Blockchain Double-Chain Cold Chain Traceability Management System for Fluctuations in Meat Quality

Abstract

The blockchain dual-chain cold chain traceability management system for meat quality fluctuations belongs to the technical field of traceability management systems. It includes a data collection and preprocessing module, a composite index module, a security verification module, a federated learning and AI resource management module, a sharding and verification architecture module, and an automated error correction module. Through the blockchain dual-chain architecture, the present invention innovatively separates the data chain for storing environmental data hashes and the transaction chain for recording operation logs, combines the anchoring technology to synchronize hash fingerprints, ensures cross-chain consistency, and realizes long-term immutability of data through anti-quantum hash algorithms. The hierarchical storage design optimizes resource allocation. The composite index technology integrates the hypergraph dynamic index to trace production batches, vehicles, warehouses, and the spatio-temporal quadtree to dynamically subdivide regions, significantly compresses the complex query response time, gives real-time warnings for path deviations, improves the cache hit rate of edge nodes, and significantly enhances the security and query efficiency of the entire link data.

Description

Technical Field

[0001] The present invention relates to the technical field of traceability management systems, and particularly to a blockchain double-chain cold chain traceability management system for meat quality fluctuations. Background Art

[0002] Meat traceability management records, tracks, and verifies key data throughout the entire life cycle of meat products from breeding, slaughtering, processing, transportation to sales through technical means, enabling rapid location of pollution sources, ensuring food safety, identifying abnormal links such as cold chain breaks and transportation overtimes, preventing quality risks, and clarifying the responsible entities in the production and circulation links through the data chain. Consumers can view the meat source, inspection reports, and logistics tracks by scanning the code.

[0003] Published Patent: A Meat Traceability Management System and Method (Publication No.: CN118446708B), which includes a data repository, and also includes a production-transportation-sales information binding module, an information classification and visualization module, an information group rearrangement module, an information traceability verification module, and a traceability processing and analysis module. The present invention can classify various types of data and generate a tree diagram, making the data structured and visualized, and facilitating the quick finding of meats that may have the same quality problems and timely processing. By shuffling and reordering multi-step data, generating random numbers and determining the reorganized data, the traceability and reliability of the data are enhanced. This binding method of random numbers to finished meats gives each finished meat a unique identifier, improving the traceability efficiency. And by calculating the verification judgment value, the meat source of the finished meat is traced. Based on the comparison method of distance and time interval, the discrimination accuracy is improved, avoiding ineffective traceability and improving the traceability efficiency.

[0004] The above patent relies on traditional data reorganization and tree classification, lacks modern technology support, does not cover dynamic tracking of temperature, humidity, and geographical location, has a lag in environmental anomaly response, low efficiency of the tree traversal mechanism, insufficient spatio-temporal correlation analysis ability, relies on fixed rules for determination, has no dynamic risk prediction and resource optimization, and depends on manual investigation for data errors, lacking an automated repair process. Summary of the Invention

[0005] The purpose of the present invention is to provide a blockchain double-chain cold chain traceability management system for meat quality fluctuations, which solves the problems of defects in traditional data processing technology, insufficient dynamic environment tracking, inefficient response mechanism, and rigid analysis and rules.

[0006] To achieve the above object, the present invention provides the following technical solutions: A blockchain double-chain cold chain traceability management system for meat quality fluctuations, including a data collection and preprocessing module, a composite index module, a security verification module, a federated learning and AI resource management module, a sharding and verification architecture module, and an automated error correction module. The data collection and preprocessing module includes an Internet of Things sensor unit, a data cleaning unit, and a check code attachment unit. The Internet of Things sensor unit is deployed in a transport vehicle to collect location data, temperature data, humidity data, and timestamp data in real time. The data cleaning unit standardizes the format of the original data and eliminates redundant information. The check code attachment unit attaches an error correction code to the data label. The composite index module includes a hypergraph dynamic index and a spatio-temporal quadtree. The hypergraph dynamic index is modeled based on vertices and hyperedges, where vertices represent entities and hyperedges represent multi-dimensional association relationships. The spatio-temporal quadtree stores cold chain transportation data in layers according to geographical regions and time windows. The security verification module includes a quantum-resistant hash acceleration unit and a fully homomorphic encryption verification unit. The quantum-resistant hash acceleration unit uses a quantum-resistant algorithm combined with a hierarchical hash pool to implement hash calculation. The fully homomorphic encryption verification unit performs homomorphic operations on consumer query data and outputs a verification decision value. The federated learning and AI resource management module includes a federated learning model and an AI-driven data preloading unit. The federated learning model aggregates the model parameters locally trained by each participant. The AI-driven data preloading unit predicts high-frequency query patterns based on a neural network. The sharding and verification architecture module includes a spatio-temporal sharding unit, a lightweight cross-shard consensus unit, and a parallel verification unit. The spatio-temporal sharding unit divides data shards according to time windows and geographical regions. The lightweight cross-shard consensus unit generates a cross-shard data consistency proof in combination with zero-knowledge proof. The parallel verification unit supports parallel processing of multiple transactions within a single shard. The automated error correction module attaches a check code to the data label, and the smart contract automatically detects and repairs input errors.

[0007] Further, the hypergraph dynamic conditional index generates a subgraph index through the following steps:

[0008] a. Extract high-frequency conditional combinations from the user's historical query logs;

[0009] b. Generate a subgraph index based on the hyperedge association relationship;

[0010] c. Cache the subgraph index to the edge node;

[0011] The spatio-temporal range query optimization of the spatio-temporal quadtree includes:

[0012] a. Divide the spatial dimension by geographical region levels;

[0013] b. Divide the time level by time windows;

[0014] c. Each tree node stores data within the corresponding spatio-temporal range;

[0015] d. Locate abnormal events through tree traversal.

[0016] Furthermore, the hierarchical hash pool of the quantum-resistant hash acceleration unit includes: a unit for real-time processing of the hash of the current data block, a unit for parallel caching of the hashes of recent data, and a unit for asynchronous archiving of the hashes of historical data; the fully homomorphic encryption verification unit performs the following steps when a consumer scans the code for query:

[0017] a. Perform homomorphic operations on the encrypted retail point label;

[0018] b. Output a verification decision value;

[0019] c. Trigger a traceability process when the verification decision value exceeds a preset threshold.

