A full-configuration low-code development platform and implementation method for the public transportation industry

Abstract

The application belongs to the technical field of information platform development, and more particularly relates to a full-configuration low-code development platform for the public transportation industry and an implementation method. The platform comprises a data model generation engine, a front-end model generation engine, an APP analysis engine, a metadata analysis engine, a visual business rule engine, a visual report tool and a low-code platform base. The platform supports users to generate intelligent public transportation system functions with zero code through a four-step standardized process of selecting a scene, configuring a page, associating rules and publishing functions. Meanwhile, through modular design, collaborative optimization and security protection mechanisms, the platform achieves a balance between functions and performance, flexibility and stability. The application solves the problems of adaptability and development efficiency, function expansion and performance imbalance, and demand change and system stability conflict.

Description

Technical Field

[0001] This invention belongs to the technical field of information platform development, and more specifically, relates to a fully configurable low-code development platform and its implementation method for the public transportation industry. Background Technology

[0002] In the process of information technology construction in the public transportation industry, the intelligent public transportation system, as the core management carrier, needs to fully cover multiple key business scenarios such as vehicle dispatching, maintenance, ticketing management, personnel management, and safety. Its development efficiency and adaptability directly affect the operational management efficiency of public transportation companies.

[0003] Chinese patent document CN117873881A discloses a software development system and method based on an intelligent transportation platform, including hardware and software architecture. The hardware includes a rich text editor, a server, and a parser. The software architecture includes a perception layer, a data access layer, a data resource layer, an application layer, a presentation layer, and a user layer. The development method first configures the rich text editor, uses it to generate dynamic templates, sets traffic variable pit locations, and replaces them with traffic variable values ​​of the same name. It then retrieves the required rich text source code from the server database, parses it using the parser configured on the client, mounts the corresponding components to their respective locations, and applies the saved attribute data to the corresponding components. Finally, the client generates a complete page.

[0004] Currently, the development of traditional intelligent public transportation systems and the application of general low-code platforms in the public transportation industry face several significant pain points:

[0005] The contradiction between adaptability and development efficiency: General low-code platforms adopt a "general scenario adaptability" architecture, which lacks targeted design for the business characteristics of the public transportation industry. This results in a large amount of customized coding during the development process, which cannot meet the needs of rapid delivery in the public transportation industry. On the other hand, some industry-specific low-code platforms focus on adaptability, but their functions are limited and they are difficult to cover the multi-scenario and complex business needs of the public transportation industry.

[0006] Functional expansion and performance imbalance: With the intelligent upgrade of the public transportation industry and the frequent adjustment of business logic, the system needs to support multi-module collaboration, massive data processing and complex business rule configuration. However, the existing platform is prone to performance problems such as response delay, excessive resource consumption and data processing bottlenecks when expanding functions, making it difficult to achieve a balance between rich functionality and efficient operation.

[0007] Conflict between requirement changes and system stability: Traditional development models involve cumbersome system expansion and update processes. Although low-code platforms support configuration-based modifications, complex requirement changes can easily lead to configuration conflicts, data inconsistency breaches, and process interruptions, resulting in decreased system stability. At the same time, the public transportation industry, which handles ticket revenue data and personnel privacy information, has extremely high requirements for data security and business continuity. Existing platforms lack targeted security protection and fault self-healing mechanisms.

[0008] Loose module collaboration and technical architecture: The metadata model of the existing low-code platform is too general, the front-end template has low adaptability to the specific scenarios of the public transportation industry, the business rule configuration and the front-end and back-end linkage mechanism are inflexible, and there is a lack of close connection between the various functional modules, data storage and technical architecture. As a result, business personnel cannot independently complete the system function configuration and still need to rely on developers for secondary development. The system iteration efficiency is low and the maintenance cost is high. Summary of the Invention

[0009] This invention aims to overcome at least one of the defects of the prior art and provide a fully configurable low-code development platform and implementation method for the public transportation industry. It solves the basic technical problems of poor adaptability, low delivery efficiency, and high operating threshold for business personnel in the development of smart bus systems in the public transportation industry by existing general low-code platforms. At the same time, it tackles advanced problems such as functional expansion and performance balance, rapid response to complex requirements, and system stability and security.

[0010] The detailed technical solution of this invention is as follows:

[0011] A fully configurable low-code development platform for the public transportation industry, the platform comprising: a data model generation engine, a front-end model generation engine, an APP parsing engine, a metadata parsing engine, a visual business rules engine, a visual reporting tool, and a low-code platform base;

[0012] The data model generation engine provides the metadata parsing engine with verified raw configuration data and business data;

[0013] The front-end model generation engine generates interface configuration data based on the standardized metadata model output by the metadata parsing engine and combined with the user's page layout configuration. At the same time, it receives metadata relationship data from the metadata parsing engine and filters out metadata components that are directly related to the current configuration scenario, allowing users to quickly drag and drop to design the page layout. It also links with the theme setting unit of the low-code platform base to achieve a unified interface style.

[0014] The APP parsing engine and the front-end model generation engine work together to achieve one-time configuration and multi-platform generation; at the same time, it receives permission configuration information from the low-code platform base to achieve consistent permission control on the APP side and the PC side.

[0015] The metadata parsing engine receives the verified raw configuration data and business data, performs structured processing, generates standardized metadata models and structured business data, verifies them, and synchronizes them to the visualization business rule engine as the basis for data field association. It also provides structured metadata and relational data for the front-end model generation engine, visualization business rule engine, and visualization reporting tools; and collaborates with the platform's unified data storage layer to realize the storage and retrieval of different types of data.

[0016] The visualization business rule engine receives structured metadata and relational data from the metadata parsing engine. By associating metadata fields with business data, the business rules can be accurately applied to the business data, thereby achieving automated execution of business logic and realizing business processes.

[0017] The visualization reporting tool receives structured business data output by the metadata parsing engine, supports report data statistics, generates adapted tables and charts based on the fields defined in the metadata model, and links with the data permission unit of the low-code platform to achieve permission isolation of report data; and works with Elasticsearch in the data storage layer to realize data query and analysis.

[0018] The low-code platform base provides access control, security protection, and configuration management support for the above modules; it supports the collaborative work between modules through event-driven architecture and RocketMQ message queue technology, and the modules transmit data and instructions through standardized message formats and monitor the running status of each module in real time.

[0019] Furthermore, the data model generation engine includes: a table structure configuration unit, a table relationship definition unit, an industry template library unit, and a model validation unit;

[0020] Industry Template Library Unit: As the basic entry point for data model configuration, it provides pre-set data model templates exclusive to the public transportation industry, supports template import, modification and saving, and allows users to select templates or modify them based on existing templates;

[0021] Table structure configuration unit: It takes over the template data of the industry template library unit or user-defined requirements. Based on a visual form interface, it supports user-defined field types, lengths and constraints. It uses JSON Schema format to store configuration information to ensure structural standardization.

[0022] Inter-table relationship definition unit: Based on the single-table structure determined by the table structure configuration unit, it provides a visual drag-and-drop interface, supports one-to-one, one-to-many, and many-to-many relationship configuration, has a built-in relationship verification algorithm, automatically detects abnormal situations, and builds a complete data model association system;

[0023] Model Validation Unit: Integrates data type validation, constraint validation, and relationship rationality validation logic to comprehensively validate the table structure configuration and inter-table relationship definition results, and the validation results are fed back to the front end in real time;

[0024] Each unit works collaboratively in the order of "industry template library unit → table structure configuration unit → inter-table relationship definition unit → model verification unit" to generate original configuration data, and achieves data model permission isolation through linkage with the low-code platform base permission management unit.

[0025] Furthermore, the front-end model generation engine includes: a metadata filtering unit, a component drag-and-drop orchestration unit, a multi-terminal adaptation unit, and an interface preview unit.

[0026] Related Metadata Filtering Unit: As a pre-configuration step, it uses RPC to call the metadata relationship identification results of the metadata parsing engine, and employs a greedy algorithm to filter metadata with related relationships in the current configuration scenario, while filtering out irrelevant components;

[0027] Component drag-and-drop orchestration unit: It takes over the metadata component output by the associated metadata filtering unit. It is based on the vuedraggabIe component encapsulation, supports component position adjustment and hierarchical nesting, and uses the DOM Diff algorithm to optimize the interface rendering efficiency after dragging. Users can complete the page layout design by dragging and dropping.

[0028] Multi-terminal adaptation unit: Based on the page layout determined by the component drag-and-drop arrangement unit, it has a built-in responsive layout rule library and automatically adjusts the component size and layout structure according to the screen size of PC and APP.

[0029] Interface Preview Unit: It receives the interface configuration data after multi-terminal adaptation, integrates iframe embedding technology, loads the configured interface effect in real time, and supports online editing and preview synchronization;

[0030] Each unit works collaboratively in the order of "Associated Metadata Filtering Unit → Component Drag and Drop Arrangement Unit → Multi-Terminal Adaptation Unit → Interface Preview Unit" to generate interface configuration data. This data is shared with the APP parsing engine to achieve synchronous generation across multiple terminals. At the same time, it receives metadata relationship data from the metadata parsing engine to support intelligent component filtering and works in conjunction with the theme setting unit of the low-code platform base to achieve a unified interface style.

[0031] Furthermore, the APP parsing engine includes: an interface configuration parsing unit, a terminal adaptation and conversion unit, an interaction logic synchronization unit, and a compatibility verification unit;

[0032] Interface configuration parsing unit: As the data processing entry point, it receives interface configuration data shared by the front-end model generation engine, uses a JSON parser to extract component type, attributes, and layout information, and generates terminal adaptation intermediate files to provide basic data for subsequent adaptation and conversion;

[0033] Terminal Adaptation and Conversion Unit: Receives the intermediate files output by the Interface Configuration Parsing Unit and performs differentiated conversions for Android / iOS system characteristics. On Android, the intermediate files are converted into XML format rendering data, and on iOS, they are converted into Storyboard format rendering data. Cross-platform adaptation plugins ensure consistent interaction logic.

[0034] Interactive Logic Synchronization Unit: Based on the rendering data of the terminal adaptation and conversion unit, the interactive logic configured on the PC is synchronized to the APP through an event mapping mechanism, using a unified event triggering protocol;

[0035] Compatibility verification unit: Verifies the rendering data after terminal adaptation and conversion. It has a built-in compatibility rule library for mainstream Android and iOS versions, automatically detects component adaptation anomalies, outputs a compatibility report, and supports one-click repair.

[0036] Each unit works collaboratively in the order of "interface configuration parsing unit → terminal adaptation and conversion unit → interaction logic synchronization unit → compatibility verification unit", sharing interface configuration data with the front-end model generation engine and receiving permission configuration information from the low-code platform base to achieve multi-terminal synchronous generation function with consistent permissions between the APP and PC.

[0037] Furthermore, the metadata parsing engine includes: a metadata modeling unit, a data format conversion unit, a data manipulation unit, a metadata relationship identification unit, and an import / export unit;

[0038] Metadata Modeling Unit: Based on the business characteristics of the public transportation industry, the metadata model structure is designed, the metadata attributes are encapsulated using object-oriented thinking, metadata version management is supported, the raw configuration data of the data model generation engine is received, the basic metadata model is generated, and structural support is provided for subsequent data processing.

[0039] Data Format Conversion Unit: It inherits the basic metadata model of the metadata modeling unit, adopts the Jackson + Fastjson dual parsing framework, and converts unstructured and semi-structured metadata and business data into relational data that conforms to the MySQL table structure specification. It ensures the accuracy of conversion through data type mapping table, supports custom mapping rules, and synchronizes the converted data to the data storage layer.

[0040] Data Operation Unit: Based on the transformed structured data, it has built-in CRUD core logic, optimizes query performance for massive data scenarios in the public transportation industry, supports multi-condition combined queries, related queries, and paginated queries. The query SQL is dynamically generated by MyBatis-Plus and is processed using a database sharding and table partitioning + read-write separation architecture to provide data read and write services for other modules.

[0041] Metadata Relationship Identification Unit: Based on the structured data processed by the data operation unit, a graph theory algorithm is used to construct a metadata relationship graph. The relationship chain is traversed by depth-first search to identify one-to-one, one-to-many, and many-to-many relationships between tables, and output structured relationship data to support intelligent filtering and rule configuration of front-end components;

[0042] Import / Export Unit: This unit receives structured data from the data manipulation unit and supports importing and exporting custom templates in Excel and CSV formats. During import, it uses POI to parse the file and matches the data through field mapping rules, with built-in data validation rules. During export, it uses SXSSF to stream large data files, enabling data reuse and backup across systems.

[0043] Each unit works collaboratively in the order of "metadata modeling unit → data format conversion unit → data operation unit → metadata relationship identification unit → import and export unit" to provide validation support for the data model generation engine, provide structured metadata and relational data for the front-end model generation engine, visualization business rule engine, and visualization reporting tools, and work with the data storage layer to achieve full lifecycle management of data.

