A closed-loop data processing method and system for the entire vehicle integrated business process

Abstract

This application provides a closed-loop data processing method and system for the entire process of comprehensive vehicle business. The method includes: acquiring comprehensive vehicle business requests and identifying business type and service requirement information; matching with a preset quotation rule base to generate scheme quotation data and obtain scheme verification status; dispatching vehicle pickup and extracting baseline feature vectors from initial multimodal vehicle status data; collecting post-construction multimodal data to extract post-construction feature vectors, calculating the post-construction comprehensive similarity with the baseline feature vectors, and outputting the post-construction verification result; dispatching vehicle delivery after payment, calculating the delivery feature vector and the post-construction feature vector's comprehensive similarity at the delivery stage, and outputting the delivery verification result; executing vehicle confirmation and triggering a feedback process based on the delivery verification result; thereby achieving closed-loop verification and risk control throughout the entire process through weighted summation of multimodal dimension similarities and adaptive verification parameters based on business type.

Description

Technical Field

[0001] The present application relates to the field of vehicle business data processing, and specifically, to a closed-loop data processing method and system for the entire process of comprehensive vehicle services. Background Art

[0002] With the continuous growth of car ownership, the market scale of comprehensive vehicle service businesses (including maintenance, repair, beauty treatment, inspection, etc.) has been expanding continuously. In comprehensive vehicle service businesses, from the customer initiating a service request to the final delivery of the vehicle, it involves multiple stages such as quotation, vehicle pickup, construction, payment, vehicle delivery and delivery feedback, and there are close business logic correlations and data dependencies between each stage. How to ensure the accuracy and consistency of service data at each stage and realize closed-loop business control of the whole process has become a key issue in the management of comprehensive vehicle service businesses.

[0003] The invention patent CN110852290A discloses a large-scale vehicle feature vector similarity comparison method, which improves the calculation speed of large-scale vehicle feature vector similarity comparison by replacing pairwise cosine calculation with matrix multiplication calculation on vehicle feature vectors. However, this method only focuses on the optimization of similarity calculation efficiency between vehicle feature vectors, does not combine the comparison of vehicle feature similarity with the verification requirements of specific business stages, lacks a multi-modal data dimension decomposition and business type adaptive weight configuration mechanism, and cannot assign differentiated weights to similarities of each dimension according to different business types (such as appearance-related businesses and non-appearance-related businesses), which results in the verification result being out of touch with the actual business needs.

[0004] The invention patent CN116071082A discloses a vehicle after-sales service system, method, device, equipment and storage medium. The system jointly processes after-sales requests through a terminal application, a service management center subsystem, a vehicle position data management subsystem and a data visualization subsystem, and realizes the scheduling of after-sales requests and vehicle position tracking. However, this system only focuses on the scheduling management of after-sales requests and vehicle position tracking, lacks a multi-modal data verification mechanism for state changes of the vehicle before and after construction, and fails to achieve closed-loop verification through feature vector similarity comparison at key stages such as vehicle pickup, after construction and vehicle delivery; meanwhile, the system also fails to establish a risk assessment and hierarchical exception handling mechanism based on the degree of similarity deviation, as well as an automatic complaint escalation and red mark elimination closed-loop mechanism, which leads to delayed exception discovery in business processes, low complaint handling efficiency, and lack of linkage update between evaluation feedback and business parameters.

[0005] In summary, the existing technologies have the following shortcomings in the closed-loop data processing of the entire vehicle business process: (1) lack of dimensional decomposition of multimodal vehicle status data and adaptive verification parameter configuration for business types; (2) lack of closed-loop verification mechanism based on feature vector similarity for each stage of vehicle pickup, post-construction and delivery; (3) lack of risk assessment and graded anomaly handling mechanism based on similarity deviation; (4) lack of automatic escalation of complaint priority and red-out closed-loop mechanism; (5) lack of linkage update mechanism for evaluation feedback and preferential rights and construction personnel task allocation weight.

[0006] Effective technical solutions are urgently needed to address the above problems. Summary of the Invention

[0007] The purpose of this application is to provide a closed-loop data processing method and system for the entire process of vehicle integrated business. This method involves: acquiring vehicle integrated business requests and identifying business types and service requirements; matching these requests with a preset quotation rule base to generate scheme quotation data and obtain scheme verification status; dispatching vehicle pickup and extracting baseline feature vectors from initial multimodal vehicle status data; collecting post-construction multimodal data to extract post-construction feature vectors, calculating the post-construction comprehensive similarity with the baseline feature vectors, and outputting the post-construction verification result; dispatching vehicle delivery after payment, calculating the post-delivery feature vector and the post-construction feature vector's comprehensive similarity at the delivery stage, and outputting the delivery verification result; executing vehicle confirmation and triggering a feedback process based on the delivery verification result; and thus achieving closed-loop verification and risk control throughout the entire process through weighted summation of multimodal dimension similarities and adaptive verification parameters based on business type.

[0008] This application also provides a closed-loop data processing method for the entire vehicle integrated business process, including the following steps:

[0009] Obtain comprehensive vehicle service requests and identify service types and service requirements.

[0010] Based on the business type and service requirements information, the system matches the pre-set quotation rule library to generate a solution quotation data, pushes it to the customer's terminal, and obtains the solution verification status.

[0011] According to the scheme, the vehicle retrieval is scheduled based on the verification status. The initial multimodal vehicle status data of the vehicle retrieval stage is obtained, and the initial feature vector is extracted and stored as the baseline feature vector.

[0012] Acquire construction status data, collect post-construction multimodal vehicle status data based on construction status and extract post-construction feature vectors, calculate the post-construction comprehensive similarity between the feature vectors and the baseline feature vectors, and output the post-construction verification results in combination with the preset construction similarity threshold corresponding to the business type.

[0013] Based on the verification results, a payment request is sent to the customer terminal and the payment status is confirmed. The vehicle is dispatched according to the payment status. Multimodal vehicle status data during the delivery stage is collected and the delivery feature vector is extracted. The comprehensive similarity between the feature vector and the delivery stage feature vector after construction is calculated. The delivery verification result is output in combination with the preset delivery similarity threshold corresponding to the business type.

[0014] Based on the vehicle delivery verification results, vehicle delivery confirmation will be executed and a feedback process will be triggered.

[0015] Optionally, in the closed-loop data processing method for the entire vehicle integrated business process described in this application, the step of matching the business type and service requirement information with a preset quotation rule base, generating scheme quotation data, pushing it to the customer terminal, and obtaining the scheme verification status specifically includes:

[0016] According to the business type, a corresponding quotation template is matched from the preset quotation rule library. The quotation template includes basic service items, optional additional service items and corresponding fee calculation rules.

[0017] Based on the vehicle identification information in the service request information, obtain the vehicle's historical service data, including historical maintenance records, historical repair records, and historical evaluation data;

[0018] Input service demand information and historical service data into the pricing calculation model to generate a main service price and a recommended price for additional services.

[0019] The main service quote and the recommended additional service quotes are combined to generate the solution quote data, which includes cost details and estimated construction period;

[0020] The status of the solution verification is obtained based on the customer's input, including whether they agree or disagree.

[0021] Optionally, in the closed-loop data processing method for the entire vehicle integrated business process described in this application, the step of scheduling vehicle retrieval according to the scheme verification status, obtaining initial multimodal vehicle status data for the retrieval stage, and extracting initial feature vectors and storing them as baseline feature vectors specifically includes:

[0022] If the verification status of the plan is confirmed and agreed, the vehicle will be dispatched and the initial multimodal vehicle status data will be obtained, including the initial image data of the vehicle's outer ring inspection, the initial mileage of the vehicle, and the initial photos of the front and rear of the vehicle.

[0023] Keyframes are extracted from the initial image data of the vehicle's outer ring inspection, and appearance features are calculated from the keyframes to generate the initial feature subvector of the outer ring image.

[0024] The initial mileage of the vehicle is numerically normalized to generate an initial mileage feature sub-vector.

[0025] Interior features are calculated from the initial front and rear photos of the vehicle interior to generate initial feature sub-vectors for the vehicle interior.

[0026] The initial feature vectors of the outer ring image, the initial feature vector of the mileage, and the initial feature vector of the vehicle interior are combined according to a preset dimension to generate an initial feature vector and store it as a baseline feature vector.

[0027] Optionally, in the closed-loop data processing method for the entire vehicle integrated business process described in this application, the steps of acquiring construction status data, collecting post-construction multimodal vehicle status data based on the construction status and extracting post-construction feature vectors, calculating the post-construction comprehensive similarity between the feature vectors and the baseline feature vectors, and outputting the post-construction verification result in combination with a preset construction similarity threshold corresponding to the business type, specifically include:

[0028] Obtain construction status data, including whether construction is completed or not;

[0029] If the construction status data indicates that construction is completed, then obtain the post-construction multimodal vehicle status data and extract the post-construction feature vector.

[0030] The post-construction feature vector and the baseline feature vector are respectively decomposed into outer circle image dimension sub-vector, mileage dimension sub-vector and in-vehicle image dimension sub-vector;

[0031] Calculate the similarity in the outer ring image dimension, the similarity in the mileage dimension, and the similarity in the in-vehicle image dimension respectively;

[0032] Based on the preset dimension weights corresponding to the business type, the similarity of each dimension is weighted and summed to obtain the comprehensive similarity after construction.

[0033] The post-construction comprehensive similarity is compared with the preset construction similarity threshold corresponding to the business type to obtain the post-construction verification result;

[0034] When the overall similarity after construction is greater than or equal to the preset construction similarity threshold, the post-construction verification result is determined to be passed; otherwise, the post-construction verification is deemed to have failed.