[0020] Furthermore, the federated learning model optimizes the caching policy through the following steps:

[0021] a. Each participating party locally trains a risk prediction model;

[0022] b. Upload the model parameters to the blockchain aggregation node;

[0023] c. Adjust the caching priority of the edge nodes according to the risk level;

[0024] The AI-driven data preloading unit optimizes the cache replacement policy through the following steps:

[0025] a. Train a cache replacement model based on historical query logs;

[0026] b. Adjust the cache weights according to the query frequency and data timeliness;

[0027] c. Retain high-value data.

[0028] Furthermore, the lightweight cross-shard consensus unit verifies the data consistency between shards through zero-knowledge proof.

[0029] Furthermore, the smart contract of the automated error correction module detects the check code in the data label. When the verification fails, it repairs the incorrect label based on redundant coding and updates the repaired data to the blockchain shard.

[0030] Further, the Internet of Things sensor unit of the data acquisition and preprocessing module transmits cold chain transportation data to the blockchain network in real time. The cold chain transportation data includes location data, temperature data, humidity data, and timestamps. After being encrypted, the data is temporarily stored through edge nodes. The data cleaning unit unifies the time format to Coordinated Universal Time (UTC), converts geographical coordinates into latitude and longitude codes, and eliminates duplicate or invalid data entries. The checksum attachment unit attaches redundant error correction codes to the production batch number, transportation vehicle number, and retail point number.

[0031] Further, the data acquisition and preprocessing module works in coordination with the composite index module, and the cleaned data is written into the blockchain double chain. The blockchain double chain includes a production chain and a logistics chain. The spatio-temporal quadtree constructs an index based on the cleaned spatio-temporal data. The hypergraph dynamic index generates subgraphs based on the cleaned entities and association relationships.

[0032] Further, the operation process of the blockchain double-chain cold chain traceability management system for meat quality fluctuations is as follows:

[0033] S1. Data Acquisition and Preprocessing

[0034] S1.1: Data Acquisition

[0035] Cold chain transportation data: The location, temperature, and humidity data of the transportation vehicle are collected in real time through sensors.

[0036] Production-end data: Information on the production process is recorded.

[0037] Sales-end data: Dealer information and the sales date are entered.

[0038] S1.2: Data Preprocessing: The collected raw data is cleaned, standardized, and checksum is attached to support subsequent error correction.

[0039] S2. Data Upload to the Blockchain and Index Construction

[0040] S2.1: Writing to the Blockchain Double Chain

[0041] Production chain: The production-end data is written into the production chain to generate a unique hash identifier.

[0042] Logistics chain: The cold chain transportation data is written into the logistics chain and stored by spatio-temporal sharding.

[0043] S2.2: Generation of Composite Index

[0044] Spatio-temporal quadtree: Build an index by stratifying according to geographical regions and time windows to quickly locate abnormal events.

[0045] Hypergraph dynamic index: Model entities as vertices and multi-dimensional association relationships as hyperedges to generate a dynamic subgraph index to support complex queries.

[0046] S3. Resource Optimization Driven by Federated Learning

[0047] S3.1: Local Model Training: Each participant trains a risk prediction model based on local data;

[0048] S3.2: Global Model Aggregation: Integrate model parameters to generate a global risk assessment model;

[0049] S3.3: Dynamic Cache Preloading: Dynamically adjust the edge node cache policy according to the risk level;

[0050] S4. Consumer Query and Security Verification

[0051] S4.1: Scan Code to Trigger Query: Consumers scan the product label to submit a query request;

[0052] S4.2: Fully Homomorphic Encryption Verification: Perform homomorphic operations on the encrypted data to output a verification decision value;

[0053] S4.3: Automatic Error Correction: If the verification fails, automatically repair the error label;

[0054] S5. Temporal and Spatial Sharding and Traceability Verification

[0055] S5.1: Parallel Verification within Shards: Process transactions in parallel within shards;

[0056] S5.2: Cross-Shard Consistency Verification: Verify the data consistency between shards through zero-knowledge proof;

[0057] S5.3: Composite Index Tracing: Locate the multi-dimensional associations of the problem batch and retrieve abnormal events;

[0058] S6. Report Generation and Feedback

[0059] S6.1: Full-Link Report Generation: Integrate data to generate a traceability report;

[0060] S6.2: Encrypted Feedback to the Terminal: The report is encrypted and returned to the consumer terminal;

[0061] S6.3: Closed-Loop Optimization: Dynamically optimize the processing flow according to the feedback.

[0062] Compared with the prior art, the beneficial effects of the present invention are:

[0063] 1. The present invention innovatively separates the data hash of the storage environment data and the operation log of the transaction chain record through the blockchain double-chain architecture, combines the anchoring technology to synchronize the hash fingerprint, ensures cross-chain consistency, and realizes the long-term immutability of data through the quantum-resistant hash algorithm. The hierarchical storage design optimizes resource allocation. The hot data layer has high processing capabilities, and the cold data layer ensures the integrity of historical data through distributed nodes and erasure coding technology. The quantum security and fully homomorphic encryption technology use hash signatures and dynamic salt value rotation to resist quantum attacks, support ciphertext verification, and combine the three-party key sharding mechanism to meet privacy compliance requirements. The composite index technology integrates the hypergraph dynamic index to trace the production batch, vehicle, warehouse, and spatio-temporal quadtree dynamic sub-regions, significantly compresses the complex query response time, gives real-time alerts for path deviation, improves the cache hit rate of edge nodes, and significantly enhances the security and query efficiency of the entire link data.

[0064] 2. The present invention supports multi-party local modeling, integrates the global risk assessment model, dynamically classifies high-risk data and preferentially caches it to edge nodes, significantly improving the response speed. The spatio-temporal sharding and parallel verification technology dynamically shards by time and space, combines zero-knowledge proof to compress the cross-shard verification data, reduces network latency, and breaks through the single-shard throughput by parallel processing transactions. Automatically allocate computing resources during peak periods to maximize resource utilization. The automatic error correction mechanism is based on redundant coding and smart contracts, realizes real-time error location and repair, synchronously updates data to all link nodes, and combines the error log optimization algorithm to ensure data integrity and system robustness.