[0044] Furthermore, the visual business rule engine includes: a rule component management unit, a function library unit, a rule configuration orchestration unit, a rule execution unit, and a conflict detection unit;

[0045] Rule Component Management Unit: As the basic support for rule configuration, it has multiple built-in rule components, adopts a plug-in architecture design, supports dynamic registration and uninstallation of components, and stores component configuration data in JSON format to form an optional rule component library. It is linked with the rule configuration orchestration unit and allows users to drag and drop to combine them.

[0046] Function Library Unit: Provides logical calculation support for rule configuration, integrates the required built-in functions, supports the uploading and calling of custom functions, and uses reflection mechanism for function execution and parameter verification to ensure security, forming an optional function library, which works in conjunction with the rule configuration orchestration unit to improve rule logic;

[0047] Rule configuration orchestration unit: It takes over the filtered rule components and functions, and is a visual flowchart editor based on antv / x6. It supports drag-and-drop combination of rule components and conditional branch settings. It uses the BPMN2.0 specification to describe the rule process, automatically generates executable rule scripts, and links with the conflict detection unit.

[0048] Conflict Detection Unit: Performs logical verification on the rule scripts output by the rule configuration orchestration unit, uses logical reasoning algorithms to detect condition conflicts and priority conflicts, provides real-time feedback on the conflict location and cause and offers solution suggestions, obtains verification results, and synchronizes the rule scripts to the rule execution unit after the verification is passed;

[0049] Rule Execution Unit: Based on the LiteFlow rule engine kernel, it parses and executes validated rule scripts, supports synchronous / asynchronous execution modes, and improves concurrent processing capabilities; the input is validated rule scripts and business data, and the output is the rule execution result. It works in conjunction with the business flow engine and OA workflow engine to support the advancement of business processes.

[0050] Each unit works collaboratively in the order of "rule component management unit → function library unit → rule configuration orchestration unit → conflict detection unit → rule execution unit". It receives structured metadata and relational data from the metadata parsing engine, supports the association between rules and business data, links with the OA workflow engine of the low-code platform to realize the binding of approval processes, and collaborates with the business flow engine to provide rule execution capabilities for business processes.

[0051] Furthermore, the visualization reporting tool includes: a report template design unit, a data query unit, a visualization rendering unit, a report export unit, and a real-time refresh unit;

[0052] Report template design unit: Provides a visual drag-and-drop interface, supports configuration of table and chart components, uses XML format to store template structure, supports template saving and reuse, forms report templates, and links with data query unit to clearly define the display structure of report data;

[0053] Data Query Unit: Connects to multiple databases via JDBC, supports custom SQL queries and multi-data source association queries, and allows query conditions to be configured through a visual interface. It automatically generates optimized query statements and uses a query result caching mechanism to improve efficiency. The input consists of the query requirements corresponding to the report template and the structured business data from the metadata parsing engine, and the output is the query result data.

[0054] Visualization rendering unit: Based on antv / x6 and ECharts, it encapsulates chart components, supports real-time data rendering and interactive operations, and adopts a Canvas+SVG hybrid rendering scheme to balance efficiency and visual effects. Based on the query result data, it generates a visual report interface, which is linked with the real-time refresh unit and the report export unit.

[0055] Real-time refresh unit: Supports setting the refresh frequency, pushes the latest data to the report interface via WebSocket, adopts an incremental data update mechanism to refresh only the changed data, and outputs a visual report after incremental update in real time to ensure the real-time nature of report data;

[0056] Report Export Unit: Supports exporting in Excel, PDF, and image formats. Excel export uses POI streaming processing, PDF export is based on iText7, and image export uses HtmI2Canvas. It supports batch export and scheduled export. Based on the visually rendered report data, it generates an export file for users to download and back up or use across systems.

[0057] The report template design unit, data query unit, visualization rendering unit, report export unit, and real-time refresh unit work together to support the overall functionality of the visualization reporting tool. It receives structured business data from the metadata parsing engine, links with the data permission unit of the low-code platform base to achieve permission isolation, and works with the EIasticsearch component to improve the efficiency of querying massive amounts of data.

[0058] Furthermore, the low-code platform base provides access control, security protection, and configuration management support for all core modules, including: access control unit, system integration unit, cross-module interaction unit, security protection unit, operation and maintenance monitoring unit, and configuration management unit;

[0059] Access Control Unit: It adopts a hybrid model of role-based access control (RBAC) and attribute-based access control (ABAC) to achieve precise control of page permissions, button permissions, data permissions, and field permissions. Permission configuration data is stored in MySQL and hot permission data is cached through Redis.

[0060] System Integration Unit: Provides standardized API interfaces to support integration with various application systems, uses the OAuth2.0 authorization protocol for secure authentication, and resolves interface differences between different systems through an interface adaptation layer;

[0061] Cross-module interaction unit: Based on an event-driven architecture, it uses RocketMQ to realize cross-module and cross-system message communication, supports synchronous / asynchronous message delivery, and has a built-in message retry mechanism and dead letter queue to ensure reliable message transmission;

[0062] Security Protection Unit: Integrates multiple security mechanisms, adopts the Spring Security framework to achieve overall security control, and supports security log auditing;

[0063] Operation and maintenance monitoring unit: integrates Prometheus+Grafana monitoring solution, collects system operation indicators in real time, supports custom monitoring thresholds and alarm rules, and pushes alarm information through multiple channels;

[0064] Configuration Management Unit: Based on Nacos Configuration Center, it enables dynamic management of system parameters, rule configurations, and theme settings. It supports canary releases and rollbacks of configurations, and configuration changes are synchronized to all modules through a message notification mechanism without requiring a system restart.

[0065] This platform is built around seven core modules: a data model generation engine, a front-end model generation engine, an APP parsing engine, a metadata parsing engine, a visual business rules engine, a visual reporting tool, and a low-code platform base. These modules enable the construction of public transportation industry solution application systems, which are the final products. These systems include, but are not limited to, various sub-business systems of the public transportation system. Through the low-code platform base, they achieve data interoperability and business collaboration, directly serving the daily operations of public transportation companies.

[0066] Furthermore, this platform also includes a front-end engine architecture, which is a functional extension of the core front-end modules, including: a front-end metadata visualization generation engine, a front-end PC parsing engine, a front-end APP parsing engine, and an OA workflow engine. The front-end engine architecture is directly linked with the front-end model generation engine and the APP parsing engine. The specific front-end technologies involved include: page module design, page module integration of core function plugins, component parameter configuration, and interface parsing and preview.

[0067] The page module design is supported by the front-end model generation engine's page templates, with four preset standardized page templates: listPage basic table page, treeAndTable left tree right table page, formTable basic form page, and tabFormTable tab form page;

[0068] The core functional plugins integrated into the page module are the technical support plugins for the front-end model generation engine component drag-and-drop arrangement unit and the interface preview unit. These plugins are directly integrated into the component library and preview module, including:

[0069] It uses vue3-highlightjs to implement syntax highlighting for JSON code preview, supports code folding and copying, and renders code snippets through a virtual DOM;

[0070] VueDragGable is used to implement drag-and-drop arrangement of page elements, and CSS3 transformations are used instead of DOM manipulation during the drag-and-drop process;

[0071] It uses the antv / x6 visualization chart library to display the table structure and table relationships graphically, supports chart zooming, panning, and node dragging, uses Canvas to render charts, and also supports the export of chart data;

[0072] We use sortablejs, a lightweight JavaScript library, to implement draggable sorting functionality.

[0073] The component parameter configuration is the implementation method of secondary encapsulation of components by the front-end model generation engine. The encapsulated hy-series components are directly used as optional components in the component drag-and-drop orchestration unit. The component parameter configuration performs secondary encapsulation of the Ant Design Vue basic components. All encapsulated components are uniformly named with hy- as a prefix. The encapsulation follows the principle of "retaining the core, extending the general, and solidifying the logic".

[0074] Retain the core properties of Ant Design Vue native components;

[0075] Extend common business attributes, including unified interface request parameters, data formatting rules, and style theme parameters, supporting global configuration and local overriding;

[0076] Solidify common interaction logic;

[0077] The component uses an on-demand loading mode;

[0078] Components communicate with each other via Props and Events, and support custom event extensions.

[0079] The interface parsing preview is the front-end interactive support for the front-end model generation engine interface preview unit and the APP parsing engine compatibility verification unit, realizing multi-terminal preview and online editing synchronization functions:

[0080] The preview function is implemented based on an iframe, with the iframe's src pointing to an independent preview service, and the preview data generated through a Mock service;

[0081] It supports real-time synchronization between "edit" and "preview". Editing operations are pushed to the preview service in real time via WebSocket, and the preview interface can display the latest effect without refreshing.

[0082] It provides multi-terminal preview switching, and intuitively demonstrates the interface adaptation effect by simulating the viewport size and interaction method of different terminals;

[0083] It supports screenshotting and exporting of the preview interface. Screenshots are implemented using html2canvas, and exported files are stored locally or as a platform file service.

[0084] Furthermore, the data storage layer involved in this platform is a globally shared support layer, providing data storage and retrieval services for all core modules. In response to the different data storage needs of the public transportation industry, a multi-database collaborative storage architecture is adopted. Each database performs its own function and works collaboratively, closely related to the front-end and back-end modules, to ensure the efficiency, reliability and security of data storage, including: relational databases, cache databases and non-relational databases.

[0085] MySQL, a relational database:

[0086] ① Positioning: The main storage database for core business data and metadata models, storing structured business data (such as personnel information, basic vehicle data, route information, approval records, and rule configurations);

[0087] ② Associated Modules: Directly associated with the metadata parsing engine, business flow engine, workflow engine, low-code platform base, and other core modules, providing transaction support and strong consistency guarantees;

[0088] ③ Optimization solution: Adopt a master-slave replication architecture, with the master database responsible for write operations and the slave database responsible for read operations, to improve read-write separation efficiency; use a database sharding and table partitioning strategy for core business tables, splitting them by time or business dimension; enable slow query logs and regularly optimize SQL and indexes; use the InnoDB storage engine, supporting transactions and row-level locks to ensure data consistency.

[0089] Redis, a caching database:

[0090] ① Positioning: Cache storage for frequently accessed data, reducing the pressure on the MySQL database and improving system response speed;

[0091] ② Related Modules: Provides caching support for all core modules. The cached data includes: hotspot line scheduling information, real-time ticket statistics, user permission information, metadata relationship data, and configuration parameters;

[0092] ③ Optimization solution: Adopt a cluster deployment mode to ensure high availability; support data expiration policies to avoid cache avalanche; use caching technology and select appropriate Redis data structures for different data types to improve caching efficiency.

[0093] MongoDB, a NoSQL database:

[0094] ① Positioning: Storage of unstructured and semi-structured data, adapting to the needs of massive, multi-dimensional, and dynamically changing data storage;

[0095] ② Association Module: Associated with the data format conversion unit of the metadata parsing engine, storing raw unstructured data;

[0096] ③ Optimization solution: Adopt a replica set architecture to ensure high data availability; support sharded clusters to handle massive data storage; create appropriate indexes for query scenarios; use the WiredTiger storage engine to improve read / write performance and compression ratio.

[0097] 4) Elasticsearch Search Engine: Built on MySQL, MongoDB and Redis databases in the data storage layer, it collects business data from the three databases in real time through data synchronization tools, builds full-text search indexes, and is specifically used for fast retrieval and multi-dimensional analysis of massive data, forming a "storage-retrieval" collaborative architecture with the three databases.

[0098] ① Positioning: Full-text search and rapid analysis of massive business data, improving data query and analysis efficiency;

[0099] ② Associated Modules: Associated with the visualization reporting tool, operation and maintenance monitoring unit, and business flow engine process monitoring unit, supporting multiple scenarios such as multi-condition retrieval of ticketing data, keyword query of maintenance records, statistical analysis of vehicle operation data, retrieval of system logs, and process trajectory query;

[0100] ③ Optimization Solution: A cluster deployment with at least 3 nodes is adopted. For example, one master node is responsible for cluster management, and two data nodes are responsible for data storage and retrieval. If the business data volume exceeds 10TB, it can be expanded to 3 data nodes + 1 coordinating node to improve concurrent processing capabilities and support horizontal scaling. The index design adopts a reasonable sharding and replication strategy, with the data size of each shard controlled within 50GB to avoid excessively large shards affecting query efficiency. Replication settings: Each shard is configured with 1-2 replicas, distributed across different data nodes, ensuring no data loss in the event of a single node failure, while also improving query concurrency. Dynamic adjustment: Through EIasticsearch's index lifecycle management, cold data shards are automatically shrunk and archived to optimize cluster storage resources and improve query and write performance. The word segmenter is optimized for public transportation industry business scenarios to improve retrieval accuracy. Aggregate analysis and filtering queries are supported to meet complex report statistics needs.

