A low-code application construction method, system and product for industrial scenarios

Abstract

The application discloses an industrial scene-oriented low-code application construction method, system and product, and relates to the technical field of industrial software development. The method comprises the following steps: receiving a natural language business requirement description, performing semantic analysis and intention recognition by using an industrial vertical large model engine, and automatically generating an adaptive basic system template; responding to a user's drag operation in a visual configuration interface, configuring a business form and a production process by using a form engine and a process engine, and generating a customized industrial application system; establishing a communication connection with a physical device by configuring a device access protocol through an Internet of Things engine, and constructing a digital twin model to map real-time operation data of the device; after starting the system, collecting production data and business data, analyzing the data by using the industrial vertical large model engine, and generating optimization suggestions or early warning information. The application can shorten the industrial software development cycle, realize flexible adaptation and seamless integration of devices in complex industrial scenes, and improve the intelligent level.

Description

Technical Field

[0001] This application relates to the field of industrial software development technology, and in particular to a method, system, and product for building low-code applications for industrial scenarios. Background Technology

[0002] With the rapid development of Industry 4.0 and smart manufacturing, the demand for industrial software from manufacturing enterprises is increasing daily. Traditional industrial software development primarily employs customized coding methods, requiring developers to go through the entire software development lifecycle from requirements analysis, system design, coding implementation to testing and deployment, a development cycle that typically lasts months or even years. This traditional approach is not only time-consuming but also requires a large number of specialized developers, resulting in high labor costs and continuously increasing maintenance and upgrade expenses. More importantly, traditional software architectures are relatively rigid; when business processes change, system modifications require recoding, leading to slow response times and difficulty in meeting the rapidly changing business needs of manufacturing enterprises. Furthermore, the production processes and management procedures vary significantly across different manufacturing industries, making it difficult for general-purpose software to meet the specific needs of these industries, further increasing development difficulty and costs.

[0003] While some low-code development platforms have emerged in the market, lowering the development threshold through visual configuration, these platforms primarily target general business applications and have significant shortcomings in industrial applications. Firstly, existing low-code platforms mainly support simple approval processes and data forms, lacking support for complex production processes, equipment management, quality traceability, and other industry-specific scenarios. Secondly, these platforms lack the ability to integrate with industrial equipment, failing to enable rapid access, data collection, and remote control of IoT devices, thus failing to meet the real-time equipment status monitoring needs in industrial scenarios. Furthermore, existing platforms have a low level of intelligence, unable to utilize artificial intelligence technology to automatically generate business modules and optimization suggestions, resulting in limited improvements in development efficiency. Summary of the Invention

[0004] This application provides a method, system, and product for building low-code applications for industrial scenarios, aiming to solve the problems of long development cycles, high costs, and difficulties in industry adaptation of existing industrial software, as well as the lack of support for complex industrial scenarios, inability to integrate with industrial equipment, and low level of intelligence of existing low-code platforms.

[0005] Firstly, a low-code application building method for industrial scenarios is provided, including:

[0006] S1 receives a natural language business requirement description input by the user, and uses the built-in industrial vertical big model engine to perform semantic parsing and intent recognition on the business requirement description. Based on the recognition results, it matches the standard business functions of the corresponding industrial sub-fields from the industry knowledge base and automatically generates an adapted basic system template.

[0007] S2, the basic system template is displayed in the visual configuration interface, responds to the user's "drag, drop, drag" operation, uses the form engine to dynamically adjust the business forms in the basic system template, and uses the process engine to configure the process nodes and flow logic of the production process to generate a customized industrial application system.

[0008] S3. Based on the data interaction requirements of the customized industrial application system, configure the industrial equipment access protocol through the Internet of Things engine, establish a communication connection with the physical equipment in the workshop, and construct a corresponding digital twin model according to the type of the physical equipment, and map the real-time operating data of the physical equipment into the digital twin model.

[0009] S4. Start the customized industrial application system, collect the production data of the physical equipment and the business data of the customized industrial application system in real time, analyze the production data and the business data using the industrial vertical large model engine, generate production optimization suggestions or abnormal early warning information, and display them on the monitoring interface of the customized industrial application system.

