Electric vehicle loss assessment data security management and sharing system based on blockchain

Abstract

This invention discloses a blockchain-integrated electric vehicle damage assessment data security management and sharing system, relating to the field of data management and processing technology. It includes: a data acquisition module for acquiring damage assessment data of accident vehicles, the damage assessment data including at least three-dimensional scan point clouds, damage geometric features, posterior damage probabilities of each component, damage assessment confidence intervals, repair plans, and repair costs; a data storage module; a hierarchical permission management module, which classifies users into different permission levels based on their reputation values ​​and configures differentiated data access permissions for different levels of users; a privacy calculation module; a sharing incentive module; and a business verification module. This invention integrates blockchain technology with the entire electric vehicle damage assessment process, employing on-chain and off-chain collaborative storage to ensure data immutability and traceability, improving data security. Repair plans, after being signed by multiple parties, are uploaded to the blockchain to form legally valid electronic certificates, effectively guaranteeing the authenticity and integrity of the data.

Description

Technical Field

[0001] This invention relates to the field of data management and processing technology, specifically to a data security management and sharing system for electric vehicle damage assessment that integrates blockchain. Background Technology

[0002] With the continued expansion of the new energy vehicle market, post-accident damage detection and compensation for electric vehicles has become a major focus of industry attention. Currently, the authenticity of post-accident investigation data for new energy vehicles is increasingly problematic. Disputes frequently arise between car owners, automakers, and insurance companies regarding data reliability. Car owners often lack the professional means to obtain sufficient raw data, while data provided by automakers is often perceived as biased. Consequently, disputes frequently arise between accident car owners and automakers or insurance companies regarding damage assessment and compensation. Therefore, based on these prevalent industry issues, some third-party platforms have begun exploring vehicle data storage services.

[0003] Regarding data sharing, traditional new energy vehicle manufacturers' data collection models have limitations in terms of credibility and security. While blockchain technology can ensure data security and make new energy vehicle data more transparent and authentic, it still has some shortcomings. Firstly, it lacks a comprehensive data management solution for the damage set characteristics and 3D scan point clouds of electric vehicles; secondly, it has not established a data hierarchical permission mechanism that is deeply integrated with damage detection technology. Therefore, it is impossible to use historical repair plans and monetary data to provide objective evidence and similarity guidance for the repair and damage assessment plans of current accident vehicles.

[0004] Therefore, this application proposes a blockchain-integrated electric vehicle damage assessment data security management and sharing system to overcome the above-mentioned shortcomings. Summary of the Invention

[0005] To address the aforementioned technical issues, this technical solution provides a blockchain-integrated electric vehicle damage assessment data security management and sharing system. This solution resolves the problems of integrated data management and the lack of integration with damage detection technology.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The electric vehicle damage assessment data security management and sharing system integrating blockchain includes: a data acquisition module for acquiring damage assessment data of accident vehicles, wherein the damage assessment data includes at least three-dimensional scan point cloud, damage geometric features, posterior damage probability of each component, damage assessment confidence interval, repair plan and repair amount; The data storage module, based on blockchain technology, stores the core summary information of the loss assessment data on the blockchain to ensure the immutability and traceability of the data; The hierarchical permission management module divides users into different permission levels based on their reputation scores and configures differentiated data access permissions for users of different levels. The reputation score is dynamically calculated based on the quantity and quality of user-uploaded data and historical behavior. The privacy computing module employs privacy protection technologies, including zero-knowledge proofs, homomorphic encryption, or trusted execution environments, to support data verification and statistical analysis without disclosing the original sensitive data. The shared incentive module uses smart contracts to reward data contributors with reputation points and share revenue, and supports high-privilege users to share data with low-privilege users a limited number of times. The business evidence module provides statistical reference for current vehicle repair decisions and total loss determination based on historical damage assessment data from similar accidents.

[0007] Preferably, the hierarchical permission management module further includes: The reputation score calculation unit calculates the reputation score based on the completeness of the user-uploaded data, the consistency score with subsequent analysis results, peer reviews, and the timeliness of the data. The permission dynamic adjustment unit automatically adjusts the user's permission level based on the reputation value, and implements reputation value deduction, permission downgrade or permanent ban for fraudulent or serious deviation behaviors; The shared control unit records the number of times high-privilege users share data, verifies whether the recipient's reputation value meets the acceptance threshold, and generates an on-chain evidence record for each share.

[0008] Preferably, the repair plan in the damage assessment data includes at least the repair method, the list of replacement parts, the repair process and labor hour information; the repair amount includes at least the cost of parts, labor hour, auxiliary materials and total repair cost; the damage assessment data also includes the final handling method and the total loss compensation amount.

[0009] Preferably, the business verification module includes: The similar accident matching unit retrieves accident cases with similarity exceeding a preset threshold from the historical database based on the current vehicle's model, collision direction, energy level, and probability of damage to key components. The decision statistics unit is used to statistically analyze the repair-total-loss ratio, repair cost distribution range, and total loss compensation range for similar cases. The supporting report generation unit outputs supporting reports that include statistical results of similar cases and current decision-making recommendations, in order to help car owners, insurance companies and repair shops reach a consensus.

