Railway engineering metadata updating and publishing method and device

By introducing role-based access control, hash routing, and natural language processing models, combined with a railway professional rule base and encrypted transmission, the problems of uncontrollable data sources and low audit efficiency in the railway engineering metadata management system have been solved, achieving efficient and secure metadata publishing.

CN121387335APending Publication Date: 2026-01-23INST OF COMPUTING TECH CHINA ACAD OF RAILWAY SCI +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511650781.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-11-12
Publication Date
2026-01-23

AI Technical Summary

Technical Problem

The existing railway engineering metadata management system suffers from problems such as uncontrollable data sources, difficulty in labeling cross-professional differences, low review efficiency, high data conflict rate, and insufficient security during the update and release process, especially in the case of concurrent processing of TB-level metadata and distributed environment.

Method used

A role-based access control mechanism is adopted, combined with hash routing, natural language processing models, and railway professional rule bases to achieve fine-grained management and intelligent review of change requests; through node-level version management and package building strategies, combined with encrypted transmission and canary release, data security and stability are ensured.

Benefits of technology

It enables precise control and intelligent review of railway engineering metadata, improves the reliability and review efficiency of change requests, reduces data conflict rate, and ensures the security of releases and system stability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121387335A_ABST
    Figure CN121387335A_ABST
Patent Text Reader

Abstract

The invention relates to the field of railway engineering, and discloses a railway engineering metadata updating and publishing method and device. Comprising the following steps: performing authority management based on a role access control model; node fragmentation is carried out according to projects, lines and professions, modifiers are distributed, and a metadata element relation tree and content are modified and submitted; an auditor audits the modification opinions, judges conflicts in combination with a semantic analysis method and a weight rule base and outputs an auditing result; a version number is given to the metadata unit passing the auditing, and the metadata unit is packaged to form a metadata packet; overall or staged release is carried out on the metadata packet, and gray release and A / B testing are supported; a digital signature and encryption mechanism is adopted in the publishing process, so that the integrity and security of the metadata in the transmission and storage process are guaranteed; therefore, the metadata updating and publishing efficiency is improved, the multi-specialty cooperation capability is enhanced, and the data security and credibility are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of railway engineering, and specifically to a method and apparatus for updating and publishing railway engineering metadata. Background Technology

[0002] With the continuous improvement of information technology in railway engineering, railway engineering metadata, as a crucial foundation for digital delivery and operation throughout the entire project lifecycle, is being widely applied in design, construction, and operation phases. Existing metadata management largely relies on centralized databases or distributed storage platforms, typically employing manual import, form editing, or synchronization via external system interfaces for updates. However, in actual engineering projects, railway metadata is characterized by its large scale, diverse professional categories, and complex business relationships, highlighting the shortcomings of existing technologies in updating and publishing.

[0003] Existing systems lack granular control over the source of change requests, and common permission mechanisms are insufficient to meet the management needs of multi-role and multi-unit collaboration. For example, the lack of unified permission constraints on modification opinions submitted by different design institutes, construction units, or review agencies leads to uncontrollable data sources, unauthorized operations, and unclear responsibilities. Railway engineering metadata models are mostly built based on the IFD standard, involving multiple disciplines such as lines, stations, and electrical systems, resulting in complex formats and massive volumes. Existing methods often only perform field-level or form-level difference comparisons, making it difficult to achieve accurate difference labeling and modification traceability across disciplines and versions. This leads to low review efficiency and high data conflict rates. Existing opinion merging and review processes usually rely on manual processing, lacking semantic-level intelligent aggregation of multi-source modification opinions at the same node, easily resulting in duplicate submissions and difficulty in resolving conflicting opinions, thereby prolonging the review cycle and affecting cross-departmental collaboration efficiency.

