A blockchain-based vehicle data processing system and method

Abstract

This invention relates to the field of data processing technology, specifically disclosing a blockchain-based vehicle data processing system and method. The system includes a data acquisition module and a cloud platform. The cloud platform includes a data processing module, a data uploading module, and a blockchain network. The data acquisition module is used to collect raw vehicle data. The data processing module is used to clean and denoise the raw vehicle data and also to extract advanced driving behavior features, including trip recognition, driving event recognition, and dimensional score aggregation. The data uploading module is used to upload the processed data to the blockchain network. The technical solution of this invention can extract features from multi-source, heterogeneous raw vehicle data and store them securely.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a blockchain-based vehicle data processing system and method. Background Technology

[0002] With the rapid development of automotive intelligence and connectivity, the data generated during vehicle operation is growing explosively. This data covers multiple dimensions such as driving behavior, vehicle status, and environmental information, and has enormous potential value, especially in the financial services sector, where it shows broad application prospects in areas such as precise pricing of vehicle insurance, credit risk assessment, and residual value evaluation of used cars.

[0003] However, existing technologies still face numerous bottlenecks and challenges in the collection, processing, and value extraction of vehicle data. Firstly, vehicle data originates from diverse sources, including in-vehicle terminals from different manufacturers, various sensors, and diagnostic interfaces, resulting in heterogeneous data formats and inconsistent standards, creating significant data silos. This fragmentation makes effective integration and interoperability of multi-source data difficult, hindering deep data fusion and comprehensive analysis.

[0004] Secondly, current vehicle data is mostly stored and managed in a centralized manner. Under this model, data faces the risk of being tampered with, leaked, or forged during transmission, storage, and use, making it difficult to reliably guarantee its authenticity and integrity. For data users such as insurance companies and financial institutions, the lack of reliable and comprehensive vehicle data means that they often rely on experience-based models or limited information in key business operations such as risk assessment, premium pricing, and claims decisions, resulting in a need to improve both accuracy and efficiency.

[0005] Therefore, there is a need for a blockchain-based vehicle data processing system and method that can extract features from multi-source heterogeneous raw vehicle data and store them securely. Summary of the Invention

[0006] One of the objectives of this invention is to provide a blockchain-based vehicle data processing system that can extract features from multi-source heterogeneous raw vehicle data and store them securely.

[0007] To solve the above-mentioned technical problems, this application provides the following technical solution: A blockchain-based vehicle data processing system includes a data acquisition module and a cloud platform. The cloud platform includes a data processing module, a data uploading module, and a blockchain network. The data acquisition module is used to collect raw vehicle data; The data processing module is used to clean and denoise the raw vehicle data; it is also used to extract advanced driving behavior features, including trip recognition, driving event recognition, and dimensional score aggregation. The data upload module is used to upload the processed data to the blockchain network.

[0008] Furthermore, the original vehicle data includes one or more of the following: vehicle speed, acceleration, braking frequency, steering angle, GPS location, engine speed, fault codes, tire pressure, battery management system data, and driver monitoring system data.

[0009] Furthermore, when the data processing module performs cleaning and noise reduction, it is used to identify and remove outliers, redundant data, and transmission errors, and also to standardize and normalize vehicle data.

[0010] Furthermore, the data processing module is used to obtain GPS trajectory data using GPS geographic location when identifying trip characteristics, and to identify the start and end points, mileage and travel time of each trip based on the GPS trajectory data and vehicle start and stop status. It is also used to identify dangerous driving events by using threshold judgment and pattern recognition algorithms, and to calculate the number of occurrences, intensity and duration of such events; to acquire or identify distracted driving and fatigued driving states based on driver monitoring system data; and to identify specific driving conditions and driving habits of vehicles, including: high-speed driving, low battery driving, driving with faults, night driving, peak driving, number of long-distance driving trips, cross-regional driving, driving in environmental risks, driving in high-risk areas, number of stops in high-risk areas, and the number of nighttime charging trips, daytime charging trips, slow charging trips and fast charging trips involving energy replenishment behavior.

[0011] Furthermore, the data processing module is used to generate a comprehensive risk score during dimensional score aggregation: .