[0101] In another aspect of the present invention, a method for implementing a fully configurable low-code development platform for the public transportation industry is provided, the method comprising:

[0102] S1, Platform Deployment;

[0103] 1. Environment preparation: Set up a Docker+K8s containerized deployment environment and configure basic middleware; initialize network configuration, storage configuration, and security policies;

[0104] 2. Base Deployment: Package each microservice of the low-code platform base into a Docker image, deploy it to the cluster via Kubernetes, configure service relationships and resource quotas; complete the initial configuration of basic services;

[0105] 3. Master Data Integration: Integrate MDM master data in the public transportation sector, import it in batches into a MySQL database using a data import tool, synchronize it to MongoDB to retain the original data, generate a standardized metadata model through a metadata parsing engine, and cache it in Redis;

[0106] 4. Monitoring Deployment: Deploy monitoring components, configure monitoring metrics, alarm rules, and tracing rules to ensure that the platform's operating status is monitorable and traceable.

[0107] S2, Scene Configuration Process;

[0108] 1. Scenario Selection: After logging into the platform, users perform permission verification through the low-code platform base and select the target business scenario from the preset public transportation industry scenario library; the platform loads the basic data model template, business components and process templates corresponding to the scenario through the metadata parsing engine and synchronizes them to the front-end model generation engine;

[0109] 2. Configuration page:

[0110] Users can customize the data table structure and inter-table relationships through the front-end metadata visualization generation engine. The configuration commands are submitted to the data model generation engine and generated after being verified by the model verification unit.

[0111] The metadata parsing engine receives the original configuration, generates a standardized metadata model through the metadata modeling unit, identifies the relationships between tables through the metadata relationship identification unit, and synchronizes it to the front-end model generation engine.

[0112] The front-end model generation engine's associated metadata filtering unit only displays metadata components that have a relationship. Users can design the page layout and configure multi-terminal adaptation rules by dragging and dropping components to arrange them.

[0113] The APP parsing engine receives page configuration data in real time, generates interface rendering data for PC and APP through the interface configuration parsing unit and terminal adaptation conversion unit, and then synchronizes and previews the data after verification by the compatibility verification unit.

[0114] S3, Association Rules:

[0115] Users can configure business rules by selecting appropriate functions from the built-in rule components and built-in functions through the rule configuration and orchestration unit of the visual business rule engine.

[0116] After the rule configuration is submitted, the rule execution unit of the business flow engine performs rule parsing and conflict detection. The conflict detection unit returns conflict information, and the user corrects the rules to generate an executable rule script.

[0117] By linking to the OA workflow, users can select the corresponding template from the workflow template library, adjust nodes, approval roles, and workflow rules through a visual interface, bind the configured business rules, and set the business data processing logic for each stage of the approval process.

[0118] The workflow engine uses a multi-engine collaborative unit to link the metadata model of the metadata parsing engine with the rule scripts of the business flow engine, ensuring the linkage between the approval process and business data and rule logic.

[0119] S4. Publishing Function:

[0120] After the user completes the page and rule configuration, submits a publishing request. The platform uses a metadata parsing engine to convert the configuration information into relational data stored in MySQL, the raw data stored in MongoDB, and the configuration data cached in Redis.

[0121] The metadata parsing engine's data format conversion unit is associated with cleaning the configuration data, removing invalid and duplicate data.

[0122] The platform generates a permission control unit through the low-code platform base, configures the corresponding page permissions, button permissions, data permissions and field permissions, and synchronizes them to various terminal applications;

[0123] Once published, users can preview the final result through the preview module. Online editing and adjustments are supported, and the adjusted configurations are updated in real time.

[0124] Furthermore, after building multiple public transportation industry solution application systems, this platform also includes: S5 system integration and expansion, specifically including:

[0125] S51. System Integration: Through the system integration unit of the low-code platform, it integrates with the existing application systems of the public transportation company, uses the OAuth2.0 authorization protocol to achieve secure authentication, and achieves data interoperability through standardized API interfaces; for example, the OA approval process and maintenance business data are automatically linked. After approval, the business flow engine triggers the generation of maintenance work orders and synchronizes them to the machine management system.

[0126] S52. Requirements Changes and System Iteration: When business requirements change, users can directly modify data models, page layouts, or business rules through the platform; the modified configurations are verified and parsed, and then synchronized to the relevant modules in real time to achieve rapid iteration of system functions; the version management unit of the metadata parsing engine records the configuration change history and supports rollback to historical versions.

[0127] S53. Report Analysis and Data Export: Users can design custom reports based on business data using visual reporting tools, and configure data query conditions, visualization chart types, and refresh frequency; reports support real-time refresh and multi-dimensional drill-down analysis, and can be exported to Excel, PDF, and image formats; report data permissions are linked with platform permissions to ensure data access security.

[0128] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0129] (1) The present invention provides a fully configurable low-code development platform and implementation method for the public transportation industry. Through modular and unitized core engine design, combined with technologies such as database sharding, caching optimization, asynchronous processing, and containerized deployment, it supports complex business functions and massive data processing while ensuring high response speed, low resource consumption, and stable operation of the system. It is customized for the business characteristics of the public transportation industry, covering core scenarios such as vehicle dispatching, maintenance, and ticketing management. It has pre-set industry-specific data models and process templates, which can meet the personalized needs of the industry without a large amount of customized coding. At the same time, it supports flexible expansion and combination of functions, which can adapt to the full life cycle needs of public transportation companies from basic informatization to intelligent upgrading, achieving dual leadership in adaptability and functionality.

[0130] (2) The present invention provides a fully configurable low-code development platform and implementation method for the public transportation industry. The platform supports visual modification of data models, page layouts, business rules and workflows without refactoring the code. The changed configuration takes effect in real time and can quickly respond to business logic adjustments and policy changes in the public transportation industry. Built-in conflict detection, data verification and compatibility verification mechanisms ensure the stability of the system and data consistency during the process of demand changes, and can quickly respond to complex demand changes.

[0131] (3) The present invention provides a fully configurable low-code development platform and implementation method for the public transportation industry. It realizes the transformation of non-relational data into relational data through a metadata parsing engine. Combined with a visualization reporting tool, it realizes real-time statistical analysis and multi-dimensional display of business data, providing accurate data support for public transportation companies' decision-making. At the same time, the platform has a built-in multi-layer security protection mechanism (data encryption, access control, interface anti-scraping, SQL injection protection, etc.) to ensure the security of sensitive data such as ticket revenue and personnel privacy. The comprehensive monitoring and fault tolerance mechanism ensures stable system operation, unaffected business continuity, and guarantees in-depth data value mining and security protection.

[0132] (4) The present invention provides a fully configurable low-code development platform and implementation method for the public transportation industry. Through the four-step process of “selecting a scenario → configuring the page → associating rules → publishing functions” and multi-engine collaboration, zero-code / low-code development is achieved. One interface design supports simultaneous generation on PC and APP terminals, which greatly shortens the delivery cycle of the smart public transportation system. The visual configuration interface, intelligent drag-and-drop filtering, and rich built-in rule components allow business personnel to independently complete the system function configuration and iteration without professional development knowledge, reducing the dependence on developers. The platform provides complete verification, preview, and testing functions to improve delivery quality. The platform base supports cross-system and cross-module business interaction and can be efficiently integrated with enterprise OA, human resources, and equipment management systems to achieve data interoperability and business collaboration. It adopts a microservice architecture and plug-in design to support the dynamic expansion and upgrade of core components. According to the business development needs of public transportation companies, functional modules can be flexibly added or third-party systems can be integrated to reduce system maintenance and upgrade costs and achieve strong system integration and scalability. Attached Figure Description

[0133] Figure 1 This is the business architecture diagram of the fully configurable low-code smart bus development platform for the public transportation industry described in this invention.

[0134] Figure 2 This is a diagram of the front-end platform technology architecture in Embodiment 1 of the present invention.

[0135] Figure 3 This is a business flow component diagram of the development platform in Embodiment 1 of the present invention.

[0136] Figure 4 This is a diagram of the built-in function components of the development platform in Embodiment 1 of the present invention.

[0137] Figure 5 This is a screenshot of the left side of the interface after the development platform has been loaded and configured in Embodiment 1 of the present invention.

[0138] Figure 6 This is an intermediate screenshot of the interface after the development platform has been loaded and configured in Embodiment 1 of the present invention.

[0139] Figure 7 This is a screenshot of the right side of the interface after the development platform has been loaded and configured in Embodiment 1 of the present invention. Detailed Implementation

[0140] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0141] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0142] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they specify the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0143] In the case of no conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0144] Embodiment 1

[0145] Refer to Figure 1 , this embodiment provides a fully configured low-code development platform for the bus industry, named the intelligent bus low-code platform architecture. The platform includes seven modules: a data model generation engine, a front-end model generation engine, an APP parsing engine, a metadata parsing engine, a visual business rule engine, a visual report tool, and a low-code platform base.

[0146] The core goal of the seven modules is to jointly build a "configured, highly adaptable, high-performance, and highly secure" low-code development platform. Through standardized processes and technical collaboration, it quickly generates an intelligent bus system (including sub-systems such as vehicle scheduling, maintenance, and ticketing management) covering the core business scenarios of the bus industry, reducing the development threshold and operation and maintenance costs. Each module achieves the goal through a collaborative mechanism of "data flow - logical linkage - function output": the data model generation engine provides an industry-based data foundation, the metadata parsing engine realizes data standardization and association recognition, the front-end / APP parsing engine completes the generation of multi-terminal interfaces, the visual business rule engine configures business logic, the visual report tool realizes data value mining, and the low-code platform base provides global support, ultimately forming a full-link closed loop of "data - interface - rule - report - operation and maintenance" to ensure the rapid construction, flexible iteration, and stable operation of the intelligent bus system.

[0147] Specifically, the data model generation engine provides the verified original configuration data and business data for the metadata parsing engine;

[0148] Specifically, the metadata parsing engine receives and performs structured processing on the verified raw configuration data and business data, generating a standardized metadata model and structured business data, which are then verified. Simultaneously, this data is synchronized to the visualization business rule engine as the basis for data field association, ensuring consistency between subsequent rule configurations and the data model (ensuring data consistency and accuracy used by subsequent modules). It provides structured metadata and relational data to the front-end model generation engine, visualization business rule engine, and visualization reporting tools, supporting the configuration functions of each module. It also collaborates with the permission management unit of the low-code platform base to achieve permission isolation of the data model; and with the platform's unified data storage layer (not an independent module, composed of MySQL, MongoDB, Redis, and Elasticsearch, with operation and maintenance monitoring handled by the low-code platform base) to achieve storage and retrieval of different types of data.

[0149] The data field association foundation refers to the standardized data fields (including field types, constraints, and inter-table relationships) output by the data model generation engine, which are completely consistent with the structured metadata and relational data received by the visualization business rule engine. This foundation supports the visualization business rule engine in binding business rules with specific data fields. For example, the approval rules for maintenance processes can be directly associated with data fields such as "maintenance record status" and "approver role," ensuring that the target data can be accurately read, judged, and modified when the rules are executed.

[0150] Specifically, the front-end model generation engine generates interface configuration data (including component types, attributes, layout structure, and interaction logic) based on the standardized metadata model output by the metadata parsing engine and combined with user page layout configurations (such as component drag-and-drop and multi-terminal adaptation rules). Simultaneously, it receives metadata relationship data from the metadata parsing engine and uses a greedy algorithm to filter out metadata components directly related to the current configuration scenario (e.g., when configuring a maintenance page, it automatically filters out related components such as vehicle information, maintenance items, and approvers), while filtering out irrelevant components (such as ticketing settlement fields). The filtered components allow users to quickly drag and drop to design page layouts, avoiding operational confusion caused by too many components, while ensuring the correlation between page components and the data model, reducing configuration errors. The front-end model generation engine also works in conjunction with the theme setting unit of the low-code platform base to achieve a unified interface style.

[0151] The original configuration data consists of the basic configuration information of the data model (such as table structure and field constraints), while the interface configuration data is the front-end display and interaction configuration data derived from the standardized metadata model. The two are linked through a metadata parsing engine to ensure that the data displayed on the interface is consistent with the underlying data model and to collaboratively achieve synchronous generation of interfaces on multiple terminals.

[0152] Specifically, the APP parsing engine and the front-end model generation engine work together to achieve "one-time configuration, multi-terminal generation", receiving permission configuration information from the low-code platform base to ensure that the permission control on the APP side is consistent with that on the PC side.