[0004] In the version control and release phase, most existing technologies adopt a whole-body replacement approach, lacking node-level version management and package building strategies, resulting in excessively large package sizes and inconvenient rollback. Meanwhile, the pre-release verification phase often remains at the level of basic format checks, lacking systematic verification of the integrity of key fields, the rationality of parameters, and cross-version discrepancies, making it easy to push erroneous data to downstream systems. Regarding release security, existing systems mostly use generic protocols in the transmission phase, failing to establish end-to-end security mechanisms for railway engineering data, lacking field-level encryption, signature verification, and canary release strategies, posing risks of data leakage and system instability. These shortcomings are particularly prominent in the concurrent processing and distributed environment of TB-level metadata, hindering the efficient and secure release of railway engineering metadata. Therefore, this paper proposes a method and apparatus for updating and releasing railway engineering metadata. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and apparatus for updating and publishing railway engineering metadata, thereby resolving the technical problems existing in the prior art.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: A method and apparatus for updating and publishing railway engineering metadata, comprising the following steps: S1: Obtain a change request for the target metadata node, wherein the target metadata node is constructed into a hierarchical management unit according to project, line and profession, and a joint sharding key is formed by the corresponding project identifier, line identifier and profession code; The change request consists of change items; the change items are generated based on the source pre-configured by the role's access control policy and written into the review queue as a source record of multiple source modification opinions within the same review cycle; S2: Perform hash routing on the target metadata node based on the combined sharding key, and route it to the corresponding physical shard; S3: Within the target segment located in S2, each change item in the review queue of S1 is compared with the historical version and a difference label is generated to form a set of changes to be reviewed with source identifier and associated with historical version records; S4: Extract multi-source modification opinions for the target metadata node within the same review period from the review queue, calculate semantic similarity based on a natural language processing model to merge duplicate opinions, and make decisions on potentially conflicting opinions based on a preset weight set to obtain a review conclusion; S5: Perform machine-readable verification on the data corresponding to the audit conclusion based on the railway professional rule base; if the data does not comply with the rules, stop the current process and return a verification failure message; if the data complies with the rules, form an audit pass set. S6: Generate metadata unit versions of the approved set at the node granularity; then, according to the predetermined package building strategy, under the conditions of satisfying the integrity of dependency relationships, reference relationships and release granularity constraints, package multiple approved metadata units into a metadata data package and assign a package version number; S7: Perform key item checks on the metadata data packet according to the pre-release checklist; if any check fails, release is aborted, otherwise a release packet is formed. S8: Calculate the digest of the packet to be released using a hash algorithm, and digitally sign it using an asymmetric encryption algorithm; write the signature, packet version number and algorithm identifier into the release version repository; S9: Distribute signed metadata packets to downstream systems through an encrypted communication channel based on a transport layer security protocol.

[0007] Preferably, the pre-configured source is configured based on role-based access control, and only the design unit, construction unit or review unit corresponding to the role is authorized to submit change items through the system interface; The roles include cataloger, reviewer, and administrator; The bibliographer only has metadata entry and submission for review permissions; The auditors have the authority to view metadata and fill in audit comments. The administrator has all permissions and can intervene in abnormal processes; The changes are structured data that conforms to the railway IFD standard and the metadata form schema specification.

[0008] Preferably, the semantic similarity determination is performed by a pre-trained semantic model fine-tuned based on railway engineering specification text and historical review opinions, wherein the pre-trained semantic model is preferably a BERT-based model; and a first similarity threshold and a second similarity threshold are set for classification. Opinions with a similarity greater than or equal to the first similarity threshold are considered duplicates and are merged. Opinions with a similarity between the second and first similarity thresholds and the first similarity threshold are considered potentially conflicting opinions and enter the weighted decision process. Opinions with a similarity between the second and first similarity thresholds are considered irrelevant opinions.

[0009] Preferably, the preferred values ​​for the similarity thresholds are: a first similarity threshold of not less than 85% and a second similarity threshold of not more than 30%. The preset weight set includes at least authority weight, frequency of occurrence weight, and standard matching weight, and is dynamically adjusted according to the project stage so that the weight that matches the business priority of the current stage of the project can obtain a higher coefficient.

[0010] Preferably, the machine-readable verification of the railway professional rule base includes at least: conditional judgment of the minimum curve radius corresponding to the line type and design speed, and regular matching verification of the mileage K in the format of kilometers plus meters; After the review is approved and before the packaging is completed, the relevant metadata elements or form models are configured to be in use through the system. The package building strategy aggregates according to three levels of nodes: project identifier, line identifier, and professional code. Only approved metadata units with business relationships under the same project are included in the package. If it is confirmed that there are business relationships and parameters match between different professions, the metadata units are included in the package.

[0011] Preferably, the pre-release verification list includes at least the following checks: key fields are not empty, data types and value ranges conform to preset rules, completeness of required fields, consistency of cross-unit references, and detection of differences and conflicts with the most recent historical packages; If any check fails, the publication will be suspended.

[0012] Preferably, the packet to be released uses the SHA-256 hash algorithm to calculate the digest and uses the RSA-2048 asymmetric encryption algorithm to perform digital signature; the signature, packet version number and algorithm identifier are written together into the release version repository; The encrypted communication channel uses the HTTPS protocol and enables TLS 1.3, and appends SHA-256 checksums to the transmitted metadata for integrity verification. During the encrypted communication distribution, sensitive fields in the metadata database are encrypted using AES-256 at the field level and decryption access is controlled based on permissions; user passwords are stored using salted PBKDF2 hashes with a salt value length of not less than 16 bytes and an iteration count of not less than 10,000; session tokens are JSON Web Tokens signed based on the certificate authority's private key, with a validity period of not more than 2 hours; the keys used for signing and transmission are generated by the railway bureau's unified certificate issuing authority based on SM2 and centrally managed, with a rotation cycle of not more than 90 days.

[0013] Preferably, after the metadata is signed and written to the release repository but before it is officially released, a canary release and A / B testing are performed. When any metric exceeds the threshold or a major defect is identified, roll back to the previous package version and mark the failed version as unusable; After the metadata is officially released, a compatibility period is set to allow multiple versions to access it in parallel, and access priorities and time windows are specified.