[0010] Preferably, the privacy computing module includes: Zero-knowledge proof units are used to verify the condition of a vehicle without revealing the specific extent of damage. Homomorphic encryption unit supports the aggregation and calculation of statistical features of historical loss assessment data in an encrypted state, which is used to update the prior probability model without leaking case data; The Trusted Execution Environment (TEE) unit is used to perform sensitive computational tasks such as approximate accident matching, ensuring that the original data is not accessed externally during the computation process.

[0011] Preferably, the data storage module adopts an on-chain and off-chain collaborative storage architecture: The original damage assessment data is encrypted and stored in a distributed file system, and its hash value and access address are stored on the blockchain for evidence. Each data access, sharing, and permission change generates an immutable blockchain transaction record; The repair plan and repair amount are uploaded to the blockchain after being digitally signed by multiple parties, serving as a legally valid electronic certificate.

[0012] Preferably, it also includes a data quality assessment module for identifying the authenticity of uploaded data: The automatic verification unit performs device fingerprint verification, spatiotemporal consistency check and physical constraint check on point cloud data to identify obvious abnormal data. The cross-validation unit compares the uploaded data with subsequent breakdown results, data from multiple sources, and insurance company claims records to calculate a consistency score. The behavior analysis unit constructs user behavior profiles based on user upload frequency, case similarity, and social network relationships to identify mass forgery or collusive fraud behavior. The comprehensive credibility scoring unit integrates the above-mentioned multi-dimensional indicators to generate a data credibility score. Data below the threshold will be subject to manual review or marked as unreliable.

[0013] Preferably, the physical constraint verification includes: The impact energy is estimated based on the product of the maximum intrusion depth and the indentation area, and compared with the energy and damage experience curves of the same vehicle model under the same working conditions. Verify whether the intrusion rates of multiple components satisfy the spatial logical relationship; Anomaly detection is performed by comparing the geometric feature vector of the current data with the distribution of the component in historical data, and outliers located at the tail of the distribution are identified.

[0014] Preferably, it also includes a data closed-loop optimization module: Based on historical damage assessment data shared in safety, the prior damage probability and likelihood function of each vehicle model under each working condition are updated regularly. Data access permissions are dynamically adjusted based on the sample size, and the permission threshold is automatically tightened when the sample size is insufficient. The decision results from the business evidence module are fed back to the loss assessment model to achieve continuous optimization of the detection algorithm.

[0015] Preferably, the revenue-sharing mechanism of the shared incentive module includes: When loss assessment data uploaded by a data contributor is queried by others, the smart contract automatically distributes the query revenue to the contributor. Revenue sharing is linked to data credibility scores and contributor reputation scores; Rewards are distributed in the form of platform tokens, which can be used to pay for inquiry fees or redeem value-added services; All profit distribution records are stored on the blockchain to ensure transparency and traceability. Compared with the prior art, the beneficial effects of the present invention are as follows: This invention integrates blockchain technology with the entire electric vehicle damage assessment process. It employs on-chain and off-chain collaborative notarization to ensure the immutability and traceability of data such as 3D scan point clouds, damage geometric features, repair plans, and repair costs, thereby improving data security. Repair plans, after being signed by multiple parties, are uploaded to the blockchain to form legally valid electronic certificates, effectively guaranteeing the authenticity and integrity of the data. Furthermore, this invention discloses a unique reputation-driven hierarchical permission-sharing mechanism. Users accumulate reputation points by uploading high-quality data to gain higher permissions. Higher-permission users can view more data to improve their work efficiency and can also earn rewards by sharing data with lower-permission users, thus promoting continuous data updates and improving system accuracy, forming a positive cycle. The system also includes punitive measures against falsified data to maintain data integrity. Simultaneously, the system integrates privacy measures such as zero-knowledge proofs, homomorphic encryption, and trusted execution environments to effectively protect sensitive data from leakage.

[0016] This invention constructs a business evidence module based on weighted Euclidean distance to retrieve similar accident cases from historical databases, statistically analyze the repair-total loss ratio and cost distribution range, and uses objective data as the basis for judging repair or total loss, significantly reducing decision-making disputes. A revenue-sharing mechanism based on smart contracts is established, where data contributors receive platform token rewards based on data quality. These tokens can be used to pay for inquiries, exchange services, and are linked to reputation scores, incentivizing continuous contributions of high-quality data. Attached Figure Description

[0017] Figure 1 A schematic diagram of each module of the system; Figure 2 This is a schematic diagram of the data storage module architecture. Detailed Implementation

