Research and development data traceability and collaborative management system based on block chain

Abstract

The invention belongs to the technical field of data management, and discloses a research and development data traceability and collaborative management system based on a block chain, and the system comprises a sensing layer which is used for collecting multi-source research and development data on various research and development tool chains through a multi-protocol collection agent, carrying out the structural packaging of the data to generate metadata, and embedding a data blood relationship tracking mark in the metadata; false data filtering is carried out on the metadata by deploying a trusted execution environment; the chain network layer is used for dividing the filtered metadata into data fragments and respectively storing the data fragments in a distributed storage platform; applying a Hash algorithm to each data fragment to generate a corresponding Hash value; taking the hash values corresponding to all the data fragments as leaf nodes, constructing a fragment hash tree, and further obtaining global root hash; the security of research and development data is ensured, and the cooperation efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of data management technology, and more specifically, to a blockchain-based R&D data traceability and collaborative management system. Background Art

[0002] The patent publication number CN118484840A discloses a blockchain-based credible data asset security management and traceability system, including: extracting keyword pairs from credible data, identifying sensitive character segments in the credible data according to the keyword pairs, obtaining sensitive character segments, desensitizing all sensitive character segments to form desensitized credible data; performing a hash process on the desensitized credible data after data sharding to obtain merged blocks, cloned merged blocks and unique identifiers; performing a secondary hash process on the merged blocks and cloned merged blocks to obtain a storage hash table; restoring the merged blocks or cloned merged blocks according to the storage hash table and the unique identifier, and performing traceability analysis on the restored merged blocks or cloned merged blocks to determine the authenticity of the stored desensitized credible data or the tampering node; the present invention can ensure the security of credible data transmission and storage.

[0003] The existing blockchain-based R&D data traceability and collaborative management system has the following main problems:

[0004] Trust deficits during the data collection phase: Blockchain can only guarantee that data cannot be tampered with after it is uploaded, but it cannot verify the authenticity of the data source. R&D personnel may submit false information by forging sensor data or tampering with experimental records, resulting in a "garbage in, garbage out" problem.

[0005] Blockchain performance bottlenecks: Most blockchains face issues such as slow transaction processing, small block capacity, and long confirmation times. This results in inefficient data upload and query, making it difficult to support high-frequency R&D data interaction scenarios.

[0006] Poor cross-chain interoperability: Data integration between different blockchains is difficult. R&D data exists on one chain, while the approval process is on another. Interoperability relies on inconsistent standards and limited maturity of cross-chain protocols.

[0007] Rigid configuration of the number of repair shards: Traditional erasure coding technologies typically use a fixed number of repair shards and cannot be dynamically adjusted based on actual conditions. This static redundant shard configuration can lead to wasted resources or insufficient data recovery capabilities.

[0008] Slow recovery from data loss and corruption: Existing distributed storage systems may rely on complete shard reconstruction when data loss or corruption occurs. This reconstruction process is time-consuming and inefficient, especially when a large number of shards are lost.

[0009] Lack of intelligent behavior monitoring and risk assessment: The lack of quantitative analysis of the dynamic behavior of remaining data shards (such as access patterns, modification frequency, and historical credibility) makes it impossible to distinguish between high-trust and low-trust data shards, leading to rigid resource allocation and a lack of trust mechanisms.

[0010] The problem of difficulty in verifying the integrity of data after repair. During the data repair process, it is difficult to ensure that the repaired data is accurate and not tampered with. Especially in a decentralized environment, the lack of effective means to verify the repair process may lead to inconsistencies or errors in the repaired data.

[0011] In view of this, the present invention proposes a blockchain-based R&D data traceability and collaborative management system to solve the above problems. Summary of the Invention

[0012] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned objectives, the present invention provides the following technical solution: a blockchain-based R&D data traceability and collaborative management system, comprising:

[0013] The perception layer collects multi-source R&D data from various R&D tool chains through multi-protocol collection agents, performs structured packaging to generate metadata, and embeds data lineage tracking tags in the metadata. By deploying a trusted execution environment, the metadata is filtered for false data.

[0014] The chain network layer divides the filtered metadata into data shards and stores them separately in the distributed storage platform; applies the hash algorithm to each data shard to generate the corresponding hash value; uses the hash values corresponding to all data shards as leaf nodes to build a shard hash tree, and then obtains the global root hash;

[0015] Use the global root hash as a summary, integrate all data shards and their corresponding hash values to form a global data package, and submit the global data package to the blockchain in the form of a blockchain transaction through a smart contract for dual-chain storage;

[0016] The intelligent layer uses the blockchain behavior monitoring component to perform real-time monitoring of blockchain transaction records. If any abnormal operation is detected in the blockchain transaction record, the relevant data shards are frozen. The remaining data shards are quickly reconstructed using erasure coding technology to generate repair shards, which are then verified through a trusted execution environment. The verified repair shards replace the frozen data shards, and the blockchain transaction record is updated.

[0017] At the application layer, the R&D tool chain and the intelligent R&D terminal are connected through a unified API gateway, and the data of the chain network layer and the intelligent R&D terminal are synchronized through a dual-write data synchronization adapter.

[0018] Preferably, the method for acquiring multi-source R&D data includes:

[0019] According to the actual layout of the R&D environment, n distributed edge nodes are planned, and a multi-protocol collection agent is deployed on each edge node. A standardized entry is provided for various R&D tool chains through a unified adaptation interface. The unified adaptation interface adopts a plug-in design. Each plug-in is independently developed and deployed for a type of R&D tool chain, and the data interface of various R&D tool chains is adapted by dynamically loading protocol parsing plug-ins. For different types of R&D tool chains, a mechanism combining scheduled pull and event-driven push is adopted to pull data, thereby obtaining multi-source R&D data.

[0020] Preferably, the method of embedding a data lineage tracking tag in metadata includes:

[0021] When collecting multi-source R&D data on various R&D tool chains through a multi-protocol collection agent, the initial source information of the multi-source R&D data is recorded, and an identification ID is assigned to each collected multi-source R&D data; the collected multi-source R&D data is encapsulated into metadata in a unified structured format, and a special field area for lineage tracking tags is reserved in the metadata structure; the initial source information and identification ID are embedded in the reserved special field area for lineage tracking tags, thereby obtaining a complete and tamper-proof lineage tracking tag.