[0014] A railway engineering metadata update and publishing device, comprising: The sharding routing unit is used to perform hash routing on the target metadata node and locate the physical shard based on the composite sharding key; The difference annotation unit is used to perform historical version comparison within the physical segment and generate difference annotations with source identifiers; The review aggregation unit is used to extract modification opinions from multiple sources from the review queue and call a natural language processing model to calculate semantic similarity. The weighted decision-making unit is used to make decisions on potentially conflicting opinions based on a preset set of weights and generate review conclusions. The rule verification unit is used to perform machine-readable verification on the data corresponding to the audit conclusion based on the railway professional rule base; Version building unit is used to generate metadata unit versions at the node granularity, package them into metadata data packages according to the package building strategy, and assign package version numbers; A pre-release verification unit is used to perform key item checks on the metadata packet and generate a release packet. The signing and publishing unit is used to calculate a digest of the package to be published, perform a digital signature, and write the result into the publishing version repository. The secure distribution unit is used to distribute signed metadata data packets through an encrypted communication channel, and to perform transmission integrity verification and encryption of sensitive fields during the distribution process; The canary release and rollback unit is used to perform canary releases and A / B testing before distribution, and to set compatibility periods and trigger version rollbacks after distribution.

[0015] In summary, the present invention has the following main beneficial effects: This invention introduces a role-based access control mechanism to configure change sources, allowing only design units, construction units, or review agencies to submit structured change items that conform to railway IFD standards and pre-defined form schema specifications within their authorized scope. This achieves precise control over data sources and uniformity of input formats, avoiding unauthorized modifications and disorganized data, and improving the reliability and traceability of change requests.

[0016] This invention establishes a historical version comparison and difference labeling mechanism, and combines semantic similarity judgment and weight decision of natural language processing model to merge repeated modification opinions from multiple sources and adjudicate conflicts. This enables intelligent aggregation of modification opinions and automatic dispute handling capabilities under multi-department collaboration, reduces repeated opinions and manual arbitration, improves review efficiency and significantly reduces conflict rate, and achieves clear traceability of modification trajectory.

[0017] This invention employs node-level version management and package building strategies to include metadata units with only business relevance and matching parameters into a package. Combined with a pre-release verification mechanism that ensures the integrity of key fields, standardized data formats, appropriate value ranges, and cross-version conflict resolution, it achieves accuracy in version building and security guarantees before release. Furthermore, through security and fault-tolerance measures such as digest signing, encrypted transmission, field-level encryption, canary releases, and compatibility rollback, it ensures the integrity and security of metadata data transmission, as well as the stability of system updates in a large-scale distributed environment. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1 refer to Figure 1A method for updating and publishing railway engineering metadata, comprising: System Architecture This method is deployed on a B / S architecture metadata management platform: the front end uses JavaScript to implement dynamic forms, tree structure visualization, and modification highlighting; the back end uses a Java framework to carry out permission, modification, approval, version, and release logic; the database uses MySQL 8.0 with database sharding and parallel querying; the source code and version files are uniformly managed using SVN to ensure the traceability of data and code; System configuration (for high concurrency and large data volumes): Server basic configuration: Memory ≥ 32GB (64GB recommended), CPU ≥ 8 cores (clock speed ≥ 2.8GHz, supports Hyper-threading), Hard drive ≥ 1TB SSD (using RAID5 array, read / write speed ≥ 500MB / s), Network card ≥ 10Gbps. Batch processing interface: Supports batch import of tens of thousands of metadata from Excel (response time < 5s).

[0021] Distributed scaling: Supports cluster deployment with a minimum of 3 nodes (including 1 master and 2 slaves), and each node has the same configuration; adopts a distributed storage architecture, and the total storage capacity can be elastically expanded to the PB level, which can support PB-level engineering projects such as the Sichuan-Tibet Railway.

[0022] Dedicated component configuration: Deploy a separate Redis cache server (memory ≥ 16G) and a message queue server (such as RabbitMQ, hard disk ≥ 500G) for request buffering and asynchronous processing in high-concurrency scenarios.

[0023] Database optimization uses MySQL 8.0, enabling parallel query functionality of the InnoDB engine to support multi-threaded processing of complex queries; a database sharding and table partitioning strategy is implemented: the metadata main table (e.g., md_metadatainfo) is split by line ID hash + professional type, with the data volume of a single table controlled within 5 million rows; multi-level indexes are created: composite indexes are created for frequently queried fields (e.g., ifd_code, line_id, create_time), with index selectivity ≥ 0.8.

[0024] The caching strategy adopts a Redis caching design, setting an expiration time of 30 minutes for hot data (such as line parameters accessed frequently in the past 30 days), and automatically evicting cold data; cache penetration protection: for non-existent metadata IDs, cache null values ​​(expiration time of 5 minutes) to avoid database pressure; for frequently used metadata, the hit rate is higher than 95%.

[0025] To address high-concurrency scenarios during peak periods of railway engineering (such as design review and data submission phases), an Nginx cluster (≥2 nodes) is deployed, employing a "weighted round-robin + IP hash" algorithm to distribute requests, ensuring that requests from the same user are routed to a fixed node, thus reducing distributed session overhead. Simultaneously, the Nginx rate limiting module is enabled: the maximum number of requests per IP per second is ≤20, with any exceeding this returning a 429 status code to prevent malicious attacks. Frontend static resources (such as form templates and metadata classification trees) are distributed via CDN, with a cache validity period set to 7 days.