[0018] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0019] Reference Figure 1As shown, the electric vehicle damage assessment data security management and sharing system integrating blockchain includes: a data acquisition module, used to acquire damage assessment data of accident vehicles, wherein the damage assessment data includes at least three-dimensional scan point cloud, damage geometric features, posterior damage probability of each component, damage assessment confidence interval, repair plan and repair amount; The data storage module, based on blockchain technology, stores the core summary information of the loss assessment data on the blockchain to ensure the immutability and traceability of the data; The hierarchical permission management module divides users into different permission levels based on their reputation scores and configures differentiated data access permissions for users of different levels. The reputation score is dynamically calculated based on the quantity and quality of user-uploaded data and historical behavior. The privacy computing module employs privacy protection technologies, including zero-knowledge proofs, homomorphic encryption, or trusted execution environments, to support data verification and statistical analysis without disclosing the original sensitive data. The shared incentive module uses smart contracts to reward data contributors with reputation points and share revenue, and supports high-privilege users to share data with low-privilege users a limited number of times. The business evidence module provides statistical reference for current vehicle repair decisions and total loss determination based on historical damage assessment data from similar accidents.

[0020] The data acquisition module obtains comprehensive damage assessment data for accident vehicles throughout the entire process, including all elements from the original scanned point cloud to the final repair plan, ensuring the integrity of the data source. The data storage module leverages the immutability of blockchain to ensure data authenticity and reliability, while recording all access operations to create a traceable audit log. The hierarchical access control module dynamically assigns access levels based on user reputation values, with different levels corresponding to differentiated data access scopes and sharing permissions, achieving hierarchical and categorized data protection. The privacy computing module supports data verification and statistical analysis without leaking original sensitive data. The business corroboration module provides historical data from similar accidents to offer statistical references for current vehicle repair and total loss decisions, reducing disputes arising from subjective judgments.

[0021] The system also includes a historical accident database, storing anonymized historical damage assessment data, including damage geometry features, posterior damage probabilities, repair plans, repair costs, and final handling methods, for use in similar case matching and statistical analysis. Additionally, there is a user reputation database for recording user reputation scores, permission levels, sharing records, and historical behavior.

[0022] like Figure 1 As shown, the hierarchical permission management module further includes: The reputation score calculation unit calculates the reputation score based on the completeness of the user-uploaded data, the consistency score with subsequent analysis results, peer reviews, and the timeliness of the data. The permission dynamic adjustment unit automatically adjusts the user's permission level based on the reputation value, and implements reputation value deduction, permission downgrade or permanent ban for fraudulent or serious deviation behaviors; The shared control unit records the number of times high-privilege users share data, verifies whether the recipient's reputation value meets the acceptance threshold, and generates an on-chain evidence record for each share.

[0023] In this embodiment, tiered access control is crucial for ensuring the healthy operation of the data ecosystem. By managing the reputation score of users uploading data, user permissions are dynamically adjusted, thereby enabling subsequent permission sharing. The reputation score R is an important indicator for judging the trustworthiness and contribution of a user. Calculating the reputation score R requires comprehensive consideration from multiple perspectives, the most important of which include data quantity, data accuracy, timeliness, and penalties for violations and falsification. Therefore, the following formula for calculating the reputation score is derived: Reputation value The formula comprehensively reflects user credibility and contribution, and is as follows: in: For the number of data, Total number of cases uploaded by users; For accuracy factor, The average consistency score between user historical data and the breakdown results; As a time-sensitive factor, This refers to the number of cases uploaded in the past 30 days; Points will be deducted as punishment for violations such as forgery; These are weighting coefficients, which can be dynamically adjusted.

[0024] The above weighting coefficients meet the normalization requirements, that is... The weighting factors are optimized and adjusted based on the user's stage of time. These stages are roughly divided into the initial launch phase, the growth phase, and the maturity phase. In the initial launch phase, since users are just entering the data contribution stage, their total number of cases is relatively small. During this phase, to incentivize users to build a data infrastructure more quickly, the aforementioned data volume weighting factors are adjusted. This should be increased to the highest level, for example, the maximum value of 0.5. This weighting coefficient will gradually decrease as the total amount of data uploaded by users increases. Once the total amount of data uploaded by users reaches the first-stage threshold, the system enters the growth stage, where the weighting system... Gradually decrease, while increasing the accuracy weighting coefficient. and timeliness weighting coefficient Accuracy weighting coefficient and timeliness weighting coefficient The maximum values ​​can reach 0.4 and 0.2 respectively, guiding users to focus on and improve data quality. Once the total amount of data uploaded by users reaches the second-stage threshold, the system enters the maturity stage, where the dominance of accuracy weighting further increases, for example... =0.5, and strengthen the dominance of punitive weights to constrain violations.

[0025] Data quality is the core of building a data sharing system. Its level directly affects the purity of the overall data. Regardless of the stage, it is necessary to regularly check the data provided by users and calculate the average accuracy score of all users. If the accuracy score remains consistently lower than the average score of all users If the accuracy weight is increased at the current stage of the user, it will incentivize the processor to improve data quality.