[0022] Preferably, the method for filtering false data from metadata includes:

[0023] Install and start the trusted execution environment hardware module on the edge node, isolate the protected TEE memory area, and generate a root certificate and key pair. Inject the signature-verified AI verification model into the TEE memory area. The multi-protocol collection agent pushes metadata embedded with data lineage tracking tags to the TEE memory area through the TLS channel, performs denoising and standard deviation normalization, and obtains pre-processed metadata. Call the AI verification model to infer the pre-processed metadata and obtain a credibility score.

[0024] The obtained credibility score is compared with the preset credibility score threshold. If the obtained credibility score is lower than the preset credibility score threshold, the metadata corresponding to the credibility score is marked as false data and filtered and discarded in the TEE memory area; if the obtained credibility score is higher than or equal to the preset credibility score threshold, the metadata corresponding to the credibility score is encapsulated into a trusted data packet and digitally signed using a key pair; the signed trusted data packet consists of the filtered metadata.

[0025] Preferably, the method for dividing the filtered metadata into data fragments includes: performing data parsing on the filtered metadata and identifying the internal logical structure to obtain structured tags; using the structured tags as segmentation points, dividing the filtered metadata into multiple data fragments, and independently storing each generated data fragment in a distributed storage platform.

[0026] Preferably, the method for obtaining the global root hash includes:

[0027] After arranging each data shard in ascending order according to the size of the identification ID, a cryptographic hash function is executed to generate a set of ordered hash value sequences. This set of hash value sequences is used as leaf nodes, and a shard hash tree is recursively constructed. If the number of leaf nodes is odd, the hash value corresponding to the last leaf node in the hash value sequence is copied and filled at the end of the hash value sequence to make the number of leaf nodes even, thereby obtaining leaf node groups.

[0028] Starting from the first leaf node in the leaf node group, combine two adjacent leaf nodes in pairs; concatenate the hash values of the two adjacent leaf nodes, and execute the encrypted hash function again to calculate the hash value of the parent node of the two adjacent leaf nodes; repeat the above steps, building the shard hash tree layer by layer until only a unique hash value can be calculated. The hash value at this time is the global root hash.

[0029] Preferably, the method of submitting the global data package to the blockchain in the form of a blockchain transaction for dual-chain storage through a smart contract includes:

[0030] The global data package is encapsulated as transaction data as input for blockchain transactions, triggering the data verification process. After receiving the blockchain transaction request, the smart contract automatically parses and extracts the global data package in the transaction data, and verifies the global data package according to the preset rule set. If the verification fails, the blockchain transaction request is rejected and the incorrect data is rolled back. If the verification passes, the smart contract marks the transaction data as available for on-chain storage and submits the global data package to the blockchain for dual-chain storage. The preset rule set includes format verification, data integrity verification, lineage legitimacy verification, and authority subject verification.

[0031] Dual-chain storage adopts a dual-chain architecture of main chain and side chain. The smart contract divides the verified global data packets into high-value transaction data and high-frequency, low-sensitivity transaction data; the high-value transaction data is pushed to the main chain as global ownership confirmation and final evidence; the high-frequency, low-sensitivity transaction data is pushed to the side chain for access and business interaction; after the main chain and side chain complete the data writing, an on-chain receipt is automatically generated and synchronized to the R&D tool chain and the intelligent R&D terminal.

[0032] Preferably, the transaction records of the blockchain include data on-chain, metadata anchoring, access control permission changes, shard status changes and smart contract execution logs.

[0033] Preferably, the method for real-time detection of transaction records on the blockchain includes:

[0034] Construct an anomaly detection model, which includes an input layer, a hidden layer, and an output layer. The input layer is used to input historical blockchain transaction records. The output layer is used to output corresponding detection labels. Detection labels include abnormal operation labels and normal operation labels. The sigmoid function is used as the activation function. The anomaly detection model is a multi-layer perceptron (MLP) model. Binary cross entropy is used as the loss function to measure the error of the model prediction. The transaction records of the current blockchain are input into the trained anomaly detection model to obtain the current corresponding detection labels.

[0035] Preferably, the method for obtaining the repair fragment includes:

[0036] Assuming there are k remaining data shards, these k remaining data shards are encoded using erasure coding technology to obtain r verification shards. The r verification shards are used as repair shards and verified through a trusted execution environment. The verified repair shards replace the frozen data shards and are integrated with the remaining data shards to obtain valid data shards. The valid data shards are encoded and stored, and updated in the transaction records of the blockchain.

[0037] Valid data fragment D={d1,d2,...,d k}∪{p1,p2,...,p r},m=k+r;where,{d1,d2,...,d k} represents k remaining data shards; d k represents the kth remaining data fragment; {p1,p2,...,p r} represents r check fragments; p r represents the rth check shard; k represents the number of remaining data shards; r represents the number of check shards; m represents the total number of valid data shards;

[0038] Calculate the sharding behavior scoring factor of the remaining data shards and dynamically adjust the number of obtained verification shards; the sharding behavior scoring factor is ω i =α·fas(d i )+β·fm(d i )+γ·fru(d i ); among them, fas(d i ) represents the remaining data fragment d i The normalized value of the access frequency; fm(d i) represents the remaining data fragment d i Frequency of change; fru(d i ) represents the remaining data fragment d i Trust score; α represents the access frequency weight coefficient; β represents the change frequency weight coefficient; γ represents the trust score weight coefficient; i represents the index of the remaining data shards;

[0039] Perform weighted summation of the shard behavior scoring factors of all remaining data shards to obtain the behavioral risk index; based on the behavioral risk index, use the linear mapping function to dynamically adjust the number of verification shards Among them, R min Indicates the preset minimum number of check fragments; R max Indicates the preset maximum number of verification shards; Q represents the behavioral risk indicator; Indicates rounding up.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] This invention uses a multi-protocol collection agent to collect multi-source R&D data from various R&D tool chains, performs structured packaging to generate metadata, and embeds data lineage tracking tags. This design enables the system to be widely adapted to different R&D tool chains, achieving unified collection and standardized processing of multi-source data, laying the foundation for subsequent data traceability and collaborative management. A trusted execution environment is deployed at edge nodes to filter out false data from metadata, ensuring data authenticity and reliability, improving data quality, and reducing errors in subsequent processing and analysis.