[0010] Optionally, in the above scheme, step S1 utilizes the built-in industrial vertical large model engine to generate a basic system template, specifically including:

[0011] S11, perform word segmentation and entity extraction on the natural language business requirement description input by the user, and extract key business entities and logical relationships;

[0012] S12, Match the extracted key business entities with standard business objects in the pre-stored industry knowledge base, which integrates expert experience data from subdivided industrial sectors;

[0013] S13. Based on the matching results, the corresponding functional module interfaces are automatically called to generate an initial system architecture that includes production management, process management, quality management and equipment management modules.

[0014] Optionally, in the above scheme, step S2 involves configuring a form engine and a process engine, specifically including:

[0015] S21 responds to the user's component drag-and-drop operation in the form designer, dynamically generates business forms containing text, numbers, dates and industry-specific components, and establishes data linkage relationships between forms;

[0016] S22 responds to the user's node connection operation in the process designer and constructs a production process flow diagram containing process nodes, exception handling nodes and parallel gateways based on the BPMN2.0 specification.

[0017] S23, Based on the user-configured flow logic, generate configuration information for controlling workshop execution.

[0018] Optionally, in the above scheme, step S3 involves configuring the industrial equipment access protocol through the IoT engine, specifically including:

[0019] S31. Select the corresponding communication protocol adapter according to the type of physical equipment in the workshop. The communication protocol includes MQTT, CoAP or HTTP.

[0020] S32, establish the connection between the edge computing node and the physical device, and perform edge-side computing processing on the collected device data;

[0021] S33 stores the processed time-series data in a time-series database and establishes a binding relationship between physical device attributes and digital twin model parameters.

[0022] Optionally, in the above scheme, step S3 involves synchronously configuring the digital twin model of the customized industrial application system, specifically including:

[0023] S34 receives CAD drawings or 3D models of physical equipment uploaded by users and performs lightweight processing and format conversion using WebGL rendering technology;

[0024] S35: Import the converted 3D model into the visual scene editor and configure the real-time data binding channel between the model and the physical device.

[0025] S36 overlays key indicator charts in a visualization scenario to generate a real-time monitoring panel.

[0026] Optionally, in the above scheme, step S4 utilizes the industrial vertical large model engine for analysis, specifically including:

[0027] S41 cleans and extracts features from the real-time collected production data to identify key quality data and equipment status data.

[0028] S42, compare the extracted data with historical normal operating condition data, and use the preset anomaly detection algorithm to identify potential equipment failures or quality anomalies;

[0029] S43, when an anomaly or area for optimization is detected, generates an alert message containing processing suggestions and pushes it to the user display layer.

[0030] Secondly, a low-code application building system for industrial scenarios is provided, which implements the steps of the above method, including:

[0031] The requirements parsing and generation module is used to receive natural language input and generate basic system templates.

[0032] The visual configuration module is used to configure forms and processes in the basic system template by dragging and dropping.

[0033] The IoT engine module is used to configure device access protocols and establish communication with physical devices;

[0034] The digital twin building block is used to create virtual models of physical devices and perform data mapping.

[0035] The intelligent monitoring module is used to collect data and generate optimization suggestions based on AI models;

[0036] The output of the requirement parsing and generation module is connected to the visualization configuration module. The configuration information generated by the visualization configuration module is transmitted to the IoT engine module and the digital twin construction module. The intelligent monitoring module obtains data from the IoT engine module and the digital twin construction module respectively. The modules work together through a distributed service architecture to realize the full lifecycle management of industrial applications.

[0037] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described above.

[0038] Fourthly, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the method as described above.

[0039] Fifthly, a computer program product is provided, including a computer program / instructions that, when executed by a processor, implement the steps of the method described above.

[0040] Compared with the prior art, this application has at least the following beneficial effects:

[0041] This application, based on further analysis and research into existing technological problems, recognizes that existing technologies suffer from long development cycles, high costs, and difficulties in industry adaptation for industrial software. It also addresses the shortcomings of existing low-code platforms, such as a lack of support for complex industrial scenarios, inability to integrate with industrial equipment, and low levels of intelligence. By utilizing a built-in industrial vertical large-scale model engine to perform semantic parsing and intent recognition of user natural language business requirements, it automatically generates adapted basic system templates. This eliminates the need for manual coding from scratch in the initial construction of industrial software, significantly shortening the development cycle and reducing reliance on professional developers. Furthermore, by responding to user drag-and-drop operations in a visual configuration interface, dynamically adjusting business forms using a form engine, and utilizing a process engine to manage production process flows... By configuring sequence nodes and flow logic, flexible adaptation to complex industrial scenarios is achieved, enabling manufacturing enterprises to quickly customize systems according to their own process characteristics. By utilizing an IoT engine to configure industrial equipment access protocols to establish communication connections with physical devices, and constructing corresponding digital twin models based on equipment type to map real-time operational data into the models, seamless integration of industrial equipment and software systems is achieved, solving the pain point of traditional low-code platforms being unable to access industrial IoT devices. By collecting real-time production and business data from physical devices, and using an industrial vertical large-scale model engine for analysis to generate production optimization suggestions or anomaly warning information, artificial intelligence capabilities are deeply integrated into the entire lifecycle of industrial applications, improving the system's intelligence level. Therefore, this application achieves a complete technical closed loop from demand input to intelligent monitoring, significantly improving the development efficiency, equipment integration capabilities, and intelligence level of industrial software while ensuring deep adaptation to industrial scenarios. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating a method for building low-code applications for industrial scenarios, provided as an embodiment of this application.