[0026] The reputation value obtained using the above methods can be directly used for the corresponding permission level: <A is level L1, A≤ <B represents level L2, B≤ <C represents Level 3, Level C is L4. A, B, and C are the thresholds separating different levels. Each level's permissions are tied to the range of data that can be accessed: L1 can only access public data, L2 can access some anonymized information, L3 can access approximate accident point cloud distribution maps, and L4 can access raw point clouds within a limited range. Reputation scores are also affected by data credibility. Forgers will have their reputation scores significantly reduced, even to zero, and will be publicly announced on the blockchain as punishment. Similarly, multiple data discrepancies will negatively impact reputation scores, leading to point deductions or temporary demotions.

[0027] To increase the contribution of damage assessment data to the offline damage assessment process, users with high reputation scores can share data with users with low reputation scores. For example, Level 3 users can share data with Level 2 users, while Level 1 users cannot share data due to their low reputation scores. High-reputation users have a limited number of daily contributions; the higher the reputation score, the more times they can share.

[0028] The repair plan in the damage assessment data includes at least the repair method, the list of replacement parts, the repair process and labor hours; the repair amount includes at least the cost of parts, labor hours, auxiliary materials and total repair cost; the damage assessment data also includes the final handling method and the total loss compensation amount.

[0029] The aforementioned repair methods, including whether to replace or repair components, form the basis for cost accounting and provide guidance for future repairs of accident vehicles. If replacement is used, a list of replacement parts and reference unit prices should be provided for parts cost accounting and inventory management. If repair is used, a brief description of key repair processes should be provided, such as laser welding, offering guidance for repairing structural damage in similar accident vehicles. Labor hour information is used to assess labor costs and repair efficiency, to estimate the repair time for the current accident vehicle, and to inform the owner of the approximate repair time.

[0030] When a vehicle is determined to be a total loss, the final compensation amount and calculation basis should be recorded. The final decision regarding the vehicle's disposal should be recorded, including at least the handling method, identified as repair or total loss. This is the final determination of the claim. A brief explanation of the reasoning should be provided for reference in handling similar cases in the future. The final disposal decision must be timestamped to record the decision time and ensure consistency with the timeline of reporting, damage assessment, and repair. All data has been anonymized before uploading to ensure the privacy and security of the vehicle owner.

[0031] like Figure 1 As shown, the business verification module includes: The similar accident matching unit retrieves accident cases with similarity exceeding a preset threshold from the historical database based on the current vehicle's model, collision direction, energy level, and probability of damage to key components. The decision statistics unit is used to statistically analyze the repair-total-loss ratio, repair cost distribution range, and total loss compensation range for similar cases. The supporting report generation unit outputs supporting reports that include statistical results of similar cases and current decision-making recommendations, in order to help car owners, insurance companies and repair shops reach a consensus.

[0032] In this embodiment, the similar accident matching unit retrieves similar past cases from the database based on the multi-dimensional features of the current accident vehicle. The similarity calculation comprehensively considers multiple features, including vehicle model information, collision conditions, damage geometry, and the probability of damage to key components. In this embodiment, a weighted Euclidean distance is used to calculate the similarity S between the current accident and historical accidents, using the following formula: The formula is as follows: Where S represents the similarity, and the smaller the value, the higher the similarity; These are the vehicle model codes for the current accident and historical accidents, respectively (the vehicle models need to be digitized in advance). These are the collision energy levels of the current accident and historical accidents (which can be quantified as 1, 2, and 3). These are the collision location codes for the current accident and historical accidents (coded by vehicle body partition). These are the th current accidents and the th historical accidents, respectively. Posterior damage probability of critical components (such as battery module #3); The number of key components involved in the similarity calculation; These are the maximum possible differences for the corresponding features, used for normalization; The weight coefficients for each feature dimension satisfy the following conditions: The weights can be determined based on industry experience or through machine learning methods.

[0033] Calculate Then, the similarity can be normalized to obtain the final similarity value in the interval [0,1]. = This makes it easy to set interval thresholds.

[0034] If the similarity exceeds a threshold, then related cases exceeding the similarity threshold will be returned. For example, the threshold can be set to 0.8. ≥ 0.8 includes K cases, and the output includes the repair plan for each historical accident, hidden damage inside the vehicle, repair cost, and final handling method.

[0035] Statistical analysis is performed on the matched similar cases. The final processing method is determined by calculating the ratio of repair to total loss in the statistical cases. Repair cost statistics are calculated, including the median, quartiles, and 90% confidence interval of the total repair cost for each case. For example, under similar historical accident conditions for a certain vehicle, the median repair cost is 8500 yuan, while 90% of the costs fall between 6000 and 12000 yuan. A detailed summary of solutions is also provided, listing common repair methods, replacement parts lists, and labor hours in similar cases, serving as a reference for the current repair plan. This also includes key information about similar cases, such as vehicle model and damage extent, which can be viewed in detail but requires specific permissions.