[0026] The system has a built-in professional verification rule library based on railway industry standards and engineering practices, which performs real-time or batch rule checks during the metadata entry, modification and review process.

[0027] Minimum curve radius verification: Different design speed target values ​​and line grades (such as high-speed railway, passenger and freight railway, heavy-haul railway) have different minimum requirements.

[0028] Example of verification logic: IF Line type="High-speed railway" AND Design speed ≥250km / h THEN Minimum curve radius = 3500m (may be reduced under specific conditions, additional verification required); IF Line type="Passenger and freight railway" AND Design speed = 160km / h THEN Minimum curve radius = 1200m (1000m under difficult conditions, special explanation required); The system verifies whether the entered curve radius value is greater than or equal to the minimum curve radius corresponding to the current line type and design speed. If it is less than the minimum value and there is no valid "difficult condition" identifier or description, an error message is triggered.

[0029] Mileage format verification: Follow the railway industry's common mileage representation standard (such as K km + m format). For example, K123 + 456.78 represents 123 km 456.78 m. Verify whether the entered mileage field string conforms to this format specification; otherwise, trigger a format error.

[0030] This invention breaks through the "generalization" limitations of generic metadata management tools, focusing on the core needs of railway engineering—"multi-disciplinary, high-standard, and rigorously collaborative"—and achieving standardized management of metadata from collection to publication through innovations such as node-based management, collaborative review, and professionally adapted version control. Compared to existing technologies, its core advantages are reflected in: Railway-specific features: It conforms to railway industry standards (IFD, Document No. 93), solving the problem of general-purpose tools being "unsuitable" for local conditions; Collaboration efficiency: Review cycle shortened by more than 30%, metadata conflict rate reduced by 25%; Data processing capabilities: Employs distributed SQL table partitioning rules to support concurrent access to TB-level metadata; According to quality: Through professional rule verification, the metadata error rate is reduced by 40%, providing a reliable data foundation for the intelligent construction of railway engineering.

[0031] Access Control The access control of this invention adopts a three-level association of user-role-permission: the system uses RBAC to perform a three-level mapping of "user (multiple roles)-role-permission", where users and roles are mapped in many-to-many relationships through sys_user_role; The definition of a character includes: Cataloger: Responsible for entering, editing, and submitting metadata, and associated with cataloging permission sets; Auditor: Responsible for reviewing metadata and providing feedback, and associated with a set of audit permissions; Administrator: Responsible for system configuration, role assignment, and full data management, as well as the associated management permission set.

[0032] Authorization permissions: Import metadata from Excel, edit form schema, and dynamically create forms; Workflow control permissions: Submit for review, view review status.

[0033] Dynamic management of nodes The system uses "nodes" as the core to divide metadata management units (such as project, line, and specialty metadata units), supporting the assignment of modifiers, submission for review, and version release by node. Each node can be associated with multiple specialties (such as bridges and tunnels in front of the station, and power and communication in the back of the station), achieving professional focus and precise control of metadata; Metadata unit division is based on business relevance, breaking down railway engineering metadata into multi-level nodes. The railway infrastructure metadata unit division includes: personnel, organization, documents, projects, lines, tracks, roadbeds, bridges, tunnels, power, substations, overhead contact lines, communications, signaling, information, disaster prevention, stations, buildings, environmental protection, engineering economics, operations, locomotives and rolling stock, and water supply and drainage.

[0034] Each level of node directly corresponds to a sharding dimension, forming a mapping rule where a node is a sharding unit: From project nodes to top-level shards, railway engineering projects (such as "Sichuan-Tibet Railway" and "Beijing-Shanghai High-Speed ​​Railway") are the top-level nodes, and each project is an independent top-level shard. For example, the metadata of the Sichuan-Tibet Railway project is stored independently in a sharded cluster, which is physically isolated from the sharded cluster of the Beijing-Shanghai High-Speed ​​Railway project to avoid cross-project data interference, while meeting the needs of "project-level data closed-loop management" for large-scale projects.

[0035] From line nodes to secondary shards, each line under a project (such as the "Ya'an-Kangding section" and "Linzhi-Lhasa section" of the Sichuan-Tibet Railway) serves as a secondary node, corresponding to a secondary shard. Line shards are distributed across different database nodes according to the "line ID hash" rule; For example, shards with line IDs ending in 0-3 are stored in database node A, shards ending in 4-6 are stored in node B, and shards ending in 7-9 are stored in node C to ensure even data distribution.

[0036] Specialized nodes are divided into three levels of shards. Specialized nodes under the line nodes (such as bridges, tunnels, power, and communications) are sharded as three levels of shards. For example, in the "Ya'an-Kangding section" line shard, the metadata of bridges is stored in a separate sub-shard, and the metadata of tunnels is stored in another sub-shard, realizing "physical isolation of specialized data". This division not only conforms to the business logic of "pre-station / post-station professional division" in railway engineering, but also facilitates quick querying by professional dimension (such as accessing only the bridge professional segment to obtain the metadata of all bridges on the line). By combining the characteristics of metadata unit nodes, sharding technology can be implemented through database sharding and table partitioning. The sharding key is "project ID + line ID + professional code", and the data is routed to the corresponding physical shard through a hash algorithm.