[0043] Figure 2 This is a four-layer service architecture diagram of a low-code platform provided in one embodiment of this application.

[0044] Figure 3 A schematic diagram of the functional modules of the six core engines provided in one embodiment of this application.

[0045] Figure 4 This is a flowchart illustrating the working principle of an AI large model engine provided in one embodiment of this application.

[0046] Figure 5 This is a schematic diagram of the process configuration interface of the process engine provided in one embodiment of this application.

[0047] Figure 6 This is a device access architecture diagram of an IoT engine provided in one embodiment of this application.

[0048] Figure 7 A 3D visualization of a digital twin engine provided in one embodiment of this application. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0050] In the description of this application, unless otherwise stated, the terms "including", "comprising", "having", etc., also mean "not limited to" (certain units, components, materials, steps, etc.).

[0051] In one embodiment, a method for building low-code applications for industrial scenarios is provided, including:

[0052] S1 receives a natural language business requirement description input by the user, and uses the built-in industrial vertical big model engine to perform semantic parsing and intent recognition on the business requirement description. Based on the recognition results, it matches the standard business functions of the corresponding industrial sub-fields from the industry knowledge base and automatically generates an adapted basic system template.

[0053] S2, the basic system template is displayed in the visual configuration interface, responds to the user's "drag, drop, drag" operation, uses the form engine to dynamically adjust the business forms in the basic system template, and uses the process engine to configure the process nodes and flow logic of the production process to generate a customized industrial application system.

[0054] S3. Based on the data interaction requirements of the customized industrial application system, configure the industrial equipment access protocol through the Internet of Things engine, establish a communication connection with the physical equipment in the workshop, and construct a corresponding digital twin model according to the type of the physical equipment, and map the real-time operating data of the physical equipment into the digital twin model.

[0055] S4. Start the customized industrial application system, collect the production data of the physical equipment and the business data of the customized industrial application system in real time, analyze the production data and the business data using the industrial vertical large model engine, generate production optimization suggestions or abnormal early warning information, and display them on the monitoring interface of the customized industrial application system.

[0056] In one embodiment, a method for building low-code applications for industrial scenarios is provided. This method can be implemented based on the low-code platform described in this invention. This platform adopts a four-layer service architecture, including an engine service layer, a module service layer, an interface service layer, and a user presentation layer, and integrates six core engine tools: a data engine, a form engine, a process engine, an AI large-scale model engine, an IoT engine, and a digital twin engine.

[0057] In step S1, the platform receives a natural language description of the user's business requirements, such as "Automotive Parts Manufacturing MES System". The built-in industrial vertical large-scale model engine, based on a pre-trained large-scale language model, is fine-tuned using an industrial domain knowledge base to form a specialized model for the industrial vertical domain. This engine performs semantic understanding and intent recognition on the user's input business requirements, matches standard business functions from the industry knowledge base, and automatically generates a basic system template containing modules such as production management, process management, and equipment management. This process deeply integrates AI large-scale model technology with a low-code platform, enabling automatic generation of industry templates and reducing manual configuration workload.

[0058] In step S2, the platform displays the basic system template generated in step S1 on the visual configuration interface. The form engine provides a rich library of field components, including basic components such as text, numbers, dates, and dropdown selections, as well as specialized components specific to industrial scenarios such as equipment selection, process parameters, and quality inspection. Users can add components to the form design area by dragging and dropping, and generate a complete form interface after configuring the field attributes. The process engine is implemented based on the BPMN 2.0 specification, supporting various process elements such as sequential flow, parallel gateway, and exclusive gateway, and adds extended elements such as process nodes, exception handling nodes, and equipment linkage nodes for industrial scenarios. Users can build production process flow diagrams through the visual process designer, configure the processing rules and trigger conditions of each node, and generate customized industrial application systems.