[0042] Submit global data packages to the blockchain in the form of blockchain transactions for dual-chain storage. Using a dual-chain architecture of main and side chains, high-value transaction data is pushed to the main chain, while high-frequency, low-sensitivity transaction data is pushed to the side chain based on data value. This dual-chain storage method not only ensures global data ownership and final proof of ownership, but also meets the access and business interaction needs of different types of data, improving data availability and security.

[0043] The blockchain behavior monitoring component monitors blockchain transaction records in real time, uses anomaly detection models to identify abnormal operations, and freezes the relevant data shards. This real-time detection mechanism can promptly detect data anomalies and take appropriate measures to address them, ensuring data security and stability. Using erasure coding technology to reconstruct the remaining data shards allows for rapid data recovery, improving the system's fault tolerance and reliability, reducing losses caused by data loss, and ensuring the normal operation of the system even when problems occur in some data shards.

[0044] By calculating the behavioral risk indicators of the remaining data shards and dynamically adjusting the number of verification shards using a linear mapping function, the storage strategy for the verification shards is optimized, improving the resource utilization efficiency and performance of the system. The R&D tool chain and the intelligent R&D terminal are connected through a unified API gateway, and the data of the chain network layer and the intelligent R&D terminal are synchronized through a dual-write data synchronization adapter, achieving efficient interaction and data sharing between the various layers of the system, facilitating R&D personnel to work together on different terminals, and improving R&D efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a schematic diagram of the blockchain-based R&D data traceability and collaborative management system of the present invention;

[0046] Figure 2 This is a flowchart of the blockchain-based R&D data traceability and collaborative management method of the present invention;

[0047] Figure 3 This is a flow chart of the intelligent layer method provided by the present invention. DETAILED DESCRIPTION

[0048] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0049] Example 1

[0050] See also Figure 1 and Figure 3 As shown, this embodiment 1 further illustrates the blockchain-based R&D data traceability and collaborative management system proposed by the present invention, including:

[0051] With the continuous deepening of complex R&D activities in software engineering, chip design, intelligent manufacturing, biomedicine, and other fields, enterprises and research institutions generate massive amounts of R&D data throughout the project lifecycle. This data typically encompasses multi-dimensional and heterogeneous content, including code, models, configurations, version changes, design files, test results, simulation logs, and more. These data are strongly interconnected, characterized by causal dependencies and version inheritance.

[0052] In current applications, R&D data management often relies on traditional centralized storage or loosely coupled collaborative platforms. While these systems initially possessed certain management capabilities, they have exposed the following technical bottlenecks and issues in the context of large-scale collaboration, multi-tool integration, and interaction between heterogeneous systems:

[0053] The R&D data traceability capability is weak, making it difficult to build a trusted chain throughout the entire life cycle:

[0054] Existing technologies primarily rely on local logs, database records, or version control tools (such as Git) for data tracking. However, these mechanisms lack a unified, verifiable traceability path. Once data flows between different tool chains and nodes, its operation traces are easily tampered with, lost, or forged, making it difficult to support the responsibility definition, intellectual property protection, and audit requirements of key R&D results.

[0055] Lack of reliable trust mechanisms in decentralized collaboration scenarios:

[0056] Cross-organizational R&D collaboration is increasingly common, but existing systems often rely on centralized services or multi-system integration for data synchronization, making it difficult to establish a verifiable and tamper-proof data consensus mechanism. Once a node is tampered with or becomes inoperable, the data trust chain is broken, seriously threatening the continuity and security of R&D.

[0057] Rigid redundant storage strategies and low resource utilization efficiency:

[0058] To ensure data reliability, existing systems typically use static erasure coding for redundant backups. However, this approach fails to dynamically optimize data based on its importance, access behavior, or security risks. The result is often insufficient redundancy and poor recovery capabilities for critical data, while non-critical data occupies a significant amount of storage resources, resulting in low overall system resource utilization and recovery efficiency.

[0059] The recovery efficiency after data loss is low, making it difficult to support high-frequency R&D activities:

[0060] When faced with situations such as storage node damage, network interruption or malicious attack, the existing system recovery mechanism mostly relies on "full reconstruction" or "static fault tolerance strategy", lacking the ability to intelligently schedule recovery paths, data priorities, and available shards, resulting in long data recovery cycles and poor performance, making it unsuitable for high-speed iterative R&D scenarios.

[0061] The data repair integrity verification capability is weak, resulting in security blind spots:

[0062] Especially in a decentralized environment, when the system repairs data through redundant sharding, traditional systems lack a trusted verification mechanism to ensure the correctness, integrity, and consistency of the repaired data. This can lead to data tampering, contamination, or forgery during the repair process, leaving security vulnerabilities in R&D results.

[0063] Lack of behavioral perception mechanisms, making it impossible to intelligently assess risks and adjust strategies:

[0064] Existing systems generally lack the ability to monitor and analyze data access behaviors (such as access frequency, change frequency, abnormal operations, etc.) in real time, making it difficult to identify potential risks or data importance levels. They are unable to dynamically optimize redundant configurations or adjust policies based on behavioral intelligent scoring, hindering the improvement of the system's intelligence level.

[0065] Therefore, designing an R&D data management system that supports trusted traceability, intelligent redundancy, rapid recovery, secure verification, and risk awareness has become a critical issue urgently needed in both industry and research. Blockchain technology provides core mechanisms such as immutability, network-wide consensus, and smart contract-driven security. The integration of a trusted execution environment, erasure coding, and intelligent behavioral scoring models also offers a new technical path for efficient collaboration and secure management of R&D data.

[0066] It is in this context that the present invention is proposed. The present invention proposes a blockchain-based R&D data traceability and collaborative management system, including:

[0067] The perception layer collects multi-source R&D data from various R&D tool chains through multi-protocol collection agents, performs structured packaging to generate metadata, and embeds data lineage tracking tags in the metadata. By deploying a trusted execution environment, the metadata is filtered for false data.