[0065] 3. The system of the present invention realizes the full life cycle management of data through hierarchical storage and dynamic sharding. The hot data is processed in real time, the warm data responds quickly, and the cold data is archived for a long time. Combining the spatio-temporal sharding technology optimizes the storage efficiency. The security verification module integrates quantum-resistant hash and fully homomorphic encryption, supports high-frequency ciphertext verification requests, and generates judicial evidence through smart contracts to meet high-standard compliance requirements. The intelligent resource scheduling and preloading strategy predict high-frequency query scenarios, dynamically allocate cache weights, significantly reducing the main chain load, ensuring the end-to-end traceability efficiency and stability, and providing a highly reliable, highly intelligent, and high-throughput integrated solution for the entire cold chain link. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is the system architecture diagram of the present invention;

[0067] Figure 2 It is the flowchart of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0068] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0069] To solve the technical problems of traditional data processing technology defects, insufficient dynamic environment tracking, inefficient response mechanisms, and rigid analysis and rules, as Figure 1-2 shown, the following preferred technical solutions are provided:

[0070] The blockchain double-chain cold chain traceability management system for meat quality fluctuations includes a data collection and preprocessing module, a composite index module, a security verification module, a federated learning and AI resource management module, a sharding and verification architecture module, and an automated error correction module. The data collection and preprocessing module includes an Internet of Things sensor unit, a data cleaning unit, and a check code attachment unit. The Internet of Things sensor unit is deployed in the transport vehicle to collect location data, temperature data, humidity data, and timestamp data in real time. The data cleaning unit standardizes the format of the original data and eliminates redundant information. The check code attachment unit attaches an error correction code to the data label. The composite index module includes a hypergraph dynamic index and a spatio-temporal quadtree. The hypergraph dynamic index is modeled based on vertices and hyperedges, where vertices represent entities and hyperedges represent multi-dimensional association relationships. The spatio-temporal quadtree stores cold chain transportation data in layers according to geographical regions and time windows. The security verification module includes a quantum-resistant hash acceleration unit and a fully homomorphic encryption verification unit. The quantum-resistant hash acceleration unit uses a quantum-resistant algorithm combined with a hierarchical hash pool to implement hash calculation. The fully homomorphic encryption verification unit performs homomorphic operations on the consumer query data and outputs a verification judgment value. The federated learning and AI resource management module includes a federated learning model and an AI-driven data preloading unit. The federated learning model aggregates the model parameters locally trained by each participant. The AI-driven data preloading unit predicts high-frequency query patterns based on neural networks. The sharding and verification architecture module includes a spatio-temporal sharding unit, a lightweight cross-shard consensus unit, and a parallel verification unit. The spatio-temporal sharding unit divides data shards according to time windows and geographical regions. The lightweight cross-shard consensus unit generates a cross-shard data consistency proof in combination with zero-knowledge proofs. The parallel verification unit supports parallel processing of multiple transactions within a single shard. The automated error correction module attaches a check code to the data label, and the smart contract automatically detects and corrects input errors.

[0071] The dynamic conditional index of the hypergraph dynamic index generates a subgraph index through the following steps:

[0072] a. Extract the high-frequency condition combinations in the user's historical query logs;

[0073] b. Generate a subgraph index based on the hyperedge association relationship;

[0074] c. Cache the subgraph index to the edge nodes;

[0075] The spatio-temporal range query optimization of the spatio-temporal quadtree includes:

[0076] a. Divide the spatial dimension by geographical region levels;

[0077] b. Divide the time level by time windows;

[0078] c. Each tree node stores the data within the corresponding spatio-temporal range;

[0079] d. Locate abnormal events through tree traversal.

[0080] The hierarchical hash pool of the quantum-resistant hash acceleration unit includes: a unit for real-time processing of the hash of the current data block, a unit for parallel caching of the hashes of recent data, and a unit for asynchronous archiving of the hashes of historical data; the fully homomorphic encryption verification unit performs the following steps when the consumer scans the code for query:

[0081] a. Perform a homomorphic operation on the encrypted retail location label;

[0082] b. Output a verification decision value;

[0083] c. Trigger a traceability process when the verification decision value exceeds a preset threshold.

[0084] The federated learning model optimizes the caching strategy through the following steps:

[0085] a. Each participating party locally trains a risk prediction model;

[0086] b. Upload the model parameters to the blockchain aggregation node;

[0087] c. Adjust the caching priority of the edge nodes according to the risk level;

[0088] The AI-driven data preloading unit optimizes the cache replacement strategy through the following steps:

[0089] a. Train a cache replacement model based on historical query logs;

[0090] b. Adjust the cache weights according to the query frequency and data timeliness;

[0091] c. Retain high-value data.

[0092] The lightweight cross-shard consensus unit verifies the data consistency between shards through zero-knowledge proof.

[0093] The intelligent contract of the automated error correction module detects the check code in the data label. When the check fails, it repairs the incorrect label based on redundant coding and updates the repaired data to the blockchain shard.

[0094] The IoT sensor unit of the data collection and preprocessing module transmits cold chain transportation data to the blockchain network in real time. The cold chain transportation data includes location data, temperature data, humidity data, and timestamps. After being encrypted, the data is temporarily stored through edge nodes. The data cleaning unit unifies the time format to Coordinated Universal Time (UTC), converts geographical coordinates into latitude and longitude codes, and eliminates duplicate or invalid data entries. The check code attachment unit attaches redundant error correction codes to the production batch, transportation vehicle, and retail point labels.

[0095] The data collection and preprocessing module works in coordination with the composite index module. The cleaned data is written into the blockchain double chain. The blockchain double chain includes a production chain and a logistics chain. The spatio-temporal quadtree constructs an index based on the cleaned spatio-temporal data. The hypergraph dynamic index generates subgraphs based on the cleaned entities and association relationships.