[0037] Update process Assignment Modifier Administrators assign modifiers based on nodes (projects / lines / specialties); after authentication, users can only operate within their authorized scope and must make modifications based on the last uploaded version to ensure consistency and continuity of multi-user collaboration.

[0038] Modify metadata elements After the person making the edit switches to "edit mode," they will only see the content of this node and can make the following changes: 1) Metadata element relationship tree (add / move / delete); 2) Modify the content of metadata elements; modify color differentiation and strikethrough annotations, which can be undone, and saved and submitted on a node-by-node basis.

[0039] 3) Version Impact (Unit Version / Package Version) Metadata versions are divided into: Unit Version: A major / project / route is a "metadata unit", and a version number is assigned once it is approved; Metadata package version: Package all approved units and assign a package version number for release and management.

[0040] 4) Modification history and accuracy assurance: The system provides historical record viewing and file download. 5) After completing the metadata cataloging, the design unit can submit it to the review unit for review. Once submitted at the node level (as a specific metadata category tree node), it cannot be cataloged again until the review is completed. The review unit can export the changes to an Excel spreadsheet and submit a description. After the design unit confirms that everything is correct, the review unit will make the final confirmation, and the confirmed version will serve as the release baseline. If there are any questions, the design unit can fill out a problem description according to the process and resubmit it, with the final confirmation coming from the review unit.

[0041] Review and Conflict Handling Review process and semantic analysis Text preprocessing removes punctuation and unifies units ("meter" → "m") to obtain standardized text; The system summarizes multiple modification suggestions for the same element and assigns reviewers; it also uses the BERT model to convert the text into a 768-dimensional vector and calculates cosine similarity. Using the cosine similarity formula, the semantic similarity was calculated to be 92%. If the judgment result is: If ≥85%, it is considered a duplicate opinion (no further processing required); 30%–85% are identified as potential conflicts (requiring rule base intervention); <30% of opinions were deemed irrelevant (recorded separately); and a semantic analysis example was provided: Comment 1: "The curve radius must be ≥1000m" Opinion 2: "Minimum curve radius 1000 meters" The system highlights / icons for conflicts that cannot be resolved automatically in the UI, showing differences for online collaborative discussion among units to confirm each point; if disagreements still exist, offline meetings are reserved as the final solution, and the final result is entered into the system by the administrator to form the final version.

[0042] Rule base weight decision (can be dynamically adjusted) The rule base is weighted by authority, frequency, and standard matching; and can be dynamically adjusted according to project stage and metadata type (e.g., design institutes have higher authority weight in the design stage, construction units have higher weight in the construction stage, etc.). Example calculations are provided in the explanation: Authority: A=0.9, B=0.6; Frequency: A appears 2 times → 0.6, B appears 1 time → 0.3; Standard match: A=1.0, B=0.2; The total weight of A is 0.9 × 0.4 + 0.6 × 0.3 + 1.0 × 0.3 = 0.84 > the total weight of B is 0.39, so A is adopted.

[0043] Real-world conflict case (curve radius 1200m vs 1000m) Case: There are two opinions regarding the "curve radius" in the metadata of a certain bridge: Opinion A (Design Institute): "The curve radius should be ≥1200m (in accordance with TB10002.2-2017 Railway Bridge and Culvert Design Code)"; Opinion B (Construction Unit): "The curve radius can be set to 1000m (if site conditions permit)."

[0044] The system makes decisions based on the weighting of authority, frequency, and standards, and automatically adopts option A. If there is still a dispute, it can be finally confirmed through online discussion or offline meeting, and the administrator will enter it as the final result.

[0045] Online-offline multi-level review model: It adopts a combination of "online process + offline confirmation" to retain the systematic collection, recording and tracking of opinions, while taking into account the efficiency and accuracy of cross-departmental and cross-professional collaboration.

[0046] Version management and release mechanism Packaging and Version Number: Approved metadata units are packaged into metadata data packages, written with the publisher / organization / time / digital signature, stored in the published version repository, and the metadata data package version list is updated; the system supports access / query / management by version number, and allows multiple versions to run in parallel to balance compatibility and flexibility; Published data: Elements or form schemas are marked as "Published" after modification / review / necessary validation is completed, and assigned a unique version number; providing the ability to view / download and compare multiple versions for traceability; Compatibility rules: differentiate between backward compatibility (fields can be added) and potential incompatibility (field deletion / type modification requires strict restrictions); provide forward compatibility strategies: mark fields to be deprecated in advance rather than remove them immediately, ensuring that existing systems can recognize them; Transition period settings: Allow the old and new versions to coexist for a period of time; automatically deactivate the old version or retain multiple versions after the period expires according to a preset strategy; Usage Status and New Version Release: Once the approved data or form schema is set to "Used", it can be submitted for release; when releasing, the entire document is released and the version number and version description are set (they cannot be repeated and must include the start execution time); when releasing a form schema, the corresponding metadata version must be specified for easy association and retrieval.