[0035] Optionally, in the closed-loop data processing method for the entire vehicle integrated business process described in this application, the steps of sending a payment request to the customer terminal based on the verification result and confirming the payment status, scheduling vehicle delivery based on the payment status, collecting multimodal vehicle status data during the delivery stage and extracting delivery feature vectors, calculating the comprehensive similarity between the feature vectors and the delivery stage feature vectors after construction, and outputting the delivery verification result in combination with the preset delivery similarity threshold corresponding to the business type, specifically include:

[0036] If the post-construction verification result is that the post-construction verification passed, a payment request will be sent to the customer terminal.

[0037] Obtain the user's payment status, including whether payment is completed or not. If the payment status is completed, dispatch a vehicle.

[0038] Collect multimodal vehicle status data during the vehicle delivery phase and extract the vehicle delivery feature vector. Decompose the vehicle delivery feature vector and the post-construction feature vector into dimensions respectively, and calculate the similarity of the outer ring image dimension, the similarity of the mileage dimension, and the similarity of the in-vehicle image dimension respectively.

[0039] Based on the preset dimension weights of the vehicle delivery stage corresponding to the business type, the similarity of the outer circle image dimension, the similarity of the mileage dimension, and the similarity of the in-vehicle image dimension are weighted and summed to obtain the comprehensive similarity of the vehicle delivery stage.

[0040] The overall similarity of the vehicle delivery stage is compared with the preset vehicle delivery similarity threshold corresponding to the business type;

[0041] When the overall similarity during the vehicle delivery stage is greater than or equal to the preset vehicle delivery similarity threshold, the vehicle delivery verification result is determined to be successful; otherwise, the vehicle delivery verification is deemed unsuccessful.

[0042] Optionally, in the closed-loop data processing method for the entire vehicle integrated business process described in this application, the method further includes a business type adaptive verification parameter configuration step:

[0043] Based on the business type, retrieve the corresponding construction dimension weight configuration and construction similarity threshold configuration from the preset business verification parameter library;

[0044] The dimension weight configuration defines the weight coefficient of each dimension sub-vector in the weighted summation, and different business types correspond to different weight coefficient allocations.

[0045] When the business type is appearance-related, the weight coefficient of the outer ring image dimension sub-vector is higher than the weight coefficient of other dimension sub-vectors;

[0046] When the business type is not related to appearance, the weight coefficient of the mileage dimension sub-vector is higher than the weight coefficient of other dimension sub-vectors.

[0047] Optionally, the closed-loop data processing method for the entire vehicle integrated business process described in this application also includes anomaly handling based on risk assessment, specifically including:

[0048] When the post-construction verification result is that the post-construction verification fails, the construction risk score is calculated based on the degree of similarity deviation between the post-construction feature vector and the baseline feature vector.

[0049] When the construction risk score is in the preset high-risk range, construction anomaly alarm data is generated and pushed to the preset management personnel terminal to wait for anomaly handling instructions.

[0050] When the construction risk score is in the preset medium risk range, a construction review request is generated, and the post-construction multimodal vehicle status data is pushed to the preset review terminal to wait for manual review instructions.

[0051] When the vehicle delivery verification result is that the vehicle delivery verification fails, the vehicle delivery risk score is calculated based on the degree of similarity deviation between the vehicle delivery feature vector and the post-construction feature vector, and the corresponding risk level handling strategy is executed according to the risk range in which the vehicle delivery risk score is located.

[0052] Secondly, this application provides a closed-loop data processing system for the entire process of integrated vehicle business. The system includes a memory and a processor. The memory stores a program for a closed-loop data processing method for the entire process of integrated vehicle business. When the program for the closed-loop data processing method for the entire process of integrated vehicle business is executed by the processor, it implements the following steps:

[0053] Obtain comprehensive vehicle service requests and identify service types and service requirements.

[0054] Based on the business type and service requirements information, the system matches the pre-set quotation rule library to generate a solution quotation data, pushes it to the customer's terminal, and obtains the solution verification status.

[0055] According to the scheme, the vehicle retrieval is scheduled based on the verification status. The initial multimodal vehicle status data of the vehicle retrieval stage is obtained, and the initial feature vector is extracted and stored as the baseline feature vector.

[0056] Acquire construction status data, collect post-construction multimodal vehicle status data based on construction status and extract post-construction feature vectors, calculate the post-construction comprehensive similarity between the feature vectors and the baseline feature vectors, and output the post-construction verification results in combination with the preset construction similarity threshold corresponding to the business type.

[0057] Based on the verification results, a payment request is sent to the customer terminal and the payment status is confirmed. The vehicle is dispatched according to the payment status. Multimodal vehicle status data during the delivery stage is collected and the delivery feature vector is extracted. The comprehensive similarity between the feature vector and the delivery stage feature vector after construction is calculated. The delivery verification result is output in combination with the preset delivery similarity threshold corresponding to the business type.

[0058] Based on the vehicle delivery verification results, vehicle delivery confirmation will be executed and a feedback process will be triggered.

[0059] Optionally, in the closed-loop data processing system for the entire vehicle integrated business process described in this application, the step of matching business type and service requirement information with a preset quotation rule base, generating scheme quotation data, pushing it to the customer terminal, and obtaining the scheme verification status specifically includes:

[0060] According to the business type, a corresponding quotation template is matched from the preset quotation rule library. The quotation template includes basic service items, optional additional service items and corresponding fee calculation rules.

[0061] Based on the vehicle identification information in the service request information, obtain the vehicle's historical service data, including historical maintenance records, historical repair records, and historical evaluation data;

[0062] Input service demand information and historical service data into the pricing calculation model to generate a main service price and a recommended price for additional services.

[0063] The main service quote and the recommended additional service quotes are combined to generate the solution quote data, which includes cost details and estimated construction period;

[0064] The status of the solution verification is obtained based on the customer's input, including whether they agree or disagree.

[0065] Optionally, in the vehicle integrated business closed-loop data processing system described in this application, the step of scheduling vehicle retrieval according to the scheme verification status, obtaining initial multimodal vehicle status data during the retrieval stage, and extracting initial feature vectors and storing them as baseline feature vectors specifically includes:

[0066] If the verification status of the plan is confirmed and agreed, the vehicle will be dispatched and the initial multimodal vehicle status data will be obtained, including the initial image data of the vehicle's outer ring inspection, the initial mileage of the vehicle, and the initial photos of the front and rear of the vehicle.

[0067] Keyframes are extracted from the initial image data of the vehicle's outer ring inspection, and appearance features are calculated from the keyframes to generate the initial feature subvector of the outer ring image.

[0068] The initial mileage of the vehicle is numerically normalized to generate an initial mileage feature sub-vector.

[0069] Interior features are calculated from the initial front and rear photos of the vehicle interior to generate initial feature sub-vectors for the vehicle interior.

[0070] The initial feature vectors of the outer ring image, the initial feature vector of the mileage, and the initial feature vector of the vehicle interior are combined according to a preset dimension to generate an initial feature vector and store it as a baseline feature vector.

[0071] As can be seen from the above, the vehicle integrated business closed-loop data processing method and system provided in this application obtain vehicle integrated business requests and identify business types and service demand information; match with a preset quotation rule base to generate scheme quotation data and obtain scheme verification status; dispatch vehicle pickup and extract baseline feature vectors of initial multimodal vehicle status data; collect post-construction multimodal data to extract post-construction feature vectors, calculate the post-construction comprehensive similarity with the baseline feature vectors and output the post-construction verification result; dispatch vehicle delivery after payment, calculate the post-delivery feature vector and the post-construction feature vector of the delivery stage comprehensive similarity and output the delivery verification result; execute vehicle confirmation and trigger feedback process based on the delivery verification result; thereby, through multimodal dimension similarity weighted summation and business type adaptive verification parameters, the entire process closed-loop verification and risk control are realized.

[0072] Other features and advantages of this application will be set forth in the following description and will be apparent in part from the description or may be learned by practicing embodiments of this application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings. Attached Figure Description

[0073] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0074] Figure 1 A flowchart of the closed-loop data processing method for the entire vehicle integrated business process provided in this application embodiment;

[0075] Figure 2 A flowchart illustrating the quotation rule matching process of the closed-loop data processing method for the entire vehicle integrated business process provided in this application embodiment;

[0076] Figure 3 A detailed flowchart of the comprehensive similarity calculation for the closed-loop data processing method for the entire vehicle business process provided in this application embodiment. Detailed Implementation

[0077] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0078] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0079] Please refer to Figure 1 , Figure 1 This is a flowchart of a closed-loop data processing method for the entire vehicle integrated business process according to some embodiments of this application. This closed-loop data processing method for the entire vehicle integrated business process is used in a terminal device. The closed-loop data processing method for the entire vehicle integrated business process includes the following steps:

[0080] S11. Obtain comprehensive vehicle service requests and identify service types and service requirements information;

[0081] S12. Based on the business type and service requirement information, match it with the preset quotation rule library, generate solution quotation data, push it to the customer terminal, and obtain the solution verification status.

[0082] S13. According to the scheme, verify the status of vehicle retrieval, obtain the initial multimodal vehicle status data of the vehicle retrieval stage, and extract the initial feature vector and store it as the baseline feature vector.

[0083] S14. Obtain construction status data, collect multimodal vehicle status data after construction based on construction status and extract post-construction feature vectors, calculate the post-construction comprehensive similarity between the feature vectors and the baseline feature vectors, and output the post-construction verification results in combination with the preset construction similarity threshold corresponding to the business type.

[0084] S15. Based on the verification result, send a payment request to the customer terminal and confirm the payment status. Based on the payment status, schedule the vehicle delivery. Collect multimodal vehicle status data during the delivery stage and extract the delivery feature vector. Calculate the comprehensive similarity between the feature vector and the delivery stage feature vector after construction. Combine the preset delivery similarity threshold corresponding to the business type to output the delivery verification result.