[0059] In step S3, the platform configures the industrial equipment access protocol through the IoT engine based on the data interaction requirements of the customized industrial application system. The IoT engine supports multiple IoT communication protocols such as MQTT, CoAP, and HTTP, and can connect to various industrial sensors, PLCs, SCADA systems, and other devices. Users can define device types, attributes, commands, and alarm rules to establish communication connections with physical equipment in the workshop. Simultaneously, the platform utilizes a digital twin engine to construct a corresponding digital twin model. The digital twin engine uses WebGL and Three.js technologies to achieve 3D visualization rendering. After users upload CAD drawings or 3D models of the physical equipment, the engine automatically performs lightweight processing and format conversion. By binding IoT data, it achieves real-time synchronization between the 3D model and the actual equipment status, mapping the real-time operating data of the physical equipment to the digital twin model.

[0060] In step S4, after the platform launches the customized industrial application system, it collects production and business data from physical equipment in real time. The industrial vertical large model engine analyzes the collected data and can provide intelligent services such as production optimization suggestions, equipment maintenance early warnings, and quality anomaly detection. When an anomaly or optimization space is identified, the engine generates an early warning message containing handling suggestions and displays it on the monitoring interface of the customized industrial application system, for example, as a chart overlaid on a 3D scene, to achieve real-time monitoring and intelligent decision support for the production process.

[0061] Through the above steps, this embodiment realizes a complete technical closed loop from natural language requirement input to industrial application system generation, IoT device access, digital twin construction, and intelligent monitoring and analysis, effectively improving the development efficiency and intelligence level of industrial software.

[0062] In one embodiment, step S1, which utilizes a built-in industrial vertical large model engine to generate a basic system template, specifically includes:

[0063] S11, perform word segmentation and entity extraction on the natural language business requirement description input by the user, and extract key business entities and logical relationships;

[0064] S12, Match the extracted key business entities with standard business objects in the pre-stored industry knowledge base, which integrates expert experience data from subdivided industrial sectors;

[0065] S13. Based on the matching results, the corresponding functional module interfaces are automatically called to generate an initial system architecture that includes production management, process management, quality management and equipment management modules.

[0066] In one embodiment, step S1, which utilizes a built-in industrial vertical large model engine to generate a basic system template, specifically includes:

[0067] S11, the industrial vertical large model engine, upon receiving the user's business requirements described in natural language, first performs semantic parsing. The engine segments the input text and extracts entities, identifying key business entities such as "production plan," "process route," "equipment ledger," and "quality inspection," and extracts the logical relationships between entities, such as "the stamping process is followed by the welding process" and "rework is triggered when quality inspection fails." This process transforms the user's unstructured requirement description into structured business elements, laying the foundation for subsequent matching.

[0068] In step S12, the platform has a pre-installed industry knowledge base that integrates expert experience data and standard business objects from various industrial sub-sectors such as automotive parts, electronics manufacturing, and equipment processing. The engine matches the key business entities extracted in step S11 with the standard business objects in the knowledge base. For example, it matches "production scheduling" with standard objects in the planning management module and "equipment inspection" with standard objects in the equipment management module. This matching process ensures that the generated system template conforms to the business specifications and practical experience of specific industrial sectors.

[0069] S13, based on the matching results of step S12, the engine automatically calls the interfaces of the corresponding functional modules in the module service layer. The module service layer encapsulates the business logic of commonly used functional modules in industrial software such as production management, process management, quality management, equipment management, and supply chain management. The engine dynamically combines these modules according to the matching results to generate an initial system architecture containing the above modules, forming a basic system template adapted to the user's industry. This template can serve as the starting point for subsequent visual configuration, and users can make further personalized adjustments based on it.

[0070] In one embodiment, step S2, which utilizes a form engine and a process engine for configuration, specifically includes:

[0071] S21 responds to the user's component drag-and-drop operation in the form designer, dynamically generates business forms containing text, numbers, dates and industry-specific components, and establishes data linkage relationships between forms;

[0072] S22 responds to the user's node connection operation in the process designer and constructs a production process flow diagram containing process nodes, exception handling nodes and parallel gateways based on the BPMN2.0 specification.