[0153] Specifically, the visualization business rule engine receives structured metadata and relational data from the metadata parsing engine. By associating metadata fields with business data, business rules can be accurately applied to business data, thereby automating the execution of business logic and realizing business processes (such as automatic settlement of ticketing data according to rules, automatic review of maintenance processes according to rules, and automatic reminders of scheduling plans according to rules).

[0154] Business data originates from operational data generated during the operation of the intelligent public transportation system (such as ticket transaction data, maintenance record data, and dispatch execution data, which are stored in the data storage layer using data structures defined by the data model generation engine). Business rules refer to logical rules set based on the business needs of the public transportation industry (such as maintenance work order approval conditions, ticket revenue calculation standards, and dispatch plan reminder thresholds), which are configured by the visual business rule engine to generate executable scripts. The visual business rule engine, also known as the business flow engine, is a process execution engine developed based on LiteFIow and is responsible for parsing and executing business rules. The OA workflow engine of the visual business rule engine is an approval process engine developed based on FIowabIe and integrated into the low-code platform base, responsible for the management of approval nodes and flow rules. Approval processes are a subcategory of business processes, specifically referring to processes requiring multi-role approval and confirmation (such as repair cost approval and route adjustment approval). Business processes also include automated processes that do not require approval (such as real-time ticketing data statistics and vehicle operation status monitoring). The two are supported by different engines and can be flexibly combined and configured independently. The business flow engine works in conjunction with the OA workflow engine of the low-code platform to bind approval processes with business rules (such as approval nodes triggering corresponding business rule verification). The OA workflow engine and the business flow engine collaborate to provide executable rule scripts for business processes, ensuring that business processes proceed automatically according to preset rules (such as repair processes automatically matching repair personnel and scheduling processes automatically avoiding conflicting routes).

[0155] Specifically, the visualization reporting tool receives structured business data output by the metadata parsing engine (the metadata parsing engine converts unstructured / semi-structured business data in the data storage layer into structured data through a data format conversion unit before outputting it), supports report data statistics, and the report data is statistical analysis data of structured business data, including quantitative indicator data (such as total ticket revenue and vehicle maintenance frequency), trend change data (such as monthly scheduling efficiency changes and quarterly failure rate trends), etc., which are generated based on fields defined by the metadata model and adapted to visualization display formats such as tables and charts; it also links with the data permission unit of the low-code platform base to achieve permission isolation of report data; and it collaborates with Elasticsearch in the data storage layer to improve the efficiency of querying and analyzing massive amounts of data.

[0156] Among them, structured business data refers to business data generated by public transportation; structured metadata refers to descriptive data about interfaces, rules, and business processes.

[0157] Specifically, the low-code platform foundation provides access control, security protection, and configuration management support for all core modules: Access control examples (e.g., restricting ordinary users to only view ticketing statistics, while administrators can modify ticketing calculation rules); security protection examples (e.g., storing ticketing revenue data or other data with AES-256 encryption to intercept SQL injection attack requests); configuration management examples (e.g., dynamically adjusting report refresh frequency and modifying parameter configurations of business rule components via Nacos); integration with external systems is achieved through a system integration unit; collaborative work between core modules is supported through an event-driven architecture and RocketMQ message queue technology, with modules transmitting data and instructions through standardized message formats (e.g., after the data model generation engine completes configuration, it notifies the metadata parsing engine to process data via a message queue; after the visualization business rule engine generates rule scripts, they are synchronized to the business flow engine via a message queue), ensuring efficient collaboration and reliable data; and the operation and maintenance monitoring unit monitors the operating status of each module in real time to ensure system stability.

[0158] Furthermore, the data model generation engine includes: a table structure configuration unit, a table relationship definition unit, an industry template library unit, and a model validation unit. The data model generation engine achieves its overall function through a logical flow of 'template selection / custom configuration → relationship definition → model validation → output configuration,' with the four core units working collaboratively to support the output of original configuration data and the goal of permission isolation. The logic and technical means of the linkage between each unit are as follows:

[0159] Industry Template Library Unit: As the basic entry point for data model configuration, it provides pre-set data model templates specific to the public transportation industry, such as vehicle information tables, route scheduling tables, and ticketing data tables. It supports template import, modification, and saving (template data is encapsulated in XML format for easy cross-system reuse). Users can directly select templates or modify them based on existing templates, reducing redundant configuration.

[0160] Table structure configuration unit: It takes over the template data of the industry template library unit or user-defined requirements. Based on the visual form interface, it supports user-defined field types (including public transportation industry-specific field types such as "license plate number format" and "route numbering rules"), lengths, and constraints (non-empty, unique, and regular expression validation). It uses JSON Schema format to store configuration information to ensure standardized structure.

[0161] Inter-table relationship definition unit: Based on the single-table structure determined by the table structure configuration unit, it provides a visual drag-and-drop interface, supports one-to-one, one-to-many, and many-to-many relationship configuration, has a built-in relationship verification algorithm, automatically detects anomalies such as circular references and field mismatches, and builds a complete data model association system.

[0162] Model Validation Unit: Integrates data type validation, constraint validation, and relationship rationality validation logic. It performs comprehensive validation of table structure configuration and inter-table relationship definition results. The validation results are fed back to the front end in real time. It supports one-click repair of common configuration errors and ensures that the output original configuration data is accurate and usable.

[0163] The four units work together in the order of "industry template library unit → table structure configuration unit → inter-table relationship definition unit → model verification unit" to generate raw configuration data that meets the requirements. This provides input for the metadata parsing engine, provides the data field association basis for the visualization business rule engine, and achieves permission isolation of the data model through linkage with the low-code platform base permission management unit.

[0164] The front-end model generation engine includes: a metadata filtering unit, a component drag-and-drop orchestration unit, a multi-terminal adaptation unit, and an interface preview unit. Through a logical flow of metadata filtering → component orchestration → multi-terminal adaptation → preview verification, the four core units collaboratively achieve the goals of interface configuration data generation and multi-terminal adaptation. The logic and technical means of the interconnection between each unit are as follows:

[0165] Related Metadata Filtering Unit: As a pre-configuration step, it uses RPC to call the metadata relationship identification results of the metadata parsing engine, and employs a greedy algorithm to filter metadata with related relationships in the current configuration scenario, filtering irrelevant components (such as when configuring the maintenance page, only filtering related metadata such as vehicles and maintenance items), reducing the error rate of operation, and providing accurate materials for component orchestration.

[0166] Component drag-and-drop orchestration unit: This unit receives metadata components from the associated metadata filtering unit. It is based on VueDragBarrier components and supports component position adjustment and nested levels. It uses the DOM Diff algorithm to optimize the rendering efficiency of the interface after dragging and dropping, avoiding frequent refreshes. Users can complete the page layout design by dragging and dropping.

[0167] Among them, business flow components such as Figure 3 As shown, it includes, but is not limited to, query components, add components, delete components, conditional components, modify components, message components, loop components, site messages, unrelated queries, and native code call components. Modification methods include copying and pasting.

[0168] Built-in functional components such as Figure 4 As shown, including but not limited to:

[0169] Logical functions: AND (conditional AND operation), OR (conditional OR operation);

[0170] Comparison functions: greaterThan (greater than), lessThan (less than), equals (equal to), greaterThanOrEqual (greater than or equal to), lessThanOrEqual (to or equal to) (ternary operator);

[0171] Date functions: computationDate (add / subtract date), temporalComparison (compare the interval between two dates), getCurrentDateTime (format time string);

[0172] Character functions: empty (empty check), size (array length).

[0173] Multi-terminal adaptation unit: Based on the page layout determined by the component drag-and-drop orchestration unit, it has a built-in responsive layout rule library and automatically adjusts the component size and layout structure according to the screen size of PC and APP (Android / iOS). It adopts a combination of media query and flexible layout to ensure consistent display across multiple terminals.

[0174] Interface Preview Unit: It receives the interface configuration data after multi-terminal adaptation, integrates iframe embedding technology, loads the configured interface effect in real time, and supports online editing and preview synchronization (the preview data is simulated through Mock service and does not occupy real business data resources), which makes it convenient for users to make timely adjustments.

[0175] The four units work together in the order of "Associated Metadata Filtering Unit → Component Drag and Drop Arrangement Unit → Multi-Terminal Adaptation Unit → Interface Preview Unit" to generate interface configuration data. This data is shared with the APP parsing engine to achieve synchronous generation across multiple terminals. At the same time, it receives metadata relationship data from the metadata parsing engine to support intelligent component filtering and works in conjunction with the theme setting unit of the low-code platform base to achieve a unified interface style.

[0176] The interface effect after loading the configuration is as follows: Figure 5 , 6 As shown in Figures 7 from left to right, this diagram presents the core components and operational functions of the platform's visual configuration interface.

[0177] The left side of the interface after loading the configuration is as follows Figure 5 As shown, this is the component category area, which includes input components (single-line input boxes, multi-line input boxes, number input boxes, etc.), selection components (drop-down selections, radio button groups, checkbox groups, etc.), layout components, and OA components (approval comments, signatures, handwritten signature feedback, etc.).

[0178] The middle of the interface after loading the configuration is as follows Figure 6 As shown, this is the operation function area (in this embodiment, it is a vehicle and tire equipment template), which covers operations such as template setting, importing configuration, viewing configuration, previewing, generating JSON, and restoring. It also includes page information display and quick access to commonly used components, intuitively presenting the entire process of interface configuration. In this embodiment, the template is set as a vehicle and tire equipment template, including but not limited to: primary key ID, vehicle number, license plate number, number of locked equipment, self-number, number of already equipped, tire number, tire brand, tire specifications, rated mileage, etc.

[0179] The right side of the interface after loading the configuration is as follows Figure 7 As shown, this is the component attribute configuration area, where you can set details such as label alignment, label, label alignment, app hiding, field selection, placeholder hints, reminder content, form grid, label width, and default values.

[0180] Furthermore, the APP parsing engine includes: an interface configuration parsing unit, a terminal adaptation and conversion unit, an interaction logic synchronization unit, and a compatibility verification unit. The APP parsing engine achieves the goal of 'one-time configuration, multi-terminal generation' through a logical flow of "parsing configuration data → terminal adaptation and conversion → synchronizing interaction logic → compatibility verification," with the four core units working collaboratively. The logic and technical means of the interconnection between each unit are as follows:

[0181] Interface configuration parsing unit: As the data processing entry point, it receives interface configuration data shared by the front-end model generation engine, uses a JSON parser to extract component type, attributes, and layout information, and generates terminal adaptation intermediate files to provide basic data for subsequent adaptation and conversion;

[0182] Terminal Adaptation and Conversion Unit: This unit receives the intermediate files output by the interface configuration parsing unit and performs differentiated conversions for Android and iOS systems. On Android, the intermediate files are converted into XML format rendering data, while on iOS, they are converted into Storyboard format rendering data. Cross-platform adaptation plugins ensure consistent interaction logic.

[0183] Interaction Logic Synchronization Unit: Based on the rendering data of the terminal adaptation and conversion unit, the interaction logic configured on the PC, such as click, submit, and filter, is synchronized to the APP through an event mapping mechanism. A unified event triggering protocol (such as a custom 'bus-event' protocol) is adopted to avoid differences in multi-terminal interaction.

[0184] Compatibility verification unit: Verifies the rendered data after terminal adaptation and conversion. It has a built-in compatibility rule library for mainstream Android (9.0 and above) and iOS (13.0 and above) versions, automatically detects component adaptation anomalies (such as misaligned buttons or unresponsive interactions), outputs a compatibility report and supports one-click repair, ensuring stable operation on multiple terminals.

[0185] The four units work together in the order of 'interface configuration parsing unit → terminal adaptation conversion unit → interaction logic synchronization unit → compatibility verification unit', sharing interface configuration data with the front-end model generation engine and receiving permission configuration information from the low-code platform base to achieve multi-terminal synchronous generation function with consistent permissions between the APP and PC.

[0186] Furthermore, the metadata parsing engine (self-developed core) includes: a metadata modeling unit, a data format conversion unit, a data manipulation unit, a metadata relationship identification unit, and an import / export unit. As the platform's core data processing engine, the metadata parsing engine, through a logical flow of 'modeling → format conversion → data manipulation → relationship identification → import / export,' collaboratively achieves standardized metadata generation and data management functions through the collaboration of these five core units. The interconnected logic and technical means of each unit are as follows:

[0187] Metadata Modeling Unit: Based on the business characteristics of the public transportation industry, the metadata model structure is designed. It adopts object-oriented thinking to encapsulate metadata attributes (field names, types, relationships, constraints), supports metadata version management (Git-like branch management mechanism), receives the raw configuration data of the data model generation engine, generates the basic metadata model, and provides structural support for subsequent data processing. The basic metadata model is the preliminary stage of the standardized metadata model.