[0068] The chain network layer divides the filtered metadata into data shards and stores them separately in the distributed storage platform; applies the hash algorithm to each data shard to generate the corresponding hash value; uses the hash values corresponding to all data shards as leaf nodes to build a shard hash tree, and then obtains the global root hash;

[0069] Use the global root hash as a summary, integrate all data shards and their corresponding hash values to form a global data package, and submit the global data package to the blockchain in the form of a blockchain transaction through a smart contract for dual-chain storage;

[0070] The intelligent layer uses the blockchain behavior monitoring component to perform real-time monitoring of blockchain transaction records. If any abnormal operation is detected in the blockchain transaction record, the relevant data shards are frozen. The remaining data shards are quickly reconstructed using erasure coding technology to generate repair shards, which are then verified through a trusted execution environment. The verified repair shards replace the frozen data shards, and the blockchain transaction record is updated.

[0071] At the application layer, the R&D tool chain and the intelligent R&D terminal are connected through a unified API gateway, and the data of the chain network layer and the intelligent R&D terminal are synchronized through a dual-write data synchronization adapter.

[0072] Methods for obtaining multi-source R&D data include:

[0073] Based on the actual layout of the R&D environment, n distributed edge nodes are planned, and multi-protocol collection agents are deployed on each edge node. A unified adapter interface provides a standardized entry point for various R&D tool chains. The unified adapter interface adopts a plug-in design. Each plug-in is independently developed and deployed for a type of R&D tool chain (such as code version management, CI / CD platform, defect tracking, or project management system). The data interface of various R&D tool chains is adapted by dynamically loading protocol parsing plug-ins. For different types of R&D tool chains, a combination of scheduled pull and event-driven push is adopted to pull data, thereby obtaining multi-source R&D data.

[0074] Multi-source R&D data includes code development data, project management data, build and deployment data, quality testing data, collaboration data, and operations and maintenance data; code development data includes code submission records (change content, submitter, timestamp), branch and merge information (branch name, merge request), and code differences (changed file list, additions and deletions); project management data includes project task assignments, progress updates, and requirements documents; build and deployment data includes build logs (compilation output, dependency installation status, error messages) and pipeline status (stage execution results: build, test, deployment; time statistics); quality testing data includes test reports, code coverage, and defect details; collaboration data includes shared documents and review records (review comments, modification suggestions, and resolution results); operations and maintenance data includes run logs, user feedback, and security scan results (vulnerability type, repair suggestions, and scan time);

[0075] Collecting and integrating multi-source R&D data can bring significant value to enterprises in R&D management, quality assurance, collaborative optimization, and risk control.

[0076] Improve R&D efficiency and product quality: Through refined multi-source R&D data monitoring and traceability, problems can be quickly located and resolved, thereby reducing downtime and rework time, improving R&D efficiency and product stability;

[0077] Enhanced transparency and accountability: The collection of multi-source R&D data makes the entire R&D process transparent, clarifies relevant responsibilities, and facilitates daily management and performance assessment.

[0078] Support for intelligent applications and automated operations: Leveraging artificial intelligence and big data technologies, it mines, predicts, and optimizes multi-source R&D data, providing R&D teams with precise decision-making support and risk warnings. Combined with real-time feedback, it can automatically trigger processes such as build, testing, deployment, and exception recovery, achieving automated and intelligent R&D operations.

[0079] Reduce management costs and risks: Multi-source R&D data, after being structured and encapsulated, can be seamlessly connected between different systems and teams, reducing information transmission errors and communication barriers. Full data recording makes the discovery and handling of abnormal behavior faster, reducing the systemic risks caused by incomplete or opaque information.

[0080] Scheduled pull: For R&D tool chains that support active queries (such as code hosting platforms and build servers), configure task scheduling within the multi-protocol acquisition agent to call the API interface at preset time intervals (for example, every minute or every 5 minutes) to pull the latest data;

[0081] Event-driven push: For R&D tool chains that support event-triggered events (such as code submission, review, and fault status changes), data is proactively pushed to the collection agent when an event occurs by registering a webhook or subscribing to a push service.

[0082] For example:

[0083] Git toolchain: monitor code commit events through Git Hook or Git API;

[0084] CI / CD tools (such as Jenkins): parse the build logs pushed by webhooks;

[0085] Project management tools (such as Jira): call the REST API to synchronize task status changes;

[0086] IoT device development: Subscribe to embedded device debugging data streams via the MQTT protocol.

[0087] Methods for embedding data lineage tracking markers in metadata include:

[0088] When multi-source R&D data is collected on various R&D tool chains through a multi-protocol collection agent, the initial source information of the multi-source R&D data is recorded, and an identification ID is assigned to each collected multi-source R&D data; the initial source information includes the data source identifier, collection timestamp, operation type, and operator; the collected multi-source R&D data is encapsulated into metadata in a unified structured format, and a dedicated field area for lineage tracking tags is reserved in the metadata structure; the initial source information and identification ID are embedded in the reserved dedicated field area for lineage tracking tags, thereby obtaining a complete and tamper-proof lineage tracking tag, providing a solid basis for subsequent R&D data verification, traceability, and auditing;

[0089] For example, use JSON or XML format to define fixed fields, and embed the initial source information and identification ID of multi-source R&D data into the dedicated field area for lineage tracking tags in the metadata structure:

[0090] sourceID: data source identifier (data source number or name); timestamp: collection timestamp (data collection time); uniqueID: identification ID; operationType: operation type (such as submission, construction, test, etc.); operator: operator.

[0091] Methods for filtering metadata for false data include:

[0092] Install and start the trusted execution environment hardware module on the edge node, isolate the protected TEE memory area, and generate a root certificate and key pair for subsequent remote attestation and communication encryption. Inject the signature-verified AI verification model into the TEE memory area. The multi-protocol collection agent pushes metadata embedded with data lineage tracking tags to the TEE memory area through the TLS channel, performs denoising and standard deviation normalization, and obtains pre-processed metadata. Call the AI verification model to infer the pre-processed metadata and obtain a credibility score.