[0012] A second objective of this invention is to provide a blockchain-based vehicle data processing method, which, using the aforementioned system, includes the following steps: S1. Collect raw vehicle data in real time and upload it to the cloud platform in real time; S2, The cloud platform performs real-time streaming processing on the uploaded data; S3. Extract advanced driving behavior features from the processed data and calculate a comprehensive risk score; S4. Generate a data fingerprint from the data processed in steps S2 and S3, package the data fingerprint with metadata, and digitally sign the data packet. S5. Upload the data packet to the blockchain network and complete consensus storage through a smart contract.

[0013] Furthermore, the real-time stream processing in S2 specifically includes: Data cleaning and noise reduction: identifying and removing outliers, redundant data, and transmission errors from the data. Standardization and normalization involve unifying the dimensions and normalizing the range of data from different sensors to eliminate differences in heterogeneous data. Multi-dimensional aggregation aggregates the cleaned data according to time granularity to generate multi-dimensional statistical data.

[0014] Furthermore, step S3 specifically includes: Trip feature recognition utilizes GPS geolocation to obtain GPS trajectory data, and identifies the start and end points, mileage, and travel time of each trip based on the GPS trajectory data and vehicle start and stop status. Driving event feature recognition utilizes threshold judgment and pattern recognition algorithms to identify dangerous driving events and calculate their frequency, intensity, and duration; it acquires or identifies distracted driving and fatigued driving states based on driver monitoring system data; it is also used to identify specific vehicle driving conditions and usage habits, including: high-speed driving, low battery driving, driving with faults, night driving, peak driving, number of long-distance driving trips, cross-regional driving, driving in environmental risks, driving in high-risk areas, number of stops in high-risk areas, as well as nighttime charging, daytime charging, slow charging, and fast charging related to energy replenishment behavior; The dimensional scores are aggregated to generate a comprehensive risk score: .

[0015] Furthermore, step S4 specifically includes: The multi-dimensional statistical data generated in step S2 and the trip identification, driving event identification feature data and comprehensive risk score generated in step S3 are organized into a structured JSON object; The JSON object is serialized into a string, and a hash algorithm is applied to generate a 256-bit hash value as a data fingerprint. The data fingerprint is packaged with metadata, which includes: anonymized vehicle identification number, data type, time granularity, generation timestamp, evaluation level, data source identifier, and optional zero-knowledge proof credential. Digitally sign the data packet using a preset private key.

[0016] This solution acquires heterogeneous data from multiple sources and performs cleaning, noise reduction, standardization, and normalization. It eliminates format differences and noise between different sensors and data sources, transforming fragmented raw data into standardized, usable data, thus solving the problems of data silos and low quality. Unlike simple raw data storage, this solution extracts advanced driving behavior features, identifies trips and driving events, and aggregates multi-dimensional scores, transforming low-density raw data into high-value behavioral features and risk scores, supporting subsequent risk assessment and refined management. The processed data is then uploaded to a blockchain network. Utilizing the distributed ledger and consensus mechanism of blockchain, it ensures that vehicle historical data, once generated, cannot be maliciously tampered with or forged, providing credible data endorsement for scenarios such as insurance claims and used car evaluations. Attached Figure Description

[0017] Figure 1 This is a logic block diagram of a first embodiment of a blockchain-based vehicle data processing system; Figure 2 This is a schematic diagram of data processing in an embodiment of a blockchain-based vehicle data processing system. Figure 3 This is a schematic diagram of data uploading to the blockchain in an embodiment of a blockchain-based vehicle data processing system. Detailed Implementation

[0018] The following detailed description illustrates the specific implementation method: Example 1 like Figure 1 As shown in the figure, a vehicle data processing system based on blockchain in this embodiment includes a data acquisition module and a cloud platform. The cloud platform includes a data processing module, a data uploading module, and a blockchain network.

[0019] The data acquisition module is used to collect static data and raw vehicle data related to the vehicle. In this embodiment, the data acquisition module, through in-depth cooperation with the automaker, pre-installs a customized TBOX device in mass-produced models, configures the data acquisition frequency to once every 5 seconds, collects raw vehicle data through the car's CAN bus, and uploads encrypted data packets to the cloud platform in real time via 4G or 5G networks.