[0085] S16. Based on the vehicle delivery verification results, perform vehicle delivery confirmation and trigger the feedback process.

[0086] It should be emphasized that the core design concept of this application is to build a closed-loop data verification link for the entire process of vehicle integrated business, from quotation to delivery feedback. The system acquires business requests and identifies business types and service requirements, providing business context parameters for subsequent steps (business types guide parameter selection, and service requirements are used for quote matching). Based on the business type and service requirements, it matches the data with a pre-defined quote rule base, generating a quote that is pushed to the customer's terminal to obtain verification status. When the verification status is "Confirmed and Agreed," vehicle pickup scheduling is triggered. During the pickup phase, initial multimodal vehicle status data is collected, and a baseline feature vector is extracted. This baseline feature vector serves as the basis for similarity comparisons during post-construction and delivery phases. After construction is completed, post-construction multimodal vehicle status data is collected, and post-construction feature vectors are extracted. The post-construction comprehensive similarity to the baseline feature vector is calculated, and a verification result is output based on a pre-defined construction similarity threshold. When the post-construction verification result is "Passed," a payment request is triggered. After payment, vehicle delivery is scheduled, and delivery multimodal data is collected, delivery feature vectors are extracted, and the post-construction feature vector's delivery phase comprehensive similarity is calculated. This delivery phase similarity is then calculated, and a delivery verification result is output based on a pre-defined delivery similarity threshold. Based on the delivery verification result, vehicle confirmation is executed, and a feedback process (including complaint handling and evaluation feedback) is triggered, completing the entire closed-loop process.

[0087] "Vehicle Comprehensive Service Request" refers to a request data initiated by a customer through a customer terminal that includes vehicle service needs, including but not limited to maintenance requests, repair requests, detailing requests, and inspection requests; "Service Type" refers to the service category label determined based on the service content in the vehicle comprehensive service request, such as "Appearance-Related Services" (including detailing, painting, and other services involving changes to the vehicle's appearance) and "Non-Appearance-Related Services" (including maintenance, inspection, and other services that do not involve significant changes to the vehicle's appearance); "Service Request Information" refers to the specific service request parameters extracted from the vehicle comprehensive service request, including vehicle identification information, service item details, and customer expected work. The terms "multimodal vehicle status data" and "baseline feature vector" refer to vehicle status description data acquired from multiple data acquisition modalities, including vehicle outer ring inspection image data, vehicle mileage data, and front and rear interior photos. "Baseline feature vector" refers to the feature vector extracted and stored from the initial multimodal vehicle status data during the vehicle retrieval phase, used as a reference benchmark for feature vector similarity comparison in subsequent stages. "Post-construction comprehensive similarity" refers to the weighted similarity sum of the post-construction feature vector and the baseline feature vector across multiple dimensional sub-vectors. "Delivery phase comprehensive similarity" refers to the weighted similarity sum of the delivery feature vector and the post-construction feature vector across multiple dimensional sub-vectors. By collecting multimodal vehicle status data and extracting feature vectors at three key stages—vehicle retrieval, post-construction, and delivery—a three-stage feature vector comparison chain is formed, connecting the baseline feature vector to the post-construction feature vector and the delivery feature vector. This achieves closed-loop verification of status data at each key stage of the comprehensive vehicle business process, effectively ensuring the traceability and consistency of vehicle status changes at each stage, and solving the problem of the lack of a multi-stage closed-loop verification mechanism in existing technologies.

[0088] Please refer to Figure 2 , Figure 2 This application provides a flowchart of a pricing rule matching process for a closed-loop data processing method for the entire vehicle integrated business process. According to an embodiment of the invention, the step of matching business type and service requirement information with a preset pricing rule database, generating a scheme pricing data, pushing it to the customer terminal, and obtaining the scheme verification status specifically includes:

[0089] According to the business type, a corresponding quotation template is matched from the preset quotation rule library. The quotation template includes basic service items, optional additional service items and corresponding fee calculation rules.

[0090] Based on the vehicle identification information in the service request information, obtain the vehicle's historical service data, including historical maintenance records, historical repair records, and historical evaluation data;

[0091] Input service demand information and historical service data into the pricing calculation model to generate a main service price and a recommended price for additional services.

[0092] The main service quote and the recommended additional service quotes are combined to generate the solution quote data, which includes cost details and estimated construction period;

[0093] The status of the solution verification is obtained based on the customer's input, including whether they agree or disagree.

[0094] It is important to emphasize that, based on the business type, a quotation template is matched from a pre-defined quotation rule library. This template defines the services available for that business type and their cost calculation rules, providing a framework for quotation calculation. Historical service data for the vehicle is obtained based on vehicle identification information (historical maintenance records reflect vehicle maintenance cycles, historical repair records reflect vehicle fault tendencies, and historical evaluation data reflect customer satisfaction with past services), providing a personalized reference for quotation calculation. The obtained service demand information and historical service data are input into the quotation calculation model. The model then performs a comprehensive calculation based on cost calculation rules, historical service preferences, and customer satisfaction data to generate a quote. The system generates service quotes (costs for basic services) and recommended quotes for additional services (costs and reasons for recommendation for optional additional services). These quotes are combined into a single pricing plan, including a detailed breakdown of costs (each cost item listed) and an estimated construction period (construction time estimated based on service items and historical service data). The pricing plan is then sent to the customer's terminal. The customer selects and confirms or rejects the pricing plan based on their selection. The system obtains the plan verification status ("Confirm Agree" or "Disagree") based on the customer's input. A "Confirm Agree" status triggers subsequent vehicle dispatch; a "Disagree" status terminates the process or requires a new quote.

[0095] "Quotation template" refers to the quotation configuration data corresponding to each business type stored in the preset quotation rule base. It includes basic service items (service items that must be included in the business type and their base fees), optional supplementary service items (service items that customers can choose additionally and their fees), and fee calculation rules (rules defining how the fees for each service item are calculated based on vehicle information and service parameters). "Quotation calculation model" refers to the data processing model that calculates the main service quotation and the recommended supplementary service quotation based on the service item parameters in the service demand information and the customer preferences and vehicle characteristics in historical service data. This model calculates the main service quotation according to the fee calculation rules for basic service items and recommends and calculates the recommended supplementary service quotation for optional supplementary service items based on the customer's historical preferences. By introducing historical vehicle service data (historical maintenance records, historical repair records, and historical evaluation data) as a personalized reference for quotation calculation, and by extracting vehicle identification information from the service demand information, this application achieves personalized customization of the quotation scheme, making the quotation scheme fit the actual needs of the customer and the vehicle, and improving the relevance and customer acceptance of the quotation scheme.

[0096] In this embodiment, when the business type is "Appearance-related business" (painting service), and the vehicle identification information is Shanghai A-12345, a paint quote template is matched from the preset quote rule library: the basic service item is "full vehicle paint" (base price 2000 yuan), and the optional additional service items are "coating protection" (cost 500 yuan) and "partial paint touch-up" (cost 300 yuan / site). Historical service data is obtained: historical maintenance records show that the vehicle is maintained every 6 months, historical repair records show no major accident repairs, and historical evaluation data shows that the customer's rating for past detailing services is 4.2 out of 5. The quote calculation model generates a main service quote of 2000 yuan (base price for full vehicle paint) based on the above data, and recommends additional service quotes: "coating protection" is recommended at 500 yuan (a high-end additional service recommended based on the customer's historical high rating), and "partial paint touch-up" is optional at 300 yuan / site (based on historical repair records showing no major accidents, only preventative paint touch-up is recommended). Combined solution quotation data: Cost details are: full car painting 2000 yuan + coating protection 500 yuan = 2500 yuan, estimated construction period 3 days.

[0097] According to an embodiment of the present invention, the step of scheduling vehicle retrieval based on scheme verification status, obtaining initial multimodal vehicle state data during the vehicle retrieval phase, and extracting initial feature vectors and storing them as baseline feature vectors specifically includes:

[0098] If the verification status of the plan is confirmed and agreed, the vehicle will be dispatched and the initial multimodal vehicle status data will be obtained, including the initial image data of the vehicle's outer ring inspection, the initial mileage of the vehicle, and the initial photos of the front and rear of the vehicle.

[0099] Keyframes are extracted from the initial image data of the vehicle's outer ring inspection, and appearance features are calculated from the keyframes to generate the initial feature subvector of the outer ring image.

[0100] The initial mileage of the vehicle is numerically normalized to generate an initial mileage feature sub-vector.

[0101] Interior features are calculated from the initial front and rear photos of the vehicle interior to generate initial feature sub-vectors for the vehicle interior.

[0102] The initial feature vectors of the outer ring image, the initial feature vector of the mileage, and the initial feature vector of the vehicle interior are combined according to a preset dimension to generate an initial feature vector and store it as a baseline feature vector.

[0103] It is important to emphasize that vehicle retrieval dispatch is triggered when the scheme verification status is "confirmed and agreed". The retrieval personnel will proceed to the customer's designated location to pick up the vehicle according to the dispatch instructions. During retrieval, initial multimodal vehicle status data is collected through vehicle inspection equipment: initial outer-circle inspection image data (360-degree outer-circle video data captured by the outer-circle inspection camera), initial vehicle mileage (mileage values ​​read from the vehicle's dashboard), and initial interior photos (front and rear interior photos captured by the in-vehicle camera). Keyframe extraction is performed on the initial outer-circle inspection image data (representative frames are selected from the video at preset intervals or based on image change detection). For each keyframe, an appearance feature calculation algorithm (such as a deep learning-based vehicle appearance feature extraction network) is used to extract appearance features. The appearance features of each keyframe are aggregated (e.g., by averaging or stitching) to generate the initial feature sub-vector of the outer-circle image. ; Initial mileage of the vehicle The numerical values ​​are normalized using the following formula:

[0104] ;

[0105] in, To preset the lower bound value of mileage normalization, To preset the upper bound value of mileage normalization, The initial feature vector for mileage is generated (here, a one-dimensional feature value); interior features are calculated from the initial front and rear interior photos (e.g., using a deep learning-based interior feature extraction network) to generate the initial interior feature vector. The three sub-vectors are concatenated into an initial feature vector according to a preset dimension combination rule.