[0188] Data Format Conversion Unit: This unit builds upon the basic metadata model of the metadata modeling unit. It uses the Jackson + Fastjson dual parsing framework to convert unstructured (JSON) and semi-structured (CSV) metadata and business data into relational data that conforms to MySQL table structure specifications and performs data cleaning. It ensures the accuracy of the conversion through a data type mapping table, supports custom mapping rules, and synchronizes the converted data to the data storage layer.

[0189] Data Operation Unit: Based on the transformed structured data, it has built-in CRUD core logic, optimizes query performance for massive data scenarios in the public transportation industry, supports multi-condition combined queries, relational queries, and paginated queries. The query SQL is dynamically generated by MyBatis-Plus, and adopts a database sharding and table partitioning + read-write separation architecture to improve processing efficiency and provide data read and write services for other modules.

[0190] Metadata Relationship Identification Unit: Based on the structured data processed by the data operation unit, a graph theory algorithm is used to construct a metadata relationship graph. The relationship chain is traversed by depth-first search (DFS) to identify one-to-one, one-to-many, and many-to-many relationships between tables, and output structured relationship data to support intelligent filtering and rule configuration of front-end components.

[0191] Import / Export Unit: This unit receives structured data from the data manipulation unit and supports importing and exporting custom templates in Excel and CSV formats. During import, it uses POI to parse the file and matches data using field mapping rules, with built-in data validation rules. During export, it uses SXSSF streaming to process large data files, avoiding memory overflow and enabling cross-system data reuse and backup. SXSSF (Streaming Usermodel API) is a streaming processing solution provided by Apache POI. It effectively avoids OOM (Out of Memory) errors by retaining only a portion of the data in memory (rows exceeding a threshold are written to a temporary file), making it particularly suitable for exporting hundreds of thousands or even millions of data entries.

[0192] The five units work together in the order of 'metadata modeling unit → data format conversion unit → data operation unit → metadata relationship identification unit → import and export unit' to provide validation support for the data model generation engine, provide structured metadata and relational data for the front-end model generation engine, visualization business rule engine, and visualization reporting tools, and work with the data storage layer to achieve full lifecycle management of data.

[0193] Furthermore, the visual business rule engine includes: a rule component management unit, a function library unit, a rule configuration and orchestration unit, a rule execution unit, and a conflict detection unit. The visual business rule engine achieves business rule configuration and execution functions through the logical flow of 'component / function provision → rule orchestration → conflict detection → rule execution', with these five core units working collaboratively. The logic and technical means of the interconnection between each unit are as follows:

[0194] Rule Component Management Unit: As the basic support for rule configuration, it has 10 built-in core rule components (specifically including encoding rules, calculation rules, reminder rules, business verification rules, audit rules, data filtering rules, process jump rules, access control rules, message push rules, and timed trigger rules). It adopts a plug-in architecture design, supports dynamic registration and uninstallation of components, and stores component configuration data in JSON format. The output is an optional rule component library, which is linked with the rule configuration orchestration unit and can be dragged and dropped by users for combination.

[0195] Function Library Unit: Provides logical calculation support for rule configuration, integrates nearly a hundred built-in functions (including comparison functions, date functions, character functions, metadata functions, mathematical calculation functions, data conversion functions, etc.), supports the uploading and calling of custom functions, and the function execution adopts a reflection mechanism and ensures security through parameter verification; the output is an optional function library, which is linked with the rule configuration orchestration unit to improve the rule logic (such as setting reminder time through date functions and setting review conditions through comparison functions).

[0196] Rule Configuration and Orchestration Unit: This unit integrates rule components and functions, and is a visual flowchart editor based on antv / x6. It supports drag-and-drop combination of rule components and conditional branch settings (conditional branches are implemented by dragging and dropping rule components and functions; for example, "Repair cost > 1000 yuan" requires dragging and dropping the "Comparison Function" and "Audit Rule" components for linked configuration). It uses the BPMN 2.0 specification to describe rule processes and automatically generates executable rule scripts. The input is the filtered rule components and functions, and the output is the rule script. The rule script is a form of business rule and is linked with the conflict detection unit.

[0197] Conflict Detection Unit: Performs logical verification on the rule scripts output by the rule configuration orchestration unit, and uses logical reasoning algorithms to detect condition conflicts (such as "repair costs are required to be >1000 and <500 at the same time") and priority conflicts (such as two rules having the same priority and the same triggering conditions). It provides real-time feedback on the conflict location and cause and offers solution suggestions. The input is the rule script, and the output is the verification result (pass / conflict and correction suggestions). After the verification is passed, the rule script is synchronized to the rule execution unit.

[0198] Rule Execution Unit: Based on the LiteFlow rule engine kernel, it parses and executes validated rule scripts, supporting synchronous / asynchronous execution modes (asynchronous execution uses RocketMQ message queues for decoupling), improving concurrent processing capabilities; the input is validated rule scripts and business data (business data is the data input by the customer), and the output is the rule execution result (such as approval / rejection, reminder message sending, data update instructions), which works in conjunction with the business flow engine and OA workflow engine to support the advancement of business processes.

[0199] The five units work together in the order of "rule component management unit → function library unit → rule configuration and orchestration unit → conflict detection unit → rule execution unit". They receive structured metadata and relational data from the metadata parsing engine, support the association between rules and business data, link with the OA workflow engine of the low-code platform to realize the binding of approval processes, and collaborate with the business flow engine to provide rule execution capabilities for business processes.

[0200] Furthermore, the visualization reporting tool includes: a report template design unit, a data query unit, a visualization rendering unit, a report export unit, and a real-time refresh unit. The visualization reporting tool, through a logical flow of "template design → data query → visualization rendering → real-time refresh → export," utilizes five core units to collaboratively implement report configuration, generation, and export functions. The correspondence between each unit and the overall description above, as well as the technical means employed, are as follows:

[0201] Report Template Design Unit: Corresponding to the basic configuration section "Supporting Report Data Statistics" mentioned earlier, it provides a visual drag-and-drop interface, supports the configuration of table and chart (line chart, bar chart, pie chart, radar chart, etc.) components, uses XML format to store the template structure, and supports template saving and reuse; the output is a report template, which is linked with the data query unit to clarify the display structure of the report data.

[0202] Data Query Unit: Corresponding to the core step of "receiving structured business data from the metadata parsing engine" mentioned earlier, it connects to databases such as MySQL and Elasticsearch via JDBC, supports custom SQL queries and multi-data source association queries, and allows query conditions to be configured through a visual interface. It automatically generates optimized query statements and adopts a query result caching mechanism to improve efficiency. The input is the query requirements corresponding to the report template and the structured business data from the metadata parsing engine, and the output is the query result data, which is linked with the visual rendering unit.

[0203] Visualization Rendering Unit: Corresponding to the core display function of the "Visual Reporting Tool" mentioned above, it encapsulates chart components based on antv / x6 and ECharts, supports real-time data rendering and interactive operations (zooming, filtering, drill-down), and adopts a Canvas+SVG hybrid rendering scheme to balance efficiency and visual effects; the input is query result data, and the output is a visual report interface, which is linked with the real-time refresh unit and the report export unit.

[0204] Real-time refresh unit: Corresponding to the "Supporting dynamic report updates" function mentioned earlier, it supports setting the refresh frequency (minimum 1 minute), pushes the latest data to the report interface via WebSocket, and adopts an incremental data update mechanism to refresh only the changed data; the input is the updated data in the data storage layer, and the output is the incrementally updated visual report to ensure the real-time nature of the report data.

[0205] Report Export Unit: Corresponding to the "Report Data Export" function mentioned above, it supports exporting in Excel, PDF, and image formats. Excel export uses POI streaming processing, PDF export is based on iText7, and image export uses HtmI2Canvas. It supports batch export and scheduled export. The input is the visually rendered report data, and the output is the exported file for users to download and back up or use across systems.

[0206] Each unit works together to support the overall functionality of the visualization reporting tool, receives structured business data from the metadata parsing engine, links with the data permission unit of the low-code platform base to achieve permission isolation, and collaborates with EIasticsearch to improve the efficiency of querying massive amounts of data.

[0207] Furthermore, the low-code platform foundation provides access control, security protection, and configuration management support for all core modules, including: access control unit, system integration unit, cross-module interaction unit, security protection unit, operation and maintenance monitoring unit, and configuration management unit; the technical means of each unit are as follows:

[0208] Access Control Unit: Employs a hybrid model of RBAC (Role-Based Access Control) and ABAC (Attribute-Based Access Control) to achieve precise control over page permissions, button permissions, data permissions, and field permissions. Permission configuration data is stored in MySQL, and frequently accessed permission data is cached in Redis to improve permission verification efficiency.

[0209] System Integration Unit: Provides standardized API interfaces (RESTful, SOAP) to support integration with enterprise OA systems, human resources systems, aircraft management systems, etc. It adopts the OAuth2.0 authorization protocol to achieve secure authentication and resolves interface differences between different systems through an interface adaptation layer.

[0210] Cross-module interaction unit: Based on an event-driven architecture, it uses RocketMQ to realize cross-module and cross-system message communication, supports synchronous / asynchronous message passing, and has a built-in message retry mechanism and dead letter queue to ensure reliable message transmission.

[0211] Security Protection Unit: Integrates security mechanisms such as data encryption (AES-256), interface anti-scraping (rate limiting and circuit breaking), SQL injection protection (parameterized query), and XSS protection (input filtering). It adopts the Spring Security framework to achieve overall security management and supports security log auditing.

[0212] Operation and Maintenance Monitoring Unit: Integrates Prometheus+Grafana monitoring solution to collect system operation indicators in real time (interface response time, data processing throughput, error rate, server resource usage), supports custom monitoring thresholds and alarm rules, and pushes alarm information through multiple channels such as SMS, email, and in-site messages.

[0213] Configuration Management Unit: Based on Nacos Configuration Center, it enables dynamic management of system parameters, rule configurations, and theme settings. It supports canary releases and rollbacks of configurations, and configuration changes are synchronized to all modules through a message notification mechanism without requiring a system restart.

[0214] This platform is built around seven core modules: a data model generation engine, a front-end model generation engine, an APP parsing engine, a metadata parsing engine, a visual business rules engine, a visual reporting tool, and a low-code platform base. These modules enable the construction of public transportation industry solution application systems, which are the final products. These include, but are not limited to, subdivided business systems such as human resources systems and aircraft maintenance management systems. Through the low-code platform base, data interoperability and business collaboration are achieved, directly serving the daily operations of public transportation companies.

[0215] Furthermore, this platform also includes a front-end engine architecture: which is a functional extension of the core front-end modules, including a front-end metadata visualization generation engine (front-end interactive support for the data model generation engine), a front-end PC parsing engine (PC-side interactive part of the front-end model generation engine), a front-end APP parsing engine (front-end interactive part of the APP parsing engine), and an OA workflow engine (system integration unit association engine of the low-code platform base), which directly links with the front-end model generation engine and the APP parsing engine.

[0216] Specifically, the following front-end technology descriptions supplement the technical implementation details of the front-end model generation engine and APP parsing engine among the seven core modules. The front-end platform technical architecture is as follows: Figure 2 As shown, the front-end technical architecture is broken down in detail, divided into four layers from top to bottom: display layer, business layer, core functional modules, and core technology foundation.

[0217] Display layer: Covers multi-terminal deployment scenarios, including PC, client (H5), Android, and iOS, ensuring platform adaptability across different terminals.

[0218] Business layer: Focuses on core business scenarios, covering core business needs such as list display, data management, document entry, and OA process initiation.

[0219] The core functional modules are the core support for the front-end engine architecture development, including the page template module (preset four templates: listPage, treeAndTable, formTable, and tabFormTable), the enhancement plugin module (integrating plugins such as vue3-highlightjs, vuedraggable, and antv / x6), secondary encapsulation components (components prefixed with hy-, such as hy-form and hy-table), and general UI components. It also has the ability to parse and render interfaces and configure interface rendering.

[0220] Core technology foundation: The front-end technology stack is clearly defined as Vue3 + Ant Design, Vue + Vite + Less, providing basic technical support for each front-end module.

[0221] Specifically, it clarifies how the front-end will support the implementation of core functional modules:

[0222] 1. Page Module Design: This section provides page template support for the front-end model generation engine. Four standardized page templates are directly available as optional templates for the front-end model generation engine, supporting rapid page configuration.

[0223] The page module offers four pre-set standardized page templates, covering mainstream business scenarios, and each template has a built-in performance optimization mechanism:

[0224] listPage is a basic table page: based on the Ant Design Vue Table component, it includes built-in common logic such as pagination, filtering, sorting, and data rendering. It uses virtual scrolling technology (vue-virtual-scroller) to handle large lists of data, avoiding page lag caused by too many DOM nodes. It supports dynamic display / hiding of table columns and uses a partial refresh mechanism to update only the data in the changed columns.