[0020] The collected raw vehicle data includes vehicle speed, acceleration, braking frequency, steering angle, GPS location, engine speed, fault codes, tire pressure, and data specific to new energy vehicles, such as battery management system data (e.g., SOC, SOH), and driver monitoring system data.

[0021] Vehicle-related static data, including vehicle age, driving experience, vehicle model, and new vehicle purchase price.

[0022] like Figure 2As shown, the data processing module is used to clean and denoise the raw vehicle data reported in real time by deploying a data processing engine based on Kafka and Flink. For example, it uses a combination of statistical methods (such as moving average and median filtering) and machine learning methods (such as Isolation Forest and One-Class SVM) to identify and remove outliers, redundant data, and transmission errors. Outliers, such as 9999, and transmission errors are determined according to preset rules. It is also used to standardize and normalize the vehicle data, eliminating differences in the units and ranges of data from different sensors, laying the foundation for subsequent in-depth analysis. In this embodiment, the data processing module is also used to process the raw vehicle data into multi-dimensional statistical data with granularity of hours, days, months, and trips.

[0023] The data processing module is also used for feature extraction of advanced driving behaviors, specifically: When identifying trip characteristics, GPS trajectory data is obtained using GPS geographic location. Based on the GPS trajectory data and vehicle start and stop status, the start and end points, mileage, and travel time of each trip are identified. When identifying driving event characteristics, feature extraction is performed based on raw vehicle data such as vehicle speed, acceleration, GPS location, and battery management system data, combined with pre-set judgment conditions. Specifically, threshold judgment and pattern recognition algorithms (such as Hidden Markov Models) are used to identify dangerous driving events such as rapid acceleration, rapid deceleration, sharp turns, and speeding, and to calculate their frequency, intensity, and duration. Simultaneously, distracted driving and fatigued driving states are acquired or identified based on driver monitoring system data. The system is also used to identify specific vehicle driving conditions and usage habits, including: high-speed driving, low battery driving, driving with malfunctions, nighttime driving, peak-hour driving, number of long-distance trips, cross-regional driving, driving in environmentally risky areas, driving in high-risk areas, number of stops in high-risk areas, and energy replenishment behaviors such as nighttime charging, daytime charging, slow charging, and fast charging. The judgment conditions for all the above identification results are pre-set; in this embodiment, to improve the accuracy and adaptability of the identification, the system sets differentiated judgment conditions for the above identification items based on the vehicle's geographical location.

[0024] For example, the engineering specifications for some features are as follows:

[0025] When quantifying driving habit characteristics, driving style profiles are constructed by analyzing long-term series of driving behavior data, such as aggressive, steady, and conservative driving styles, and their risk levels are quantified.

[0026] The data processing module is also used for dimensional score aggregation, considering data at the time granularity of hours, days, months, and trips, and incorporating environmental factors such as driving scenarios (e.g., highways, cities, rural areas), road conditions (e.g., congestion, smooth traffic), and weather, as well as individual factors such as vehicle type and driver age. Through reinforcement learning, the weights of each dimension's score are dynamically adjusted to more accurately reflect the actual usage of the vehicle and driving risks. For example, the risk weight for sudden braking in congested urban areas should differ from that on highways, ultimately generating a comprehensive driving risk score.

[0027] Traditional algorithms are typically as follows: ; Traditional methods rely heavily on manual experience or static settings, lacking data-driven support. Once implemented, they struggle to handle the non-linear relationship between high-frequency time-series vehicle data and actual accident outcomes. Weights cannot be dynamically adjusted based on different vehicles, drivers, and usage scenarios, resulting in insufficient personalization. Furthermore, models struggle to continuously optimize themselves with new data, limiting the accuracy of risk assessment.

[0028] This embodiment aggregates dimensional scores: ; The key feature is the transformation of high-frequency time-series data into structured risk characteristic data with business implications. This includes data such as monthly average number of rapid accelerations, monthly average number of distracted drivings, monthly average number of fatigued drivings, monthly average number of rapid decelerations, monthly average number of high-speed drivings, monthly average number of low-battery drivings, monthly average number of drivings with malfunctions, monthly average number of nighttime drivings, monthly average number of peak-hour drivings, monthly average number of trips, monthly average driving time, monthly average mileage, monthly average number of long-distance drivings, monthly average number of cross-regional drivings, monthly average number of drivings at environmental risks, monthly average mileage at environmental risks, monthly average number of trips in high-risk areas, monthly average average driving speed, monthly average driving time, monthly average number of nighttime chargings, monthly average number of daytime chargings, monthly average number of slow chargings, monthly average number of fast chargings, and monthly average number of stays in high-risk areas. This data is presented in slices of hours, days, and trips, along with static data such as vehicle age, driving experience, vehicle type, and new vehicle purchase price.