[0047] Phased and grayscale: During the phased process, continuously monitor the differences between the old and new versions in data processing, performance, and business results; if stable, expand the scope to the full version; if abnormal, quickly roll back.

[0048] Prioritize gray-scale deployment scenarios: high-risk changes (such as speed limits / critical safety parameters like power supply parameters), and scenarios with strong regional / business isolation (facilitating gray-scale deployment by region / line / system). A / B testing: Prioritize A / B scenarios: experience / performance optimization, comparison of multiple solutions (such as the impact of different data models on response time), and scenarios where the impact is dispersed and difficult to categorize by region / business. Allow random or rule-based traffic allocation, compare metrics such as accuracy / latency / error rate / experience, and proceed with the best performance and terminate and roll back if the performance is poor.

[0049] Pre-release verification: Automatically verifies key fields according to the checklist, ensuring they are not empty, have the correct format, are complete in all required fields, have the correct value range, and have historical differences; if any field fails to pass, the release will be blocked and a detailed report will be provided; only if all fields pass will the release be allowed to proceed.

[0050] Digital signature: Generates a digest and signature for the new version's metadata and stores it together with the version number / version description. Subsequent calls can verify the signature to confirm integrity and trustworthiness of the source.

[0051] Security Mechanism Extended encryption mechanisms and key management Building upon the basic encryption mechanism, and considering the sensitive nature of railway engineering metadata, the following encryption scenarios are extended: User authentication encryption: User passwords are stored using the PBKDF2 algorithm with salt hash (salt value length 16 bytes, iteration count 10,000 times) to avoid password leakage caused by plaintext or weak hash; Session tokens are in JWT format, containing user role and permission information, and are signed with a private key issued by the CA center. The validity period is set to 2 hours, after which they will automatically expire.

[0052] Operation log encryption: Sensitive operations recorded in the system operation log (such as metadata publishing and permission changes) must be stored in encrypted form (AES-256-GCM mode). Only users with the "audit administrator" role can decrypt and view the data to ensure that the operation trajectory is traceable and that sensitive information is not leaked.

[0053] Temporary data encryption: Temporary files generated during metadata import / export (such as Excel cache) are stored in memory with encryption. The files are automatically encrypted when they are written to memory and are deleted immediately after use (overwriting file sectors) to prevent temporary data leakage. To ensure the confidentiality and integrity of metadata during network transmission, an end-to-end encryption scheme is adopted: Transport Protocol: Enable HTTPS protocol across the entire link, enforce the use of TLS 1.3 encryption suite (such as TLS_AES_256_GCM_SHA384), and disable insecure TLS 1.0 / 1.1 and SSL protocols to prevent risks such as man-in-the-middle attacks and replay attacks.

[0054] Certificate Management: The server-side certificate is issued by the railway bureau's CA center and contains the domain name and IP information of the metadata management system. It is valid for one year. The client verifies the server's identity through the built-in CA root certificate to prevent certificate forgery.

[0055] Transmission verification: SHA-256 hash values ​​are appended to the transmitted metadata blocks (such as batch-imported Excel files), and the receiving end verifies the hash consistency to ensure that the data transmission process has not been tampered with.

[0056] For sensitive information in the metadata database, "field-level fine-grained encryption" is implemented, with the following specific rules: The AES-256 algorithm (CBC mode) is used to encrypt sensitive fields, including: business-sensitive fields such as power supply system parameters (e.g., power_supply_params) and catenary voltage levels (e.g., catenary_voltage), which are related to railway operation safety; and management-sensitive fields such as user contact information (e.g., sys_user.phone) and confidential content in audit opinions (e.g., md_audit.opinion). Non-sensitive fields (e.g., metadata names and public codes) are not encrypted, balancing security and query efficiency.

[0057] When writing metadata to the database, the system automatically identifies sensitive fields and calls the encryption interface (implemented based on JavaCryptographyArchitecture) to encrypt and store the data; when reading, decryption is triggered according to user permissions (e.g., "advanced administrators" can decrypt all fields, while ordinary users can only view non-sensitive fields) to ensure that the data is "usable but not visible".

[0058] Professional verification rules Based on the railway IFD standard, this invention expands and constructs a metadata model covering all disciplines before and after the station (bridges, tunnels) and after the station (electrical, electrical, and station yard) to support deep coupling of design, construction and operation and maintenance processes. During data entry and review, it supports real-time / batch verification of key indicators such as curve radius and mileage format (K km + m, such as K123+456.78 means 123 km 456.78 m). Incorrect format will trigger an error.