[0106] ,in For the initial feature sub-vectors of the outer circle image, For the initial feature subvector of mileage, The initial feature vector inside the vehicle, and Stored as a baseline feature vector, serving as a reference benchmark for similarity comparison during subsequent construction and vehicle delivery phases.

[0107] "Keyframe extraction" refers to the process of selecting representative frames from the vehicle's outer perimeter inspection image data (video sequence). Selection methods include uniformly extracting frames at preset time intervals or extracting frames at points of significant change based on the difference between adjacent frames. The number of keyframes and the selection strategy are configured according to actual deployment requirements. "Exterior feature calculation" refers to calculating the exterior feature vector from the keyframe images using feature extraction algorithms (such as a vehicle exterior feature extraction model based on convolutional neural networks). This feature vector encodes information such as the vehicle's texture, color, and shape. "Numerical normalization" refers to mapping mileage from its original numerical range to the [0,1] interval, eliminating the impact of different vehicle mileage magnitudes on similarity calculation. "Interior feature calculation" refers to calculating the interior feature vector from interior photos using feature extraction algorithms. This feature vector encodes information such as the vehicle's interior material, color, and cleanliness. "Preset dimension combination" refers to the configuration rules defining the splicing order and dimensional proportion of each sub-vector in the initial feature vector. The initial vehicle outer perimeter inspection image data refers to the data obtained by taking images around the vehicle.

[0108] In this embodiment, the initial multimodal vehicle state data collected during the vehicle retrieval phase includes: initial image data of the vehicle's outer perimeter inspection (360-degree outer perimeter video, approximately 2 minutes in length) and initial vehicle mileage. =56,000 km, initial interior and exterior photos (one photo each for the front and rear seats). Keyframe extraction was performed on the outer perimeter video, with one frame extracted every 10 seconds to obtain 12 keyframes. For each keyframe, an appearance feature computation network was used to extract a 256-dimensional appearance feature vector. The average of the appearance feature vectors from the 12 keyframes was then used to generate the initial feature sub-vectors for the outer perimeter image. (256 dimensions); Normalize the initial mileage, assuming a preset lower bound for mileage normalization. =0, upper bound value =200000, then =(56000-0) / (200000-0)=0.28 (1D); A 128-dimensional interior feature vector is extracted from the front and rear interior photos using an interior feature computation network. The average of the interior feature vectors from the front and rear photos is used to generate the initial interior feature sub-vector. (128 dimensions); Concatenate the three sub-vectors according to a preset dimension combination rule: The total dimensions are 385, and the data is stored as a baseline feature vector.

[0109] Please refer to Figure 3 , Figure 3This is a detailed flowchart of the comprehensive similarity calculation for the closed-loop data processing method for the entire vehicle integrated business process in some embodiments of this application. According to an embodiment of the present invention, the steps of acquiring construction status data, collecting post-construction multimodal vehicle status data based on the construction status and extracting post-construction feature vectors, calculating the post-construction comprehensive similarity between the feature vectors and the baseline feature vectors, and outputting the post-construction verification result in conjunction with a preset construction similarity threshold corresponding to the business type, specifically include:

[0110] Obtain construction status data, including whether construction is completed or not;

[0111] If the construction status data indicates that construction is completed, then obtain the post-construction multimodal vehicle status data and extract the post-construction feature vector.

[0112] The post-construction feature vector and the baseline feature vector are respectively decomposed into outer circle image dimension sub-vector, mileage dimension sub-vector and in-vehicle image dimension sub-vector;

[0113] Calculate the similarity in the outer ring image dimension, the similarity in the mileage dimension, and the similarity in the in-vehicle image dimension respectively;

[0114] Based on the preset dimension weights corresponding to the business type, the similarity of each dimension is weighted and summed to obtain the comprehensive similarity after construction.

[0115] The post-construction comprehensive similarity is compared with the preset construction similarity threshold corresponding to the business type to obtain the post-construction verification result;

[0116] When the overall similarity after construction is greater than or equal to the preset construction similarity threshold, the post-construction verification result is determined to be passed; otherwise, the post-construction verification is deemed to have failed.

[0117] It is important to emphasize that the construction status data is acquired by construction personnel who mark the work as "completed" after completion or "incomplete" during construction. Post-construction multimodal data collection is triggered only when the construction status is "completed," and the collection method is the same as during the vehicle retrieval phase (outer ring inspection images, mileage, and front and rear interior photos). The resulting post-construction feature vector is then extracted. The extraction method is consistent with the baseline feature vector extraction method (keyframe extraction → appearance feature calculation → feature sub-vectors after construction of outer ring image). Mileage normalization → Feature subvectors after mileage construction Interior feature calculation → Feature sub-vectors after interior construction splicing → =[ , , ]); and the feature vector after construction and baseline feature vector Decompose them into corresponding dimension sub-vectors: Decomposed into , , , Decomposed into , , Calculate the dimensional similarity between sub-vectors of each dimension, using cosine similarity:

[0118] Regarding the outer ring image dimensions: ;

[0119] For the mileage dimension: ;

[0120] Regarding in-vehicle imaging: ;

[0121] The similarity scores of each dimension are weighted and summed according to the preset dimension weights corresponding to the business type. The formula for calculating the overall similarity after construction is as follows:

[0122] ;

[0123] in, The outer ring image dimension weighting coefficient. These are the weighting coefficients for the mileage dimension. For the in-vehicle image dimension weighting coefficient, and + + =1, To determine the dimensional similarity of the outer ring images during the construction phase, For similarity in the mileage dimension during the construction phase, The similarity is calculated for in-vehicle images during the construction phase. Weighting coefficients are retrieved from a pre-defined business verification parameter library based on the business type; the overall similarity is then calculated after construction. Preset construction similarity thresholds corresponding to business types To make a comparison, when ≥ If the post-construction similarity is not met, the post-construction verification is considered successful; otherwise, it is considered unsuccessful. A preset construction similarity threshold is set. The parameters are retrieved from the preset business verification parameter library based on the business type.

[0124] "Dimensional sub-vector" refers to the sub-part of a feature vector corresponding to a specific data acquisition modality. The outer ring image dimensional sub-vector encodes vehicle exterior features, the mileage dimensional sub-vector encodes vehicle mileage values, and the interior image dimensional sub-vector encodes vehicle interior features. "Dimensional similarity" refers to the similarity value between two feature vectors on the same dimensional sub-vector, using cosine similarity measurement with a value range of [-1, 1]. A value closer to 1 indicates greater similarity in that dimension. "Preset dimensional weight" refers to the weight coefficients of each dimensional sub-vector configured according to the business type in the weighted summation of the comprehensive similarity. Different business types correspond to different weight allocation schemes. "Preset construction similarity threshold" refers to the threshold for determining the comprehensive similarity after construction, configured according to the business type. Different business types correspond to different threshold values. By decomposing feature vectors into dimensional sub-vectors, calculating dimensional similarity separately, and then weighted summing to obtain the comprehensive similarity, this application achieves fine-grained comparison of multimodal data. The weight coefficients of each dimension are adaptively configured according to the business type, ensuring that the verification results align with the actual focus of different business types. This solves the problem in existing technologies where verification parameters are fixed and cannot be dynamically adjusted according to business needs.

[0125] In this embodiment, when the business type is "appearance-related business" (painting), the preset dimension weight is configured as follows: =0.6、 =0.1、 =0.3, preset construction similarity threshold =0.85. Multimodal data was collected after construction to extract post-construction feature vectors. , and baseline feature vector Dimensional similarity is calculated after decomposition: =0.92 (The appearance after painting is highly similar to the baseline appearance, with only the color of the painted area changing). =1.0 (mileage unchanged) =0.98 (Interior design unaffected by construction). Overall similarity after construction. =0.6×0.92+0.1×1.0+0.3×0.98=0.552+0.1+0.294=0.946. Since =0.946≥ =0.85, indicating that the post-construction verification has passed.

[0126] According to an embodiment of the present invention, the steps of sending a payment request to the customer terminal based on the verification result and confirming the payment status, scheduling vehicle delivery based on the payment status, collecting multimodal vehicle status data during the delivery phase and extracting delivery feature vectors, calculating the comprehensive similarity between the feature vectors and the delivery phase feature vectors after construction, and outputting the delivery verification result in combination with a preset delivery similarity threshold corresponding to the business type, specifically include:

[0127] If the post-construction verification result is that the post-construction verification passed, a payment request will be sent to the customer terminal.

[0128] Obtain the user's payment status, including whether payment is completed or not. If the payment status is completed, dispatch a vehicle.

[0129] Collect multimodal vehicle status data during the vehicle delivery phase and extract the vehicle delivery feature vector. Decompose the vehicle delivery feature vector and the post-construction feature vector into dimensions respectively, and calculate the similarity of the outer ring image dimension, the similarity of the mileage dimension, and the similarity of the in-vehicle image dimension respectively.

[0130] Based on the preset dimension weights of the vehicle delivery stage corresponding to the business type, the similarity of the outer circle image dimension, the similarity of the mileage dimension, and the similarity of the in-vehicle image dimension are weighted and summed to obtain the comprehensive similarity of the vehicle delivery stage.

[0131] The overall similarity of the vehicle delivery stage is compared with the preset vehicle delivery similarity threshold corresponding to the business type;

[0132] When the overall similarity during the vehicle delivery stage is greater than or equal to the preset vehicle delivery similarity threshold, the vehicle delivery verification result is determined to be successful; otherwise, the vehicle delivery verification is deemed unsuccessful.