[0029] The weights are calculated using a machine learning model to dynamically analyze the correlation between features and actual vehicle accident data, thereby determining the importance coefficient for each feature. In this embodiment, the actual vehicle accident data is provided by the cooperating insurance company and can be obtained in real time via a data interface or periodically via data packets.

[0030] Specifically, the steps for obtaining the weight value are as follows: Step 1, Construct the training sample set: Construct training samples, each sample containing: feature vector X = {feature1, feature2, …, feature…} n}; Label Y = Whether the vehicle had an accident / Number of accidents / Amount of loss during the corresponding period; The sample comes from the matching results of features obtained from the original vehicle data and the actual accident data of the vehicle.

[0031] Step 2, Correlation Modeling between Features and Outcomes: A machine learning model is used to establish the mapping relationship between features and outcomes. The model includes, but is not limited to, logistic regression, gradient boosting decision tree (GBDT), random forest, XGBoost / LightGBM, and neural network models. During model training, the contribution of each feature to the risk of an accident is automatically learned.

[0032] Step 3, Calculation of feature weights: Depending on the model type, calculate the feature weights using one or a combination of the following methods: Method 1: Weight extraction based on model parameters. For linear models, the feature coefficients obtained from model training are directly used as weights. The weights are normalized to ensure that each weight is within a comparable range.

[0033] Method 2: Weight calculation based on feature importance. For tree models, feature importance is calculated based on indicators such as information gain, number of splits, and SHAP value; feature importance is then mapped to corresponding weight values.

[0034] Step 4, Weight Normalization and Stability Constraints: To avoid excessive or abnormally amplified weight fluctuations, further measures include: weight normalization processing, upper and lower limit constraints on weights, weight smoothing processing (such as sliding window, exponential decay); and grouping constraints for different feature categories (behavioral, environmental, and vehicle attribute categories).

[0035] Step 5, the feedback-based adaptive weight update mechanism: During the actual operation of the model, new vehicle operation data is continuously collected; the latest accident results are continuously incorporated; the model is retrained periodically or in real time; and the weights are adaptively corrected based on the deviation between the predicted results and the actual accident results. This achieves dynamic adjustment of weights according to changes in driving behavior; differentiated weight combinations for different vehicles and drivers; and continuous improvement in risk assessment accuracy through long-term model evolution.

[0036] Through intelligent processing by the data processing module, raw, discrete vehicle data is transformed into structured, high-value, multi-dimensional statistical and feature data, laying a solid foundation for subsequent data on-chaining and financial value mining.

[0037] like Figure 3 As shown, the data on-chain module is used to generate data fingerprints from processed data using a high-strength hash algorithm (such as SHA-256); it is also used to package the data fingerprints together with metadata such as timestamps, data source identifiers, and data types into a data packet, and upload it to the blockchain network through a smart contract.

[0038] In this embodiment, the information contained in each data block in the blockchain includes: vin, Vehicle Identification Number (anonymized hash value); data_fingerprint, data fingerprint (raw data hash value); data_type, data type (e.g., driving behavior statistics, vehicle health data); time_granularity, time granularity (e.g., hourly, daily, monthly, trip); timestamp, the timestamp used to generate the data; level, driving behavior evaluation level; data_source_id, the identifier of the data source (e.g., the hash value of the TBOX device ID); zkp_proof, zero-knowledge proof credential (optional, used for privacy-preserving verification).

[0039] In this embodiment, the blockchain network is built based on the AntChain platform. Node deployment utilizes cloud services or private deployment solutions provided by AntChain.

[0040] Before data is uploaded, the data on-chain module uses its private key to digitally sign the data packet to ensure the authenticity and non-repudiation of the data source.