[0036] like Figure 1 As shown, the privacy computing module includes: Zero-knowledge proof units are used to verify the condition of a vehicle without revealing the specific extent of damage. Homomorphic encryption unit supports the aggregation and calculation of statistical features of historical loss assessment data in an encrypted state, which is used to update the prior probability model without leaking case data; The Trusted Execution Environment (TEE) unit is used to perform sensitive computational tasks such as approximate accident matching, ensuring that the original data is not accessed externally during the computation process.

[0037] Zero-knowledge proof units are used for damage existence verification. Repair technicians can prove to insurance companies that a vehicle has experienced battery pack damage, without disclosing the specific location, depth, or probability of the damage. The prover generates the proof using a hash value of the damage's geometric features, and the verifier knows the result (yes or no). Alternatively, it can be used for cost range verification. Vehicle owners can prove to a third party that the cost is within a reasonable range without disclosing the specific amount. The system generates a range for the proof based on the historical distribution of repair costs, and the third party only verifies whether the current cost falls within the confidence interval of historical data.

[0038] Furthermore, it can be used by high-reputation users to prove that their reputation value meets the sharing threshold, without revealing the actual score; it is only used for quick verification during permission sharing. The aforementioned zero-knowledge proof adopts the Bulletproofs scheme, ensuring security while also considering the efficiency of proof generation and verification.

[0039] Homomorphic encryption units allow direct computation on the ciphertext, and the decrypted result is identical to the result of the same computation on the plaintext. In this embodiment, this unit is used in aggregated statistical scenarios. For example, in repair cost distribution statistics, if regulatory authorities need to understand the industry's repair cost levels without involving individual case privacy, each repair shop uploads encrypted repair amounts, and the system aggregates and generates quantile information from a cost histogram, keeping the original data confidential at all times. Furthermore, it can also be used for similarity calculation protection, calculating the similarity between the current incident and historical incidents without disclosing the specific feature vectors of both parties, for matching similar cases for protection.

[0040] The Trusted Execution Environment (TEE) ensures, through hardware-level isolation, that sensitive data is not accessed by the operating system or other applications during computation. In this embodiment, this unit is used to perform complex computational tasks that require combining sensitive data from multiple sources. When performing approximate incident matching, when a technician queries past cases similar to the current incident, both the current incident's point cloud features and the complete data from historical incidents are decrypted within the TEE for similarity calculation, and only a list of matching results is returned. The original data is never exposed to the external environment.

[0041] If a user with higher privileges (L3 or above) views the intrusion geometry diagram, a TEE (Translation Engine) generates the data in real time based on historical data. The output statistical charts are anonymized, but the original data points cannot be deduced.

[0042] The three sub-units mentioned above are used in combination according to the characteristics of the business scenario, covering three types of scenarios: verification, statistics, and calculation. For example, zero-knowledge proof unit is used when auditing qualifications; homomorphic encryption unit is used when calculating the mean; and trusted execution environment unit is used when matching similarity.

[0043] like Figure 2As shown, the data storage module adopts an on-chain and off-chain collaborative storage architecture: The original damage assessment data is encrypted and stored in a distributed file system, and its hash value and access address are stored on the blockchain for evidence. Each data access, sharing, and permission change generates an immutable blockchain transaction record; The repair plan and repair amount are uploaded to the blockchain after being digitally signed by multiple parties, serving as a legally valid electronic certificate.

[0044] In this system, the raw data generated during the damage assessment process, such as 3D scan point clouds, damage geometry features, repair plans, and repair costs, is too large to be directly stored on the blockchain. Therefore, we encrypt the raw data and store it in a distributed file system (such as IPFS), generating a unique content-addressed hash value. This hash value, along with the storage address, upload time, and uploader's digital signature, is written into the blockchain transaction. Any tampering with the raw data will result in a change in the hash value, which can be detected by comparing it with the on-chain record, ensuring data integrity. To ensure the traceability of the flow process, all data access operations need to be recorded on the blockchain. Each data query, download, or sharing generates a transaction record containing the visitor's identity, data ID, timestamp, and operation type. For sharing events, the sharing party, recipient, credential hash, and validity period are additionally recorded. These records are permanently stored, forming an immutable audit log for post-event traceability and dispute resolution.

[0045] To address disputes arising from repair plans and financial information, a multi-party digital signature mechanism is introduced. After the repair plan is confirmed by the repair technician, the insurance company's claims adjuster, and the vehicle owner, a joint digital signature is generated. The signed plan and its financial hash value are then uploaded to the blockchain as a legally valid electronic certificate. Authorized users can query the corresponding on-chain evidence records by entering a data ID or accident number in the client application. This includes the original data hash value for comparison and verification with locally downloaded data; complete access logs to understand who accessed the data, when, and how; and the multi-party signature status to confirm whether the repair plan has been approved by all parties. Through this on-chain and off-chain collaborative architecture, the system ensures both the immutability and traceability of damage assessment data while also accommodating the storage and retrieval of large volumes of data, providing reliable technical support for claims processing, supervision, auditing, and dispute resolution.