[0096] The data collection and preprocessing module is an important part of the blockchain double-chain cold chain traceability management system for meat quality fluctuations. It is responsible for collecting raw data from cold chain transportation, production, and sales links and processing it to provide a basis for subsequent operations. The IoT sensor unit is deployed in cold chain transportation vehicles to collect key data in real time. It obtains the vehicle's geographical location information through the BeiDou Navigation Satellite System (BDS), records the temperature changes of the goods and the humidity level of the transportation environment using temperature and humidity sensors, adds an accurate timestamp to each piece of data to ensure the integrity of the time series. The collected data is encrypted and transmitted to the blockchain network and temporarily stored in edge nodes to ensure security. The data cleaning unit processes the originally collected data, unifies all time data into Coordinated Universal Time (UTC) to ensure consistency, converts geographical location information into latitude and longitude codes to facilitate subsequent spatial index construction, removes duplicate data entries and obviously invalid data to improve data quality. The check code attachment unit attaches redundant error correction codes to the labels of each production batch, transportation vehicle, and retail point, and attaches a Reed-Solomon check code at the end of each data label to support the automated error correction function, ensure the integrity of the data during transmission and storage, and reduce the impact of errors. Through the coordinated work of these three units, the data collection and preprocessing module effectively collects, cleans, and prepares high-quality cold chain transportation data, provides a reliable data foundation for the entire traceability management system, and significantly improves the system's fault tolerance through check code technology, providing high-precision and high-reliability data support for blockchain double-chain storage, composite index construction, and security verification.

[0097] As the core part of the blockchain double-chain cold chain traceability management system for meat quality fluctuations, the composite index module integrates the hypergraph dynamic index and the spatio-temporal quadtree technology to achieve multi-dimensional association modeling and efficient spatio-temporal query of cold chain data, solve the problem of efficient retrieval and verification of heterogeneous data in cold chain traceability scenarios. Its core principle is to fuse the labels, hash values, and spatio-temporal information in the blockchain double-chain to construct a hybrid index structure that supports cross-chain queries, spatio-temporal range screening, and anomaly traceability. The hypergraph dynamic index is based on hypergraph theory to model the entities and association relationships in cold chain data. Entities such as production batches, transport vehicles, and distributors are set as vertices, and multi-dimensional associations are used as hyperedges. By analyzing the high-frequency condition combinations in the user query log, sub-graph indexes are automatically generated and cached to edge nodes to provide a quick response to complex condition queries and reduce the main chain query load. In the entire cold chain, the entities of production batches, vehicles, and warehouses are abstracted into hypergraph nodes, and the loading, transportation, and storage business relationships are defined as hyperedges to construct a hypergraph network covering all links of production, transportation, and sales. Each hyperedge is dynamically bound to the hash pointer of the blockchain double-chain to achieve cross-chain data association. When new cold chain data appears, the system automatically parses the hyperedges affected by the event, updates the hypergraph weights, and generates incremental indexes to support multi-hop queries. With the hypergraph path traversal algorithm, it can quickly locate associated data blocks. For the abnormal nodes marked in the hypergraph, it can trace the upstream and downstream influence ranges along the hyperedge direction, generate a risk diffusion map, and pre-load it into the index to speed up subsequent query speeds. The spatio-temporal quadtree stores cold chain data in layers according to geographical regions and time windows. The geographical dimension is divided by administrative region levels, and the time dimension is divided by time windows. Each tree node stores the data hash pointer within the corresponding spatio-temporal range. Through the hierarchical division of the quadtree structure and the tree traversal algorithm, abnormal events within the spatio-temporal range can be quickly located. For spatio-temporal range queries, the target node can be quickly located through quadtree hierarchical traversal to improve query efficiency. The BDS trajectory points of transport vehicles are divided into quadtree spaces according to time windows and geographical regions and aggregated into spatio-temporal cube units. Each cube unit is associated with the original environmental data hash set in the blockchain data chain, and a quadtree hierarchical index is established. The system will dynamically identify the spatio-temporal regions with high-frequency access, refine and divide the corresponding quadtree nodes in sub-spaces to improve local query accuracy. For low-frequency historical data regions, the quadtree nodes are merged to reduce index storage overhead. Based on the quadtree index, the spatio-temporal coverage relationship between the preset transport route and the actual trajectory can be quickly compared to identify deviation regions, trigger index marking, and generate warning events. It also supports composite spatio-temporal condition queries and can terminate invalid search branches in advance through the quadtree hierarchical filtering mechanism.

[0098] Based on the type of user requests, the query engine dynamically selects either a hypergraph or a quadtree index and uses a caching strategy to reduce redundant calculations. If the query contains multi-entity association conditions, it preferentially calls the hypergraph dynamic index. If the query involves spatio-temporal ranges, it preferentially calls the spatio-temporal quadtree. By caching high-frequency subgraph indexes and quadtree node data at edge nodes, the number of main-chain accesses is reduced, and the overall query response time is significantly reduced. In complex queries, the hypergraph is used to locate associated batches, and the spatio-temporal quadtree is used to filter area and time information, enabling a rapid collaborative response. When an abnormal event occurs, the quadtree quickly locates the abnormal event, and the hypergraph traces the associated production and transportation links to form a complete evidence chain. When a new block is added to the data chain, the spatio-temporal quadtree updates the index according to the timestamp. When the transaction chain records operation changes, the hypergraph dynamic index synchronously adjusts the states of associated hyperedges to ensure strong consistency between the index and the data on the chain. The hypergraph dynamic index is responsible for processing complex relationship retrievals, and the quadtree focuses on spatio-temporal range filtering. The two collaborate to respond to different types of queries through a weight-based priority scheduler, avoiding resource competition. The composite index module, while ensuring the immutability of blockchain data, breaks through the performance bottleneck of traditional on-chain traversal and realizes the core capabilities of quickly locating abnormal batches in cold-chain data and rapidly generating full-link traceability reports, providing efficient and accurate data support for full-link traceability.