[0041] The original signed data is hashed to generate a data fingerprint. Specifically, the processed multi-dimensional statistical data and advanced driving behavior feature data are organized into a structured JSON object. Then, the JSON object is serialized into a string. Finally, the SHA-256 hash algorithm is applied to the JSON string to generate a 256-bit hash value (64 characters in hexadecimal representation) as the data fingerprint.

[0042] Invoke a pre-deployed smart contract (such as "storeDataFingerprint") to submit the data fingerprint and related metadata to the blockchain.

[0043] In a blockchain network, each node performs consensus verification on the received data packets (e.g., using the PBFT consensus algorithm) to verify the validity of the data signature and the uniqueness of the data fingerprint. Once verification is successful, the data fingerprint and its metadata are written into the blockchain ledger, forming an immutable record.

[0044] This solution leverages the data processing engines of Kafka and Flink, combined with statistical and machine learning methods for data cleaning. It can handle the real-time streaming processing needs of massive amounts of vehicle data, significantly improving data throughput and anomaly detection accuracy. When calculating the comprehensive risk score, this solution abandons the traditional fixed weights and instead introduces machine learning. It adaptively adjusts the weights of each factor based on feedback from actual accident data, allowing the risk scoring model to continuously evolve with environmental changes and data accumulation, reflecting actual driving risks more accurately than traditional experience-based models. This solution does not directly upload all raw plaintext data to the blockchain. Instead, it organizes the data into JSON objects and generates data fingerprints using the SHA-256 algorithm. These fingerprints, along with anonymized metadata, are then packaged and uploaded to the blockchain. This leverages blockchain to ensure data non-repudiation and tamper-proofing, effectively protecting users' detailed privacy data from public disclosure, while also reducing the storage pressure on the blockchain.

[0045] Example 2 The difference between this embodiment and Embodiment 1 is that this embodiment also includes a vehicle credit assessment module, an insurance support module, and a transaction management module; The vehicle credit assessment module acquires trusted driving behavior fingerprints, vehicle mileage fingerprints, and vehicle maintenance record fingerprints from the blockchain. It employs a deep learning model to perform time-series analysis on historical vehicle behavior data and combines this with graph neural networks to analyze the relationship between the vehicle and the driver, constructing a multi-factor vehicle credit scoring model. This model considers driving risk, comprehensively assessing vehicle usage intensity, maintenance condition, and potential malfunction risks to generate a comprehensive and dynamic vehicle credit score. This provides financial institutions (such as banks and auto finance companies) with accurate vehicle loan amount suggestions, interest rate pricing basis, used car residual value assessment, and leasing risk assessment services, significantly reducing the credit risk for financial institutions.

[0046] The insurance support module is based on trusted driving risk scores, vehicle fault data fingerprints, accident history data fingerprints, and driving area data fingerprints on the blockchain. It utilizes an actuarial AI model, combined with Bayesian networks and decision tree algorithms, to accurately quantify the risks associated with different driving behavior patterns and vehicle usage habits. Based on real-time and historical vehicle data, it dynamically adjusts insurance rates to achieve personalized pricing for UBI (Usage-Based Insurance) products.

[0047] When a vehicle accident occurs, the insurance company can authorize access to vehicle behavior data fingerprints (such as speed, braking, and collision sensor data) on the blockchain before and after the accident through the platform. This data, combined with AI image recognition technology, is used to analyze accident scene photos, enabling intelligent reconstruction of the accident scenario and rapid determination of liability. Smart contracts can preset claims rules, automatically triggering the claims process when specific conditions are met, improving claims efficiency and reducing fraud.

[0048] It can provide insurance companies with more scientific and personalized insurance product design solutions (such as mileage-based insurance and driving behavior-based insurance), optimize renewal pricing, and significantly improve the accuracy and efficiency of claims processing.

[0049] The transaction management module is used to authorize and charge data users for specific types of data fingerprints through smart contracts. All data access records are transparently verifiable on the blockchain, ensuring the compliance of data usage.

[0050] It is also used to build a data asset trading market, allowing data providers to monetize the analytical value of their data fingerprints while protecting privacy, thereby promoting the circulation and maximizing the value of data elements.