[0046] like Figure 1 As shown, it also includes a data quality assessment module, used to identify the authenticity of uploaded data: The automatic verification unit performs device fingerprint verification, spatiotemporal consistency check and physical constraint check on point cloud data to identify obvious abnormal data. The cross-validation unit compares the uploaded data with subsequent breakdown results, data from multiple sources, and insurance company claims records to calculate a consistency score. The behavior analysis unit constructs user behavior profiles based on user upload frequency, case similarity, and social network relationships to identify mass forgery or collusive fraud behavior. The comprehensive credibility scoring unit integrates the above-mentioned multi-dimensional indicators to generate a data credibility score. Data below the threshold will be subject to manual review or marked as unreliable.

[0047] In this embodiment, the automatic verification unit can perform real-time verification during data upload to identify obvious anomalies or forged data. This includes device verification, spatiotemporal consistency verification, physical constraint verification, and outlier verification. Each scanning device has a unique ID, and a device signature is automatically added during data upload. If a device uploads a large amount of data from different vehicles within a short period, anomaly detection is triggered. Point clouds from multiple scans of the same accident vehicle should be spatiotemporally consistent. The presence of numerous discrete isolated points or unreasonable holes indicates suspicion and may be deemed incomplete scanning, resulting in low-quality data. Physical constraint verification verifies whether the component intrusion rate meets spatial logic requirements. For example, if the battery pack intrusion rate is as high as 25%, there must be a corresponding indentation on its upper base plate; or it is based on the maximum intrusion depth. With the area of ​​the depression Estimate impact energy The energy-damage experience curve of the same vehicle model under the same working conditions is compared. If the deviation is too large, anomaly detection is triggered. For outlier detection, the geometric feature vector of the current data is... By comparing the component's historical distribution, an isolated forest or autoencoder is used to identify outliers located at the tail end of the distribution. After anomaly detection is initiated, a real photo must be uploaded and verified through multi-party joint signature verification.

[0048] The cross-validation unit verifies the authenticity of the data, including disassembly closed-loop verification. Technicians are required to upload the actual disassembly results (real damaged parts, damage level, actual cost) after completing the repair, which are automatically compared with the previously uploaded damage assessment data. The lower the deviation, the higher the data accuracy score. It also includes multi-source comparison. The same accident may be recorded by multiple parties, such as traffic police, insurance companies, and repair shops. The system can automatically cross-compare damage scans and point cloud features from different sources; significant differences trigger an anomaly review.

[0049] The positional analysis unit monitors user upload frequency and counts the average number of cases uploaded by users per day. If there is a sudden surge on a certain day or during a certain period, an abnormal review is triggered. Case similarity analysis calculates the entropy value of the geometric feature distribution in user-uploaded cases. If a large number of cases have highly similar features, such as an intrusion rate that is always 0.3±0.01, it may be a copy forgery. Social network analysis constructs a user interaction graph, including sharing relationships and rating relationships, and finds closely connected but isolated behavioral patterns in subgraphs to identify collusive fraud groups.

[0050] By combining the above multi-dimensional indicators, the overall credibility score T for each data point is calculated: in: For automatic score verification, the value range is 0~100; The cross-validation consistency score; The user's behavior is scored normally; Let be the weighting coefficient, satisfying .

[0051] Scoring threshold mechanism: Automatic data entry, which can be used for prior probability updates; : Added to the database but marked, not used for model training; : Enter the manual review queue. Only after the review is passed can it be put into the warehouse.

[0052] For users confirmed to have committed fraud, the penalty mechanism described in claim 2 is triggered, including punitive measures such as resetting reputation value to zero, downgrading privileges, and publicizing the information on the blockchain. Through the above multi-layered screening mechanism, the system possesses self-purification capabilities, effectively maintaining the long-term credibility of the database.

[0053] The physical constraint verification includes: The impact energy is estimated based on the product of the maximum intrusion depth and the indentation area, and compared with the energy and damage experience curves of the same vehicle model under the same working conditions. Verify whether the intrusion rates of multiple components satisfy the spatial logical relationship; Anomaly detection is performed by comparing the geometric feature vector of the current data with the distribution of the component in historical data, and outliers located at the tail of the distribution are identified.

[0054] According to the principles of impact, the impact energy absorbed by the dent and the maximum penetration depth... and the area of ​​the depression Positive correlation, For the same vehicle model and under the same collision conditions, the energy and damage characteristics should satisfy the empirical curve. ,in This represents the regional stiffness coefficient, with different values ​​for different parts; for example, 0.8 for the door and 1.2 for the fender. Extracted from the current data... and ,calculate Value, query historical data of the same vehicle model and operating conditions Statistical distribution, calculate the mean Standard deviation .like Data that exceeds the normal fluctuation range is marked as suspicious.