[0093] The obtained credibility score is compared with the preset credibility score threshold. If the obtained credibility score is lower than the preset credibility score threshold, the metadata corresponding to the credibility score is marked as false data and filtered and discarded in the TEE memory area; if the obtained credibility score is higher than or equal to the preset credibility score threshold, the metadata corresponding to the credibility score is encapsulated into a trusted data packet and digitally signed using a key pair; the signed trusted data packet consists of the filtered metadata.

[0094] The method for segmenting filtered metadata into data shards includes: parsing the filtered metadata and identifying its internal logical structure to obtain structured tags; using the structured tags as segmentation points, segmenting the filtered metadata into multiple data shards, and independently storing each generated data shard in a distributed storage platform. For example, a document can be segmented into multiple data shards by chapter; JSON data can be segmented into multiple data shards by record object. If each data shard contains multiple sub-objects, each data shard is further segmented using the sub-objects as segmentation points.

[0095] The methods for obtaining the global root hash include:

[0096] After arranging each data shard in ascending order according to the size of the identification ID, a cryptographic hash function is executed to generate a set of ordered hash value sequences. This set of hash value sequences is used as leaf nodes, and a shard hash tree is recursively constructed. If the number of leaf nodes is odd, the hash value corresponding to the last leaf node in the hash value sequence is copied and filled at the end of the hash value sequence to make the number of leaf nodes even, thereby obtaining leaf node groups.

[0097] For example, there is an ordered set of hash value sequences consisting of leaf nodes {H1,H2,...,H5,}. The number of leaf nodes is odd. The hash value corresponding to the last leaf node needs to be copied and filled at the end of this set of hash value sequences: {H1,H2,...,H5,} → {H1,H2,...,H5,H5}.

[0098] Starting from the first leaf node in the leaf node group, combine two adjacent leaf nodes in pairs; concatenate the hash values of the two adjacent leaf nodes, and execute the encrypted hash function again to calculate the hash value of the parent node of the two adjacent leaf nodes; repeat the above steps, building the shard hash tree layer by layer until only a unique hash value can be calculated. The hash value at this time is the global root hash, which is the summary result of the entire shard hash tree.

[0099] Methods for submitting global data packages to the blockchain for dual-chain storage in the form of blockchain transactions through smart contracts include:

[0100] The global data package is encapsulated as transaction data as input for blockchain transactions, triggering the data verification process. After receiving the blockchain transaction request, the smart contract automatically parses and extracts the global data package in the transaction data, and verifies the global data package according to the preset rule set. If the verification fails, the blockchain transaction request is rejected and the incorrect data is rolled back. If the verification passes, the smart contract marks the transaction data as available for on-chain storage and submits the global data package to the blockchain for dual-chain storage. The preset rule set includes format verification, data integrity verification, lineage legitimacy verification, and authority subject verification.

[0101] Data integrity verification verifies whether the metadata structure conforms to the standard model defined by the platform, including field integrity, required fields, and structural legitimacy.

[0102] Lineage legitimacy verification means calling the on-chain lineage mapping table to verify whether the data is correctly inherited from existing data objects (such as the previous task data in a certain R&D process) to ensure the logical closure of the data evolution chain;

[0103] Permission subject verification means verifying whether the operating subject who submits the data has write permission and complies with the bound organizational identity policy (such as local regulations, data level classification, etc.).

[0104] Dual-chain storage adopts a main-chain and side-chain dual-chain architecture. The smart contract divides the verified global data packets into high-value transaction data and high-frequency, low-sensitivity transaction data; the high-value transaction data is pushed to the main chain as global right confirmation and final evidence; high-value transaction data involves ownership, approval status, and core asset changes, including global root hash, core metadata anchoring, smart contract key logs, identity credentials and permissions, approval records, and ownership change records; the main chain is deployed in the form of a consortium chain, and a BFT-type consensus algorithm is used to ensure immutability and finality; high-frequency, low-sensitivity transaction data is pushed to the side chain for access and business interaction; high-frequency, low-sensitivity transaction data does not involve core assets and has a lower risk of leakage. It is used for secondary scenarios such as behavioral analysis and audit tracking, including user operation logs, version change flow, system monitoring indicators, off-chain event indexes, and audit trails; after the main chain and side chain complete data writing, an on-chain receipt is automatically generated and synchronized to the R&D tool chain and smart R&D terminal.

[0105] The transaction records of the blockchain include data on-chain, metadata anchoring, access control permission changes, shard status changes and smart contract execution logs.

[0106] Data on-chain refers to the process of submitting and storing the global data package verified by the smart contract to the blockchain, ensuring the integrity and immutability of the global data package, and all subsequent versions, evolutions and references are traceable;

[0107] Metadata anchoring is the process of associating filtered metadata with specific transactions or blocks on the blockchain, facilitating the rapid location and retrieval of relevant data on the chain. It provides complete metadata structure support for the business layer and anchors metadata with its upstream and downstream relationships and version history, facilitating traceability, comparison, and rollback.

[0108] Access control permission changes refer to recording the access authorization or permission change operations of users or entities on the blockchain, such as adding read and write permissions, revoking access rights, and switching roles. This ensures that every access or operation behavior is auditable and tamper-proof, and cooperates with smart contracts to achieve dynamic access control.

[0109] Shard state changes refer to updates to the state of a data shard, such as operations like creating, moving, copying, or deleting a data shard. Recording these changes on the blockchain helps monitor data integrity and availability, allowing for rapid responses to suspicious data behavior. Erasure codes and a trusted execution environment enable automated repair and rollback.

[0110] Contract execution log transactions record the call, execution status, and execution results of smart contracts, and are system log transactions on the chain. They provide transparent, tamper-proof execution records for subsequent security audits, system debugging, and accountability.

[0111] Methods for real-time detection of blockchain transaction records include:

[0112] Construct an anomaly detection model, which includes an input layer, a hidden layer, and an output layer. The input layer is used to input historical blockchain transaction records. The output layer is used to output corresponding detection labels. Detection labels include abnormal operation labels and normal operation labels. The sigmoid function is used as the activation function. The anomaly detection model is a multi-layer perceptron (MLP) model. Binary cross entropy is used as the loss function to measure the error of the model prediction. The transaction records of the current blockchain are input into the trained anomaly detection model to obtain the current corresponding detection labels.