[0051] Example 3 This embodiment provides a blockchain-based vehicle data processing method, including the following steps: S1. Raw vehicle data is collected in real time via a customized TBOX device pre-installed in the vehicle and using the automotive CAN bus interface; the optimal collection frequency is once every 5 seconds. The collected data is encrypted and then uploaded to the cloud platform in real time via a 4G or 5G network.

[0052] The original vehicle data includes: vehicle speed, acceleration, braking frequency, steering angle, GPS location, engine speed, fault codes, tire pressure, and battery management system data unique to new energy vehicles (including SOC remaining charge and SOH battery health status).

[0053] S2, the cloud platform utilizes a data processing engine based on Kafka and Flink to perform real-time stream processing on uploaded data, specifically including: Data cleaning and denoising employs statistical methods (such as moving average and median filtering) combined with machine learning algorithms (such as Isolation Forest and One-Class SVM) to identify and remove outliers, redundant data, and transmission errors from the data.

[0054] Standardization and normalization involve unifying the dimensions and normalizing the range of data from different sensors to eliminate differences in heterogeneous data.

[0055] Multi-dimensional aggregation aggregates the cleaned data according to time granularity such as hour, day, month, and itinerary to generate multi-dimensional statistical data.

[0056] S3. Extract advanced driving behavior features from the processed data and calculate a comprehensive risk score, specifically including: Trip recognition uses GPS location data to obtain GPS trajectory data. Based on the GPS trajectory data and vehicle start / stop status, it identifies the start and end points, mileage, and duration of each trip.

[0057] Driving event recognition, based on acceleration sensor and vehicle speed data, combined with preset thresholds and pattern recognition algorithms (such as Hidden Markov Model), identifies dangerous driving events such as rapid acceleration, rapid deceleration, sharp turns, and speeding, and records the number of occurrences, intensity, and duration of these events.

[0058] Quantify driving habits by analyzing long-term driving data to construct driving style profiles, including aggressive, steady, and conservative driving styles.

[0059] The dimensional score aggregation incorporates environmental factors such as driving scenarios (e.g., highways, cities), road conditions (e.g., congestion), and weather, as well as individual factors related to the vehicle and driver. A reinforcement learning model is used to dynamically adjust the nonlinear aggregation function of the weights for each dimension to calculate the comprehensive risk score. ; Where w is the factor weight, which is determined by the vehicle data factor and the actual vehicle accident data. The factor with the highest correlation is dynamically found through machine learning, and the weight is adaptively adjusted based on the actual data feedback. S4. Generate data fingerprints and digitally sign data packets, specifically including: The data is structured by organizing the multi-dimensional statistical data generated in step S2 and the trip identification, driving event identification feature data and comprehensive risk score generated in step S3 into a structured JSON object.

[0060] To generate a fingerprint, serialize the JSON object into a string and apply a high-strength hash algorithm (such as SHA-256) to generate a 256-bit hash value as the data fingerprint.

[0061] Data packaging involves bundling the data fingerprint with metadata. Metadata includes: anonymized vehicle identification number (VIN hash), data type, time granularity, generation timestamp, rating level, data source identifier (device ID hash), and optional zero-knowledge proof credential (ZKP).

[0062] Digital signatures use the private key of the data on-chain module to digitally sign data packets, ensuring the authenticity and non-repudiation of the data.

[0063] S5. Upload the data packet to the blockchain network and complete consensus storage via a smart contract. Specifically, a pre-deployed smart contract on the blockchain (such as "storeDataFingerprint") is invoked to submit the signed data packet to the blockchain network. Each node in the blockchain network uses a consensus algorithm (such as PBFT, the Practical Byzantine Fault Tolerance algorithm) to verify the received data packet, verifying the validity of the digital signature and the uniqueness of the data fingerprint. After successful verification, the data fingerprint and its metadata are written into the blockchain distributed ledger, forming an immutable evidence record.

[0064] The above are merely embodiments of the present invention. The invention is not limited to the fields covered by these embodiments. Commonly known structures and characteristics in the solutions are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are able to access all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A blockchain-based vehicle data processing system, comprising a data acquisition module and a cloud platform, characterized in that, The cloud platform includes a data processing module, a data upload module, and a blockchain network; The data acquisition module is used to collect raw vehicle data; The data processing module is used to clean and denoise the raw vehicle data; It is also used for advanced driving behavior feature extraction, including: trip recognition, driving event recognition, and dimensional score aggregation; The data upload module is used to upload the processed data to the blockchain network.