[0055] For the analysis of spatial logic relationships, the degree of damage to multiple components must satisfy the laws of spatial location and mechanical transmission. For example, if the battery pack intrusion rate is as high as 25%, the vehicle floor directly above it will inevitably have a corresponding dent; or if the front bumper has a deep intrusion depth, the energy-absorbing box behind it should have visible deformation. The system will automatically query the damage characteristics of other components that are spatially related to the current component, determine the correlation, and if the correlation is low and clearly does not conform to the energy transmission logic, a consistency warning will be triggered.

[0056] The geometric feature distribution is detected by comparing the geometric feature vector of the current data with the distribution of the component in the historical database. The isolated forest algorithm is used to train a normal data distribution model and calculate the anomaly score of the current data point. If the anomaly score is large, it will be marked as a potential anomaly.

[0057] It also includes a data closed-loop optimization module: Based on historical damage assessment data shared in safety, the prior damage probability and likelihood function of each vehicle model under each working condition are updated regularly. Data access permissions are dynamically adjusted based on the sample size, and the permission threshold is automatically tightened when the sample size is insufficient. The decision results from the business evidence module are fed back to the loss assessment model to achieve continuous optimization of the detection algorithm.

[0058] In this embodiment, the aim is to feed securely shared historical damage assessment data back into the damage detection model, forming a positive cycle of detection, evidence storage, sharing, and optimization. After verification, historical damage assessment data with high reliability is periodically recalculated for the prior damage probability and likelihood function for each vehicle model and operating condition. The updated prior probability directly optimizes the confidence calculation step, and the updated likelihood function makes the correlation between geometric features and damage state closer to reality, improving the accuracy of the posterior damage probability.

[0059] Data access permissions are dynamically adjusted based on sample size. This means that the access permission threshold for data related to the same working condition is dynamically adjusted based on the number N of samples in the historical database. When the sample size is small, typically N < 20, access permissions are tightened, allowing only users at level L3 and above to view the information to prevent misleading judgments due to sparse data. When N ≥ 20, access permissions are relaxed, allowing more users to view and refer to the data.

[0060] The decision results of the business evidence module are fed back to the damage assessment model. The repair or total loss decision results output by the evidence module and the disassembly verification data after actual repair are periodically fed back to the historical accident database for updating and optimizing the database.

[0061] The revenue-sharing mechanism of the shared incentive module includes: When loss assessment data uploaded by a data contributor is queried by others, the smart contract automatically distributes the query revenue to the contributor. Revenue sharing is linked to data credibility scores and contributor reputation scores; Rewards are distributed in the form of platform tokens, which can be used to pay for inquiry fees or redeem value-added services; All profit distribution records are stored on the blockchain to ensure transparency and traceability.

[0062] When a user queries similar cases through the business verification module, the smart contract automatically triggers revenue sharing for each historical loss assessment data query: the contributor of the queried data receives the query revenue, the amount of which is dynamically adjusted based on the confidence level of the queried data and the query frequency; generally, the higher the confidence level, the higher the revenue. Platform tokens serve as equity certificates circulating within the ecosystem. They can be directly consumed to pay query fees or exchanged for other value-added services. For example, high-privilege users can use tokens to redeem in-depth analysis reports and priority technical support. Furthermore, when the total token supply exceeds a threshold, the user's reputation score calculation receives a small bonus, such as 5%. This creates a positive incentive cycle.

[0063] Furthermore, all revenue distributions are transparent and traceable, with distribution records stored on the blockchain, including the query address, contributor address, data ID, query time, base fee, quality coefficient, actual distribution amount, and transaction hash. Contributors can view their personal revenue details and historical distribution records at any time through the client, ensuring that the distribution process is open, transparent, and tamper-proof.

[0064] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A blockchain-integrated electric vehicle damage assessment data security management and sharing system, characterized in that, include: The data acquisition module is used to acquire damage assessment data of the accident vehicle. The damage assessment data includes at least three-dimensional scan point cloud, damage geometric features, posterior damage probability of each component, damage assessment confidence interval, repair plan and repair cost. The data storage module, based on blockchain technology, stores the core summary information of the loss assessment data on the blockchain to ensure the immutability and traceability of the data; The hierarchical permission management module divides users into different permission levels based on their reputation scores and configures differentiated data access permissions for users of different levels. The reputation score is dynamically calculated based on the quantity and quality of user-uploaded data and historical behavior. The privacy computing module employs privacy protection technologies, including zero-knowledge proofs, homomorphic encryption, or trusted execution environments, to support data verification and statistical analysis without disclosing the original sensitive data. The shared incentive module uses smart contracts to reward data contributors with reputation points and share revenue, and supports high-privilege users to share data with low-privilege users a limited number of times. The business evidence module provides statistical reference for current vehicle repair decisions and total loss determination based on historical damage assessment data from similar accidents.