[0225] The treeAndTable page integrates the Tree and Table components from Ant Design Vue to enable the linkage between tree-like categories and table-like data. The Tree component uses a lazy loading mode, loading only the child node data of the currently expanded node. When the table data is linked with the Tree node, an event delegation mechanism is used to reduce the number of event bindings and improve the linkage response speed.

[0226] The formTable basic form page is a wrapper around the Form component of Ant Design Vue, integrating common logic such as form validation, data submission, and reset. Form validation adopts a combination of front-end real-time validation and back-end asynchronous validation to reduce invalid submissions; data submission uses debouncing (default 300ms) to avoid duplicate submissions; form data is cached locally and supports data recovery after page refresh.

[0227] tabFormTable tab form page: Based on the basic form page, it integrates the Ant Design Vue Tabs component to realize the switching of multi-tab forms and independent data management. Tab switching adopts a lazy loading mode, only initializing the form component of the currently active tab; the form data of each tab is stored independently and isolated by namespace to avoid data conflicts.

[0228] 2. Page module integration of core function plugins: This part provides technical support for the drag-and-drop arrangement unit and interface preview unit of the front-end model generation engine components. The plugin functions are directly integrated into the component library and preview module to improve configuration efficiency and preview experience.

[0229] This code uses vue3-highlightjs to implement syntax highlighting for JSON code preview, supports code folding and copying, and improves preview performance for large amounts of JSON data by rendering code snippets through a virtual DOM.

[0230] VueDragGable is used to implement drag-and-drop arrangement of page elements, supporting drag boundary restrictions and snapping alignment. During the drag process, CSS3 transform is used to replace DOM operations, reducing reflow and repaint, and improving drag smoothness.

[0231] It uses the antv / x6 visualization chart library to graphically display table structure and relationships, supports chart zooming, panning, and node dragging, and uses Canvas to render charts to reduce memory usage; it also supports exporting chart data (PNG / SVG format) for easy document organization.

[0232] It uses sortablejs, a lightweight JavaScript library, to implement draggable sorting functionality, supporting multiple input methods such as mouse, touchscreen, and keyboard.

[0233] 3. Component parameter configuration: This part describes the implementation method of secondary encapsulation of components in the front-end model generation engine. The encapsulated hy-series components are directly used as optional components in the component drag-and-drop orchestration unit, ensuring component compatibility and ease of use.

[0234] The component parameter configuration performs a secondary encapsulation of the basic Ant Design Vue components. All encapsulated components are uniformly named with the prefix "hy-" (such as hy-table, hy-form, hy-tree, hy-select, hy-title, etc.). The encapsulation follows the principle of "preserving the core, extending the general, and solidifying the logic".

[0235] (1) Retain the core properties of Ant Design Vue native components to ensure component functionality compatibility;

[0236] (2) Extend common business attributes, including unified interface request parameters (such as request URL, request method, timeout), data formatting rules (such as date formatting, number thousands display), style theme parameters (such as color, font size), and support global configuration and local overriding;

[0237] (3) Solidify common interaction logic, such as form submission debouncing, lazy loading of table data, display of component loading status, and exception handling (network error, empty data), to reduce redundant development;

[0238] (4) The components adopt an on-demand loading mode (based on Vite's Tree Shaking), only importing the components used in the current page to reduce the package size; the components communicate with each other through Props and Events, and support custom event extensions.

[0239] 4. Interface Parsing Preview: This part provides front-end interaction support for the front-end model generation engine interface preview unit and the APP parsing engine compatibility verification unit, enabling multi-terminal preview and online editing synchronization.

[0240] Developers can directly preview the pages, forms, and other content they have written within the platform, and edit and adjust them online without needing to download additional files.

[0241] The preview function is implemented based on iframe. The src of the iframe points to an independent preview service. The preview data is generated through a mock service and does not depend on the real business database, thus avoiding data pollution.

[0242] It supports real-time synchronization between "edit" and "preview". Editing operations are pushed to the preview service in real time via WebSocket, and the preview interface can display the latest effect without refreshing.

[0243] It provides multi-terminal preview switching (PC and APP), and intuitively demonstrates the interface adaptation effect by simulating the viewport size and interaction method of different terminals;

[0244] It supports screenshotting and exporting of the preview interface. Screenshots are implemented using html2canvas, and exported files are stored locally or as a platform file service.

[0245] This platform also includes a backend engine architecture: which is a functional mapping of the backend related core modules, including a report engine (corresponding to a visual report tool), a data modeling and generation engine (corresponding to a data model generation engine), a metadata parsing engine (a core self-developed module), a business flow engine (corresponding to a visual business rules engine), and an OA workflow engine (corresponding to a system integration unit association engine of the low-code platform base). Each engine is the backend functional carrier of the seven major modules, and data and logic linkage are realized through cross-module interaction units.

[0246] The following core engine technology implementation details the technical implementation of the key backend engines (metadata parsing engine, visualization business rules engine, and association engine) among the seven core modules. It is consistent with the composition units of the seven modules, only the focus of the description is different: the seven modules focus on unit division, functional positioning, and module linkage; this part focuses on the specific technical implementation details of each unit, data processing flow, and performance optimization methods. The two complement each other and fully present the engine functions.

[0247] Basic services: This is the general functional support layer for the seven modules, covering theme settings (interface style configuration associated with the front-end model generation engine and APP parsing engine), message template settings (reminder rule unit associated with the visual business rule engine), organization / role / permission management (permission control unit associated with the low-code platform base), etc., providing standardized basic functions for the core modules and reducing redundant development.

[0248] Platform Base: Fully corresponding to the "Low-code Platform Base" among the seven modules, it is the foundation for the entire platform's operation. It integrates core support capabilities such as MDM master data, authentication services, and data security services, providing global support for all core modules, including access control, security protection, and configuration management.

[0249] The front end of this platform refers to the user interface layer (including the configuration and usage interface of all functional modules), and the back end refers to the server processing layer (including the microservice cluster of the low-code platform base).

[0250] Front-end and back-end connection: The front-end (user interface layer) obtains data such as permission configuration and system parameters through HTTP requests, and the back-end (microservice unit of low-code platform base) processes and returns the results; cross-module interaction is realized through RocketMQ message queue, without the need for direct calls between modules; security protection mechanism runs through the entire process of front-end and back-end data transmission and processing, and operation and maintenance monitoring data is collected through Prometheus and visualized to users through Grafana.

[0251] This invention also includes a data storage technology solution:

[0252] The data storage layer of this platform is a globally shared support layer, not part of the seven core modules. It consists of multiple databases working together to provide data storage and retrieval services for all core modules. To address the different data storage needs of the public transportation industry, a multi-database collaborative storage architecture is adopted. Each database performs its specific function and works collaboratively, closely integrated with the front-end and back-end modules to ensure efficient, reliable, and secure data storage.

[0253] Relational database (MySQL):

[0254] ① Positioning: The main storage database for core business data and metadata models, storing structured business data (such as personnel information, basic vehicle data, route information, approval records, and rule configurations);

[0255] ② Associated Modules: Directly associated with core modules such as the metadata parsing engine, business flow engine, workflow engine, and low-code platform base, providing transaction support and strong consistency guarantees;

[0256] ③ Optimization solution: Adopt a master-slave replication architecture (one master and multiple slaves), with the master database responsible for write operations and the slave databases responsible for read operations, to improve read-write separation efficiency; core business tables adopt a database sharding and table partitioning strategy (Sharding-JDBC), splitting them by time or business dimension; enable slow query logs and regularly optimize SQL and indexes; use the InnoDB storage engine, supporting transactions and row-level locks to ensure data consistency.

[0257] Cache database (Redis):

[0258] ① Positioning: Cache storage for frequently accessed data, reducing the pressure on the MySQL database and improving system response speed;

[0259] ② Related Modules: Provides caching support for all core modules. The cached data includes hotspot line scheduling information, real-time ticket statistics, user permission information, metadata relationship data, configuration parameters, etc.

[0260] ③ Optimization solution: Adopt a cluster deployment mode (master-slave + sentinel) to ensure high availability; support data expiration policies (such as TTL expiration, lazy deletion) to avoid cache avalanche; adopt technologies such as cache preheating, cache degradation, and cache penetration protection (Bloom filter); select appropriate Redis data structures (such as strings, hashes, lists, sorted sets) for different data types to improve cache efficiency.

[0261] No relational database (MongoDB):

[0262] ① Positioning: Storage of unstructured and semi-structured data, adapting to the needs of massive, multi-dimensional, and dynamically changing data storage;

[0263] ② Association Module: Associated with the data format conversion unit of the metadata parsing engine, storing raw unstructured data (such as vehicle operation trajectory data, real-time monitoring logs, user operation behavior data, and importing / exporting raw files).

[0264] ③ Optimization solution: Adopt a replica set architecture to ensure high data availability; support sharded clusters to handle massive data storage; create appropriate indexes for query scenarios (such as geospatial indexes and text indexes); and use the WiredTiger storage engine to improve read / write performance and compression ratio.

[0265] The search engine (Elasticsearch) is built on MySQL, MongoDB, and Redis databases in the data storage layer. It collects business data (such as ticketing data, maintenance records, and scheduling information) from the three databases in real time through data synchronization tools (such as Logstash and CanaI), builds a full-text search index, and is specifically used for the rapid retrieval and multi-dimensional analysis of massive data, forming a "storage-retrieval" collaborative architecture with the three databases.

[0266] ① Positioning: Full-text search and rapid analysis of massive business data, improving data query and analysis efficiency;

[0267] ② Associated Modules: Associated with the visualization reporting tool, operation and maintenance monitoring unit, and business flow engine process monitoring unit, supporting scenarios such as multi-condition retrieval of ticketing data, keyword query of maintenance records, statistical analysis of vehicle operation data, retrieval of system logs, and process trajectory query;

[0268] ③ Optimization Solution: A cluster deployment is adopted (at least 3 nodes, e.g., 1 master node for cluster management and 2 data nodes for data storage and retrieval; if the business data volume exceeds 10TB, it can be expanded to 3 data nodes + 1 coordinating node to improve concurrent processing capabilities), supporting horizontal scaling; the index design adopts a reasonable sharding and replication strategy (sharding: index shards are divided according to business scenarios (e.g., 1 shard for vehicle dispatch data, 2 shards for ticketing data, and 1 shard for maintenance data), with each shard's data size controlled within 50GB to avoid excessively large shards affecting query efficiency; replication settings: each shard is configured with 1-2 replicas, distributed across different data nodes, ensuring no data loss in case of a single node failure, while also improving query concurrency; dynamic adjustment: through EIasticsearch's Index Lifecycle Management (ILM), cold data shards are automatically shrunk and archived to optimize cluster storage resources), improving query and write performance; the word segmenter is optimized for public transportation industry business scenarios (e.g., a custom industry dictionary) to improve retrieval accuracy; aggregation analysis and filtering queries are supported to meet complex report statistics needs.

[0269] This invention also includes deployment and monitoring technologies:

[0270] Through containerized deployment and comprehensive monitoring technologies, the platform ensures stable operation, efficient maintenance, and rapid expansion, while supporting a balance between system functionality and performance.

[0271] Containerized deployment:

[0272] (1) Docker is used to implement containerized packaging and deployment of application services. Each microservice is packaged independently as a Docker image and the images are stored in a private image repository (Harbor).

[0273] (2) Work with Kubernetes (K8s) to automate service deployment, scaling and maintenance management. K8s is responsible for container scheduling, load balancing, health checks and fault recovery.

[0274] (3) Supports environment isolation (development, testing, production), and resource isolation is achieved through Namespace; ConfigMap and Secret are used to manage configuration information and sensitive data to avoid hard coding; Supports rolling updates and rollbacks to ensure that services are not interrupted during service upgrades.

[0275] System monitoring technology:

[0276] (1) Integrate the Grafana monitoring platform and use monitoring components such as Prometheus, Node Exporter, MySQL Exporter, and Redis Exporter to achieve comprehensive monitoring of microservice clusters, databases, middleware, and server resources (CPU, memory, disk, network).

[0277] (2) Supports custom monitoring metrics (such as interface response time, data processing throughput, error rate, rule execution success rate, and approval process completion rate), and uses PromQL query language to aggregate and filter metrics;

[0278] (3) The visual dashboard displays the system's operating status in real time and supports multi-dimensional drill-down analysis; alarm rules are configured (such as CPU utilization exceeding 80% or interface response time exceeding 500ms), and alarm information is pushed via SMS, email, DingTalk / WeChat Work to promptly warn of abnormal situations;

[0279] (4) Integrate the link tracing tool (SkyWalking) to realize the visual tracing of microservice call links, record the entire process of the request from the entry point to the exit point, including the call order, time consumption and status of each service, which is convenient for troubleshooting cross-service problems; support log aggregation analysis (ELK Stack), centrally collect logs of each service, and support searching by keywords, time range, service name and other conditions to improve the efficiency of problem investigation.