[0133] It is important to emphasize that payment requests are only sent when post-construction verification passes; no payment requests are sent if post-construction verification fails. Payment status is retrieved; vehicle dispatch is triggered when the payment status is "Payment Completed," and the process waits or terminates when the payment status is "Payment Unpaid." Multimodal data is collected and vehicle dispatch feature vectors are extracted during the vehicle dispatch phase. ,Will With post-construction feature vectors Perform dimensional decomposition separately ( Decomposed into outer ring image vehicle feature subvectors Mileage delivery feature subvector In-vehicle delivery feature subvector , Decomposed into , , ), calculate the similarity in three dimensions respectively:

[0134] ; ;

[0135] ;

[0136] The weighted summation is performed based on the preset dimension weights of the vehicle delivery stage corresponding to the business type. The formula for calculating the comprehensive similarity of the vehicle delivery stages is as follows:

[0137]

[0138] in, For the similarity of the outer ring images during the vehicle delivery stage, For similarity in mileage during vehicle delivery, To assess the dimensional similarity of in-vehicle images during the vehicle delivery phase, , , These are the dimension weight coefficients for the vehicle delivery stage, and + + =1. The dimension weighting coefficients for the vehicle delivery stage may differ from those for the construction stage (based on the vehicle delivery stage verification parameter configuration according to the business type) to accommodate the differences in verification focus between the vehicle delivery stage and the construction stage; the overall similarity of the vehicle delivery stage will be considered. Similarity threshold with preset delivery vehicle To make a comparison, when ≥ If the vehicle delivery verification is successful, it is considered successful; otherwise, it is considered unsuccessful. The delivery stage verification uses the post-construction feature vector as the comparison benchmark (rather than the baseline feature vector) because the delivery stage needs to verify whether the vehicle has undergone unexpected state changes (such as transportation damage, additional mileage, etc.) from the time of construction to delivery, rather than verifying whether construction changed the vehicle's state (construction changing the vehicle's state is an expected behavior). Therefore, the comprehensive similarity of the delivery stage measures the degree of consistency between the vehicle's state at delivery and its state after construction, with the post-construction feature vector serving as the comparison benchmark for the delivery stage. In this embodiment, if the business type is "appearance-related business" (painting), the preset dimension weight configuration for the delivery stage is as follows: =0.5、 =0.2、 =0.3, preset delivery vehicle similarity threshold =0.90. Multimodal data was collected during vehicle delivery to extract delivery feature vectors, and the dimensional similarity was calculated between these vectors and the post-construction feature vectors. =0.95 (The appearance of the vehicle after delivery is basically the same as that after construction, with only slight dust during transportation). =0.97 (The delivery distance increased by about 20 kilometers, and the similarity after normalization is close to 1). =0.99 (Interior unchanged). Overall similarity at the delivery stage. =0.5×0.95+0.2×0.97+0.3×0.99=0.475+0.194+0.297=0.966. Since =0.966≥ =0.90, indicating that the vehicle delivery verification has passed.

[0139] According to an embodiment of the present invention, the method further includes a service type adaptive verification parameter configuration step:

[0140] Based on the business type, retrieve the corresponding construction dimension weight configuration and construction similarity threshold configuration from the preset business verification parameter library;

[0141] The dimension weight configuration defines the weight coefficient of each dimension sub-vector in the weighted summation, and different business types correspond to different weight coefficient allocations.

[0142] When the business type is appearance-related, the weight coefficient of the outer ring image dimension sub-vector is higher than the weight coefficient of other dimension sub-vectors;

[0143] When the business type is not related to appearance, the weight coefficient of the mileage dimension sub-vector is higher than the weight coefficient of other dimension sub-vectors.

[0144] It is important to emphasize that, based on the identified business type, the corresponding verification parameter configuration is retrieved from the preset business verification parameter library. The preset business verification parameter library stores the mapping relationship between each business type and the verification parameters, including the construction dimension weight configuration (defining the weight coefficients of each dimension in the construction stage) and the construction similarity threshold configuration (defining the similarity judgment threshold in the construction stage), as well as the vehicle delivery dimension weight configuration and the vehicle delivery similarity threshold configuration. The meaning of the dimension weight configuration is—the weight coefficient of each dimension sub-vector in the weighted sum of the comprehensive similarity. The sum of the weight coefficients equals 1. Different business types correspond to different weight coefficient allocation schemes, reflecting the different degrees of attention paid to the verification results of each dimension by different business types. The weight allocation rules for two typical business types are as follows: Appearance-related businesses (such as painting, detailing, etc.) focus on changes in vehicle appearance, so the weight coefficient of the outer ring image dimension sub-vector is higher than the weight coefficients of other dimension sub-vectors; Non-appearance-related businesses (such as maintenance, inspection, etc.) focus on changes in vehicle functional indicators (mileage changes reflect vehicle usage status), so the weight coefficient of the mileage dimension sub-vector is higher than the weight coefficients of other dimension sub-vectors. "Preset Business Verification Parameter Library" refers to a data configuration library that stores the mapping relationship between various business types and verification parameters. It includes fields such as business type identifier, construction dimension weight configuration (weight coefficients of each dimension in the construction stage), construction similarity threshold configuration (judgment threshold in the construction stage), vehicle delivery dimension weight configuration (weight coefficients of each dimension in the delivery stage), and vehicle delivery similarity threshold configuration (judgment threshold in the delivery stage). "Appearance-related business" refers to business types that involve significant changes to the vehicle's appearance, such as painting, detailing, and window tinting. For these types of business, changes to the vehicle's appearance before and after construction are expected, and the verification focuses on the reasonable range of appearance changes. "Non-appearance-related business" refers to business types that do not involve significant changes to the vehicle's appearance, such as maintenance, inspection, and parts replacement. For these types of business, the vehicle's appearance should remain basically unchanged before and after construction, and the verification focuses on changes in functional indicators (such as mileage).

[0145] By adaptively configuring verification parameters (dimensional weights and similarity thresholds) according to business type, this application ensures that the verification results align with the actual focus of different business types, solving the problem of fixed verification parameters in existing technologies that cannot be dynamically adjusted according to business needs. For appearance-related businesses, higher weight is given to the outer ring image dimension to highlight appearance change verification; for non-appearance-related businesses, higher weight is given to the mileage dimension to highlight functional indicator verification, achieving precise matching between verification parameters and business requirements. For example, the parameter configurations stored in the preset business verification parameter library are shown in the table below:

[0146] The parameter configuration table stored in the preset business verification parameter library

[0147] Business type Construction dimension weight ( , , ) Construction similarity threshold ( ) Vehicle delivery dimension weight ( , , ) Vehicle delivery similarity threshold Spray painting (appearance-related category) (0.6,0.1,0.3) 0.85 (0.5,0.2,0.3) 0.90 Maintenance (excluding appearance-related items) (0.2,0.6,0.2) 0.90 (0.2,0.6,0.2) 0.88 Inspection (non-appearance related) (0.1,0.7,0.2) 0.92 (0.1,0.7,0.2) 0.90

[0148] When the business type is "spray painting", the construction dimension weight configuration is retrieved. =0.6、 =0.1、 =0.3, construction similarity threshold =0.85, the outer ring image dimension weight coefficient of 0.6 is higher than that of the mileage dimension (0.1) and the in-vehicle image dimension (0.3); when the business type is "maintenance", the construction dimension weight configuration is retrieved. =0.2、 =0.6、 =0.2, the weighting coefficient of the mileage dimension is 0.6, which is higher than that of the outer ring image dimension (0.2) and the in-vehicle image dimension (0.2).

[0149] According to embodiments of the present invention, it further includes anomaly handling based on risk assessment, specifically including:

[0150] When the post-construction verification result is that the post-construction verification fails, the construction risk score is calculated based on the degree of similarity deviation between the post-construction feature vector and the baseline feature vector.

[0151] When the construction risk score is in the preset high-risk range, construction anomaly alarm data is generated and pushed to the preset management personnel terminal to wait for anomaly handling instructions.

[0152] When the construction risk score is in the preset medium risk range, a construction review request is generated, and the post-construction multimodal vehicle status data is pushed to the preset review terminal to wait for manual review instructions.

[0153] When the vehicle delivery verification result is that the vehicle delivery verification fails, the vehicle delivery risk score is calculated based on the degree of similarity deviation between the vehicle delivery feature vector and the post-construction feature vector, and the corresponding risk level handling strategy is executed according to the risk range in which the vehicle delivery risk score is located.

[0154] It should be emphasized that if the post-construction inspection fails (i.e.) < The construction risk score is calculated based on the degree of similarity deviation between the post-construction feature vector and the baseline feature vector. The degree of similarity deviation is defined as the difference between the post-construction comprehensive similarity score and 1 (because a comprehensive similarity score closer to 1 indicates greater similarity, and a deviation from 1 indicates greater deviation). The formula for calculating the construction risk score is: ;in, The overall similarity after construction (value range [0,1]) The construction risk score is calculated (value range [0,100]). The lower the value (the less similar it is to the baseline feature vector). The higher the risk, the greater the risk. A tiered treatment strategy is implemented based on the construction risk score range: when... Located in a pre-defined high-risk zone (e.g.) ≥60, that is When the value is ≤0.40, it indicates that the vehicle status after construction deviates significantly from the baseline status. Construction anomaly alarm data is generated (including business request identifier, construction risk score, post-construction multimodal data summary, etc.), and pushed to the preset management personnel terminal (such as store manager or operations director), awaiting the management personnel to issue anomaly handling instructions (such as suspending vehicle delivery, initiating an investigation, etc.). Within the preset medium-risk range (e.g., 20≤ <60, that is, 0.40< When the value is ≤0.80, it indicates that the vehicle status after construction deviates moderately from the baseline status. A construction review request is generated, and the multimodal vehicle status data after construction is pushed to a preset review terminal (such as a technical supervisor or quality inspector), awaiting manual review instructions (such as confirming that the construction quality is qualified or confirming the existence of anomalies). The same graded processing logic is used for cases where the vehicle delivery verification fails. The vehicle delivery risk score calculation formula is as follows:

[0155] ,in, To assess the overall similarity during the vehicle delivery phase, Assess the risk of vehicle delivery. The corresponding handling strategy is executed according to the risk range (high risk pushes alarm to the management personnel, medium risk pushes review request to the review terminal).

