Electric vehicle reconnaissance task intelligent management method based on multi-dimensional data collaboration

Abstract

The application provides an electric vehicle reconnaissance task intelligent management method based on multidimensional data cooperation. It belongs to the technical field of task management. The reconnoitering personnel submits information to the system, and the franchised repair shops are based on the preliminary repair demand pushed by the system; an automatic quotation model is constructed; surrounding comparison is conducted to generate a quotation ranking; the best franchised repair shop recommendation is automatically generated; the system detects abnormal cooperative behavior through historical quotation fluctuation rate, identifies and prevents collusive quotation behavior among franchised repair shops; for the franchised repair shops suspected of collusive quotation, the system marks them as high risk and pays special attention to them in subsequent quotation; after the repair is completed, the repair result is accepted, and the order settlement is completed after confirmation. The sliding window mechanism and Monte Carlo simulation are used to dynamically adjust the floating threshold, which avoids the false screening caused by the static threshold and improves the rationality of the quotation.

Description

TECHNICAL FIELD

[0001] The application provides an electric vehicle inspection task intelligent management method based on multi-dimensional data collaboration, and belongs to the technical field of task management. BACKGROUND

[0002] In the field of electric vehicle maintenance services, traditional inspection task management methods have many shortcomings. On the one hand, the allocation of inspection personnel and franchise repair shops often relies on manual operation, which is inefficient and prone to errors. This can cause some inspection personnel to be overburdened, while others are idle, resulting in waste of resources. On the other hand, the bidding process between franchise repair shops lacks transparency and fairness, often resulting in overpricing, underpricing, or unreasonable pricing, which can harm consumers and affect the overall quality and efficiency of electric vehicle maintenance services. In addition, franchise repair shops often lack data protection mechanisms when submitting bids, making them vulnerable to malicious competition and collusion, which can lead to unfair practices. These problems not only harm consumers but also affect the reputation and sustainable development of electric vehicle maintenance services. Finally, different franchise repair shops may have different fee calculation logic and standards, making it difficult to unify and accurately control the results of price calculation, which can affect the reduction of claims for insurance companies. SUMMARY

[0003] The application provides an electric vehicle inspection task intelligent management method based on multi-dimensional data collaboration to solve the problems mentioned in the background.

[0004] The application provides an electric vehicle inspection task intelligent management method based on multi-dimensional data collaboration, which includes the following steps:

[0005] S1, the inspection personnel submit information to the system, and the franchise repair shop submits the preliminary repair demand according to the system push;

[0006] S2, an automatic bidding model is constructed;

[0007] S3, surrounding price comparison is performed to generate a price ranking, and the best franchise repair shop is automatically recommended;

[0008] S4, the system detects abnormal collaborative behavior through historical price fluctuation rate, identifies and prevents collusion between franchise repair shops, and marks the suspected franchise repair shops as high-risk and pays special attention to them in subsequent bidding;

[0009] S5, after the repair is completed, the repair result is accepted, and the order settlement is completed after confirmation.

[0010] The application provides a multi-dimensional data cooperation-based electric vehicle reconnaissance task intelligent management system, including a memory, a processor, and a computer program stored in the memory and capable of running on the memory, and the processor executes the program to realize the multi-dimensional data cooperation-based electric vehicle reconnaissance task intelligent management method.

[0011] The application has the advantages that: the processing efficiency of the reconnaissance task is improved through automatic order allocation, quotation comparison, maintenance process monitoring and other processes; a quotation model is constructed, historical maintenance data and real-time market dynamics are combined to generate a benchmark quotation interval, reasonable quotation reference is provided for the franchise maintenance shop, the accuracy of the quotation is improved; the influence weight of each feature on the quotation is quantified through SHAP value analysis, the transparency and interpretability of the quotation model are enhanced; a sliding window mechanism and Monte Carlo simulation are used to dynamically adjust the floating threshold, the false screening caused by the static threshold is avoided, and the rationality of the quotation is improved; the quotation data of the franchise maintenance shop is transmitted in an all-process encryption mode, the safety of the data in the transmission process is ensured; the quotation hash value is stored through the alliance chain, data tampering is prevented, and a traceability basis is provided for subsequent disputes; through time series analysis, graph neural network and other technologies, collusion quotation behaviors between the franchise maintenance shops are detected, high-risk franchise maintenance shops are marked in time and given special attention, human manipulation and malicious collusion are reduced, and the accuracy and efficiency of the detection are improved; a dynamic credit score mechanism is introduced, the collusion behavior is penalized, and the order receiving authority is limited, which effectively curbs the collusion quotation behavior; the parts list and the maintenance order are automatically associated, the parts compatibility is verified through the knowledge graph, the compensation scheme is automatically generated for the false parts reporting behavior, and the user rights are maintained. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 The method steps of the application are described. DETAILED DESCRIPTION

[0013] The preferred embodiments of the application are described below with reference to the accompanying drawings, and it should be understood that the preferred embodiments described herein are only used to illustrate and explain the application, and are not used to limit the application.

[0014] An embodiment of the application is shown in Figure 1 The multi-dimensional data cooperation-based electric vehicle reconnaissance task intelligent management method includes:

[0015] S1, the total administrator creates a new surveyor and a franchise repair shop account on the web side and assigns corresponding permissions; the order administrator creates a new electric vehicle survey order on the web side, and the system automatically assigns a surveyor according to the case information and the distribution of franchise repair shops; the surveyor uses the APP to receive the order and go to the scene for survey; the surveyor uploads the vehicle photos, vehicle location information and preliminary repair demand information to the system; the franchise repair shop submits a preliminary quotation according to the preliminary repair demand pushed by the system using the APP.

[0016] S2, the system constructs an automatic quotation model according to historical repair data; compare the preliminary quotation submitted by the franchise repair shop with the quotation generated by the system automatic quotation model; the system selects the franchise repair shop with reasonable quotation according to the preset floating range;

[0017] S3, the quotation data of the franchise repair shop is encrypted in the transmission process; after the system selects the franchise repair shop with reasonable quotation, it performs surrounding price comparison and generates quotation ranking; the system applies a multi-objective decision model to automatically generate the best franchise repair shop recommendation; the order administrator confirms the winning franchise repair shop according to the system recommendation;

[0018] S4, the system detects abnormal cooperative behavior through historical quotation fluctuation rate, identifies and prevents collusive quotation behavior among franchise repair shops; for the franchise repair shop suspected of collusive quotation, the system marks it as high risk and pays special attention to it in subsequent quotation;

[0019] S5, after receiving the repair order, the franchise repair shop starts repair work and uploads real-time repair process photos; after the repair is completed, the franchise repair shop uploads the repaired photos, price and accessory list to the system; the order administrator checks the repair result and confirms that there is no error, and completes the order settlement.

[0020] The working principle of the above technical solution is as follows: the total administrator is responsible for creating and managing the accounts of surveyors and franchised repair shops on the web side, and assigns different operation permissions according to the roles to ensure the safety and controllability of the system; the order administrator creates a new electric vehicle survey order on the web side, records vehicle information, location information, repair needs, and contact information, etc., to provide basic data for subsequent survey and repair work; the system automatically assigns the most suitable surveyor to the scene according to the case information, the geographical location of the franchised repair shop, and the available resources, etc., to conduct a survey; the surveyor uses the APP to receive the order, goes to the scene to conduct a survey, and uploads vehicle photos, location information, and preliminary repair needs, etc., to the system to provide a basis for subsequent repair pricing; the franchised repair shop uses the APP to submit a preliminary offer based on the preliminary repair needs pushed by the system to provide comparison data for the automatic pricing model of the system; the system builds an automatic pricing model based on historical repair data to generate a relatively accurate repair offer; the system compares the preliminary offer submitted by the franchised repair shop with the offer generated by the automatic pricing model, and selects the franchised repair shop with a reasonable offer according to a preset floating range (for example, 15%); if the user has a self-selected shop (including a non-franchised repair shop), the repair offer of the self-selected shop is within the floating range of the offer generated by the automatic pricing model, for example, 5%, and the user's self-selected shop is preferred; the repair offer data of the franchised repair shop is encrypted throughout the transmission process to ensure data security; after the system selects the franchised repair shop with a reasonable offer, it performs a surrounding price comparison, considers factors such as offer amount, repair timeliness, and shop rating, etc., to generate an offer ranking; the system applies a multi-objective decision model, considers multiple dimensions (for example, offer amount, repair timeliness, shop rating, etc., in a 5-dimensional evaluation system), and automatically generates a recommendation of the best franchised repair shop; the order administrator confirms the winning franchised repair shop according to the system recommendation and enters the subsequent repair process; the system detects abnormal collaborative behavior through historical offer fluctuation rate, identifies and prevents collusive offer behavior among franchised repair shops; for franchised repair shops suspected of collusive offer, the system marks them as high-risk and pays special attention to them in subsequent offers to ensure the fairness and reasonableness of the offers; after receiving the repair order, the franchised repair shop starts the repair work and uploads real-time repair process photos to ensure the transparency and traceability of the repair process; after the repair is completed, the franchised repair shop uploads information such as the repaired photos, price, and parts list to the system. The order administrator inspects and accepts the repair result and completes the order settlement after confirming that there is no error.