[0113] Methods for obtaining repaired shards include:

[0114] Assuming there are k remaining data shards, these k remaining data shards are encoded using erasure coding technology to obtain r verification shards. The r verification shards are used as repair shards and verified through a trusted execution environment. The verified repair shards replace the frozen data shards and are integrated with the remaining data shards to obtain valid data shards. The valid data shards are encoded and stored, and updated in the transaction records of the blockchain.

[0115] Valid data fragment D={d1,d2,...,d k}∪{p1,p2,...,p r},m=k+r;where,{d1,d2,...,d k} represents k remaining data shards; d k represents the kth remaining data fragment; {p1,p2,...,p r} represents r check fragments; p r represents the rth check shard; k represents the number of remaining data shards; r represents the number of check shards; m represents the total number of valid data shards;

[0116] Calculate the sharding behavior scoring factor of the remaining data shards and dynamically adjust the number of obtained verification shards; the sharding behavior scoring factor is ω i =α·fas(d i )+β·fm(d i )+γ·fru(d i ); among them, fas(d i ) represents the remaining data fragment d i The normalized value of the access frequency; fm(d i ) represents the remaining data fragment d i Frequency of change; fru(d i ) represents the remaining data fragment d iTrust score; α represents the access frequency weight coefficient; β represents the change frequency weight coefficient; γ represents the trust score weight coefficient; i represents the index of the remaining data shards;

[0117] According to expert experience, the values of α, β, and γ range from 0 to 1, and α + β + γ = 1. The number of accesses to the remaining data shards within a certain period of time is divided by the maximum number of accesses to all data shards to obtain the normalized access frequency value of the remaining data shards. The number of modifications to the remaining data shards within a certain period of time is divided by the maximum number of changes to all data shards to obtain the change frequency of the remaining data shards. The total number of accesses and successful accesses to the remaining data shards in the historical records is collected, and the number of successful accesses is divided by the total number of accesses to obtain the trust score of the remaining data shards.

[0118] Perform weighted summation of the shard behavior scoring factors of all remaining data shards to obtain the behavioral risk index; based on the behavioral risk index, use the linear mapping function to dynamically adjust the number of verification shards Among them, R min Indicates the preset minimum number of check fragments; R max Indicates the preset maximum number of verification shards; Q represents the behavioral risk indicator; Indicates rounding up;

[0119] The present invention solves the following major problems existing in the prior art:

[0120] Deficiencies in the parity shard storage strategy: The number of parity shards in traditional erasure coding technologies is usually fixed and cannot be dynamically adjusted based on actual conditions (such as data importance and risk). This static parity shard configuration may lead to resource waste or insufficient data recovery capabilities.

[0121] Slow recovery from data loss and corruption: Existing distributed storage systems may rely on complete shard reconstruction when data loss or corruption occurs. This reconstruction process is time-consuming and inefficient, especially when a large number of shards are lost.

[0122] Lack of intelligent behavior monitoring and risk assessment: The lack of quantitative analysis of the dynamic behavior of remaining data shards (such as access patterns, modification frequency, and historical credibility) makes it impossible to distinguish between high-trust and low-trust data shards, leading to rigid resource allocation and a lack of trust mechanisms.

[0123] The problem of difficulty in verifying the integrity of data after repair. During the data repair process, it is difficult to ensure that the repaired data is accurate and not tampered with. Especially in a decentralized environment, the lack of effective means to verify the repair process may lead to inconsistencies or errors in the repaired data.

[0124] The innovative points of this program are:

[0125] Dynamic erasure code configuration driven by intelligent behavioral scoring. A behavioral scoring factor is constructed based on the access frequency, change frequency, and trust score of each remaining data shard, dynamically adjusting the number of parity shards. This intelligent adjustment dynamically optimizes parity shard storage, saving storage resources and improving data recovery efficiency.

[0126] Erasure coding is used to rapidly reconstruct the remaining data shards, generating repair shards. This process effectively utilizes the parity shards to recover lost data and reduces reconstruction time. The flexibility and efficiency of erasure coding overcomes the slow recovery issues inherent in traditional reconstruction processes.

[0127] Use the trusted execution environment to verify the repair shards to ensure the correctness and integrity of the repair data. Through the trusted execution environment, data verification can be performed in a trusted execution environment to prevent data tampering during the repair process, provide strong data integrity protection, and ensure the security of the repair shards in a decentralized environment;

[0128] Adjust the checksum shard configuration based on the behavior risk index. The higher the behavior risk index, the more checksum shards will be generated to enhance recoverability; the lower the behavior risk index, the fewer checksum shards will be generated to save storage space and resources;

[0129] The beneficial effects relative to the prior art are:

[0130] Dynamically adjust the parity sharding and efficient erasure code reconstruction mechanism to ensure that data can be quickly repaired when reconstruction is needed, improving the system's data recovery capabilities. Especially when dealing with large-scale data reconstruction issues, data recovery can be performed quickly and accurately without blindly increasing the amount of parity sharding data.

[0131] The number of parity shards is increased or decreased based on the level of behavioral risk indicators, reducing storage costs while improving storage efficiency. Compared with the fixed parity shard configuration in traditional strategies, this dynamic adjustment method is more flexible and cost-effective.

[0132] The repair shards are verified through a trusted execution environment to ensure the integrity and security of the repair data and prevent data tampering during the repair process. The isolation and security provided by the trusted execution environment ensure that the repair data is not maliciously tampered with, which is a guarantee that traditional storage cannot provide.

[0133] The preset credibility score thresholds are set by staff based on historical data analysis. This historical analysis involves the system collecting credibility scores from multiple data points and calculating their averages for reference. Similarly, the preset minimum and maximum validation shard numbers are also set by staff based on historical system operation data and specific application scenario requirements. These preset thresholds can be adjusted by staff during system operation based on actual conditions.

[0134] In this embodiment, a multi-protocol collection agent is used to collect multi-source R&D data from various R&D tool chains, and metadata is generated through structured packaging, while also embedding data lineage tracking tags. This design enables the system to be widely adapted to different R&D tool chains, achieving unified collection and standardized processing of multi-source data, laying the foundation for subsequent data traceability and collaborative management. A trusted execution environment is deployed at edge nodes to filter out false data from metadata, ensuring data authenticity and reliability, improving data quality, and reducing errors in subsequent processing and analysis.