[0156] "Deviation of similarity" refers to the difference between the overall similarity and perfect similarity (value 1), i.e., deviation = 1 - The greater the deviation, the greater the difference between the actual state and the reference state; "Construction Risk Score" refers to a quantitative indicator that maps the degree of similarity deviation to a percentage score, with a higher score indicating greater risk; "Preset High-Risk Zone" and "Preset Medium-Risk Zone" refer to risk score division zones configured according to actual deployment needs. The high-risk zone corresponds to scenarios with severe deviations requiring management intervention, while the medium-risk zone corresponds to scenarios with moderate deviations requiring manual review and confirmation; "Construction Anomaly Alarm Data" refers to alarm data packets containing information such as business request identifiers, construction risk scores, deviation dimension analysis, and post-construction multimodal data summaries; "Construction Review Request" refers to review request packets containing business request identifiers, construction risk scores, and post-construction multimodal vehicle status data.

[0157] By calculating risk scores based on the degree of similarity deviation and implementing a graded anomaly handling strategy according to the interval of the risk score, this application achieves precision and timeliness in anomaly handling. High-risk scenarios are directly pushed to management personnel for intervention, and medium-risk scenarios are pushed to reviewers for confirmation. This solves the problems of delayed anomaly detection and single handling strategy in the prior art, and improves the efficiency and accuracy of anomaly handling.

[0158] In this embodiment, when the business type is "appearance-related business" (spray painting), the overall similarity after construction is... =0.30 (The vehicle's appearance deviates significantly from the baseline after construction, possibly due to a construction accident), Construction Risk Score =100×(1-0.30)=70. =70 is in the preset high-risk range (≥60), generating construction anomaly alarm data and pushing it to the store manager's terminal. The alarm data includes the business request number, risk score 70, and deviation dimension analysis (outer ring image dimension). =0.15 is the main deviation dimension), waiting for the store manager to issue processing instructions.

[0159] Furthermore, the process of confirming vehicle delivery and triggering a feedback process based on the vehicle delivery verification results includes a complaint handling branch, specifically including:

[0160] When a complaint instruction is received from a customer terminal, the complaint type, customer level tag, and historical complaint frequency are extracted based on the complaint instruction. The complaint priority score is then calculated based on the complaint type, customer level tag, and historical complaint frequency.

[0161] The initial processing level is determined by matching the complaint priority score with a preset job level mapping table, and the complaint instruction is sent to the processing personnel corresponding to the initial processing level.

[0162] Set a processing time threshold for complaint instructions. The processing time threshold is dynamically determined based on the complaint priority score. The higher the priority score, the shorter the corresponding processing time.

[0163] Obtain processing time data. When the processing time data reaches the corresponding processing time limit threshold and the complaint instruction is not marked as processed, automatically escalate the complaint instruction to the next higher level of the current processing level and reset the processing time limit threshold.

[0164] Repeat the above complaint instruction escalation steps until the complaint instruction is marked as processed.

[0165] If the current processing level is already the highest level and the complaint instruction has not been marked as processed, then the complaint instruction will be marked as pending human intervention and pushed to the preset administrator terminal.

[0166] It is important to emphasize that the process is triggered when a customer initiates a complaint request. The complaint type (e.g., "construction quality issue," "service attitude issue," "construction delay issue"), customer level label (e.g., "VIP customer," "regular customer," "new customer"), and historical complaint frequency (the number of times the customer has previously filed complaints) are extracted from the complaint request. A complaint priority score is calculated based on these three factors. The formula for calculating the complaint priority score is as follows:

[0167]

[0168] in, The complaint priority is scored, where α is the weighting coefficient for each complaint type. The severity score is assigned to the complaint type (obtained from a preset complaint type score mapping table), where β is the customer level weighting coefficient. The priority score is assigned to the customer level tag (obtained from the preset customer level score mapping table), and γ is the weighting coefficient for historical complaint frequency. This represents the customer's historical complaint frequency (number of complaints), with δ being the complaint frequency amplification factor (used to prioritize customers with high-frequency complaints). Parameters α, β, γ, and δ are preset coefficients and satisfy the normalization condition α + β + γ = 1. Specific weighting coefficients are configured according to actual deployment requirements. Scoring is based on complaint priority. Match a preset job level mapping table (defining the correspondence between priority scoring ranges and processing job levels, such as...). ≥80 assigned to "Supervisor" level ≥50 assigned to the "team leader" rank. <50 complaints are assigned to the "Customer Service Specialist" job level. The initial processing level is determined, and the complaint instruction is sent to the corresponding personnel at that level. A processing time limit threshold is set for the complaint instruction; this threshold is dynamically determined based on the complaint priority score, and the calculation formula is: ;in, The processing time limit threshold (unit: hours) is used. The baseline processing time (e.g., 24 hours). This is the upper limit for the complaint priority score (e.g., 100). This is a time limit adjustment factor. The higher the priority score, the lower the priority adjustment factor. The smaller the value, the shorter the processing time, ensuring faster responses to high-priority complaints. When the processing time reaches T_limit and the complaint instruction is not marked as processed, the complaint instruction is automatically escalated to the next higher level of the current processing level (e.g., from "Customer Service Specialist" to "Team Leader"), and the processing time threshold is reset; the escalation logic is repeated until the complaint instruction is marked as processed; if the complaint instruction is still not processed at the highest level, it is marked as awaiting human intervention and pushed to the administrator terminal.

[0169] "Complaint Priority Score" refers to a quantitative score calculated based on the severity of the complaint type, customer level priority, and historical complaint frequency. A higher score indicates that the complaint requires a more urgent handling response. "Preset Job Level Mapping Table" is a configuration table that defines the correspondence between complaint priority score ranges and handling job levels. Different score ranges correspond to different handling job levels, with job levels ranging from low to high as Customer Service Specialist, Team Leader, Supervisor, Manager, etc. "Processing Time Threshold" refers to the maximum processing time limit for a complaint instruction at the current handling job level. If the complaint is not processed within this time limit, it will be automatically escalated. "Automatic Escalation" refers to the automatic processing mechanism that transfers a complaint instruction from the current handling job level to the next higher job level.

[0170] By calculating a complaint priority score based on the complaint type, customer level, and historical complaint frequency, and dynamically determining the processing time limit threshold and initial processing level based on the priority score, this application realizes an automatic priority escalation mechanism for complaint handling. This solves the problems of low complaint handling efficiency and lack of tiered response in the prior art, ensuring that high-priority complaints receive faster responses, while low-priority complaints are handled by lower-level personnel to avoid wasting resources.

[0171] In this embodiment, when the complaint type is "construction quality problem" (queried from the preset complaint type scoring mapping table)... =80), the customer level label is "VIP Customer" (queried from the preset customer level rating mapping table). =90), historical complaint frequency =2 times, preset weighting coefficients α=0.4, β=0.3, γ=0.3, complaint frequency amplification coefficient δ=0.5. Complaint priority scoring. =32+27+0.3×1.414=32+27+0.42=59.42. According to the preset job level mapping table, =59.52 falls within the range [50, 80), therefore the initial processing job level is determined to be "Team Leader". Processing time limit threshold: Assuming =24 hours =100, =1.0, =24×(1-59.42 / 100)×1.0=24×0.4058=9.74 hours. The team leader must handle the complaint within 9.74 hours, otherwise the team leader will be automatically promoted to the "supervisor" level.

[0172] Furthermore, the complaint handling branch also includes a closed-loop mechanism for eliminating red flags:

[0173] When a complaint instruction is marked as processed, a complaint processing result confirmation request is sent to the client terminal that initiated the complaint instruction.

[0174] If the complaint handling result returned by the customer terminal is "satisfied", the status flag of the complaint instruction will be switched from the abnormal status to the normal status, that is, the red flag removal operation will be performed.

[0175] If the complaint handling result returned by the customer terminal is "unsatisfied" or "not satisfied", the complaint instruction will be resent to the next higher level of the current handling level, and the processing time limit threshold will be reset.

[0176] If there are complaint instructions that have not been cleared, the closed-loop data chain corresponding to the vehicle integrated business request remains open, and the closed-loop archiving operation is prohibited.

[0177] It is important to emphasize that when a complaint instruction is marked as processed (the processing personnel confirm that the complaint has been processed), a complaint processing result confirmation request is sent to the customer terminal, requesting the customer to confirm whether they are satisfied with the processing result. If the customer returns "Confirmed Satisfied," the red flag removal operation is performed—the status flag of the complaint instruction is switched from "abnormal status" to "normal status." "Abnormal status" (red flag) indicates that there is an unresolved complaint issue in the business request, and "normal status" (red flag removed) indicates that the complaint issue has been accepted and resolved by the customer. If the customer returns "Not Confirmed Satisfied" or "Dissatisfied," it means that the current processing result has not been accepted by the customer. The complaint instruction is then resent to the next higher level of the current processing level (higher-level personnel may have stronger processing capabilities and authorization), and the processing time limit threshold is reset to start a new round of processing timer. To ensure the integrity of the closed loop—if there are any complaint instructions that have not been cleared of red flags (i.e., there are still complaint instructions in the abnormal status), the closed loop data chain corresponding to the vehicle integrated business request remains open, and the closed loop archiving operation is prohibited. This means that the closed loop data chain can only be closed and archived after all complaints have been confirmed as satisfactory by the customer and cleared of red flags.