[0021] Effects of the above technical solutions are as follows: through the cooperative work of the web end and the APP, the rapid creation, distribution and processing of the survey order are realized, the survey personnel can quickly receive the order and go to the scene for survey, the survey time is greatly shortened, the survey waiting time is reduced, and the survey process is optimized; the system automatically distributes the survey personnel and screens the franchise repair shops with reasonable quotes, reduces manual intervention, and improves work efficiency; the franchise repair shop can upload the maintenance process photos and the information after repair in real time, so that the order administrator can quickly check the maintenance result, and the order settlement speed is further accelerated; the system constructs an automatic quote model according to historical maintenance data, provides a reference standard for the preliminary quote of the franchise repair shop, compares the preliminary quote submitted by the franchise repair shop with the quote generated by the automatic quote model of the system, sets a reasonable floating range, effectively screens the franchise repair shop with reasonable quotes, improves the accuracy of the quote and the competitiveness of the franchise repair shop, and reduces the risk of quote failure; the system applies a multi-objective decision model, comprehensively considers multiple dimensions, and automatically generates the best franchise repair shop recommendation, ensuring the comprehensiveness and fairness of the quote; the quote data of the franchise repair shop is encrypted throughout the transmission process, effectively preventing data leakage and tampering, and ensuring the security of the data; the system detects abnormal cooperative behavior through historical quote fluctuation rate, can identify and prevent collusive quoting behavior among franchise repair shops, and maintains market order and a fair competitive environment; for the franchise repair shop suspected of collusive quoting, the system marks it as high risk and pays special attention to it in subsequent quoting, effectively curbing the occurrence of collusive quoting behavior; through rapid and accurate survey and maintenance services, as well as a fair and reasonable quoting mechanism, the satisfaction of customers with the electric vehicle survey and maintenance services is improved; real-time uploading of the maintenance process photos and the information after repair enables the customer to know the maintenance progress and result at any time, and enhances the transparency and credibility of the service.

[0022] In an embodiment of the present application, the S1 comprises:

[0023] S11, when the total administrator creates an account through the web end, the permissions of the survey personnel and the franchise repair shop are dynamically distributed based on the RBAC model;

[0024] S12, the real-time positions of the franchise repair shop and the survey personnel are bound by introducing the geofencing technology;

[0025] S13, the system dynamically calculates the optimal survey personnel distribution scheme based on the reinforcement learning model in combination with the real-time load of the survey personnel, the historical response speed and the traffic condition data;

[0026] S14, when the order administrator newly creates an order, the system automatically calls the Gaode / Baidu map API, analyzes the vehicle position information and generates three-dimensional geographic coordinates, and accurately matches the distribution of the franchise repair shop;

[0027] S15, when the inspector uploads data through the APP, the system has an image recognition module built-in, which automatically checks the integrity of the vehicle photo and extracts key fields;

[0028] S16, the maintenance demand adopts a structured form, and the system automatically generates a maintenance priority label.

[0029] The working principle of the above technical solution is: when the total administrator creates the account of the inspector and the franchised maintenance shop through the Web end, based on the RBAC (Role-Based Access Control) model, the rights of different roles are dynamically allocated; the rights include data access range (such as only the orders in the own area), operation rights (such as quotation submission, maintenance progress update) and data encryption level, to ensure that each role can only access and operate the data and functions within its permission range; the geo-fencing technology is introduced to bind the real-time location of the franchised maintenance shop and the inspector; it ensures that when the order is assigned, the system can preferentially match the available personnel within the preset range (such as 5 kilometers), to improve the response speed and the inspection efficiency; the system is based on a reinforcement learning model (such as DQN), combined with the real-time load (current task number) of the inspector, historical response speed and traffic condition data, to dynamically calculate the optimal inspector allocation scheme; through continuous learning and optimization, the system can more accurately predict the availability and response speed of the inspector, so as to make more reasonable allocation decisions; when the order administrator creates a new order, the system automatically calls the Gaode / Baidu map API to analyze the vehicle location information; three-dimensional geographic coordinates (longitude, latitude, altitude) are generated, to provide accurate location data for subsequent distribution matching of franchised maintenance shops and allocation of inspectors; when the inspector uploads the vehicle photo through the APP, the image recognition module (based on YOLOv5 algorithm) built-in the system automatically checks the integrity of the photo; whether the license plate and frame number are clear and identifiable, and extracts key fields (such as vehicle model code), to provide accurate information for subsequent maintenance quotation and parts preparation; the maintenance demand adopts a structured form, such as checking the "urgent" and "replace the car for compensation" options; the system automatically generates a maintenance priority label (such as P0-P3) according to the checked options, for subsequent quotation model reference and priority arrangement of maintenance progress.

[0030] The above technical scheme has the following effects: through the RBAC model, the total administrator can dynamically allocate permissions for the surveyors and the franchised repair shops, ensures that each role can only access and operate the data and functions within its permission range, improves the security and controllability of the system, and enhances the data protection and privacy control capability; the geographic fence technology is introduced to bind the real-time positions of the franchised repair shops and the surveyors, ensures that the order allocation can preferentially match the available personnel within the preset range, shortens the response time of the surveyors, improves the survey efficiency, and also reduces unnecessary long-distance travel and saves resources; the system dynamically calculates the optimal surveyor allocation scheme based on the reinforcement learning model, in combination with the real-time load of the surveyors, the historical response speed and the traffic condition data, can more accurately predict the availability and response speed of the surveyors, so as to make more reasonable allocation decisions, and improves the completion quality and efficiency of the survey task; when the order administrator newly creates an order, the system automatically calls the Gaode / Baidu map API, parses the vehicle position information and generates three-dimensional geographic coordinates; this provides accurate position data for subsequent distribution matching of the franchised repair shops and allocation of the surveyors, ensures that the order can be accurately and quickly allocated to the most suitable franchised repair shop or surveyor; when the surveyor uploads the vehicle photos through the APP, the built-in image recognition module of the system can automatically verify the integrity of the photos and extract the key fields, reduces the work burden of the surveyor, improves the efficiency and accuracy of data uploading, and also provides accurate information for subsequent repair price and spare part preparation; the repair demand adopts a structured form, such as checking the “urgent” and “car replacement compensation” options, and the system automatically generates a repair priority label; this structured management mode makes the repair demand more clear and explicit, and also provides a strong reference basis for subsequent pricing models and repair progress arrangement.

[0031] In one embodiment of the present application, the S2 comprises:

[0032] S21, inputting features in the XGBoost integrated learning model and outputting results;

[0033] S22, the XGBoost integrated learning model quantifies the influence weight of each feature on the price through SHAP value analysis;

[0034] S23, a sliding window mechanism is adopted to real-time statistics the deviation distribution of the franchised repair shop price and the system benchmark price, and the floating threshold is dynamically adjusted through Monte Carlo simulation;

[0035] S24, for the price exceeding the threshold, triggering an abnormal mark and being associated to the anti-collusion detection step of S4;

[0036] S25, in combination with the OCR technology, analyzing the spare part list picture uploaded by the franchised repair shop, comparing with the system spare part database (containing SKU code, market price), and verifying the reasonableness of the price.

[0037] The working principle of the above technical solution is: input features:

[0038] Historical maintenance type: records maintenance history data of different vehicle models and different fault types, providing reference for pricing.

[0039] Parts cost: the system dynamically captures supply chain prices to ensure the real-time and accuracy of parts cost;

[0040] Work time coefficient: according to the complexity and required time of maintenance tasks, providing reference for work time in pricing;

[0041] Regional economic level: considering the economic development level and consumption level of different regions, the pricing is adjusted appropriately;

[0042] Seasonal fluctuation factor: considering the seasonal fluctuations in maintenance demand, the pricing is seasonally adjusted;

[0043] Output result:

[0044] Benchmark pricing interval: the XGBoost integrated learning model outputs a reasonable benchmark pricing interval based on the input features, providing a reference standard for the pricing of franchise maintenance shops; the XGBoost integrated learning model analyzes the SHAP (Shapley Additive Explanations) value to quantify the influence weight of each feature on pricing; SHAP value is based on Shapley value in game theory, which can fairly allocate the contribution of each feature to the model prediction result, helping to understand how the model makes predictions; through SHAP value analysis, it can clearly see which features have the greatest impact on pricing, providing a basis for subsequent pricing adjustment and optimization; a sliding window mechanism (such as a 30-day window) is used to real-time statistics of the deviation distribution of franchise maintenance shop pricing and system benchmark price; this helps to timely discover the trend of pricing deviation, providing a basis for dynamically adjusting the floating threshold; through Monte Carlo simulation, the floating threshold (such as 10%-20%) is dynamically adjusted to avoid the misjudgment caused by static threshold; for pricing that exceeds the threshold, an abnormal flag is triggered, which helps to quickly identify possible problems in pricing, providing a basis for subsequent investigation and processing; the abnormal flag of pricing is associated with the anti-collusion detection step of S4; through further analysis and investigation, it is judged whether there is a collusive pricing behavior among franchise maintenance shops; combined with OCR technology to analyze the parts list pictures uploaded by franchise maintenance shops; compare the extracted parts list with the system parts database (including SKU code, market price); through comparison, it can be judged whether the pricing of franchise maintenance shops is reasonable, whether there is false reporting, omission, etc.