2. The electric vehicle damage assessment data security management and sharing system integrating blockchain as described in claim 1, characterized in that: The hierarchical access control module further includes: The reputation score calculation unit calculates the reputation score based on the completeness of the user-uploaded data, the consistency score with subsequent analysis results, peer reviews, and the timeliness of the data. The permission dynamic adjustment unit automatically adjusts the user's permission level based on the reputation value, and implements reputation value deduction, permission downgrade or permanent ban for fraudulent or serious deviation behaviors; The shared control unit records the number of times high-privilege users share data, verifies whether the recipient's reputation value meets the acceptance threshold, and generates an on-chain evidence record for each share.

3. The electric vehicle damage assessment data security management and sharing system integrating blockchain as described in claim 1, characterized in that: The repair plan in the damage assessment data includes at least the repair method, the list of replacement parts, the repair process and labor hours; the repair amount includes at least the cost of parts, labor hours, auxiliary materials and total repair cost; the damage assessment data also includes the final handling method and the total loss compensation amount.

4. The electric vehicle damage assessment data security management and sharing system integrating blockchain as described in claim 1, characterized in that: The business verification module includes: The similar accident matching unit retrieves accident cases with similarity exceeding a preset threshold from the historical database based on the current vehicle's model, collision direction, energy level, and probability of damage to key components. The decision statistics unit is used to statistically analyze the repair-total-loss ratio, repair cost distribution range, and total loss compensation range for similar cases. The supporting report generation unit outputs supporting reports that include statistical results of similar cases and current decision-making recommendations, in order to help car owners, insurance companies and repair shops reach a consensus.

5. The electric vehicle damage assessment data security management and sharing system integrating blockchain as described in claim 1, characterized in that: The privacy computing module includes: Zero-knowledge proof units are used to verify the condition of a vehicle without revealing the specific extent of damage. Homomorphic encryption unit supports the aggregation and calculation of statistical features of historical loss assessment data in an encrypted state, which is used to update the prior probability model without leaking case data; The Trusted Execution Environment (TEE) unit is used to perform sensitive computational tasks such as approximate accident matching, ensuring that the original data is not accessed externally during the computation process.

6. The electric vehicle damage assessment data security management and sharing system integrating blockchain as described in claim 1, characterized in that: The data storage module adopts an on-chain and off-chain collaborative storage architecture: The original damage assessment data is encrypted and stored in a distributed file system, and its hash value and access address are stored on the blockchain for evidence. Each data access, sharing, and permission change generates an immutable blockchain transaction record; The repair plan and repair amount are uploaded to the blockchain after being digitally signed by multiple parties, serving as a legally valid electronic certificate.

7. The electric vehicle damage assessment data security management and sharing system integrating blockchain according to any one of claims 1-6, characterized in that: It also includes a data quality assessment module, used to identify the authenticity of uploaded data: The automatic verification unit performs device fingerprint verification, spatiotemporal consistency check and physical constraint check on point cloud data to identify obvious abnormal data. The cross-validation unit compares the uploaded data with subsequent breakdown results, data from multiple sources, and insurance company claims records to calculate a consistency score. The behavior analysis unit constructs user behavior profiles based on user upload frequency, case similarity, and social network relationships to identify mass forgery or collusive fraud behavior. The comprehensive credibility scoring unit integrates the above-mentioned multi-dimensional indicators to generate a data credibility score. Data below the threshold will be subject to manual review or marked as unreliable.

8. The electric vehicle damage assessment data security management and sharing system integrating blockchain as described in claim 7, characterized in that: The physical constraint verification includes: The impact energy is estimated based on the product of the maximum intrusion depth and the indentation area, and compared with the energy and damage experience curves of the same vehicle model under the same working conditions. Verify whether the intrusion rates of multiple components satisfy the spatial logical relationship; Anomaly detection is performed by comparing the geometric feature vector of the current data with the distribution of the component in historical data, and outliers located at the tail of the distribution are identified.

9. The electric vehicle damage assessment data security management and sharing system integrating blockchain as described in claim 1, characterized in that: It also includes a data closed-loop optimization module: Based on historical damage assessment data shared in safety, the prior damage probability and likelihood function of each vehicle model under each working condition are updated regularly. Data access permissions are dynamically adjusted based on the sample size, and the permission threshold is automatically tightened when the sample size is insufficient. The decision results from the business evidence module are fed back to the loss assessment model to achieve continuous optimization of the detection algorithm.

10. The electric vehicle damage assessment data security management and sharing system integrating blockchain as described in claim 1, characterized in that: The revenue-sharing mechanism of the shared incentive module includes: When loss assessment data uploaded by a data contributor is queried by others, the smart contract automatically distributes the query revenue to the contributor. Revenue sharing is linked to data credibility scores and contributor reputation scores; Rewards are distributed in the form of platform tokens, which can be used to pay for inquiry fees or redeem value-added services; All profit distribution records are stored on the blockchain to ensure transparency and traceability.

Images