[0178] "Red-mark removal operation" refers to the operation of switching the status marker of a complaint instruction from an abnormal state to a normal state. A "red marker" (abnormal state) indicates that the business request has unresolved complaint issues that require attention and processing. After the red marker is removed, it means that the complaint has been resolved and accepted by the customer. "Closed-loop data chain" refers to the complete data link formed by connecting the data generated in the application according to the business time sequence. Closing the closed-loop data chain marks the completion of the entire process of the vehicle comprehensive business request. "Closed-loop archiving operation" refers to the operation of archiving and storing the closed closed-loop data chain. After archiving, the entire process data of the business request is saved as a historical record. The red-mark removal closed-loop mechanism ensures that the complaint status is switched from abnormal to normal only after the customer confirms satisfaction. If the customer is not satisfied, it is automatically escalated to a higher level for reprocessing. Complaint instructions that have not been red-marked remain in the closed-loop data chain state and are prohibited from being archived. This application realizes a true closed loop in complaint processing—not only must the complaint be marked as processed by the handling personnel, but the customer must also confirm satisfaction before the red marker is removed. This solves the problem of the lack of a red-mark removal closed loop in the complaint processing of existing technologies, effectively ensuring customer satisfaction and the integrity of the business closed loop.

[0179] In this embodiment, once a complaint instruction has been marked as processed by a "team leader" level personnel, a complaint processing result confirmation request is sent to the customer. The customer returns "Confirm Satisfied," and the complaint instruction status changes from "Abnormal" to "Normal," indicating successful removal from the red flag. If there are no other unresolved complaint instructions for this service request, the closed-loop data chain can be closed and archived.

[0180] Furthermore, the feedback process also includes an evaluation feedback branch, specifically including:

[0181] Obtain evaluation data, including service ratings and written reviews;

[0182] Query the preset rating mapping table based on the service rating to determine the discount level coefficient;

[0183] Query the preset business discount parameter library according to the business type to obtain the basic discount amount and discount cap corresponding to the business type;

[0184] Multiply the discount level coefficient by the basic discount amount to obtain the initial discount benefit value. When the initial discount benefit value exceeds the discount limit, the discount limit is recorded as the initial discount benefit value.

[0185] Based on the initial preferential benefit value, business type and the preset validity period corresponding to the business type, generate preferential benefit data and store the preferential benefit data in the customer profile;

[0186] The evaluation data is associated and stored in the evaluation file of the construction personnel corresponding to the vehicle integrated business request, and the task allocation weight coefficient of the construction personnel is updated according to the evaluation data, which is used to affect the task allocation priority when scheduling subsequent construction tasks.

[0187] It is important to emphasize that after vehicle delivery confirmation, customer feedback data should be obtained, including service ratings (e.g., integer ratings from 1 to 5 points) and written service reviews (customer's written description of service quality). Based on the service ratings, a pre-defined rating mapping table (defining the correspondence between rating values ​​and discount level coefficients) should be consulted to determine the corresponding discount level coefficient for that rating. The discount level coefficient is usually higher for higher scores (e.g., 5 points corresponds to a higher coefficient). =1.0, 4 points correspond =0.8, 3 points correspond =0.5, 2 points correspond =0.2, 1 point correspond =0.0); Query the preset business discount parameter library according to the business type to obtain the basic discount amount corresponding to that business type. (Base discount amount for this type of service) and discount cap (The maximum discount amount for this type of service); Calculate the initial discount benefit value.

[0188] ,when > Time to take As the final value of the preferential benefits (i.e. This ensures that the preferential benefits do not exceed the upper limit set for the business type; based on the initial preferential benefit value (or upper limit value), the business type, and the preset validity period corresponding to the business type, preferential benefit data (including the preferential amount, applicable business type, and start and end dates of the validity period) is generated, and the preferential benefit data is associated and stored in the customer profile (bound to the customer identifier for subsequent business use); the evaluation data is associated and stored in the construction worker's evaluation profile (bound to the construction worker identifier), and the task allocation weight coefficient of the construction worker is updated based on the evaluation data. The update method is as follows:

[0189] in, Assign weight coefficients to the updated tasks. Assign weight coefficients to the original task, where η is the weight update adjustment coefficient (preset value, such as 0.1). Rate the current service. The historical average rating of this construction worker. This is the maximum score (e.g., 5 points). When > hour, > (A score higher than average results in increased weight and higher priority for subsequent task allocation); when < hour, < (If the score is below average, the weight will be reduced, and the priority of subsequent task allocation will be lower); when = The time weight remains unchanged.

[0190] "Discount Level Coefficient" refers to the coefficient value obtained from the preset rating mapping table based on the service rating. This coefficient is positively correlated with the service rating; the higher the rating, the larger the coefficient. It is used to calculate the amount of discount benefits received by the customer. "Basic Discount Amount" refers to the benchmark discount amount set for each business type. The basic discount amount may be different for different business types (e.g., the basic discount amount for maintenance is 50 yuan, and the basic discount amount for painting is 100 yuan). "Discount Upper Limit" refers to the upper limit of the discount amount set for each business type to prevent excessive discount benefits from affecting business revenue. "Discount Benefit Data" refers to discount voucher data containing the discount amount, applicable business type, and validity period, which is associated and stored in the customer's file for use in subsequent business. "Task Allocation Weight Coefficient" refers to the priority weight of construction personnel in construction task scheduling. The higher the weight, the higher the priority in subsequent task allocation (more likely to be assigned tasks), and the lower the weight, the lower the priority. By determining the discount level coefficient based on service ratings and multiplying it by the basic discount amount to generate a discount benefit value (with an upper limit protection), and simultaneously storing the evaluation data in the construction personnel's evaluation file and updating their task allocation weight coefficients, this invention achieves the linkage update of evaluation feedback, discount benefits, and construction personnel's task allocation weights. This solves the problem of the lack of linkage update of business parameters in evaluation feedback in the prior art. On the one hand, it incentivizes customers to actively participate in evaluation feedback and promotes customer loyalty through discount benefits; on the other hand, it incentivizes construction personnel to improve service quality and promotes continuous optimization of service quality through the update of task allocation weights.

[0191] In this embodiment, if the service rating =5 points (maximum), retrieve the discount level coefficient from the preset rating mapping table. =1.0. The business type is "spray painting". The basic discount amount is retrieved from the preset business discount parameter database. =100 yuan, maximum discount =200 yuan. Initial discount value =1.0 × 100 = 100 yuan, 100 ≤ 200 (not exceeding the upper limit), the final discount value is 100 yuan. Based on the initial discount value of 100 yuan, the business type "spray painting" and the preset validity period of 30 days, discount data (discount amount 100 yuan, applicable business type "spray painting", validity period 30 days) is generated and stored in the customer file. Historical average rating of construction personnel. =4.0 points, maximum rating =5 points, weight update adjustment coefficient η=0.1, original task assigned weight coefficient =1.0. After the update =1.0×(1+0.1×(5-4) / 5)=1.0×(1+0.02)=1.02, the weight increases by 0.02, and the priority of subsequent task allocation is slightly improved.

[0192] The present invention also discloses a closed-loop data processing system for the entire process of vehicle integrated business, including a memory and a processor. The memory stores a program for a closed-loop data processing method for the entire process of vehicle integrated business. When the program for a closed-loop data processing method for the entire process of vehicle integrated business is executed by the processor, it implements the steps of the closed-loop data processing method for the entire process of vehicle integrated business.

[0193] This invention discloses a closed-loop data processing method and system for the entire process of vehicle integrated business. The method involves: acquiring vehicle integrated business requests and identifying business types and service requirements; matching these requests with a preset quotation rule base to generate scheme quotation data and obtain scheme verification status; dispatching vehicle pickup and extracting baseline feature vectors from initial multimodal vehicle status data; collecting post-construction multimodal data to extract post-construction feature vectors, calculating the post-construction comprehensive similarity with the baseline feature vectors, and outputting the post-construction verification result; dispatching vehicle delivery after payment, calculating the post-delivery feature vector and the post-construction feature vector's comprehensive similarity at the delivery stage, and outputting the delivery verification result; executing vehicle confirmation and triggering a feedback process based on the delivery verification result; and thus achieving closed-loop verification and risk control throughout the entire process through weighted summation of multimodal dimension similarities and adaptive verification parameters based on business types.

[0194] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0195] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0196] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

Claims

1. A closed-loop data processing method for the entire process of integrated vehicle business, characterized in that, Includes the following steps: Obtain comprehensive vehicle service requests and identify service types and service requirements. Based on the business type and service requirements information, the system matches the pre-set quotation rule library to generate a solution quotation data, pushes it to the customer's terminal, and obtains the solution verification status. According to the scheme, the vehicle retrieval is scheduled based on the verification status. The initial multimodal vehicle status data of the vehicle retrieval stage is obtained, and the initial feature vector is extracted and stored as the baseline feature vector. Acquire construction status data, collect post-construction multimodal vehicle status data based on construction status and extract post-construction feature vectors, calculate the post-construction comprehensive similarity between the feature vectors and the baseline feature vectors, and output the post-construction verification results in combination with the preset construction similarity threshold corresponding to the business type. Based on the verification results, a payment request is sent to the customer terminal and the payment status is confirmed. The vehicle is dispatched according to the payment status. Multimodal vehicle status data during the delivery stage is collected and the delivery feature vector is extracted. The comprehensive similarity between the feature vector and the delivery stage feature vector after construction is calculated. The delivery verification result is output in combination with the preset delivery similarity threshold corresponding to the business type. Based on the vehicle delivery verification results, vehicle delivery confirmation will be executed and a feedback process will be triggered.