[0045] The effect of the above technical solution is: through the XGBoost integrated learning model, multiple factors such as historical maintenance type, accessory cost, work time coefficient, regional economic level and seasonal fluctuation factor are comprehensively considered, and the benchmark quotation interval is output, which can more accurately reflect the actual maintenance cost and improve the accuracy of the quotation; at the same time, the SHAP value analysis is used to quantify the influence weight of each feature on the quotation, so that the quotation process is more transparent, and it is convenient for understanding and management; the sliding window mechanism is adopted to realize real-time statistics of the deviation distribution of the franchise maintenance shop quotation and the system benchmark price, which can dynamically reflect the market dynamics and the change of the franchise maintenance shop quotation behavior; the floating threshold is dynamically adjusted through Monte Carlo simulation, which avoids the mis-screening problem caused by the static threshold, so that the quotation system is more adaptable to market changes, and the flexibility of the quotation is enhanced; the quotation exceeding the threshold triggers an abnormal mark, and is associated with the anti-collusion detection step of S4, which can quickly identify and handle abnormal quotations, and prevent unreasonable quotation behaviors from affecting the market order; this mechanism helps to timely discover and handle collusive quotation behaviors among franchise maintenance shops, and maintains a fair competitive environment in the market; combined with the OCR technology, the accessory list picture uploaded by the franchise maintenance shop is analyzed, and is compared with the system accessory database, which can automatically verify the rationality of the quotation, improve the efficiency of the quotation audit, reduce human errors, and enhance the accuracy of the quotation audit; dynamically grabbing the supply chain price as the input feature of the accessory cost makes the quotation system able to timely reflect the change of the accessory price, which is helpful for the optimization of supply chain management; at the same time, through the feedback mechanism of the quotation system, the franchise maintenance shop can more reasonably control the accessory price, and the efficiency of the entire maintenance service market can be improved.

[0046] In an embodiment of the present application, the S23 comprises:

[0047] According to historical data analysis and business needs, the size of the sliding window is set; and the sliding step is selected;

[0048] The quotation data submitted by the franchise maintenance shop is preprocessed, and for each franchise maintenance shop, the relative deviation or absolute deviation between its quotation and the system benchmark price is calculated;

[0049] In the sliding window, the deviation distribution of all franchise maintenance shops is counted, and based on the historical deviation data, a large number of possible deviation distribution samples are generated using the Monte Carlo simulation method;

[0050] For each floating threshold (for example, 10%-20%), the probability of mis-screening (i.e. reasonable quotation is judged as abnormal) under the threshold is calculated; by comparing the mis-screening probabilities under different thresholds, the threshold with the minimum mis-screening probability is selected as the current optimal floating threshold;

[0051] The Monte Carlo simulation is periodically re-run (e.g., weekly or monthly) to dynamically adjust the float threshold to adapt to new market conditions based on external factors (e.g., market changes, seasonal factors, fluctuations in the price of parts, etc.).

[0052] The working principle of the above technical solution is as follows: according to historical data analysis and business requirements, the size of the sliding window is set, for example, 30 days, which means that the system will be based on the data of the past 30 days to calculate the deviation of the franchise repair shop price from the system benchmark price; the size of the sliding window needs to balance the sufficiency and timeliness of the historical data to ensure that it can reflect the recent market dynamics and not be affected by short-term fluctuations; the sliding step is selected to determine the frequency of window movement; if more detailed deviation changes are needed, a smaller sliding step is selected, such as sliding once a day or every few days; the selection of the sliding step affects the granularity and real-time of data analysis, and a smaller step can detect market changes faster; pre-process the price data submitted by the franchise repair shop to ensure the accuracy and consistency of the data; for each franchise repair shop, calculate the relative deviation or absolute deviation between its price and the system benchmark price; the relative deviation considers the percentage difference between the price and the benchmark price, which is suitable for comparing different magnitudes of prices; the absolute deviation is the difference between the price and the benchmark price, which is suitable for specific numerical comparisons; within the sliding window, the deviation distribution of all franchise repair shops is calculated, including statistical indicators such as the mean, standard deviation, maximum, and minimum of the deviation; these statistical indicators are used to describe the overall characteristics and fluctuations of the deviation, providing basic data for subsequent Monte Carlo simulation; based on historical deviation data, a large number of possible deviation distribution samples are generated using the Monte Carlo simulation method; Monte Carlo simulation reflects the possible deviation between the franchise repair shop price and the system benchmark price under different market conditions through random sampling and simulation calculations; the simulation results are used to evaluate the mis-screening probability under different floating thresholds; for each possible floating threshold (e.g. 10%-20%), the probability of mis-screening (i.e. reasonable price being judged as abnormal) under that threshold is calculated; the mis-screening probability is an important indicator for evaluating the reasonableness of the threshold, and a lower mis-screening probability means higher accuracy and reliability; by comparing the mis-screening probabilities under different thresholds, the threshold that minimizes the mis-screening probability is selected as the current optimal floating threshold; the selection of the optimal threshold needs to consider the mis-screening probability, market actual situation, and business requirements; the existing pricing system usually compares the deviation between the benchmark price and the franchise repair shop price, and may use a fixed threshold to determine whether the deviation is too large, which may not capture some dynamic market fluctuations in time; and the existing technology often uses static rules to determine whether the price is reasonable, rather than simulating the mis-screening probability under different floating thresholds, resulting in insufficient accuracy of price monitoring; based on external factors (such as market changes, seasonal factors, and fluctuations in parts prices), the Monte Carlo simulation is run regularly (e.g. weekly or monthly); regularly re-running the simulation ensures that the floating threshold adapts to the new market environment, maintaining the accuracy and effectiveness of price management; when adjusting the floating threshold, consider regional economic level factors to set different threshold ranges for different regions; introduce seasonal fluctuation factors to reflect the cyclical changes in repair demand and adjust the threshold to adapt to seasonal fluctuations.

[0053] The effect of the above technical solution is: by setting the sliding window size and calculating the deviation of the franchise repair shop price from the system benchmark price based on data from a certain period of time (such as 30 days), the recent market dynamics and changes in franchise repair shop pricing behavior can be more accurately reflected, the benchmark price adjustment strategy can be optimized, the prediction ability of the system can be improved, and the data analysis ability can be enhanced, by calculating the relative deviation and absolute deviation, the difference between the price and the benchmark price is considered comprehensively, the comprehensiveness and accuracy of deviation identification are improved, the identification ability of abnormal price fluctuations is improved, and the sensitivity to large-scale price changes is enhanced; selecting a suitable sliding step can capture the deviation changes more carefully according to business needs, such as sliding once a day or every few days, so that the price management system can more flexibly adapt to market changes, while better filtering noise and abnormal fluctuations can improve system stability; a large number of possible deviation distribution samples are generated by the Monte Carlo simulation method, which reflects the possible deviation between the franchise repair shop price and the system benchmark price under different market conditions, providing a more comprehensive perspective for price management and avoiding the limitations of single scenario analysis; for each possible floating threshold, the probability of mis-screening under the threshold is calculated, and by comparing the mis-screening probabilities under different thresholds, the threshold with the minimum mis-screening probability is selected as the current optimal floating threshold, reducing the risk of misjudging reasonable prices as abnormal; this data-driven threshold setting method makes the floating threshold more consistent with the actual market situation, improving the accuracy and fairness of price management; according to external factors, periodically re-run Monte Carlo simulation to dynamically adjust the floating threshold to adapt to the new market environment, so that the price management system can effectively respond to market changes, reduce the risk of market fluctuations, and avoid delays or errors that may occur in the manual adjustment process; considering regional economic level differences and seasonal fluctuation factors, these factors are introduced when adjusting the floating threshold, making the price management more consistent with the actual situation in different regions and different seasons, and improving the pertinence and effectiveness of the price management; by calculating the mean, standard deviation, maximum value, minimum value, etc. Statistical indicators of deviation, as well as generating a large number of possible deviation distribution samples, provide rich data support and decision-making basis for decision-makers.