[0135] Submit global data packages to the blockchain in the form of blockchain transactions for dual-chain storage. Using a dual-chain architecture of main and side chains, high-value transaction data is pushed to the main chain, while high-frequency, low-sensitivity transaction data is pushed to the side chain based on data value. This dual-chain storage method not only ensures global data ownership and final proof of ownership, but also meets the access and business interaction needs of different types of data, improving data availability and security.

[0136] The blockchain behavior monitoring component monitors blockchain transaction records in real time, uses anomaly detection models to identify abnormal operations, and freezes the relevant data shards. This real-time detection mechanism can promptly detect data anomalies and take appropriate measures to address them, ensuring data security and stability. Using erasure coding technology to reconstruct the remaining data shards allows for rapid data recovery, improving the system's fault tolerance and reliability, reducing losses caused by data loss, and ensuring the normal operation of the system even when problems occur in some data shards.

[0137] By calculating the behavioral risk indicators of the remaining data shards and dynamically adjusting the number of verification shards using a linear mapping function, the storage strategy for the verification shards is optimized, improving the resource utilization efficiency and performance of the system. The R&D tool chain and the intelligent R&D terminal are connected through a unified API gateway, and the data of the chain network layer and the intelligent R&D terminal are synchronized through a dual-write data synchronization adapter, achieving efficient interaction and data sharing between the various layers of the system, facilitating R&D personnel to work together on different terminals, and improving R&D efficiency.

[0138] Example 2

[0139] See also Figure 2As shown, for the parts not described in detail in this embodiment, please refer to the description of Example 1, which provides a blockchain-based R&D data traceability and collaborative management method, including:

[0140] S1. Collect multi-source R&D data from various R&D tool chains through multi-protocol collection agents, perform structured packaging to generate metadata, and embed data lineage tracking tags in the metadata. Filter out false data from the metadata by deploying a trusted execution environment.

[0141] S2. Split the filtered metadata into data shards and store them separately in the distributed storage platform; apply a hash algorithm to each data shard to generate a corresponding hash value; construct a shard hash tree with the hash values corresponding to all data shards as leaf nodes, and then obtain the global root hash;

[0142] Use the global root hash as a summary, integrate all data shards and their corresponding hash values to form a global data package, and submit the global data package to the blockchain in the form of a blockchain transaction through a smart contract for dual-chain storage;

[0143] S3. Use the blockchain behavior monitoring component to perform real-time monitoring of blockchain transaction records. If any abnormal operation is detected in the blockchain transaction record, the relevant data shards are frozen. Erasure coding technology is used to quickly reconstruct the remaining data shards to generate repair shards, which are verified in a trusted execution environment. The verified repair shards replace the frozen data shards, and the blockchain transaction record is updated.

[0144] S4. Connect the R&D tool chain and the intelligent R&D terminal through a unified API gateway, and synchronize the data of the chain network layer and the intelligent R&D terminal through a dual-write data synchronization adapter.

[0145] Since the electronic device introduced in this embodiment is an electronic device used to implement the blockchain-based R&D data traceability and collaborative management system in the embodiment of this application, based on the blockchain-based R&D data traceability and collaborative management system introduced in the embodiment of this application, those skilled in the art can understand the specific implementation of the electronic device of this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application will not be described in detail here. As long as those skilled in the art implement the electronic device used in the blockchain-based R&D data traceability and collaborative management system in the embodiment of this application, they all fall within the scope of protection of this application.

[0146] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0147] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the concept of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. The blockchain-based R&D data traceability and collaborative management system is characterized by: include: The perception layer collects multi-source R&D data from various R&D tool chains through multi-protocol collection agents, performs structured packaging to generate metadata, and embeds data lineage tracking tags in the metadata; By deploying a trusted execution environment, false data can be filtered from metadata; The chain network layer divides the filtered metadata into data shards and stores them separately in the distributed storage platform; Apply a hash algorithm to each data shard to generate a corresponding hash value; Use the hash values corresponding to all data shards as leaf nodes to build a shard hash tree and obtain the global root hash; Use the global root hash as a summary, integrate all data shards and their corresponding hash values to form a global data package, and submit the global data package to the blockchain in the form of a blockchain transaction through a smart contract for dual-chain storage; The intelligent layer uses the blockchain behavior monitoring component to perform real-time detection of blockchain transaction records. If any abnormal operation is detected in the blockchain transaction records, the relevant data shards will be frozen. Use erasure coding technology to quickly reconstruct the remaining data shards, generate repair shards, and verify them through a trusted execution environment. Replace the frozen data shards with the verified repair shards, and update the transaction records on the blockchain. At the application layer, the R&D tool chain and the intelligent R&D terminal are connected through a unified API gateway, and the data of the chain network layer and the intelligent R&D terminal are synchronized through a dual-write data synchronization adapter.

2. The blockchain-based R&D data traceability and collaborative management system according to claim 1 is characterized in that: The method for acquiring multi-source R&D data includes: According to the actual layout of the R&D environment, n distributed edge nodes are planned, and a multi-protocol collection agent is deployed on each edge node. A standardized entry is provided for various R&D tool chains through a unified adaptation interface. The unified adaptation interface adopts a plug-in design. Each plug-in is independently developed and deployed for a type of R&D tool chain, and the data interface of various R&D tool chains is adapted by dynamically loading protocol parsing plug-ins. For different types of R&D tool chains, a mechanism combining scheduled pull and event-driven push is adopted to pull data, thereby obtaining multi-source R&D data.

3. The blockchain-based R&D data traceability and collaborative management system according to claim 2 is characterized in that: The method of embedding a data lineage tracking mark in metadata includes: When collecting multi-source R&D data on various R&D tool chains through a multi-protocol collection agent, the initial source information of the multi-source R&D data is recorded, and an identification ID is assigned to each collected multi-source R&D data; the collected multi-source R&D data is encapsulated into metadata in a unified structured format, and a special field area for lineage tracking tags is reserved in the metadata structure; the initial source information and identification ID are embedded in the reserved special field area for lineage tracking tags, thereby obtaining a complete and tamper-proof lineage tracking tag.