[0099] The security verification module is the core security barrier of the blockchain dual-chain cold chain traceability management system. It uses cryptographic technology to ensure the integrity, non-repudiation and privacy of cold chain data. It operates based on the on-chain and off-chain collaborative verification mechanism. It performs quantum-resistant hash calculations and fully homomorphic encryption on the temperature records and transportation trajectory data of the entire cold chain, generates a verifiable ciphertext summary, and cross-verifies the ciphertext summary with the original hash value and operation log stored in the blockchain dual chain to ensure that the data has not been tampered with. It combines quantum-resistant algorithms with lightweight homomorphic encryption technology to resist traditional computing power attacks and future quantum computing threats, and supports fast batch verification in ciphertext state. The quantum-resistant hash acceleration unit uses the SPHINCS+ hash algorithm based on lattice cryptography to replace the traditional SHA-25. 6 algorithms, generate quantum-safe hash fingerprints for cold chain data labels, and bind them to the metadata of the blockchain dual chain. FPGA chips are used to implement parallel calculations of quantum-resistant hash algorithms, increasing the hash processing speed of cold chain data to more than 5 times that of traditional CPUs. Dynamic key rotation mechanism is supported, and the salt value of the hash algorithm is automatically updated every 24 hours to prevent rainbow table attacks. The quantum-resistant hash value is written into the transaction chain and the data chain at the same time to form a double verification anchor point. When the verification request is triggered, the hash value consistency in the dual chain is compared. If a deviation is found, it is immediately marked as a suspicious data block. A hierarchical hash pool architecture is used to accelerate hash calculations in a hierarchical manner. The hot pool processes the hash calculation of the current data block in real time through FPGA hardware, and the warm pool uses GPU to cache recent data in parallel. Hash, supports rapid verification of recent data integrity, cold pool asynchronously archives historical data hash through distributed nodes to ensure that long-term data cannot be tampered with, and the fully homomorphic encryption verification unit uses fully homomorphic encryption technology to encrypt sensitive fields, allowing third parties to directly execute verification logic on ciphertext without decrypting data, and optimize encryption operations through polynomial approximation algorithms, compressing the time required for a single verification from minutes to milliseconds, verifying the legitimacy of the data format in an encrypted state, and executing cold chain business rules. When the transportation time exceeds 72 hours, the integrity of the temperature curve is checked compulsorily, and encrypted verification results are generated. Abnormal patterns are identified through homomorphic comparison operators, and encryption alarm tags are triggered. Linked with the quantum-resistant hash acceleration unit, when fully homomorphic verification finds an abnormality, it automatically Automatically call the quantum-resistant hash unit to recalculate the data block fingerprint, perform dual-chain consistency review, support key sharding trusteeship, and distribute decryption permissions to regulatory departments, enterprises, and third-party audit institutions. Plaintext data can only be restored when the three parties provide key sharding. Quantum resistance and privacy protection work together. The quantum-resistant hash unit ensures the long-term security of data, and the fully homomorphic encryption unit realizes the availability of privacy data. The two maintain collaboration through shared security context. The lightweight verification protocol is aimed at cold chain mobile devices. The verification process is abstracted into a challenge-response protocol. The fully homomorphic encryption unit generates a lightweight verification code within 20 bytes. The quantum-resistant unit completes local fast verification. The verification results of the trusted audit channel automatically generate a digital evidence package through the blockchain smart contract.It contains anti-quantum hash fingerprints, homomorphic verification logs and double-chain anchoring information, meets the requirements of judicial evidence preservation, realizes data without leaving the domain and plaintext without landing, can support more than 100,000 cold-chain data verification requests per second, and ensures that it still has a security life cycle of at least 20 years in the era of quantum computing. Through the combination of hierarchical hash acceleration and fully homomorphic encryption technology, it systematically solves the shortcomings of traditional traceability systems in terms of quantum security, verification efficiency and privacy protection, and provides full-life-cycle security protection for cold-chain data.

[0100] The Federated Learning and AI Resource Management Module is the core of the system's intelligence. It enables data collaboration under privacy protection through federated learning technology and combines AI-driven resource optimization strategies to improve system efficiency. Participating parties such as production enterprises including meat processing plants and slaughterhouses, logistics providers including cold chain transportation companies, warehousing service providers, distributors including wholesalers, retailers, e-commerce platforms, and quality inspection agencies use local private cold chain transportation records including BDS trajectories, temperature control data, humidity logs, timestamps, sales data including sales dates, customer information, return records, inventory status, production data including production batches, hygiene inspection reports, raw material source information, and processing technology parameters to independently train risk prediction models. During the training process, the data does not need to leave the local area to ensure privacy compliance. Subsequently, only the model parameters are uploaded to the blockchain aggregation node. The blockchain node integrates the model parameters of all participating parties through federated averaging algorithms, weighted averaging algorithms, and gradient aggregation algorithms to generate a unified global risk assessment model. This global model is continuously updated and can reflect the dynamic risk characteristics of the entire supply chain. The risk levels are divided into high risks including temperature exceeding the threshold, path deviation exceeding 10 kilometers, medium risks including humidity fluctuation exceeding the limit, transportation delay exceeding 6 hours, and low risks including normal data and no abnormal records. According to the risk levels output by the global model, high-risk data is preferentially cached to edge nodes, which can optimize the cache hit rate and ensure that high-risk data with high-frequency access can quickly respond to queries. The global model can real-time analyze high-risk paths in the supply chain including temperature-abnormal transportation vehicles, distributors with frequent quality problems, production batches with repeated hygiene problems, and transportation nodes with excessive path deviation, and accordingly dynamically adjust the cache priorities of edge nodes. High-risk data of recent abnormal batches including batches with temperature control exceeding the standard, return rates exceeding 5%, vehicles that trigger alarms multiple times including triggering more than 3 alarms within 24 hours, and goods associated with recall events including goods associated with raw material pollution are pre-loaded. The AI-driven data pre-loading unit analyzes historical query logs based on the Transformer neural network to identify high-frequency query scenarios. At the same time, it analyzes time including accurate to the minute level, region including provincial administrative codes, municipal administrative codes, keywords including temperature exceeding the standard, vehicle A, batch recall, abnormal total number of colonies, user types including consumers, regulatory agencies, internal quality inspectors in enterprises, and logistics administrators' historical query characteristics. It captures long-distance dependencies through the self-attention mechanism, predicts possible future high-frequency queries, and combines reinforcement learning algorithms to dynamically evaluate the data value based on query frequency including times / minute, timeliness including the time difference between data generation time and the current time, risk level including high, medium, low, data relevance including the degree of association with abnormal events, and the integrity of batch traceability links, and adjusts the cache weights. High-value data is preferentially retained in the cache, and low-value data is replaced from the cache according to the strategy. The predicted high-frequency query data is pre-loaded to regional servers including the North China Data Center, East China Edge Node, and South China Distributed Storage Center Edge Node, which can reduce the number of direct accesses to the main chain and reduce the main chain load.Reduce the complex query response time from seconds to milliseconds, effectively reduce the query load on the main chain, improve resource utilization. The federated learning model works in collaboration with the AI-driven data preloading unit. The federated learning model guides the cache priority through global risk analysis. When a certain transport vehicle is determined to be at high risk, the AI preloading unit will preferentially cache the historical temperature control data of the vehicle, including the temperature fluctuation curve, the path deviation record (including the coordinates of the points deviating from the preset route), the associated quality inspection report (including the microbial test results), and the transportation compliance certificate. The AI preloading unit then ensures efficient resource allocation through prediction and optimization strategies. When consumers query, they can directly obtain the results quickly from the edge node, realizing end-to-end efficient traceability and resource optimization, achieving that data does not leave the domain (producer data does not leave the local server, logistics provider data does not share the original trajectory, risks can be collaboratively shared across participating parties for risk model parameters and jointly update the risk level), significantly improving the system response speed and stability, and solving the pain points of traditional traceability systems in terms of data islands (cross-enterprise data cannot be interconnected, formats are inconsistent, resource waste due to redundant storage occupying most of the space, inefficient queries consuming computing power, response latency with the main chain throughput below 1000 TPS, and query response exceeding 2 seconds), providing intelligent decision-making support for the whole-link quality management.