[0054] In an embodiment of the present application, the S3 comprises:

[0055] S31, transmitting the franchise repair shop price data through a symmetric encryption algorithm, and dynamically distributing the key through an SM2 asymmetric encryption algorithm;

[0056] S32, storing the price hash value through the alliance chain;

[0057] S33, divide the geographical grid based on the GeoHash algorithm, screen the franchised repair shops within the preset range, and calculate the expected value of the repair time efficiency combined with real-time traffic data;

[0058] S34, construct a repair time efficiency prediction model, input historical repair time, weather data, and traffic flow, and output time confidence interval;

[0059] S35, define five-dimensional indicators: offer amount (30% weight), repair time efficiency (25%), shop rating (20%), accessory compliance rate (15%), and user complaint rate (10%), and support dynamic adjustment of weights (for example, focus on time efficiency during accident-prone periods such as severe weather);

[0060] S36, use NSGA-II multi-objective genetic algorithm to generate a Pareto optimal solution set, and the system recommends the top 3 candidate franchised repair shops and visualizes the scores in each dimension.

[0061] The working principle of the above technical solution is as follows: using a symmetric encryption algorithm (such as AES) to encrypt and transmit the quote data submitted by the franchise repair shop ensures the security of the data during transmission; the symmetric encryption algorithm uses the same key for encryption and decryption, has the advantages of algorithm openness, small calculation amount, fast encryption speed, and high encryption efficiency; the SM2 asymmetric encryption algorithm is used to dynamically distribute the symmetric encryption key; the SM2 algorithm generates a public key and a private key pair, the public key is used to encrypt the symmetric key, and the private key is used to decrypt. Calculate the hash value of the encrypted quote data (for example, use the SHA-256 algorithm), the hash value is unique, and any modification of the original data will cause the hash value to change; store the quote hash value through the alliance chain; once the data is tampered with, it can be found by comparing the stored hash value; use the GeoHash algorithm to encode the geographic location information into a short string, which represents a geographic grid; filter the franchise repair shops within the preset range through the GeoHash algorithm, improve the service response speed and customer satisfaction; combine real-time traffic data (such as congestion index) to calculate the expected value of the repair time; real-time traffic data reflects the current traffic conditions, and by considering traffic congestion, the time it takes for a franchise repair shop to reach the repair site can be more accurately estimated; select historical repair time, weather data, and traffic flow as input features; these features have an important influence on repair time, for example, bad weather can cause traffic congestion, which in turn affects repair time; use a long short-term memory network (LSTM) to build a repair time prediction model; by training the LSTM network, the complex relationship between input features and repair time can be learned, allowing future repair times to be predicted; the model outputs a time confidence interval rather than a single time point. The time confidence interval represents the uncertainty range of the prediction result, which helps users better understand the repair time; define five dimensions, including quote amount, repair time, store rating, accessory compliance rate, and user complaint rate, as indicators for evaluating franchise repair shops; each indicator has a corresponding weight, which reflects the importance of the indicator; the weight supports dynamic adjustment, for example, during periods of high accidents such as bad weather, the time efficiency can be focused on, and the weight of repair time can be increased; by dynamically adjusting the weight, the system can ensure that in different market environments, the comprehensive performance of the franchise repair shop can be more accurately evaluated; use the NSGA-II multi-objective genetic algorithm to generate a Pareto optimal solution set. NSGA-II is a non-dominated sorting genetic algorithm, which is suitable for solving problems with multiple conflicting objectives; in this solution, the NSGA-II algorithm is used to find the optimal combination of franchise repair shops in the five dimensions; the system recommends the top 3 candidate franchise repair shops from the Pareto optimal solution set; by visualizing the scores in each dimension, users can intuitively understand the strengths and weaknesses of each candidate franchise repair shop, allowing them to make more informed choices.

[0062] The effect of the above technical scheme is that: the symmetric encryption algorithm is used for transmitting the franchised repair shop quotation data, ensuring the confidentiality of the data in the transmission process and avoiding the risk of data leakage; at the same time, the key is dynamically distributed through the SM2 asymmetric encryption algorithm, further improving the security of key management, preventing the key from being illegally obtained or intercepted, thereby ensuring the integrity and confidentiality of the quotation data; the quotation hash value is stored through the alliance chain, and the tamper resistance of the blockchain is used to effectively prevent the data from being maliciously tampered with, thereby enhancing the credibility of the data; the geographic grid is divided based on the GeoHash algorithm, the franchised repair shops within the preset range are quickly screened, the real-time traffic data is combined to calculate the maintenance time efficiency expectation value, the arrival time of the maintenance service can be accurately estimated, the service response speed is improved, and the customer satisfaction is improved; the maintenance time efficiency prediction model is constructed, multi-dimensional information such as historical maintenance time, weather data, and traffic flow is input, and a time confidence interval is output. The LSTM network is good at processing time series data, can capture the complex relationship between maintenance time efficiency and various factors, thereby improving the accuracy of prediction, and can maintain good prediction stability and not be easily affected by changes in historical data to cause the performance of the model to decrease, and can better handle the timeliness and dynamics of time series data, ensuring that the prediction result is more accurate; the five-dimensional indexes of quotation amount, maintenance time efficiency, shop rating, accessory compliance rate, and user complaint rate are defined, and corresponding weights are set, the multi-dimensional evaluation method can more comprehensively reflect the comprehensive strength of the franchised repair shop, avoid the error of a single index, and improve the accuracy of evaluation. At the same time, the weights support dynamic adjustment, such as the accident-prone period of adverse weather, which can focus on time efficiency, so that the evaluation system is more flexible and has strong adaptability; the NSGA-II multi-objective genetic algorithm is used to generate a Pareto optimal solution set, and the system recommends the top 3 candidate franchised repair shops; the NSGA-II algorithm can process multiple conflicting objectives to find a set of balanced solutions, making the recommended results more reasonable and reliable; at the same time, the scores of each dimension are visually displayed, and the user can intuitively understand the advantages and disadvantages of each candidate franchised repair shop, thereby making a more intelligent choice.

[0063] In an embodiment of the present application, the S31 comprises:

[0064] S311, according to the data characteristics and security requirements, a symmetric encryption algorithm is selected, and a key length is set;

[0065] S312, the franchised repair shop quotation data is divided into several small blocks, and the data after the division is processed in parallel through multi-threading or distributed computing resources;

[0066] S313, after encryption processing, record the processing time of each data block; and, before encryption, calculate the checksum of the data and transmit the checksum together with the encrypted data; at the receiving end, calculate the checksum of the received encrypted data and compare it with the transmitted checksum; if the checksums are consistent, it indicates that the data has not been lost or incomplete during transmission;

[0067] S314, the system generates a pair of SM2 keys, including public key and private key; the public key is used to encrypt the key of the symmetric encryption algorithm, and the private key is used to decrypt; the system sends the public key to the franchise repair shop through a secure channel;

[0068] S315, the franchise repair shop uses the public key to encrypt the key of the symmetric encryption algorithm, and sends the encrypted key to the system; the system uses the private key to decrypt to obtain the key of the symmetric encryption algorithm, which is used to decrypt the franchise repair shop's quotation data;

[0069] S316, according to the data block processing time and transmission delay, predict the key expiration time, and dynamically generate a new key pair in advance, and distribute it to the franchise repair shop through SM2 algorithm;

[0070] S317, when the key is about to expire or needs to be updated, the system notifies the franchise repair shop to update the key through a secure channel; after receiving the notification, the franchise repair shop obtains the new symmetric encryption algorithm key according to the new key distribution mechanism;

[0071] S318, during data transmission, the system monitors the transmission status in real time and detects whether there is an exception; if the transmission exception is detected, the system starts the retry mechanism to resend the data.

