset risk assessment model.

7. The vehicle price assessment method as described in claim 1, characterized in that, The process of generating a price assessment report for the target vehicle based on the predicted price range, the market pricing adjustment strategy, and the risk assessment results includes: The vehicle feature data, the market trend feature data, and the predicted price range are input into the SHAP algorithm to calculate the feature contribution value of each feature to the predicted price. Based on natural language generation technology, text conversion is performed on various features and their contribution values ​​to generate pricing text data; The predicted price range, the pricing text data, the market pricing adjustment strategy, and the risk assessment results are integrated to obtain the price assessment report.

8. A vehicle price evaluation device, characterized in that, include: The acquisition module is used to acquire the full life cycle data and market information data of the target vehicle. The full life cycle data includes at least one of the following: basic vehicle information, vehicle exterior image data, historical operation data, and historical event data. The historical event data includes at least one of the following: historical maintenance event data, historical accident event data, historical transfer event data, and historical maintenance event data. The feature extraction module is used to extract features from the full lifecycle data and the market data respectively to obtain vehicle feature data and market trend feature data; The first generation module is used to generate the predicted price range of the target vehicle and the market pricing adjustment strategy based on the vehicle feature data, the market trend feature data and the preset price prediction model. The second generation module is used to generate the risk assessment result of the target vehicle based on the vehicle feature data and the preset risk assessment model; The third generation module is used to generate a price assessment report for the target vehicle based on the predicted price range, the market pricing adjustment strategy, and the risk assessment results.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the vehicle price assessment method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the vehicle price assessment method as described in any one of claims 1 to 7.

Images