[0101] The sharding and verification architecture module is the core performance optimization component of the meat quality fluctuation blockchain dual-chain cold chain traceability management system. It divides data storage through spatio-temporal sharding, ensures global consistency through lightweight cross-shard consensus, and improves the processing efficiency of a single shard through parallel verification, systematically solving the problems of low throughput and large cross-shard communication overhead in traditional blockchain systems. The spatio-temporal sharding unit divides the blockchain network into multiple independent shards according to the spatio-temporal characteristics of cold chain data. In the time dimension, the data attribution is divided according to the fixed time window from 00:00 to 23:59 every day. In the space dimension, the geographical shards are divided according to the administrative division boundaries of provinces and cities based on BDS coordinates. Each shard corresponds to a specific 24-hour time window and geographical area, independently storing and processing cold chain data within the corresponding spatio-temporal range, including the BDS coordinates of transport vehicles, temperature control records, humidity records, and cargo status information, storing the hash pointers, timestamps, regional codes, batch numbers, and product category metadata of cold chain transportation data, supporting fast positioning and querying. The data write, query, update, and delete transactions within a shard are independently processed without cross-shard communication, reducing latency. When the data volume in a single shard exceeds the threshold of 10,000 records, it is automatically split into new shards according to hourly sub-time windows, with each hour being a window or sub-region at the district or county level, such as Chaoyang District in Beijing, Pudong New Area in Shanghai, and Tianhe District in Guangzhou, to avoid performance bottlenecks. It also dynamically shards according to the time represented by quarters Q1, Q2, Q3, Q4, months January, February, March, April, May, June, July, August, September, October, November, December, and the geographical levels of provinces and cities. The query efficiency within each shard is optimized using a B+ tree index, supporting fast retrieval of data within a single shard. The shards store data independently, reducing the storage requirements of nodes. The lightweight cross-shard consensus unit uses zero-knowledge proof technology to generate cross-shard data consistency proofs, verifying the logical relevance of data between different shards, including the matching of production batches and logistics records, the matching of production batches and warehousing records, and the matching of logistics records and sales records, without broadcasting data to all nodes, significantly reducing communication overhead. When verifying the cross-shard data relevance, the source shard generates a zero-knowledge proof containing fields such as batch number, hash value, transaction time, shipping location, and receiving location. The proof content only exposes the logical correctness of hash matching, correct time order, and reasonable location association, hiding sensitive data such as temperature control values, specific quantities of goods, and unit prices. The target shard quickly verifies the validity of the proof through a preset algorithm, verifying the hash consistency, compliance of the timestamp order, and accuracy of the location correspondence relationship. After successful verification, the cross-shard association status is updated, marking that the batch has completed the logistics binding, marking the connection between the warehousing and logistics links as completed, and marking the correspondence between sales and logistics information as correct. The cross-shard consensus communication overhead is reduced to 15% of the traditional scheme, compressed from 1MB to 150KB, supporting asynchronous verification, without real-time synchronization between shards, reducing network pressure. A Merkle root hash is generated for each shard data block as the unique identifier of data integrity. When performing cross-shard queries, only the shard summary hash is exchanged instead of the original data, and the data consistency between two shards is proven through zero-knowledge proof.Reduce the cross-shard communication original minute-level latency to the second-level through a lightweight protocol. The parallel verification unit uses the single instruction multiple data instruction set to parallelly process multiple transactions such as data writing, query requests, update operations, and delete instructions within a single shard, significantly improving the transaction throughput within the shard. Group the transactions within the shard according to data upload, query, verification, modification, and append types, and allocate independent calculation threads to each group to parallelly execute transaction verifications such as hash verification, digital signature verification, data format check, and permission verification. The throughput of a single shard reaches 10,000 TPS. For concurrent operations involving the same data, avoid conflicts through timestamp sorting and optimistic locking mechanisms. Conflicting transactions are automatically rolled back and re-queued for processing to ensure data consistency. Dynamically adjust the number of CPU threads, GPU cores, and memory allocation amount of computing resources according to the transaction queue length, task priority, and resource occupancy of the shard load. Automatically expand the capacity to increase the number of threads, enable more GPU cores, and allocate more memory during peak periods, and release redundant resources during low load to reduce energy consumption. Within a single shard, use SIMD technology to parallelly process multiple transaction verification tasks, group multiple transactions within the shard such as uploading temperature data to the chain, consumer query requests, updating humidity data, and changing the status of goods records, and dynamically allocate computing resources such as CPU threads, GPU cores, memory space, and network bandwidth, and give priority to processing high-risk or high-frequency transactions such as temperature anomaly records, queries of high-risk batches, and updates of emergency order data. During the parallel verification process, automatically isolate invalid or malicious transactions such as requests for tampering with data, illegal permission operations, and submission of data with incorrect formats to ensure the data security within the shard. Parallelly process 1,000 consumer query requests, 500 data upload operations, and 300 update tasks within a certain shard, and control the response time within 200 ms. Spatial and temporal shard isolation for data processing and parallel verification accelerate the execution of a single shard, and the overall throughput of the system is increased to 10,000 TPS, while that of the traditional system is only 1,000 TPS. In terms of global consistency guarantee, lightweight cross-shard consensus ensures the credible association of data across regions and time periods, such as verifying the matching of production batches and logistics records in different regions and the coherence of warehousing and sales data in different quarters. In terms of efficient resource utilization, dynamic shard adjustment and parallel verification resource allocation improve the hardware utilization rate and reduce the operating cost. This module deeply integrates spatial and temporal sharding, zero-knowledge proof, and parallel computing technologies to provide a high-throughput, low-latency, and strong-consistency underlying architecture support for the cold chain traceability system, breaking through the performance bottleneck of traditional blockchains and achieving breakthroughs in three dimensions: storage efficiency, cross-shard consistency, and processing speed, providing underlying support for the efficient operation of the cold chain traceability system.,