[0072] The working principle of the above technical solution is: according to the data characteristics and security requirements, select the appropriate symmetric encryption algorithm; for example, select AES (Advanced Encryption Standard) as the symmetric encryption algorithm, AES encryption speed is fast, high security, and suitable for large amount of data encryption transmission; Set the key length of AES algorithm, such as 128 bits, 192 bits or 256 bits, to meet the needs of different security levels; The franchise repair shop quotation data is divided into several small blocks, so as to use multi-threaded or distributed computing resources for parallel encryption processing; Through parallel encryption, the encryption efficiency can be significantly improved, and the encryption time can be shortened. Traditional encryption algorithms are usually single-threaded or limited parallel, and there is a performance bottleneck when processing large amounts of data, especially in large-scale data processing environment and have not been popularized; At the same time, record the processing time of each data block, and use it in the subsequent key distribution strategy; Before encryption, calculate the checksum (such as MD5 or SHA-256 hash value) of the data, and transmit the checksum together with the encrypted data; At the receiving end, calculate the checksum of the received encrypted data and compare it with the transmitted checksum; If the checksum is consistent, it means that the data has not been lost or incomplete during transmission, ensuring the integrity of the data. Existing systems usually use simple checksum mechanism, but may not perform complete verification between encryption and decryption processes. The checksum is transmitted together with the data, which is less common and more dependent on a single verification mechanism to determine whether the data is transmitted completely; The system generates a pair of SM2 key pairs, including public key and private key; The public key is used to encrypt the symmetric encryption algorithm key, and the private key is used for decryption; The system sends the public key to the franchise repair shop through a secure channel (such as HTTPS or SSL / TLS protocol), ensuring the security of the public key during transmission; The franchise repair shop uses the received public key to encrypt the symmetric encryption algorithm key, and sends the encrypted key to the system; The system uses the private key to decrypt the symmetric encryption algorithm key, which is used to decrypt the franchise repair shop quotation data; According to the data block processing time and transmission delay, predict the key expiration time, and dynamically generate a new key pair in advance, which is distributed to the franchise repair shop through SM2 algorithm; When the key is about to expire or needs to be updated, the system notifies the franchise repair shop to update the key through a secure channel, enhancing the security of use. After receiving the notification, the franchise repair shop obtains the new symmetric encryption algorithm key according to the new key distribution mechanism; Through the key update mechanism, the system can continuously use the secure key for data encryption and decryption. Most existing technologies rely on static key management mechanism. Once the key is generated, it is usually not updated regularly, or it is updated manually, without automatic update and expiration processing mechanism, which greatly increases the risk of key leakage; During data transmission, the system monitors the transmission status in real time and detects whether there are any abnormalities (such as network interruption, data transmission timeout, etc.); If transmission anomalies are detected, the system starts the retry mechanism to resend the data;Through the transmission state monitoring and retry mechanism, the reliable transmission and integrity of data are ensured.

[0073] The effect of the above technical solution is that the symmetric encryption algorithm improves the encryption speed and enhances the security performance; a large amount of data is encrypted and transmitted, meeting the dual requirements of efficiency and security for data transmission; the franchise repair shop quotation data is divided into several small blocks, and parallel encryption processing is performed using multi-threading or distributed computing resources, significantly improving the encryption efficiency, fully utilizing the computing resources, shortening the encryption time, and improving the overall processing capacity of the system; in a distributed environment, the encryption task can be completed smoothly, and the reliability and fault tolerance of the system are enhanced; the system can continue to run when a node fails, and will not be interrupted due to a single point failure, thereby enhancing the fault tolerance of the system; the checksum (such as MD5 or SHA-256 hash value) of the data is calculated before encryption, and the checksum is transmitted together with the encrypted data; at the receiving end, the checksum of the received encrypted data is calculated and compared with the transmitted checksum to ensure that the data has not been lost or incomplete during transmission, effectively ensuring the integrity and reliability of the data; the SM2 asymmetric encryption algorithm is used to dynamically distribute the key of the symmetric encryption algorithm, enhancing the security of key management; the system generates a pair of SM2 keys, the public key is used to encrypt the key of the symmetric encryption algorithm, and the private key is used for decryption, ensuring the security of the key during transmission and preventing the key from being illegally obtained or intercepted; by predicting the key expiration time and generating a new key pair in advance, security vulnerabilities caused by long-term use or expiration of the key are avoided, and the possibility of attackers cracking the key is reduced; the SM2 national encryption algorithm is used, which is based on elliptic curve cryptography (ECC) and has shorter key length and higher encryption efficiency under the same security level, effectively resisting classical attack methods before quantum computing; the dynamic key generation and distribution mechanism can smoothly transition the key update process, avoiding service suspension or data encryption failure caused by key expiration; by predicting the transmission delay, the key distribution time is planned in advance to ensure that the key update can be completed on time even when the network is unstable, reducing the key synchronization failure rate caused by network problems. By introducing the data block processing time and transmission delay, the usage and expiration time of the key can be more accurately predicted, thereby realizing more accurate and real-time key distribution, ensuring the accuracy and real-time performance of the key distribution, and improving the efficiency and security of the entire encryption system.

[0074] In one embodiment of the present application, the S312 comprises:

[0075] Before blocking, the franchise repair shop quotation data is preprocessed, and a blocking strategy is developed according to the characteristics of the data and the requirements of the encryption algorithm;

[0076] For linear data structures such as arrays or lists, a sequential blocking algorithm is used to divide the data into several small blocks from head to tail.

[0077] For complex data structures such as hash tables or graphs, a hash chunking algorithm is used to distribute the data into different chunks according to their hash values;

[0078] For large files, the memory-mapped file technique is used to map the large file into memory, and the data in memory is chunked;

[0079] According to the parallel processing capability of the system, a thread pool or process pool is created for executing parallel encryption tasks; the chunked data is distributed to each thread or process in the thread pool or process pool for encryption processing through static allocation (i.e. each thread or process handles a fixed number of chunks) or dynamic allocation (i.e. dynamically adjusts the number of allocated chunks according to the load of the thread or process);

[0080] The symmetric encryption algorithm is encapsulated into a function that accepts data chunks and keys as input and outputs encrypted data chunks; the encryption function is executed in parallel in each thread or process to encrypt the data chunks assigned to it;

[0081] The encrypted data chunks from each thread or process are merged into complete encrypted data through sequential merging or parallel merging.

[0082] The working principle of the above technical solution is as follows: before the franchise repair shop quotation data is blocked, preprocessing is first performed, including removing redundant parts in the data to reduce data volume and improve encryption efficiency; the data is compressed to further reduce the storage space occupied by the data, facilitating subsequent blocking and encryption operations; a suitable blocking strategy is developed based on the characteristics of the data and the requirements of the encryption algorithm; the block size can be determined based on factors such as the size of the data, the block size of the encryption algorithm, memory limitations, and parallel processing capabilities; fixed-size blocks can be selected, or the data can be blocked based on natural boundaries (such as records, rows, or columns) to adapt to the characteristics of different data structures; for linear data structures (such as arrays or lists), a sequential blocking algorithm is used to divide the data from head to tail into several small blocks; for complex data structures (such as hash tables or graphs), a hash blocking algorithm is used to distribute the data into different blocks based on its hash value; for large files, the mmap (memory mapping file) technique is used to map large files into memory and block the data in memory; although existing technologies use fixed block sizes or blocking methods based on natural data boundaries, they do not have special processing for complex data structures (such as hash tables, graphs, etc.), and existing parallel encryption is usually based on static task allocation and simple file reading, lacking flexible dynamic load distribution and memory optimization; and, existing technologies rarely consider the mmap file technique, which usually requires reading all data at once before processing, resulting in high memory usage and I / O overhead; a thread pool or process pool is created based on the parallel processing capabilities of the system to execute parallel encryption tasks; the size of the thread pool or process pool can be determined based on the number of CPU cores, memory size, and parallel encryption requirements; through static allocation (i.e., each thread or process handles a fixed number of blocks) or dynamic allocation (i.e., dynamically adjusting the number of blocks allocated based on the load of the thread or process), the blocked data is distributed to each thread or process in the thread pool or process pool for encryption processing; a symmetric encryption algorithm (such as AES) is encapsulated into a function that accepts data blocks and keys as input and outputs encrypted data blocks; the encryption function should have a good interface design to facilitate parallel execution in various threads or processes; the encryption function is executed in parallel in each thread or process to encrypt the data blocks assigned to it; through parallel processing, encryption efficiency can be significantly improved, and encryption time can be shortened; the encrypted data blocks from each thread or process are merged into complete encrypted data through sequential merging or parallel merging (such as using the idea of merge sort); the order and integrity of the data blocks should be ensured during the merging process to ensure that the original data can be correctly decrypted and restored at the receiving end, while traditional encryption schemes usually write encrypted data directly to the target file without involving complex merging processes.

[0083] The effect of the above technical scheme is that: through the preprocessing step to remove redundant data and compress data, the amount of data that needs to be encrypted is reduced, thereby significantly improving the efficiency of the encryption process; by formulating a reasonable blocking strategy, the blocking size is determined according to the data characteristics and the requirements of the encryption algorithm, so that the encryption process is more efficient, the parallel processing capability of encryption is improved, and the requirements of different encryption algorithms also improve the memory management and data transmission performance to some extent; the thread pool or process pool is used for parallel encryption task processing, the multi-core processing capability of the system is fully utilized, the encryption speed is further improved, and the delay of the encryption process is reduced; for linear data structures (such as arrays or lists), a sequential blocking algorithm is used, for complex data structures (such as hash tables or graphs), a hash blocking algorithm is used, and for large files, a memory mapping file (mmap) technology is used for blocking, so that the scheme can adapt to various types of data structures, improve the efficiency and response speed of data processing, and optimize the use of resources; such flexibility enables the scheme to perform outstanding performance in a wider range of application scenarios; through static allocation or dynamic allocation strategy, the blocked data is allocated to each thread or process in the thread pool or process pool for encryption processing, which optimizes the use of system resources, reduces the overhead of task scheduling, ensures that the load is evenly distributed among the threads or processes, thereby avoiding the overload of some threads or processes, which leads to a decrease in processing efficiency; the dynamic allocation strategy can dynamically adjust the number of blocks allocated according to the load of the threads or processes, avoiding the situation of idle or overloaded resources, and improving the stability and reliability of the system; the symmetric encryption algorithm (such as AES) is encapsulated into a function, which accepts data blocks and keys as input and outputs encrypted data blocks, simplifying the implementation of the encryption process, so that the encryption function can be executed in parallel in each thread or process, improving the reusability and maintainability of the code, reducing the error rate, and improving the maintainability and scalability of the system while enhancing security; the encrypted data blocks of each thread or process are merged into complete encrypted data through sequential merging or parallel merging, ensuring the integrity of the data; checksum, hash value and other methods can be used for data integrity verification during the merging process to ensure that the encrypted data is not tampered with or damaged during transmission or storage.