[0059] RESTful API Interface Specification The system uniformly defines the interface path design, request methods, data formats, status codes, and error handling rules, ensuring that downstream systems can call metadata services in a standardized and clear manner, which greatly simplifies the integration difficulty and maintenance cost between systems. To improve the real-time performance and intelligence of the system, this invention integrates a Webhook notification mechanism into the API specification, so that when metadata changes such as being published, updated or rolled back, the system can proactively push notifications to subscribed downstream systems. The notifications include information such as the event type, change time, metadata identifiers involved, version number, and change summary. With this mechanism, downstream systems no longer need to obtain change information through periodic polling, and can respond to data updates in a timely manner, thereby ensuring that business operations are always based on the latest metadata definitions; Clearly define backward compatibility / forward compatibility strategies (marking new fields / fields to be deprecated, etc.), and provide a transition period mechanism for the parallel use of new and old versions. When the transition period expires, the system will automatically deactivate or allow multiple versions to coexist according to the strategy.

[0060] Example 2 refer to Figure 1 A railway engineering metadata update and publishing device. This system enables end-to-end management of railway engineering metadata during the input, modification, review, version control, and release processes. Running on a B / S architecture, the front-end utilizes JavaScript for interaction and dynamic rendering, while the back-end employs a Java framework for business logic. The database is MySQL 8.0 with sharding and partitioning. Source code and version files are managed using SVN. The server hardware configuration includes a 16-core CPU, 64GB of RAM, and a 4TB SSD. It is configured with Redis (≥16GB RAM) for caching, RabbitMQ (≥500GB hard drive) for message queuing, and an Nginx cluster for load balancing and rate limiting. Combined with CDN nodes, it ensures smooth access across railway bureaus in different regions. The system includes the following modules: Access control module: Used for user authentication and permission allocation based on the RBAC model, supporting three roles: cataloger, auditor and administrator, to achieve hierarchical access and operation restrictions; This module creates a clear operational environment by binding roles and permissions, avoiding confusion in personnel operations and improving system security. Modification module: Used to make modifications within the node range assigned by the system administrator. The nodes include a three-level sharding system of project nodes, line nodes, and professional nodes. The modification module provides modification of the element relationship tree and element content, and saves modification identifiers and history records. This module implements data partitioning management through node-based partitioning, ensuring the locality and controllability of modifications and preventing local modifications from affecting the overall data consistency; The review module is used to assign reviewers and review modification opinions. It includes a semantic analysis unit and a rule base decision unit. The former is used to calculate semantic similarity and determine the relationship between opinions, while the latter performs weighted judgment based on authority weight, frequency weight and standard matching weight, and prompts manual confirmation when conflicts are not resolved. This module improves the efficiency and accuracy of the review process and reduces inconsistencies in human judgment through automated semantic recognition and weighted decision-making mechanisms. Version management module: Used to generate version numbers for approved metadata units, package multiple units into a metadata data package, assign package version numbers, and realize the correspondence between unit version and overall version.

[0061] This module uses a dual version number management mechanism to enable flexible local updates and unified global release of metadata, ensuring system consistency and traceability. Release module: Used to perform overall release, phased release or A / B testing, and gradually enable the new version by region, business system or user group under the canary release strategy, while performing pre-release verification; This module effectively reduces the risk of system upgrades through a gradual release strategy, ensuring the stability and safety of railway engineering operations. Security module: Used to perform digital signatures and verifications during metadata publication, enable encryption protocols during transmission, encrypt sensitive fields during storage, and uniformly manage the key lifecycle; This module combines multiple security mechanisms to ensure the confidentiality and integrity of metadata during transmission and storage, significantly improving the system's security capabilities.

[0062] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for updating and publishing railway engineering metadata, characterized in that, Includes the following steps: S1: Obtain a change request for the target metadata node, wherein the target metadata node is constructed into a hierarchical management unit according to project, line and profession, and a joint sharding key is formed by the corresponding project identifier, line identifier and profession code; The change request consists of change items; the change items are generated based on the source pre-configured by the role's access control policy and written into the review queue as a source record of multiple source modification opinions within the same review cycle; S2: Perform hash routing on the target metadata node based on the combined sharding key, and route it to the corresponding physical shard; S3: Within the target segment located in S2, each change item in the review queue of S1 is compared with the historical version and a difference label is generated to form a set of changes to be reviewed with source identifier and associated with historical version records; S4: Extract multi-source modification opinions for the target metadata node within the same review period from the review queue, calculate semantic similarity based on a natural language processing model to merge duplicate opinions, and make decisions on potentially conflicting opinions based on a preset weight set to obtain a review conclusion; S5: Perform machine-readable verification on the data corresponding to the audit conclusion based on the railway professional rule base; if the data does not comply with the rules, stop the current process and return a verification failure message; if the data complies with the rules, form an audit pass set. S6: Generate metadata unit versions of the approved set at the node granularity; then, according to the predetermined package building strategy, under the conditions of satisfying the integrity of dependency relationships, reference relationships and release granularity constraints, package multiple approved metadata units into a metadata data package and assign a package version number; S7: Perform key item checks on the metadata data packet according to the pre-release checklist; if any check fails, release is aborted, otherwise a release packet is formed. S8: Calculate the digest of the packet to be released using a hash algorithm, and digitally sign it using an asymmetric encryption algorithm; write the signature, packet version number and algorithm identifier into the release version repository; S9: Distribute the signed metadata to downstream systems through an encrypted communication channel based on a transport layer security protocol.