[0102] The automated error correction module is the core fault-tolerant component of the blockchain dual-chain cold chain traceability management system for meat quality fluctuations. It realizes the automatic detection and repair of data entry errors through redundant coding and smart contract technology, ensuring the integrity and consistency of the full-link data. An error correction code based on the Reed-Solomon algorithm is appended to the end of the data labels such as production batch numbers, transport vehicle numbers, and retail point numbers to generate a complete label containing redundant information. The Reed-Solomon algorithm is used to generate a 4-digit hexadecimal redundant check code for each label, and the check code and the original label are written into the production chain and logistics chain of the blockchain dual-chain. During the data preprocessing stage, when generating labels in the production, transportation, and sales links, the check bits are calculated through the Reed-Solomon algorithm and appended to the end of the label to achieve standardized storage, serving as the benchmark for subsequent verification. The smart contract monitors the data entry or transmission process in real time. When consumers scan the code to query or data is uploaded to the chain, the label is automatically extracted and the check code is separated to verify the data integrity. If a label error is detected, such as character missing, format error, coding mismatch, or character misalignment, an error alarm is triggered, the data is marked as "to be repaired", and the error type is recorded. Based on the error correction ability of the redundant coding, the smart contract reversely calculates the error position according to the check code in the label, locates the error field, reversely deduces the original correct label using the redundant information of the Reed-Solomon algorithm, recalculates the check code of the repaired label to ensure consistency with the original data, regenerates the hash value of the repaired label, and after re-encryption, updates it to the corresponding space-time shard of the blockchain shard logistics chain or production chain. The old version label is marked as invalid, and the new version data is synchronized to all relevant nodes, such as logistics providers and distributors. At the same time, the error type, repair time, and repaired label are recorded in the error log for subsequent analysis. The coding length is dynamically adjusted according to the error frequency, increasing the redundant bits in high-frequency error scenarios, and the complexity of the redundant coding is dynamically adjusted according to the repair success rate, increasing the length of the check bits to balance the storage overhead and fault tolerance ability. The high-frequency error types, such as manual entry mistakes and sensor transmission interference, are statistically analyzed to optimize the coding rules or the design of the entry interface. By implementing the automated error correction module, the error recall rate is reduced, the need for manual intervention is decreased, and the error correction efficiency is improved to millisecond-level response. This module systematically solves the problem of data distortion caused by manual entry or transmission errors in traditional traceability systems through an automated error correction process driven by a smart contract, significantly enhancing the robustness and operation efficiency of the traceability system while ensuring data integrity, providing technical support for the credibility of the full-chain data.

[0103] The operation process of the blockchain dual-chain cold chain traceability management system for meat quality fluctuations is as follows:

[0104] S1. Data collection and preprocessing

[0105] S1.1: Data collection

[0106] Cold chain transportation data: The location, temperature, and humidity data of the transportation vehicle are collected in real time through sensors;

[0107] Production - end data: Record information on the production process;

[0108] Sales - end data: Enter dealer information and sales date;

[0109] S1.2: Data pre - processing: Clean, standardize the collected raw data, and append a check code to support subsequent error correction;

[0110] S2. Data uploading to the blockchain and index construction

[0111] S2.1: Writing to the blockchain's dual chains

[0112] Production chain: Write the production - end data to the production chain to generate a unique hash identifier;

[0113] Logistics chain: Write the cold chain transportation data to the logistics chain and store it by spatio - temporal sharding;

[0114] S2.2: Generation of composite indexes

[0115] Spatio - temporal quadtree: Build indexes by stratifying according to geographical regions and time windows to quickly locate abnormal events;

[0116] Hypergraph dynamic index: Model entities as vertices, associate multi - dimensional relationships as hyper - edges, and generate dynamic sub - graph indexes to support complex queries;

[0117] S3. Resource optimization driven by federated learning

[0118] S3.1: Local model training: Each participant trains a risk prediction model based on local data;

[0119] S3.2: Global model aggregation: Integrate model parameters to generate a global risk assessment model;

[0120] S3.3: Dynamic cache pre - loading: Dynamically adjust the edge - node cache strategy according to the risk level;

[0121] S4. Consumer query and security verification

[0122] S4.1: Scan the code to trigger a query: Consumers scan the product label to submit a query request;

[0123] S4.2: Fully homomorphic encryption verification: Perform homomorphic operations on the encrypted data to output a verification judgment value;

[0124] S4.3: Automatic error correction: If the verification fails, automatically repair the error label;

[0125] S5. Spatio - temporal sharding and traceability verification

[0126] S5.1: Parallel Verification within a Shard: Process transactions in parallel within a shard;

[0127] S5.2: Cross-Shard Consistency Verification: Verify data consistency between shards through zero-knowledge proofs;

[0128] S5.3: Composite Index Tracing: Locate multi-dimensional associations of problem batches and retrieve abnormal events;

[0129] S6. Report Generation and Feedback

[0130] S6.1: Full-Link Report Generation: Integrate data to generate a tracing report;

[0131] S6.2: Encrypted Feedback to the Terminal: Return the encrypted report to the consumer terminal;

[0132] S6.3: Closed-Loop Optimization: Dynamically optimize the processing flow based on the feedback.