[0084] In one embodiment of the present application, the S4 comprises:

[0085] S41, based on time series analysis, calculate the volatility rate standard deviation of the franchise repair shop price, if the volatility rate synchronization rate of more than 3 shops in the region is more than 90%, it is determined as potential collusion;

[0086] S42, based on graph neural network, construct the relationship topology graph of the franchise repair shop, detect abnormal cooperative mode through node embedding feature;

[0087] S43, assign a dynamic credit score to high-risk franchise repair shops, with an initial value of 100 points, deduct 20 points for each collusion detected, and limit their order receiving rights when the score is below 60;

[0088] S44, update the detection rules in real time by fusing new data through online learning algorithms.

[0089] The working principle of the above technical solution is as follows: first, use the ARIMA model to analyze the time series data of the franchise repair shop's quotes; the ARIMA model is a time series prediction model that can effectively capture trends, seasonality, and random fluctuations in time series data; through the ARIMA model, calculate the standard deviation of the volatility rate of each franchise repair shop's quotes, which reflects the degree of volatility of the quotes; then, compare the volatility rates of multiple stores in the region. If the volatility rates of more than 3 stores in the region are synchronized (i.e., the consistency of the volatility trend) exceeds 90%, it is determined that these stores have potential collusion behavior; a synchronization rate exceeding 90% means that the price volatility trends of these stores are highly consistent, which may involve price manipulation or coordinated behavior; use a graph neural network (GNN) to construct a relationship topology graph of franchise repair shops; GNN is a deep learning model specifically designed to handle graph-structured data, which can capture complex relationships in data; in the relationship topology graph, each franchise repair shop is considered a node, and the edges between nodes represent the relationship between stores (such as geographical proximity, frequent business transactions, etc.); learn the node embedding features through GNN, and extract feature representations that can reflect the relationship between nodes; use these feature representations to detect abnormal coordination patterns in the graph, such as small groups of frequent similar quotes. The stores in these small groups may have collusion behavior and manipulate market prices through coordinated pricing; use GNN to detect abnormal coordination patterns in the relationship network to further confirm the existence of collusion behavior; assign a dynamic credit score to franchise repair shops that are determined to be high-risk (i.e., have potential collusion behavior); the initial credit score is set to 100 points; deduct 20 points for each collusion detected; when the credit score is below 60, limit the order receiving rights of the franchise repair shop to prevent it from continuing to participate in market manipulation behavior; update the detection rules in real time by fusing new data through online learning algorithms; online learning algorithms can handle streaming data and constantly adapt to changes in the market environment; as new data arrives, the algorithm can continuously optimize and adjust the detection rules to improve the accuracy and robustness of the detection.

[0090] The effect of the above technical solution is that: by calculating the volatility rate standard deviation of the franchise repair shop price through time series analysis, and comparing the volatility rate synchronization rate of multiple shops in the region, potential collusion behavior can be accurately identified; this method is based on the volatility characteristics of the data itself, avoiding the inaccuracy of subjective judgment, improving the accuracy of detection, and enhancing the risk warning ability; using a graph neural network (GNN) to construct a relationship topology graph of the franchise repair shop can reveal the potential relationship network between the shops; by detecting abnormal coordination patterns such as small groups of frequent pricing through node embedding features, the existence of collusion behavior can be further confirmed; this method can handle complex data relationships, discover hidden collusion patterns, enhance the supervision ability of collusion behavior, and improve detection efficiency and accuracy; a dynamic credit score is assigned to high-risk franchise repair shops, and the credit score is deducted according to the number of collusion behaviors, and the order receiving authority is limited when it is below a certain threshold, effectively restraining and punishing high-risk franchise repair shops, encouraging them to abide by market rules, and maintaining a fair competitive environment in the market; through an online learning algorithm, new data is fused in real time to update the detection rules, so that the detection system can continuously adapt to changes in the market environment; as new data arrives, the detection rules can be continuously optimized and adjusted to improve the accuracy and robustness of detection; this adaptability enables the detection system to remain effective for a long time; by accurately detecting collusion behavior, revealing complex relationships, enhancing credit management, and improving the adaptability of detection rules, it helps to improve the overall efficiency and fairness of the market; the reduction of collusion behavior can promote market competition and improve resource allocation efficiency; the enhancement of credit management can maintain market order and protect consumer rights; the adaptability of the detection rules can ensure that market rules keep pace with the times and meet the needs of market development.

[0091] In one embodiment of the present application, the S42 comprises:

[0092] S421, collect corresponding data between franchise repair shops, and preprocess the collected corresponding data;

[0093] S422, regarding each franchise repair shop as a node in a graph, the node containing basic information of the franchise repair shop; defining edges between nodes according to relevant factors between franchise repair shops; wherein the weight of the edge represents the correlation strength between the franchise repair shops;

[0094] S423, combining nodes and edges into a relationship topology graph of the franchise repair shop using graph theory knowledge;

[0095] S424, extracting initial features for each node, and training a GNN model using the extracted node features and the structure information of the topology graph;

[0096] S425, generate embedding vectors for each node through the trained GNN model; and perform dimension reduction processing on the generated embedding vectors to visualize the distribution of nodes in two-dimensional or three-dimensional space;

[0097] S426, calculate the similarity of embedding vectors between nodes through cosine similarity, and detect frequent quote similar small groups based on the similarity between nodes using a clustering algorithm;

[0098] S427, identify abnormal collaborative patterns in combination with business knowledge and expert knowledge base; and feed back the detection results to the system in a timely manner; at the same time, use the detection results as the input of the online learning algorithm to update the detection rules in real time.

[0099] The working principle of the above technical solution is as follows: collect the corresponding data between the franchised repair shops, including transaction data and quotation data; these data are the basis for constructing the relationship topology graph; preprocess the collected data, such as removing outliers, filling missing values, and data standardization, to ensure the quality and consistency of the data; regard each franchised repair shop as a node in the graph, and the node contains the basic information of the franchised repair shop, such as shop ID, location, scope of operation, etc.; define the edges between nodes according to the relevant factors between franchised repair shops, such as transaction relationship, quotation similarity, and interaction frequency. The weight of the edge represents the correlation strength between the franchised repair shops, reflecting the closeness between them; use the knowledge of graph theory to combine nodes and edges into the relationship topology graph of the franchised repair shops. This graph reflects the complex relationship network between the franchised repair shops; extract the initial features for each node, including the basic information of the franchised repair shop, historical transaction data, and quotation mode, etc.; use the extracted node features and the structural information of the topology graph to train the GNN model; the GNN model can capture the complex relationships between nodes and learn the embedding representation of the nodes; generate embedding vectors for each node through the trained GNN model; the embedding vectors capture the local and global structural information of the nodes in the topology graph; perform dimensionality reduction processing (such as PCA, t-SNE, etc.) on the generated embedding vectors, and visualize the distribution of nodes in two-dimensional or three-dimensional space; calculate the similarity of the embedding vectors between nodes through cosine similarity; based on the similarity between nodes, use clustering algorithms (such as K-means, DBSCAN, etc.) to detect small groups with frequent quotation similarity; traditional clustering methods, such as K-means, DBSCAN, etc. clustering algorithm, they usually have high requirements for data preprocessing and parameter setting, and are difficult to handle data with complex structure and dependency; the franchised repair shops in these small groups may have potential collusion behavior; combine business knowledge and expert knowledge base to identify abnormal collaboration patterns. For example, some franchised repair shops have highly similar quotations within a certain time period, and there is no other reasonable explanation, which may be considered as potential collusion behavior; feedback the detection results to the system in a timely manner, so that the system can take appropriate measures (such as limiting the order receiving authority, conducting investigation, etc.); at the same time, use the detection results as the input of the online learning algorithm to update the detection rules in real time to adapt to the changes of the market environment; traditional methods rely on rules or simple statistical methods for anomaly detection, which are difficult to dynamically adapt to changes in data.