[0280] Example 2

[0281] This embodiment provides an implementation method based on a fully configurable low-code development platform for the public transportation industry. The method includes:

[0282] S1, Platform Deployment;

[0283] S11. Environment Preparation: Set up a Docker+K8s containerized deployment environment and configure basic middleware; initialize network configuration, storage configuration, and security policies.

[0284] S12. Base Deployment: Package each microservice of the low-code platform base into a Docker image, deploy it to the cluster via Kubernetes, configure service relationships and resource quotas; complete the initial configuration of basic services;

[0285] S13. Master Data Integration: Integrate MDM master data in the public transportation sector, import it in batches into a MySQL database using a data import tool, synchronize it to MongoDB to retain the original data, generate a standardized metadata model through a metadata parsing engine, and cache it in Redis;

[0286] S14. Monitoring Deployment: Deploy monitoring components, configure monitoring metrics, alarm rules, and tracing rules to ensure that the platform's operating status is monitorable and traceable.

[0287] S2, Scene Configuration Process;

[0288] S21. Select Scenario: After logging into the platform, users perform permission verification through the low-code platform base and select the target business scenario from the preset public transportation industry scenario library; the platform loads the basic data model template, business components and process templates corresponding to the scenario through the metadata parsing engine and synchronizes them to the front-end model generation engine;

[0289] S22. Configuration Page:

[0290] Users can customize the data table structure and inter-table relationships through the front-end metadata visualization generation engine. The configuration commands are submitted to the data model generation engine and generated after being verified by the model verification unit.

[0291] The metadata parsing engine receives the original configuration, generates a standardized metadata model through the metadata modeling unit, identifies the relationships between tables through the metadata relationship identification unit, and synchronizes it to the front-end model generation engine.

[0292] The front-end model generation engine's associated metadata filtering unit only displays metadata components that have a relationship. Users can design the page layout and configure multi-terminal adaptation rules by dragging and dropping components to arrange them.

[0293] The APP parsing engine receives page configuration data in real time, generates interface rendering data for PC and APP through the interface configuration parsing unit and terminal adaptation conversion unit, and then synchronizes and previews the data after verification by the compatibility verification unit.

[0294] S3, Association Rules:

[0295] Users can configure business rules by selecting appropriate functions from the built-in rule components and built-in functions through the rule configuration and orchestration unit of the visual business rule engine.

[0296] After the rule configuration is submitted, the rule execution unit of the business flow engine performs rule parsing and conflict detection. The conflict detection unit returns conflict information, and the user corrects the rules to generate an executable rule script.

[0297] By linking to the OA workflow, users can select the corresponding template from the workflow template library, adjust nodes, approval roles, and workflow rules through a visual interface, bind the configured business rules, and set the business data processing logic for each stage of the approval process.

[0298] The workflow engine uses a multi-engine collaborative unit to link the metadata model of the metadata parsing engine with the rule scripts of the business flow engine, ensuring the linkage between the approval process and business data and rule logic.

[0299] S4. Publishing Function:

[0300] After the user completes the page and rule configuration, submits a publishing request. The platform uses a metadata parsing engine to convert the configuration information into relational data stored in MySQL, the raw data stored in MongoDB, and the configuration data cached in Redis.

[0301] The metadata parsing engine's data format conversion unit is associated with cleaning the configuration data, removing invalid and duplicate data.

[0302] The platform generates a permission control unit through the low-code platform base, configures the corresponding page permissions, button permissions, data permissions and field permissions, and synchronizes them to various terminal applications;

[0303] Once published, users can preview the final result through the preview module. Online editing and adjustments are supported, and the adjusted configurations are updated in real time.

[0304] Furthermore, after building multiple public transportation industry solution application systems, this platform also includes: S5 system integration and expansion, specifically including:

[0305] S51. System Integration: Through the system integration unit of the low-code platform, it integrates with the existing application systems of the public transportation company, uses the OAuth2.0 authorization protocol to achieve secure authentication, and achieves data interoperability through standardized API interfaces; for example, the OA approval process and maintenance business data are automatically linked. After approval, the business flow engine triggers the generation of maintenance work orders and synchronizes them to the machine management system.

[0306] S52. Requirements Changes and System Iteration: When business requirements change, users can directly modify the data model (data model generation engine), page layout (front-end model generation engine), or business rules (visual business rule engine) through the platform. After the modified configuration is verified and parsed, it is synchronized to the relevant modules in real time to achieve rapid iteration of system functions. The version management unit of the metadata parsing engine records the configuration change history and supports rollback to historical versions.

[0307] S53. Report Analysis and Data Export: Users can design custom reports (such as vehicle operation reports, ticket revenue reports, maintenance statistics reports, etc.) based on business data (stored in MySQL and Elasticsearch) through visual reporting tools, and configure data query conditions, visualization chart types, and refresh frequency; reports support real-time refresh and multi-dimensional drill-down analysis, and can be exported to Excel, PDF, and image formats; report data permissions are linked with platform permissions to ensure data access security.

[0308] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the technical solutions of the present invention, and are not intended to limit the specific implementation of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the claims of the present invention should be included within the protection scope of the claims of the present invention.

Claims

1. A fully configurable low-code development platform for the public transportation industry, characterized in that: The platform includes: a data model generation engine, a front-end model generation engine, an APP parsing engine, a metadata parsing engine, a visual business rules engine, a visual reporting tool, and a low-code platform base. The data model generation engine provides the verified original configuration data to the metadata parsing engine; The front-end model generation engine generates interface configuration data based on the standardized metadata model output by the metadata parsing engine and combined with the user's page layout configuration. At the same time, it receives metadata relationship data from the metadata parsing engine and filters out metadata components that are directly related to the current configuration scenario, allowing users to quickly drag and drop to design the page layout. It also links with the theme setting unit of the low-code platform base. The APP parsing engine and the front-end model generation engine work together to achieve one-time configuration and multi-platform generation; at the same time, it receives permission configuration information from the low-code platform base to achieve consistent permission control on the APP side and the PC side. The metadata parsing engine receives the verified raw configuration data and performs structured processing to generate a standardized metadata model, which is then verified and synchronized to the visualization business rule engine as the basis for data field association. It also provides structured metadata and relational data to the front-end model generation engine, visualization business rule engine, and visualization reporting tools, and collaborates with the platform's unified data storage layer to achieve the storage and retrieval of different types of data. The visualization business rules engine receives structured metadata and relational data from the metadata parsing engine. By associating metadata fields with business data, it enables the automated execution of business logic, i.e., the realization of business processes. The visualization reporting tool receives structured business data output by the metadata parsing engine, supports report data statistics, generates adapted tables and charts based on the fields defined in the metadata model, and links with the data permission unit of the low-code platform to achieve permission isolation of report data; and works with Elasticsearch in the data storage layer to realize data query and analysis. The low-code platform foundation provides access control, security protection, configuration management support, and collaborative work between modules for the data model generation engine, front-end model generation engine, APP parsing engine, metadata parsing engine, visualization business rule engine, and visualization reporting tool. Modules transmit data and instructions through standardized message formats and monitor the running status of each module in real time.

2. The fully configurable low-code development platform for the public transportation industry according to claim 1, characterized in that, The data model generation engine includes: a table structure configuration unit, a table relationship definition unit, an industry template library unit, and a model validation unit; Industry Template Library Unit: As the basic entry point for data model configuration, it provides pre-set data model templates exclusive to the public transportation industry, supports template import, modification and saving, and allows users to select templates or modify them based on existing templates; Table structure configuration unit: It takes over the template data of the industry template library unit or user-defined requirements. Based on a visual form interface, it supports user-defined field types, lengths and constraints. It uses JSON Schema format to store configuration information to ensure structural standardization. Inter-table relationship definition unit: Based on the single-table structure determined by the table structure configuration unit, it provides a visual drag-and-drop interface, supports one-to-one, one-to-many, and many-to-many relationship configuration, has a built-in relationship verification algorithm, automatically detects abnormal situations, and builds a complete data model association system; Model Validation Unit: Integrates data type validation, constraint validation, and relationship rationality validation logic to comprehensively validate the table structure configuration and inter-table relationship definition results, and provides real-time feedback on the validation results; Each unit works collaboratively in the order of "industry template library unit → table structure configuration unit → inter-table relationship definition unit → model verification unit" to generate original configuration data, and achieves data model permission isolation through linkage with the low-code platform base permission management unit.

3. The fully configurable low-code development platform for the public transportation industry according to claim 1, characterized in that, The front-end model generation engine includes: a related metadata filtering unit, a component drag-and-drop arrangement unit, a multi-terminal adaptation unit, and an interface preview unit; Related Metadata Filtering Unit: As a pre-configuration step, it uses RPC to call the metadata relationship identification results of the metadata parsing engine, and employs a greedy algorithm to filter metadata with related relationships in the current configuration scenario, while filtering out irrelevant components; Component drag-and-drop orchestration unit: It takes over the metadata component output by the associated metadata filtering unit. It is based on the vuedraggabIe component encapsulation, supports component position adjustment and hierarchical nesting, and uses the DOM Diff algorithm to optimize the interface rendering efficiency after dragging. Users can complete the page layout design by dragging and dropping. Multi-terminal adaptation unit: Based on the page layout determined by the component drag-and-drop arrangement unit, it has a built-in responsive layout rule library and automatically adjusts the component size and layout structure according to the screen size of PC and APP. Interface Preview Unit: It receives the interface configuration data after multi-terminal adaptation, integrates iframe embedding technology, loads the configured interface effect in real time, and supports online editing and preview synchronization; Each unit works collaboratively in the order of "Associated Metadata Filtering Unit → Component Drag and Drop Arrangement Unit → Multi-Terminal Adaptation Unit → Interface Preview Unit" to generate interface configuration data. This data is shared with the APP parsing engine and generated synchronously across multiple terminals. At the same time, it receives metadata relationship data from the metadata parsing engine to support intelligent component filtering and works in conjunction with the theme setting unit of the low-code platform base to achieve a unified interface style.

4. The fully configurable low-code development platform for the public transportation industry according to claim 1, characterized in that, The APP parsing engine includes: an interface configuration parsing unit, a terminal adaptation and conversion unit, an interaction logic synchronization unit, and a compatibility verification unit. Interface configuration parsing unit: As the data processing entry point, it receives interface configuration data shared by the front-end model generation engine, uses a JSON parser to extract component type, attributes, and layout information, and generates terminal adaptation intermediate files to provide basic data for subsequent adaptation and conversion; Terminal Adaptation and Conversion Unit: Receives the intermediate files output by the Interface Configuration Parsing Unit and performs differentiated conversions for Android / iOS system characteristics. On Android, the intermediate files are converted into XML format rendering data, and on iOS, they are converted into Storyboard format rendering data. Cross-platform adaptation plugins ensure consistent interaction logic. Interactive Logic Synchronization Unit: Based on the rendering data of the terminal adaptation and conversion unit, the interactive logic configured on the PC is synchronized to the APP through an event mapping mechanism, using a unified event triggering protocol; Compatibility verification unit: Verifies the rendering data after terminal adaptation and conversion. It has a built-in compatibility rule library for Android and iOS versions, automatically detects component adaptation anomalies, outputs a compatibility report and supports one-click repair. Each unit works collaboratively in the order of "interface configuration parsing unit → terminal adaptation conversion unit → interaction logic synchronization unit → compatibility verification unit", sharing interface configuration data with the front-end model generation engine and receiving permission configuration information from the low-code platform base to achieve multi-terminal synchronous generation function with consistent permissions between the APP and PC.

5. A fully configurable low-code development platform for the public transportation industry according to claim 1, characterized in that, The metadata parsing engine includes: a metadata modeling unit, a data format conversion unit, a data manipulation unit, a metadata relationship identification unit, and an import / export unit; Metadata Modeling Unit: Based on the business characteristics of the public transportation industry, the metadata model structure is designed, the metadata attributes are encapsulated using object-oriented thinking, metadata version management is supported, the raw configuration data of the data model generation engine is received, the basic metadata model is generated, and structural support is provided for subsequent data processing. Data Format Conversion Unit: It inherits the basic metadata model of the metadata modeling unit, adopts the Jackson + Fastjson dual parsing framework, and converts unstructured and semi-structured metadata and business data into relational data that conforms to the MySQL table structure specification. It ensures the accuracy of conversion through data type mapping table, supports custom mapping rules, and synchronizes the converted data to the data storage layer. Data Operation Unit: Based on the transformed structured data, it has built-in CRUD core logic, supports multi-condition combined queries, related queries, and paginated queries. The query SQL is dynamically generated by MyBatis-Plus and is processed using a database sharding and table partitioning + read-write separation architecture to provide data read and write services for other modules. Metadata Relationship Identification Unit: Based on the structured data processed by the data operation unit, a graph theory algorithm is used to construct a metadata relationship graph. The relationship chain is traversed by depth-first search to identify one-to-one, one-to-many, and many-to-many relationships between tables, and output structured relationship data to support intelligent filtering and rule configuration of front-end components; Import / Export Unit: This unit receives structured data from the data manipulation unit and supports importing and exporting custom templates in Excel and CSV formats. During import, it uses POI to parse the file and matches the data through field mapping rules, and has built-in data validation rules. During export, it uses SXSSF to stream large data files, enabling data reuse and backup across systems. Each unit works collaboratively in the order of "metadata modeling unit → data format conversion unit → data operation unit → metadata relationship identification unit → import and export unit" to provide validation support for the data model generation engine, provide structured metadata and relational data for the front-end model generation engine, visualization business rule engine, and visualization reporting tools, and work with the data storage layer to achieve full lifecycle management of data.