[0133] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0134] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. Blockchain double-chain cold chain traceability management system for meat quality fluctuations, characterized in that, It includes a data collection and preprocessing module, a composite index module, a security verification module, a federated learning and AI resource management module, a sharding and verification architecture module, and an automated error correction module. The data collection and preprocessing module includes an IoT sensor unit, a data cleaning unit, and a check code attachment unit. The composite index module includes a hypergraph dynamic index and a spatio-temporal quadtree. The security verification module includes a quantum-resistant hash acceleration unit and a fully homomorphic encryption verification unit. The federated learning and AI resource management module includes a federated learning model and an AI-driven data preloading unit. The sharding and verification architecture module includes a spatio-temporal sharding unit, a lightweight cross-shard consensus unit, and a parallel verification unit. The automated error correction module attaches a check code to the data label, and the smart contract automatically detects and corrects the input error; The dynamic condition index of the hypergraph dynamic index generates a subgraph index through the following steps: a. Extract the high-frequency condition combinations in the user's historical query logs; b. Generate a subgraph index based on the hyperedge association relationship; c. Cache the subgraph index to the edge node; The spatio-temporal range query optimization of the spatio-temporal quadtree includes: a. Divide the spatial dimension by geographical area levels; b. Divide the time level by time windows; c. Each tree node stores the data within the corresponding spatio-temporal range; d. Locate abnormal events through tree traversal; The hierarchical hash pool of the quantum-resistant hash acceleration unit includes: a unit for real-time processing of the hash of the current data block, a unit for parallel caching of the hashes of recent data, and a unit for asynchronous archiving of the hashes of historical data. The fully homomorphic encryption verification unit performs the following steps when the consumer scans the code for query: a. Perform homomorphic operations on the encrypted retail point label; b. Output a verification judgment value; c. Trigger a traceability process when the verification judgment value exceeds a preset threshold; The data collection and preprocessing module works in cooperation with the composite index module, and the cleaned data is written into the blockchain double chain. The blockchain double chain includes a production chain and a logistics chain. The spatio-temporal quadtree constructs an index based on the cleaned spatio-temporal data. The hypergraph dynamic index generates a subgraph based on the cleaned entities and association relationships.

2. The blockchain double-chain cold chain traceability management system for meat quality fluctuations according to claim 1, wherein: The federated learning model optimizes the cache policy through the following steps: a. Each participating party locally trains a risk prediction model; b. Upload the model parameters to the blockchain aggregation node; c. Adjust the cache priority of the edge node according to the risk level; The AI-driven data preloading unit optimizes the cache replacement policy through the following steps: a. Train a cache replacement model based on the historical query logs; b. Adjust the cache weights according to the query frequency and data timeliness; c. Retain high-value data.

3. The blockchain double-chain cold chain traceability management system for meat quality fluctuations as described in claim 2, characterized in that: The lightweight cross-shard consensus unit verifies the data consistency between shards through zero-knowledge proof.

4. The blockchain double-chain cold chain traceability management system for meat quality fluctuations according to claim 1, characterized in that: The smart contract of the automated error correction module detects the check code in the data label. When the verification fails, it repairs the incorrect label based on the redundant coding and updates the repaired data to the blockchain shard.

5. The blockchain double-chain cold chain traceability management system for meat quality fluctuations according to claim 1, characterized in that: The IoT sensor unit of the data collection and preprocessing module transmits cold chain transportation data to the blockchain network in real time. The cold chain transportation data includes location data, temperature data, humidity data, and timestamps. After being encrypted, the data is temporarily stored through edge nodes. The data cleaning unit standardizes the time format to Coordinated Universal Time (UTC), converts geographical coordinates into latitude and longitude codes, and eliminates duplicate or invalid data entries. The checksum attachment unit attaches redundant error correction codes to the production batch numbers, transportation vehicle numbers, and retail point numbers.

6. The blockchain double-chain cold chain traceability management system for meat quality fluctuations according to any one of claims 1-5, characterized in that: The operation process is as follows: S1. Data collection and preprocessing S1.1: Data collection Cold chain transportation data: The location, temperature, and humidity data of the transportation vehicle are collected in real time through sensors. Production - end data: Information on the production process is recorded. Sales - end data: Dealer information and the sales date are entered. S1.2: Data preprocessing: The collected raw data is cleaned, standardized, and checksum is attached to support subsequent error correction. S2. Data uploading to the blockchain and index construction S2.1: Writing to the dual - chain of the blockchain Production chain: The production - end data is written to the production chain to generate a unique hash identifier. Logistics chain: The cold chain transportation data is written to the logistics chain and stored in time - space slices. S2.2: Generation of composite index Spatio - temporal quadtree: An index is constructed in layers according to geographical regions and time windows to quickly locate abnormal events. Hypergraph dynamic index: Entities are modeled as vertices, and multi - dimensional relationships are associated as hyper - edges to generate dynamic sub - graph indexes to support complex queries. S3. Resource optimization driven by federated learning S3.1: Local model training: Each participant trains a risk prediction model based on local data. S3.2: Global model aggregation: Integrate model parameters to generate a global risk assessment model. S3.3: Dynamic cache pre - loading: Dynamically adjust the cache strategy of edge nodes according to the risk level. S4. Consumer query and security verification S4.1: Scanning code to trigger query: Consumers scan the product label to submit a query request. S4.2: Fully homomorphic encryption verification: Perform homomorphic operations on the encrypted data to output a verification judgment value. S4.3: Automatic error correction: If the verification fails, automatically repair the incorrect label. S5. Time - space slicing and traceability verification S5.1: Parallel verification within the slice: Transactions are processed in parallel within the slice. S5.2: Cross - slice consistency verification: Verify the data consistency between slices through zero - knowledge proofs. S5.3: Composite index traceability: Locate the multi - dimensional associations of the problem batch and retrieve abnormal events. S6. Report generation and feedback S6.1: Full - link report generation: Integrate data to generate a traceability report. S6.2: Encrypted feedback to the terminal: The report is encrypted and returned to the consumer terminal. S6.3: Closed - loop optimization: Dynamically optimize the processing flow according to the feedback.

Images