[0100] The effect of the above technical solution is that: by collecting transaction data and quotation data between franchise repair shops and constructing a relationship topology graph, the complex relationship network between franchise repair shops can be fully revealed, which helps to understand the interaction mode between franchise repair shops, provides a basis for subsequent anomaly detection, and improves the accuracy of anomaly detection; initial features including basic information, historical transaction data, and quotation mode are extracted for each franchise repair shop node, which can accurately reflect the operating conditions and quotation behavior of the franchise repair shop, enhance the insight into market trends and competitive situation, and reduce the misjudgment rate; through the training of the GNN model, the features are further fused and utilized to generate embedding vectors that can capture the local and global structure information of the nodes in the topology graph; by using the embedding vectors generated by the trained GNN model, the similarity between nodes is calculated by cosine similarity, and a clustering algorithm is used to detect small groups with frequent similar quotations, which can effectively identify small groups that may have potential collusion behavior, provide protection for fair competition in the market, and improve the accuracy of collusion detection; the generated embedding vectors are dimensionally reduced, and the distribution of the nodes in two-dimensional or three-dimensional space is visualized, which helps to intuitively understand the relationship network between franchise repair shops and the distribution of abnormal coordination patterns, and further deepens the understanding of the data; when identifying abnormal coordination patterns, combining business knowledge and expert knowledge base can more accurately judge whether the quotation behavior is abnormal; this combination of business practice and expert experience improves the accuracy and reliability of detection; the detection results are used as input for online learning algorithms to update detection rules in real time. This makes the detection system adapt to changes in the market environment and continuously improve the accuracy and efficiency of detection.

[0101] In one embodiment of the present application, the S426 comprises:

[0102] Before calculating the similarity, the generated embedding vectors are standardized, and for each pair of nodes in the topology graph, the cosine similarity between their embedding vectors is calculated by matrix operation or loop traversal;

[0103] According to the characteristics of the data and the requirements of the clustering task, a clustering algorithm is selected;

[0104] The similarity matrix is used as input, and the selected clustering algorithm is applied for clustering analysis;

[0105] For each cluster, analyze the nodes within the cluster, and determine whether it belongs to a small group with frequent similar quotations by calculating the statistical characteristics (such as mean, variance, etc.) of the quotation data of the nodes within the cluster;

[0106] According to the number of nodes within the cluster and the degree of similar quotations, the size of the small group is determined; combined with the actual situation of electric vehicle inspection business, the quotation behavior of the small group is analyzed to determine whether it conforms to the normal logic.

[0107] The working principle of the above technical solution is: before calculating the similarity, the embedding vector generated by the GNN model is standardized. The standardization processing is to eliminate the dimensional difference between different dimensions, so that the similarity calculation is more accurate. The existing technology usually relies on traditional similarity measurement methods (such as Euclidean distance, Manhattan distance, etc.), but these methods may be affected by data scale and noise, resulting in unstable calculation results; for each pair of nodes in the topological graph, the cosine similarity between their embedding vectors is calculated through matrix operation or loop traversal. Cosine similarity is an index to measure the direction similarity of two vectors, and its value range is [-1, 1], the closer to 1, the more similar the two vectors; according to the characteristics of the data and the requirements of the clustering task, select a suitable clustering algorithm. For example, K-means is suitable for processing spherical clusters, while DBSCAN can discover clusters of arbitrary shape and has good robustness to noise data; set parameters for the selected clustering algorithm. For the K-means algorithm, the number of clusters K needs to be determined; for the DBSCAN algorithm, the neighborhood radius Eps and the minimum sample size MinPts need to be set; input the similarity matrix and apply the selected clustering algorithm for clustering analysis. The clustering algorithm divides the nodes into different clusters according to the similarity between the nodes; evaluate the clustering results to determine whether the cluster division is reasonable. Use indicators such as silhouette coefficient, Calinski-Harabasz index to evaluate the clustering effect. These indicators can reflect the tightness of nodes within the cluster and the separation degree of nodes between clusters; for each cluster, analyze the nodes within the cluster, and determine whether they belong to the small group of similar bidding by calculating the statistical characteristics (such as mean, variance, etc.) of the bidding data within the cluster; according to the number of nodes within the cluster and the degree of similar bidding, determine the size of the small group. Small groups with larger size may be more worthy of attention because they may have more obvious collusion behavior; combined with the actual situation of electric vehicle inspection business, analyze whether the bidding behavior of the small group conforms to the normal logic. For example, some franchise repair shops have highly similar bidding in a certain time period, and there is no other reasonable explanation, which may be considered as potential collusion behavior. And the existing technology mostly uses fixed clustering algorithms (such as K-means or DBSCAN), which may not consider the characteristics of different types of data, and do not further optimize the behavior analysis after clustering, and the existing method only performs simple clustering statistics (such as cluster size, density, etc.) after clustering, lacking in-depth analysis and logical verification of the behavior of nodes within the cluster.

[0108] The effect of the above technical scheme is that: before calculating the similarity, the generated embedding vector is standardized, which eliminates the dimensional difference between different dimensions, makes the calculation of the cosine similarity more accurate and reliable, helps to more accurately reflect the similarity between nodes, provides accurate basic data for subsequent clustering analysis, improves the performance and generalization ability of the model, and enhances the reliability of the clustering analysis; according to the characteristics of the data and the requirements of the clustering task, a suitable clustering algorithm is selected, and reasonable parameters are set, and the flexible design makes the clustering algorithm better adapt to different data distribution and clustering requirements, improves the accuracy and stability of the clustering effect; the clustering results are evaluated by using indicators such as the silhouette coefficient and the Calinski-Harabasz index, and whether the cluster division is reasonable is determined, these evaluation indicators can objectively reflect the closeness of the nodes in the cluster and the separation degree of the nodes between the clusters, provide a strong basis for optimizing the clustering effect, improve the efficiency and accuracy of the clustering analysis, and promote the improvement and iteration of the model; through clustering analysis, similar nodes are divided into the same cluster, and statistical feature analysis (such as mean, variance, etc.) is performed on the nodes in each cluster, and the small group of similar frequent quotation is accurately identified; it is helpful to reveal the quotation behavior mode of the franchised repair shops; according to the number of nodes in the cluster and the degree of similar quotation, the size of the small group is determined, and the small group with larger size is focused on. The small group with larger size may have more obvious collusion behavior, so this attention is helpful to timely discover and handle the potential collusion behavior; combined with the actual situation of the electric vehicle inspection business, whether the quotation behavior of the small group conforms to the normal logic is analyzed. This analysis method combined with the actual business improves the accuracy and reliability of the collusion behavior identification, and provides strong evidence support for subsequent investigation and handling.

[0109] In one embodiment of the present application, the S5 comprises:

[0110] S51, real-time acquisition of maintenance site video clips based on maintenance progress, verification of accessory replacement authenticity through a target detection model;

[0111] S52, the system compares vehicle photos before and after maintenance, calculates the image similarity using a Siamese network, and triggers system review if the difference is below the threshold;

[0112] S53, the accessory list is automatically associated with the maintenance order, the system calls a knowledge graph to verify the compatibility of the accessories, and automatically generates a compensation scheme for the false reporting of accessories;

[0113] S54, after acceptance, the system automatically triggers settlement through a smart contract, user evaluation data flows back to the store rating module, forming a "quotation-maintenance-feedback" closed loop optimization.

[0114] The working principle of the above technical solution is as follows: during the key nodes in the maintenance process (for example, when starting maintenance, when maintenance is half), the system automatically acquires video clips of the maintenance site; the system analyzes the key frames in the video clips through a target detection model (for example, Faster R-CNN) to identify and verify whether the spare parts are truly replaced; the target detection model can accurately identify the type and position of the spare parts and compare them with the spare part information in the maintenance order to ensure the authenticity of the replacement of the spare parts; the system obtains vehicle photos before and after maintenance and uses a Siamese network to calculate the similarity between the two photos; the Siamese network is a neural network structure used to measure the similarity between two inputs and can accurately calculate the similarity between images; if the similarity between the vehicle photos before and after maintenance is lower than a set threshold (for example, 85%), the system triggers a review to further confirm the maintenance effect; the spare parts list is automatically associated with the maintenance order, and the system can accurately track the spare part information used for each maintenance order; the system calls a knowledge graph (for example, built using Neo4j) to verify the compatibility of the spare parts. The knowledge graph stores the compatibility relationship between vehicle models and spare parts, such as a certain vehicle model only supporting a specific battery model; for false reporting of spare parts, the system automatically generates a compensation plan according to the preset rules, such as deducting the deposit by 3 times the market price; the false reporting of spare parts is punished to maintain market order; after the maintenance service acceptance, the system automatically triggers the settlement process through a smart contract (for example, using Hyperledger Fabric); the smart contract is an automatically executed contract that can automatically execute preset operations such as transfer and deduction when certain conditions are met; user evaluation data flows back to the store rating module to evaluate the service quality and credibility of the store; through the closed-loop optimization mechanism of "quotation-maintenance-feedback", the service quality and user satisfaction are continuously improved.