6. A fully configurable low-code development platform for the public transportation industry according to claim 1, characterized in that, The visual business rule engine includes: a rule component management unit, a function library unit, a rule configuration and orchestration unit, a rule execution unit, and a conflict detection unit; Rule Component Management Unit: As the basic support for rule configuration, it has multiple built-in rule components, adopts a plug-in architecture design, supports dynamic registration and uninstallation of components, and stores component configuration data in JSON format to form an optional rule component library. It is linked with the rule configuration orchestration unit and allows users to drag and drop to combine them. Function Library Unit: Provides logical calculation support for rule configuration, integrates the required built-in functions, supports the uploading and calling of custom functions, and uses reflection mechanism for function execution and parameter verification to ensure security, forming an optional function library, which works in conjunction with the rule configuration orchestration unit to improve rule logic; Rule configuration orchestration unit: It takes over the filtered rule components and functions, and is a visual flowchart editor based on antv / x6. It supports drag-and-drop combination of rule components and conditional branch settings. It uses the BPMN2.0 specification to describe the rule process, automatically generates executable rule scripts, and links with the conflict detection unit. Conflict Detection Unit: Performs logical verification on the rule scripts output by the rule configuration orchestration unit, uses logical reasoning algorithms to detect condition conflicts and priority conflicts, provides real-time feedback on the conflict location and cause and offers solution suggestions, obtains verification results, and synchronizes the rule scripts to the rule execution unit after the verification is passed; Rule Execution Unit: Based on the LiteFlow rule engine kernel, it parses and executes validated rule scripts, supports synchronous / asynchronous execution modes, and improves concurrent processing capabilities; the input is validated rule scripts and business data, and the output is the rule execution result. It works in conjunction with the business flow engine and OA workflow engine to support the advancement of business processes. Each unit works collaboratively in the order of "rule component management unit → function library unit → rule configuration orchestration unit → conflict detection unit → rule execution unit". It receives structured metadata and relational data from the metadata parsing engine, supports the association between rules and business data, links with the OA workflow engine of the low-code platform to realize the binding of approval processes, and collaborates with the business flow engine to provide rule execution capabilities for business processes.

7. A fully configurable low-code development platform for the public transportation industry according to claim 1, characterized in that, The visualization reporting tool includes: a report template design unit, a data query unit, a visualization rendering unit, a report export unit, and a real-time refresh unit; Report template design unit: Provides a visual drag-and-drop interface, supports configuration of table and chart components, uses XML format to store template structure, supports template saving and reuse, forms report templates, and links with data query unit to clearly define the display structure of report data; Data Query Unit: Connects to the database via JDBC, supports custom SQL queries and multi-data source association queries. Query conditions are configured through a visual interface, and optimized query statements are automatically generated. A query result caching mechanism is used to improve efficiency. The input consists of the query requirements corresponding to the report template and the structured business data from the metadata parsing engine, and the output is the query result data. Visualization rendering unit: Based on antv / x6 and ECharts, it encapsulates chart components, supports real-time data rendering and interactive operations, and adopts a Canvas+SVG hybrid rendering scheme to balance efficiency and visual effects. Based on the query result data, it generates a visual report interface, which is linked with the real-time refresh unit and the report export unit. Real-time refresh unit: Supports setting the refresh frequency, pushes the latest data to the report interface via WebSocket, adopts an incremental data update mechanism to refresh only the changed data, and outputs a visual report after incremental update in real time to ensure the real-time nature of report data; Report Export Unit: Supports exporting in Excel, PDF, and image formats. Excel export uses POI streaming processing, PDF export is based on iText7, and image export uses HtmI2Canvas. It supports batch export and scheduled export. Based on the visually rendered report data, it generates an export file for users to download and back up or use across systems. The report template design unit, data query unit, visualization rendering unit, report export unit, and real-time refresh unit work together to support the overall functionality of the visualization reporting tool. It receives structured business data from the metadata parsing engine, links with the data permission unit of the low-code platform base to achieve permission isolation, and works with the EIasticsearch component to improve data query efficiency.

8. A fully configurable low-code development platform for the public transportation industry according to claim 1, characterized in that, The low-code platform base provides access control, security protection, and configuration management support for other modules, including: access control unit, system integration unit, cross-module interaction unit, security protection unit, operation and maintenance monitoring unit, and configuration management unit; Access Control Unit: It adopts a hybrid model of role-based access control (RBAC) and attribute-based access control (ABAC) to achieve precise control of page permissions, button permissions, data permissions, and field permissions. Permission configuration data is stored in MySQL and hot permission data is cached through Redis. System Integration Unit: Provides standardized API interfaces to support integration with various application systems, uses the OAuth2.0 authorization protocol for secure authentication, and resolves interface differences between different systems through an interface adaptation layer; Cross-module interaction unit: Based on an event-driven architecture, it uses RocketMQ to realize cross-module and cross-system message communication, supports synchronous / asynchronous message delivery, and has a built-in message retry mechanism and dead letter queue to ensure reliable message transmission; Security Protection Unit: Integrates multiple security mechanisms including data encryption, API anti-scraping, SQL injection protection, and XSS protection. It adopts the Spring Security framework to achieve overall security management and supports security log auditing. Operation and maintenance monitoring unit: integrates Prometheus+Grafana monitoring solution, collects system operation indicators in real time, supports custom monitoring thresholds and alarm rules, and pushes alarm information through multiple channels; Configuration Management Unit: Based on Nacos Configuration Center, it enables dynamic management of system parameters, rule configurations, and theme settings. It supports canary releases and rollbacks of configurations, and configuration changes are synchronized to all modules through a message notification mechanism.

9. A fully configurable low-code development platform for the public transportation industry according to claim 1, characterized in that, This platform also includes a front-end engine architecture, specifically involving front-end technologies such as: page module design, integration of core functional plugins into page modules, component parameter configuration, and interface parsing and preview. The page module design is supported by the front-end model generation engine's page templates, with four preset standardized page templates: listPage basic table page, treeAndTable left tree right table page, formTable basic form page, and tabFormTable tab form page; The core functional plugins integrated into the page module are the technical support plugins for the front-end model generation engine component drag-and-drop arrangement unit and the interface preview unit. These plugins are directly integrated into the component library and preview module, including: It uses vue3-highlightjs to implement syntax highlighting for JSON code preview, supports code folding and copying, and renders code snippets through a virtual DOM; VueDragGable is used to implement drag-and-drop arrangement of page elements, and CSS3 transformations are used instead of DOM manipulation during the drag-and-drop process; It uses the antv / x6 visualization chart library to display the table structure and table relationships graphically, supports chart zooming, panning, and node dragging, uses Canvas to render charts, and also supports the export of chart data; The component parameter configuration is the implementation method of secondary encapsulation of components by the front-end model generation engine. The encapsulated hy-series components are directly used as optional components in the component drag-and-drop orchestration unit. The component parameter configuration performs secondary encapsulation of the Ant Design Vue basic components, and all encapsulated components are uniformly named with hy- as the prefix. The interface parsing preview is the front-end interactive support for the front-end model generation engine interface preview unit and the APP parsing engine compatibility verification unit, realizing multi-terminal preview and online editing synchronization functions: The preview function is implemented based on an iframe, with the iframe's src pointing to an independent preview service, and the preview data generated through a Mock service; Supports real-time synchronization between editing and previewing; editing operations are pushed to the preview service in real time via WebSocket. It provides multi-terminal preview switching, and intuitively demonstrates the interface adaptation effect by simulating the viewport size and interaction method of different terminals; It supports screenshotting and exporting of the preview interface. Screenshots are implemented using html2canvas, and exported files are stored locally or as a platform file service.

10. An implementation method based on a fully configurable low-code development platform for the public transportation industry, characterized in that, The method includes: S1, Platform Deployment; S11. Environment Preparation: Set up a Docker+K8s containerized deployment environment and configure basic middleware; initialize network configuration, storage configuration, and security policies. S12. Base Deployment: Package each microservice of the low-code platform base into a Docker image, deploy it to the cluster via Kubernetes, configure service relationships and resource quotas; complete the initial configuration of basic services; S13. Master Data Integration: Integrate MDM master data in the public transportation sector, import it in batches into a MySQL database using a data import tool, synchronize it to MongoDB to retain the original data, generate a standardized metadata model through a metadata parsing engine, and cache it in Redis; S14. Monitoring Deployment: Deploy monitoring components, configure monitoring metrics, alarm rules, and tracing rules; S2, Scene Configuration Process; S21. Select Scenario: After logging into the platform, users perform permission verification through the low-code platform base and select the target business scenario from the preset public transportation industry scenario library; the platform loads the basic data model template, business components and process templates corresponding to the scenario through the metadata parsing engine and synchronizes them to the front-end model generation engine; S22. Configuration Page: Users can customize the data table structure and inter-table relationships through the front-end metadata visualization generation engine. The configuration commands are submitted to the data model generation engine and generated after being verified by the model verification unit. The metadata parsing engine receives the original configuration, generates a standardized metadata model through the metadata modeling unit, identifies the relationships between tables through the metadata relationship identification unit, and synchronizes it to the front-end model generation engine. The front-end model generation engine's associated metadata filtering unit only displays metadata components that have a relationship. Users can design the page layout and configure multi-terminal adaptation rules by dragging and dropping components to arrange them. The APP parsing engine receives page configuration data in real time, generates interface rendering data for PC and APP through the interface configuration parsing unit and terminal adaptation conversion unit, and then synchronizes and previews the data after verification by the compatibility verification unit. S3, Association Rules: Users can configure business rules by selecting appropriate functions from the built-in rule components and built-in functions through the rule configuration and orchestration unit of the visual business rule engine. After the rule configuration is submitted, the rule execution unit of the business flow engine performs rule parsing and conflict detection. The conflict detection unit returns conflict information, and the user corrects the rules to generate an executable rule script. By linking to the OA workflow, users can select the corresponding template from the workflow template library, adjust nodes, approval roles, and workflow rules through a visual interface, bind the configured business rules, and set the business data processing logic for each stage of the approval process. The workflow engine, through multi-engine collaboration units, connects the metadata model of the metadata parsing engine with the rule scripts of the business flow engine to ensure the linkage between the approval process and business data and rule logic. S4. Publishing Function: After the user completes the page and rule configuration, submits a publishing request. The platform uses a metadata parsing engine to convert the configuration information into relational data stored in MySQL, the raw data stored in MongoDB, and the configuration data cached in Redis. The metadata parsing engine's data format conversion unit is associated with cleaning the configuration data, removing invalid and duplicate data. The platform generates a permission control unit through the low-code platform base, configures the corresponding page permissions, button permissions, data permissions and field permissions, and synchronizes them to various terminal applications; Once published, users can preview the results through the preview module and edit and adjust them online. S5 System Integration and Expansion: S51. System Integration: Through the system integration unit of the low-code platform, it integrates with the existing application systems of the public transportation company, uses the OAuth2.0 authorization protocol to achieve secure authentication, and achieves data interoperability through standardized API interfaces; for example, the OA approval process is automatically linked with maintenance business data. After approval, the business flow engine triggers the generation of maintenance work orders, which are synchronized to the machine management system. S52. Requirements Changes and System Iteration: When business requirements change, users can directly modify data models, page layouts, or business rules through the platform; the modified configurations are verified and parsed, and then synchronized to the relevant modules in real time to achieve rapid iteration of system functions; the version management unit of the metadata parsing engine records the configuration change history and supports rollback to historical versions; S53. Report Analysis and Data Export: Users can design custom reports based on business data using visual reporting tools, and configure data query conditions, visualization chart types, and refresh frequency; reports support real-time refresh and multi-dimensional drill-down analysis, and can be exported to Excel, PDF, and image formats; report data permissions are linked with platform permissions to ensure data access security.

Images