4. The blockchain-based R&D data traceability and collaborative management system according to claim 3 is characterized in that: The method for filtering false data from metadata includes: Install and start the trusted execution environment hardware module on the edge node, isolate the protected TEE memory area, and generate a root certificate and key pair. Inject the signature-verified AI verification model into the TEE memory area. The multi-protocol collection agent pushes metadata embedded with data lineage tracking tags to the TEE memory area through the TLS channel, performs denoising and standard deviation normalization, and obtains pre-processed metadata. Call the AI verification model to infer the pre-processed metadata and obtain a credibility score. The obtained credibility score is compared with the preset credibility score threshold. If the obtained credibility score is lower than the preset credibility score threshold, the metadata corresponding to the credibility score is marked as false data and filtered and discarded in the TEE memory area; if the obtained credibility score is higher than or equal to the preset credibility score threshold, the metadata corresponding to the credibility score is encapsulated into a trusted data packet and digitally signed using a key pair; the signed trusted data packet consists of the filtered metadata.

5. The blockchain-based R&D data traceability and collaborative management system according to claim 4 is characterized in that: The method for dividing the filtered metadata into data fragments includes: performing data parsing on the filtered metadata and identifying the internal logical structure to obtain structured tags; using the structured tags as segmentation points, dividing the filtered metadata into multiple data fragments, and independently storing each generated data fragment in a distributed storage platform.

6. The blockchain-based R&D data traceability and collaborative management system according to claim 5 is characterized in that: The method for obtaining the global root hash includes: After arranging each data shard in ascending order according to the size of the identification ID, a cryptographic hash function is executed to generate a set of ordered hash value sequences. This set of hash value sequences is used as leaf nodes, and a shard hash tree is recursively constructed. If the number of leaf nodes is odd, the hash value corresponding to the last leaf node in the hash value sequence is copied and filled at the end of the hash value sequence to make the number of leaf nodes even, thereby obtaining leaf node groups. Starting from the first leaf node in the leaf node group, combine two adjacent leaf nodes in pairs; concatenate the hash values of the two adjacent leaf nodes, and execute the encrypted hash function again to calculate the hash value of the parent node of the two adjacent leaf nodes; repeat the above steps, building the shard hash tree layer by layer until only a unique hash value can be calculated. The hash value at this time is the global root hash.

7. The blockchain-based R&D data traceability and collaborative management system according to claim 6 is characterized in that: The method of submitting a global data package to a blockchain in the form of a blockchain transaction for dual-chain storage through a smart contract includes: The global data package is encapsulated as transaction data as input for blockchain transactions, triggering the data verification process. After receiving the blockchain transaction request, the smart contract automatically parses and extracts the global data package in the transaction data, and verifies the global data package according to the preset rule set. If the verification fails, the blockchain transaction request is rejected and the incorrect data is rolled back. If the verification passes, the smart contract marks the transaction data as available for on-chain storage and submits the global data package to the blockchain for dual-chain storage. The preset rule set includes format verification, data integrity verification, lineage legitimacy verification, and authority subject verification. Dual-chain storage adopts a dual-chain architecture of main chain and side chain. The smart contract divides the verified global data packets into high-value transaction data and high-frequency, low-sensitivity transaction data; the high-value transaction data is pushed to the main chain as global ownership confirmation and final evidence; the high-frequency, low-sensitivity transaction data is pushed to the side chain for access and business interaction; after the main chain and side chain complete the data writing, an on-chain receipt is automatically generated and synchronized to the R&D tool chain and the intelligent R&D terminal.

8. The blockchain-based R&D data traceability and collaborative management system according to claim 7 is characterized in that: The transaction records of the blockchain include data on-chain, metadata anchoring, access control permission changes, shard status changes and smart contract execution logs.

9. The blockchain-based R&D data traceability and collaborative management system according to claim 8 is characterized in that: The method for real-time detection of transaction records on the blockchain includes: Construct an anomaly detection model, which includes an input layer, a hidden layer, and an output layer. The input layer is used to input historical blockchain transaction records. The output layer is used to output corresponding detection labels. Detection labels include abnormal operation labels and normal operation labels. The sigmoid function is used as the activation function. The anomaly detection model is a multi-layer perceptron (MLP) model. Binary cross entropy is used as the loss function to measure the error of the model prediction. The transaction records of the current blockchain are input into the trained anomaly detection model to obtain the current corresponding detection labels.

10. The blockchain-based R&D data traceability and collaborative management system according to claim 9 is characterized in that: The method for obtaining the repair fragment includes: Assuming there are k remaining data shards, these k remaining data shards are encoded using erasure coding technology to obtain r verification shards. The r verification shards are used as repair shards and verified through a trusted execution environment. The verified repair shards replace the frozen data shards and are integrated with the remaining data shards to obtain valid data shards. The valid data shards are encoded and stored, and updated in the transaction records of the blockchain. Valid data fragment D={d1,d2,...,d k }∪{p1,p2,...,p r },m=k+r;where,{d1,d2,...,d k } represents k remaining data shards; d k represents the kth remaining data fragment; {p1,p2,...,p r } represents r check fragments; p r represents the rth check shard; k represents the number of remaining data shards; r represents the number of check shards; m represents the total number of valid data shards; Calculate the sharding behavior scoring factor of the remaining data shards and dynamically adjust the number of obtained verification shards; the sharding behavior scoring factor is ω i =α·fas(d i )+β·fm(d i )+γ·fru(d i ); among them, fas(d i ) represents the remaining data fragment d i The normalized value of the access frequency; fm(d i ) represents the remaining data fragment d i Frequency of change; fru(d i ) represents the remaining data fragment d i Trust score; α represents the access frequency weight coefficient; β represents the change frequency weight coefficient; γ represents the trust score weight coefficient; i represents the index of the remaining data shards; Perform weighted summation of the shard behavior scoring factors of all remaining data shards to obtain the behavioral risk index; based on the behavioral risk index, use the linear mapping function to dynamically adjust the number of verification shards Among them, R min Indicates the preset minimum number of check fragments; R max Indicates the preset maximum number of verification shards; Q represents the behavioral risk indicator; Indicates rounding up.

Images