[0115] The effect of the above technical scheme is that: by acquiring the maintenance site video segment in real time, and verifying the authenticity of the accessory replacement by using a target detection model (such as Faster R-CNN), the transparency of the maintenance process is significantly improved, which helps to prevent false maintenance or the case that the accessory is not replaced, and enhances the trust of consumers on the maintenance service; the system compares the vehicle photos before and after maintenance, and calculates the image similarity using a Siamese network. This objective and automatic evaluation method can accurately reflect the maintenance effect, and if the difference is lower than the threshold, the system triggers a review, further ensuring the maintenance quality; the accessory list is automatically associated with the maintenance order, the system calls the knowledge graph to verify the compatibility of the accessory, preventing maintenance failure or safety hazards caused by incompatible accessories, and at the same time, automatically generating a compensation scheme for false accessory reporting, maintaining market order and consumer rights and interests; after acceptance, the system automatically triggers the settlement process through the smart contract, reducing manual intervention, improving settlement efficiency and accuracy, and reducing the risk of human error; user evaluation data flows back to the store rating module, forming a closed-loop optimization mechanism of "quotation-maintenance-feedback". This helps the store to understand the service quality and user satisfaction in a timely manner, so as to continuously adjust and optimize the service process and improve the overall service quality; through the implementation of the above technical scheme, consumers can more intuitively understand the maintenance process and have a more objective evaluation of the maintenance effect, and at the same time, the automatic settlement and closed-loop optimization mechanism also improves the user experience and trust. This helps to enhance the competitiveness of the store and promote the healthy development of the market.

[0116] In one embodiment of the present application, the intelligent management system for electric vehicle reconnaissance tasks based on multi-dimensional data collaboration includes a memory, a processor, and a computer program stored on the memory and executable on the memory. The processor executes the program to implement the intelligent management method for electric vehicle reconnaissance tasks based on multi-dimensional data collaboration as described in any of the above embodiments.

[0117] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. An intelligent management method for electric vehicle reconnaissance tasks based on multi-dimensional data collaboration, characterized in that, The method comprises: S1, the surveyor submits information to the system, and the franchise repair shop submits a preliminary quotation according to the preliminary repair demand pushed by the system; S2, an automatic quotation model is constructed; S3, surrounding price comparison is carried out, a quotation ranking is generated, and the best franchise repair shop recommendation is automatically generated; S4, the system detects abnormal cooperative behavior through historical quotation fluctuation rate, identifies and prevents collusive quotation behavior among franchise repair shops; for the franchise repair shop suspected of collusive quotation, the system marks it as high risk and pays attention to it in subsequent quotation; S5, after the repair is completed, the repair result is accepted, and the order settlement is completed after confirmation; S2 comprises: S21, input features in the XGBoost integrated learning model and output results; S22, the influence weight of each feature on the quotation is quantified through SHAP value analysis; S23, a sliding window mechanism is used to real-time statistics of the deviation distribution of the quotation of the franchise repair shop and the system benchmark price, and the floating threshold is dynamically adjusted through Monte Carlo simulation; S24, for the quotation exceeding the threshold, an abnormal mark is triggered and is associated to the anti-collusion detection step S4; S25, combined with the OCR technology, the parts list picture uploaded by the franchise repair shop is analyzed, compared with the system parts database, and the quotation rationality is verified; S4 comprises: S41, based on time series analysis, the fluctuation rate standard deviation of the quotation of the franchise repair shop is calculated; S42, based on graph neural network, a franchise repair shop relationship topology graph is constructed, and an abnormal cooperative mode is detected through node embedding features; S43, a dynamic credit score is allocated to the high-risk franchise repair shop; S44, new data is fused in real time through an online learning algorithm to update the detection rules. 2.The method of claim 1, wherein, S1 comprises: S11, based on the RBAC model, the permissions of the surveyor and the franchise repair shop are dynamically allocated; S12, the real-time positions of the franchise repair shop and the surveyor are bound; S13, based on the reinforcement learning model, the optimal surveyor allocation scheme is dynamically calculated; S14, when the order administrator newly creates an order, the franchise repair shop distribution is accurately matched; S15, when the surveyor uploads data through the APP, the system built-in image recognition module automatically verifies the integrity of the vehicle photo and extracts the key fields; S16, a structured form is used, and the system automatically generates a maintenance priority label. 3.The method of claim 1, wherein, S23 comprises: According to historical data analysis and business needs, the size of the sliding window is set; and the sliding step is selected; For each franchise repair shop, the relative deviation or absolute deviation between its quotation and the system benchmark price is calculated; In the sliding window, the deviation distribution of all franchise repair shops is counted, and based on historical deviation data, a large number of possible deviation distribution samples are generated using the Monte Carlo simulation method; For each floating threshold, the probability of false screening under the threshold is calculated; by comparing the false screening probabilities under different thresholds, the threshold that minimizes the false screening probability is selected as the current optimal floating threshold; According to external factors, the Monte Carlo simulation is periodically re-run to dynamically adjust the floating threshold to adapt to the new market environment.

4. The method of claim 1, wherein, S3 comprises: S31, transmit the franchised repair shop quotation data by a symmetric encryption algorithm, and distribute the key dynamically by an SM2 asymmetric encryption algorithm; S32, store the quotation hash value by an alliance chain; S33, divide a geographic grid based on a GeoHash algorithm, filter the franchised repair shops within a preset range, and calculate a maintenance time efficiency expectation value in combination with real-time traffic data; S34, construct a maintenance time efficiency prediction model, input historical maintenance time, weather data, and traffic flow, and output a time confidence interval; S35, define five-dimensional indexes; S36, generate a Pareto optimal solution set by using an NSGA-II multi-objective genetic algorithm, and recommend the top three candidate franchised repair shops, and visualize the scores in each dimension.

5. The method of claim 1, wherein, The S42 comprises: S421, collect corresponding data between the franchised repair shops, and pre-process the collected corresponding data; S422, regard each franchised repair shop as a node in a graph, the node contains basic information of the franchised repair shop, define edges between the nodes according to relevant factors between the franchised repair shops, and the weight of the edge represents the correlation strength between the franchised repair shops; S423, combine the nodes and the edges into a relationship topology graph of the franchised repair shops by using related knowledge of graph theory; S424, extract initial features for each node, and train a GNN model by using the extracted node features and the structure information of the topology graph; S425, generate an embedding vector for each node by using the trained GNN model, and perform dimension reduction processing on the generated embedding vector, so that the distribution of the nodes can be visualized in a two-dimensional or three-dimensional space; S426, calculate the embedding vector similarity between the nodes by using a cosine similarity, detect a small group of frequent similar quotations based on the similarity between the nodes by using a clustering algorithm; S427, identify an abnormal collaborative mode in combination with business knowledge and an expert knowledge base, feed back the detection result to the system in a timely manner, and update the detection rules in real time by taking the detection result as an input of an online learning algorithm.

6. The method of claim 5, wherein, The S426 comprises: Before calculating the similarity, perform standardization processing on the generated embedding vector, and calculate the cosine similarity between the embedding vectors of each pair of nodes in the topology graph by using matrix operation or loop traversal; According to the characteristics of the data and the requirements of the clustering task, select a clustering algorithm, and set parameters for the selected clustering algorithm; Input the similarity matrix, and perform clustering analysis by using the selected clustering algorithm; For each cluster, analyze the nodes in the cluster, and determine whether the cluster belongs to a small group of frequent similar quotations by calculating the statistical features of the quotation data of the nodes in the cluster; According to the number of nodes in the cluster and the degree of similar quotations, determine the size of the small group, and analyze whether the quotation behavior of the small group conforms to the normal logic in combination with the actual situation of the electric vehicle inspection business.

7. The method of claim 1, wherein, The S5 comprises: S51, obtain a maintenance site video clip in real time based on a maintenance progress, and verify the authenticity of a spare part replacement by using a target detection model; S52, compare vehicle photos before and after maintenance by using the system, calculate the image similarity by using a Siamese network, and trigger a system review if the difference is lower than a threshold value; S53, the accessory list is automatically associated with the repair order, the system calls the knowledge graph to verify the compatibility of the accessories, and automatically generates a compensation scheme for false reporting of accessories; S54, after acceptance, the system automatically triggers settlement through smart contract, user evaluation data flows back to the store rating module, forming a "quotation-repair-feedback" closed loop optimization.

8. An intelligent management system for electric vehicle reconnaissance tasks based on multi-dimensional data collaboration, characterized in that, The computer program stored in the memory and executable on the memory includes a memory, a processor, and a computer program stored on the memory and executable on the memory. The processor executes the program to implement the multi-dimensional data collaboration-based electric vehicle reconnaissance task intelligent management method according to any one of claims 1 to 7.

Images