2. The closed-loop data processing method for the entire vehicle integrated business process according to claim 1, characterized in that, The process of matching business type and service requirement information with a preset quotation rule base, generating a solution quotation data, pushing it to the customer terminal, and obtaining the solution verification status specifically includes: According to the business type, a corresponding quotation template is matched from the preset quotation rule library. The quotation template includes basic service items, optional additional service items and corresponding fee calculation rules. Based on the vehicle identification information in the service request information, obtain the vehicle's historical service data, including historical maintenance records, historical repair records, and historical evaluation data; Input service demand information and historical service data into the pricing calculation model to generate a main service price and a recommended price for additional services. The main service quote and the recommended additional service quotes are combined to generate the solution quote data, which includes cost details and estimated construction period; The status of the solution verification is obtained based on the customer's input, including whether they agree or disagree.

3. The closed-loop data processing method for the entire vehicle integrated business process according to claim 1, characterized in that, The step of verifying the vehicle retrieval status according to the scheme, obtaining the initial multimodal vehicle status data during the retrieval phase, and extracting the initial feature vector and storing it as a baseline feature vector specifically includes: If the verification status of the plan is confirmed and agreed, the vehicle will be dispatched and the initial multimodal vehicle status data will be obtained, including the initial image data of the vehicle's outer ring inspection, the initial mileage of the vehicle, and the initial photos of the front and rear of the vehicle. Keyframes are extracted from the initial image data of the vehicle's outer ring inspection, and appearance features are calculated from the keyframes to generate the initial feature subvector of the outer ring image. The initial mileage of the vehicle is numerically normalized to generate an initial mileage feature sub-vector. Interior features are calculated from the initial front and rear photos of the vehicle interior to generate initial feature sub-vectors for the vehicle interior. The initial feature vectors of the outer ring image, the initial feature vector of the mileage, and the initial feature vector of the vehicle interior are combined according to a preset dimension to generate an initial feature vector and store it as a baseline feature vector.

4. The closed-loop data processing method for the entire vehicle integrated business process according to claim 1, characterized in that, The process of acquiring construction status data includes collecting post-construction multimodal vehicle status data and extracting post-construction feature vectors based on the construction status, calculating the post-construction comprehensive similarity between the feature vectors and the baseline feature vectors, and outputting the post-construction verification result based on a preset construction similarity threshold corresponding to the business type. Specifically, this includes: Obtain construction status data, including whether construction is completed or not; If the construction status data indicates that construction is completed, then obtain the post-construction multimodal vehicle status data and extract the post-construction feature vector. The post-construction feature vector and the baseline feature vector are respectively decomposed into outer circle image dimension sub-vector, mileage dimension sub-vector and in-vehicle image dimension sub-vector; Calculate the similarity in the outer ring image dimension, the similarity in the mileage dimension, and the similarity in the in-vehicle image dimension respectively; Based on the preset dimension weights corresponding to the business type, the similarity of each dimension is weighted and summed to obtain the comprehensive similarity after construction. The post-construction comprehensive similarity is compared with the preset construction similarity threshold corresponding to the business type to obtain the post-construction verification result; When the overall similarity after construction is greater than or equal to the preset construction similarity threshold, the post-construction verification result is determined to be passed; otherwise, the post-construction verification is deemed to have failed.

5. The closed-loop data processing method for the entire vehicle integrated business process according to claim 1, characterized in that, The process involves sending a payment request to the customer terminal based on the verification result and confirming the payment status, scheduling vehicle delivery based on the payment status, collecting multimodal vehicle status data during the delivery phase and extracting delivery feature vectors, calculating the comprehensive similarity between these feature vectors and the post-construction feature vectors during the delivery phase, and outputting the delivery verification result based on a preset delivery similarity threshold corresponding to the business type. Specifically, this includes: If the post-construction verification result is that the post-construction verification passed, a payment request will be sent to the customer terminal. Obtain the user's payment status, including whether payment is completed or not. If the payment status is completed, dispatch a vehicle. Collect multimodal vehicle status data during the vehicle delivery phase and extract the vehicle delivery feature vector. Decompose the vehicle delivery feature vector and the post-construction feature vector into dimensions respectively, and calculate the similarity of the outer ring image dimension, the similarity of the mileage dimension, and the similarity of the in-vehicle image dimension respectively. Based on the preset dimension weights of the vehicle delivery stage corresponding to the business type, the similarity of the outer circle image dimension, the similarity of the mileage dimension, and the similarity of the in-vehicle image dimension are weighted and summed to obtain the comprehensive similarity of the vehicle delivery stage. The overall similarity of the vehicle delivery stage is compared with the preset vehicle delivery similarity threshold corresponding to the business type; When the overall similarity during the vehicle delivery stage is greater than or equal to the preset vehicle delivery similarity threshold, the vehicle delivery verification result is determined to be successful; otherwise, the vehicle delivery verification is deemed unsuccessful.

6. The closed-loop data processing method for the entire vehicle integrated business process according to claim 1, characterized in that, The method also includes a business type adaptive validation parameter configuration step: Based on the business type, retrieve the corresponding construction dimension weight configuration and construction similarity threshold configuration from the preset business verification parameter library; The dimension weight configuration defines the weight coefficient of each dimension sub-vector in the weighted summation, and different business types correspond to different weight coefficient allocations. When the business type is appearance-related, the weight coefficient of the outer ring image dimension sub-vector is higher than the weight coefficient of other dimension sub-vectors; When the business type is not related to appearance, the weight coefficient of the mileage dimension sub-vector is higher than the weight coefficient of other dimension sub-vectors.

7. The closed-loop data processing method for the entire vehicle integrated business process according to claim 1, characterized in that, It also includes anomaly handling based on risk assessment, specifically including: When the post-construction verification result is that the post-construction verification fails, the construction risk score is calculated based on the degree of similarity deviation between the post-construction feature vector and the baseline feature vector. When the construction risk score is in the preset high-risk range, construction anomaly alarm data is generated and pushed to the preset management personnel terminal to wait for anomaly handling instructions. When the construction risk score is in the preset medium risk range, a construction review request is generated, and the post-construction multimodal vehicle status data is pushed to the preset review terminal to wait for manual review instructions. When the vehicle delivery verification result is that the vehicle delivery verification fails, the vehicle delivery risk score is calculated based on the degree of similarity deviation between the vehicle delivery feature vector and the post-construction feature vector, and the corresponding risk level handling strategy is executed according to the risk range in which the vehicle delivery risk score is located.

8. A closed-loop data processing system for the entire vehicle business process, characterized in that, The system includes a memory and a processor. The memory stores a program for a closed-loop data processing method for the entire vehicle integrated business process. When the program for the closed-loop data processing method for the entire vehicle integrated business process is executed by the processor, it performs the following steps: Obtain comprehensive vehicle service requests and identify service types and service requirements. Based on the business type and service requirements information, the system matches the pre-set quotation rule library to generate a solution quotation data, pushes it to the customer's terminal, and obtains the solution verification status. According to the scheme, the vehicle retrieval is scheduled based on the verification status. The initial multimodal vehicle status data of the vehicle retrieval stage is obtained, and the initial feature vector is extracted and stored as the baseline feature vector. Acquire construction status data, collect post-construction multimodal vehicle status data based on construction status and extract post-construction feature vectors, calculate the post-construction comprehensive similarity between the feature vectors and the baseline feature vectors, and output the post-construction verification results in combination with the preset construction similarity threshold corresponding to the business type. Based on the verification results, a payment request is sent to the customer terminal and the payment status is confirmed. The vehicle is dispatched according to the payment status. Multimodal vehicle status data during the delivery stage is collected and the delivery feature vector is extracted. The comprehensive similarity between the feature vector and the delivery stage feature vector after construction is calculated. The delivery verification result is output in combination with the preset delivery similarity threshold corresponding to the business type. Based on the vehicle delivery verification results, vehicle delivery confirmation will be executed and a feedback process will be triggered.

9. The closed-loop data processing system for the entire vehicle integrated business process according to claim 8, characterized in that, The process of matching business type and service requirement information with a preset quotation rule base, generating a solution quotation data, pushing it to the customer terminal, and obtaining the solution verification status specifically includes: According to the business type, a corresponding quotation template is matched from the preset quotation rule library. The quotation template includes basic service items, optional additional service items and corresponding fee calculation rules. Based on the vehicle identification information in the service request information, obtain the vehicle's historical service data, including historical maintenance records, historical repair records, and historical evaluation data; Input service demand information and historical service data into the pricing calculation model to generate a main service price and a recommended price for additional services. The main service quote and the recommended additional service quotes are combined to generate the solution quote data, which includes cost details and estimated construction period; The status of the solution verification is obtained based on the customer's input, including whether they agree or disagree.

10. The closed-loop data processing system for the entire vehicle integrated business process according to claim 8, characterized in that, The step of verifying the vehicle retrieval status according to the scheme, obtaining the initial multimodal vehicle status data during the retrieval phase, and extracting the initial feature vector and storing it as a baseline feature vector specifically includes: If the verification status of the plan is confirmed and agreed, the vehicle will be dispatched and the initial multimodal vehicle status data will be obtained, including the initial image data of the vehicle's outer ring inspection, the initial mileage of the vehicle, and the initial photos of the front and rear of the vehicle. Keyframes are extracted from the initial image data of the vehicle's outer ring inspection, and appearance features are calculated from the keyframes to generate the initial feature subvector of the outer ring image. The initial mileage of the vehicle is numerically normalized to generate an initial mileage feature sub-vector. Interior features are calculated from the initial front and rear photos of the vehicle interior to generate initial feature sub-vectors for the vehicle interior. The initial feature vectors of the outer ring image, the initial feature vector of the mileage, and the initial feature vector of the vehicle interior are combined according to a preset dimension to generate an initial feature vector and store it as a baseline feature vector.

Images