2. The blockchain-based vehicle data processing system according to claim 1, characterized in that: The original vehicle data includes one or more of the following: vehicle speed, acceleration, braking frequency, steering angle, GPS location, engine speed, fault codes, tire pressure, battery management system data, and driver monitoring system data.

3. The blockchain-based vehicle data processing system according to claim 2, characterized in that: When the data processing module performs cleaning and noise reduction, it is used to identify and remove outliers, redundant data, and transmission errors, and also to standardize and normalize vehicle data.

4. The blockchain-based vehicle data processing system according to claim 3, characterized in that: The data processing module is used to obtain GPS trajectory data using GPS geographic location when identifying trip characteristics, and to identify the start and end points, mileage and travel time of each trip based on the GPS trajectory data and vehicle start and stop status. It is also used to identify dangerous driving events by using threshold judgment and pattern recognition algorithms when identifying driving event characteristics, and to calculate the number of occurrences, intensity and duration of such events; Based on driver monitoring system data acquisition or identification of distracted driving and fatigued driving conditions; It is also used to identify specific driving conditions and usage habits of vehicles, including: high-speed driving, low battery driving, driving with faults, night driving, peak driving, number of long-distance driving trips, cross-regional driving, driving in environmental risks, driving in high-risk areas, number of stops in high-risk areas, as well as nighttime charging trips, daytime charging trips, slow charging trips, and fast charging trips involving energy replenishment behavior.

5. The blockchain-based vehicle data processing system according to claim 4, characterized in that: The data processing module is used to generate a comprehensive risk score during dimensional score aggregation. 。 6. A blockchain-based vehicle data processing method, using the system described in any one of claims 1-5, characterized in that, Includes the following steps: S1. Collect raw vehicle data in real time and upload it to the cloud platform in real time; S2, The cloud platform performs real-time streaming processing on the uploaded data; S3. Extract advanced driving behavior features from the processed data and calculate a comprehensive risk score; S4. Generate a data fingerprint from the data processed in steps S2 and S3, package the data fingerprint with metadata, and digitally sign the data packet. S5. Upload the data packet to the blockchain network and complete consensus storage through a smart contract.

7. The vehicle data processing method based on blockchain according to claim 6, characterized in that: The real-time stream processing in S2 specifically includes: Data cleaning and noise reduction: identifying and removing outliers, redundant data, and transmission errors from the data. Standardization and normalization involve unifying the dimensions and normalizing the range of data from different sensors to eliminate differences in heterogeneous data. Multi-dimensional aggregation aggregates the cleaned data according to time granularity to generate multi-dimensional statistical data.

8. The vehicle data processing method based on blockchain according to claim 7, characterized in that: Step S3 specifically includes: Trip feature recognition utilizes GPS geolocation to obtain GPS trajectory data, and identifies the start and end points, mileage, and travel time of each trip based on the GPS trajectory data and vehicle start and stop status. Driving event feature recognition utilizes threshold judgment and pattern recognition algorithms to identify dangerous driving events and calculate their frequency, intensity, and duration; it acquires or identifies distracted driving and fatigued driving states based on driver monitoring system data; it is also used to identify specific vehicle driving conditions and usage habits, including: high-speed driving, low battery driving, driving with faults, night driving, peak driving, number of long-distance driving trips, cross-regional driving, driving in environmental risks, driving in high-risk areas, number of stops in high-risk areas, as well as nighttime charging, daytime charging, slow charging, and fast charging related to energy replenishment behavior; The dimensional scores are aggregated to generate a comprehensive risk score: 。 9. The vehicle data processing method based on blockchain according to claim 8, characterized in that: Step S4 specifically includes: The multi-dimensional statistical data generated in step S2 and the trip identification, driving event identification feature data and comprehensive risk score generated in step S3 are organized into a structured JSON object; The JSON object is serialized into a string, and a hash algorithm is applied to generate a 256-bit hash value as a data fingerprint. The data fingerprint is packaged with metadata, which includes: anonymized vehicle identification number, data type, time granularity, generation timestamp, evaluation level, data source identifier, and optional zero-knowledge proof credential. Digitally sign the data packet using a preset private key.

Images