2. The method for updating and publishing railway engineering metadata according to claim 1, characterized in that, The pre-configured sources are configured based on role-based access control, authorizing only the design unit, construction unit, or review unit corresponding to the role to submit change items through the system interface; The roles include cataloger, reviewer, and administrator; The bibliographer only has metadata entry and submission for review permissions; The auditors have the authority to view metadata and fill in audit comments. The administrator has all permissions and can intervene in abnormal processes; The changes are structured data that conforms to the railway IFD standard and the metadata form schema specification.

3. The method for updating and publishing railway engineering metadata according to claim 2, characterized in that, The semantic similarity determination is performed by a pre-trained semantic model fine-tuned based on railway engineering specification texts and historical review opinions. The pre-trained semantic model is preferably a BERT-based model. A first similarity threshold and a second similarity threshold are set for classification. Opinions with a similarity greater than or equal to the first similarity threshold are considered duplicates and are merged. Opinions with a similarity between the second and first similarity thresholds and the first similarity threshold are considered potentially conflicting opinions and enter the weighted decision process. Opinions with a similarity between the second and first similarity thresholds are considered irrelevant opinions.

4. The method for updating and publishing railway engineering metadata according to claim 3, characterized in that, The preferred values ​​for the similarity thresholds are: a first similarity threshold of not less than 85% and a second similarity threshold of not more than 30%. The preset weight set includes at least authority weight, frequency of occurrence weight, and standard matching weight, and is dynamically adjusted according to the project stage so that the weight that matches the business priority of the current stage of the project can obtain a higher coefficient.

5. A method for updating and publishing railway engineering metadata according to claim 4, characterized in that, The machine-readable verification of the railway professional rule base includes at least: conditional judgment of the minimum curve radius corresponding to the line type and design speed, and regular matching verification of the mileage K in the format of kilometers plus meters; After the review is approved and before the packaging is completed, the relevant metadata elements or form models are configured to be in use through the system. The package building strategy aggregates according to three levels of nodes: project identifier, line identifier, and professional code. Only approved metadata units with business relationships under the same project are included in the package. If it is confirmed that there are business relationships and parameters match between different professions, the metadata units are included in the package.

6. The method for updating and publishing railway engineering metadata according to claim 5, characterized in that, The pre-release verification checklist includes at least the following checks: key fields are not empty, data types and value ranges conform to preset rules, completeness of required fields, consistency of cross-unit references, and detection of differences and conflicts with the most recent historical packages; If any check fails, the publication will be suspended.

7. A method for updating and publishing railway engineering metadata according to claim 6, characterized in that, The packet to be released uses the SHA-256 hash algorithm to calculate the digest and the RSA-2048 asymmetric encryption algorithm to perform digital signature; the signature, packet version number and algorithm identifier are written together into the release version repository; The encrypted communication channel uses the HTTPS protocol and enables TLS 1.3, and appends SHA-256 checksums to the transmitted metadata for integrity verification. During the encrypted communication distribution, sensitive fields in the metadata database are encrypted using AES-256 at the field level and decryption access is controlled based on permissions; user passwords are stored using salted PBKDF2 hashes with a salt value length of not less than 16 bytes and an iteration count of not less than 10,000; session tokens are JSON Web Tokens signed based on the certificate authority's private key, with a validity period of not more than 2 hours; the keys used for signing and transmission are generated by the railway bureau's unified certificate issuing authority based on SM2 and centrally managed, with a rotation cycle of not more than 90 days.

8. The method for updating and publishing railway engineering metadata according to claim 7, characterized in that, After the metadata is signed and written to the release repository, but before it is officially released, perform canary release and A / B testing. When any metric exceeds the threshold or a major defect is identified, roll back to the previous package version and mark the failed version as unusable; After the metadata is officially released, a compatibility period is set to allow multiple versions to access it in parallel, and access priorities and time windows are specified.

9. A railway engineering metadata update and publishing device, comprising: The sharding routing unit is used to perform hash routing on the target metadata node and locate the physical shard based on the composite sharding key; The difference annotation unit is used to perform historical version comparison within the physical segment and generate difference annotations with source identifiers; The review aggregation unit is used to extract modification opinions from multiple sources from the review queue and call a natural language processing model to calculate semantic similarity. The weighted decision-making unit is used to make decisions on potentially conflicting opinions based on a preset set of weights and generate review conclusions. The rule verification unit is used to perform machine-readable verification on the data corresponding to the audit conclusion based on the railway professional rule base; Version building unit is used to generate metadata unit versions at the node granularity, package them into metadata data packages according to the package building strategy, and assign package version numbers; A pre-release verification unit is used to perform key item checks on the metadata packet and generate a release packet. The signing and publishing unit is used to calculate a digest of the package to be published, perform a digital signature, and write the result into the publishing version repository. The secure distribution unit is used to distribute signed metadata data packets through an encrypted communication channel, and to perform transmission integrity verification and encryption of sensitive fields during the distribution process; The canary release and rollback unit is used to perform canary releases and A / B testing before distribution, and to set compatibility periods and trigger version rollbacks after distribution.