[0073] S23, Based on the user-configured flow logic, generate configuration information for controlling workshop execution.

[0074] In one embodiment, step S2, which utilizes a form engine and a process engine for configuration, specifically includes:

[0075] S21 provides a visual form designer, allowing users to drag and drop components from the component library to add to the form design area. The component library includes not only basic components such as text, number, date, dropdown selection, and file upload, but also specialized components specific to industrial scenarios, such as equipment selection, process parameters, and quality inspection. After configuring field attributes, the system automatically generates the corresponding form interface. The form engine supports data linkage between forms; for example, the selected work order number in the main form can automatically be associated with and displayed in the corresponding process parameter form, as well as data synchronization between master and child tables, fulfilling data interaction needs in complex business scenarios.

[0076] The S22 platform provides a visual process designer, implementing process modeling based on the BPMN 2.0 specification. Users can add process elements to the design area by dragging and dropping, and define the flow sequence between nodes using connecting lines. For the specific needs of industrial scenarios, the process engine adds extended elements: process nodes define specific production processes (such as stamping, welding, painting, and final assembly); exception handling nodes define the handling logic for quality inspection failures or equipment malfunctions; and a parallel gateway supports the simultaneous execution of multiple processes. Users can construct a complete production process flow diagram by configuring the handlers, processing rules, and trigger conditions for each node.

[0077] S23, the process engine automatically generates configuration information that can be parsed by the shop floor execution system based on the node sequence, flow conditions, and exception handling logic configured by the user in the process designer. When a production task starts, this configuration information drives each process in the shop floor to execute according to the predetermined process: the normal process flows in the process sequence, and when an exception is encountered, the rework, pause, or alarm logic defined in the exception handling node is triggered. The process engine supports real-time monitoring of the process and can display the current execution status of each process, the distribution of work-in-process, and the occurrence and handling records of exception events in a visual interface.

[0078] In one embodiment, configuring the industrial equipment access protocol via the IoT engine in step S3 specifically includes:

[0079] S31. Select the corresponding communication protocol adapter according to the type of physical equipment in the workshop. The communication protocol includes MQTT, CoAP or HTTP.

[0080] S32, establish the connection between the edge computing node and the physical device, and perform edge-side computing processing on the collected device data;

[0081] S33 stores the processed time-series data in a time-series database and establishes a binding relationship between physical device attributes and digital twin model parameters.

[0082] In one embodiment, configuring the industrial equipment access protocol via the IoT engine in step S3 specifically includes:

[0083] The S31 IoT engine supports multiple industrial IoT communication protocols, allowing users to select the appropriate protocol adapter based on the type of physical equipment in the workshop. For industrial control equipment such as CNC machine tools and PLCs, the MQTT protocol is typically used for data publishing and subscription; for resource-constrained sensor nodes, the CoAP protocol can be used for lightweight communication; and for smart devices supporting RESTful interfaces, the HTTP protocol is used for data interaction. The IoT engine provides a unified device modeling function, allowing users to define device types, attributes, commands, and alarm rules in a visual interface. The system automatically generates the corresponding communication adapter based on the selected protocol, enabling rapid access to industrial equipment from different manufacturers and of different types.

[0084] The S32 IoT engine supports establishing connections between edge computing nodes and physical devices. Edge computing nodes are deployed on the factory floor, responsible for real-time communication with equipment, collecting data such as equipment operating status, process parameters, and production output. The collected raw data undergoes preprocessing at the edge, including data cleaning, outlier filtering, unit conversion, and aggregation calculations. Configuring computational logic enables the calculation and storage of measurement point data; for example, calculating effective values ​​from raw vibration sensor signals or calculating energy consumption indicators from equipment current data. Edge computing processing reduces the amount of data transmitted to the cloud, lowers network latency, and enables real-time response to equipment status changes.

[0085] S33: After edge processing, device data is stored in a time-series database as time-series data. This database is optimized for time-series data, supporting high-concurrency writes and efficient querying of massive amounts of data, making it suitable for scenarios where industrial equipment continuously generates data streams. Simultaneously, the IoT engine establishes a binding relationship between physical device attributes and digital twin model parameters. By binding IoT data, real-time synchronization between the 3D model and the actual device status is achieved. When device attribute values ​​(such as temperature, speed, and operating status) change, the corresponding parameters in the digital twin model are automatically updated, thus realizing real-time mapping of the virtual model to the physical entity.

[0086] In one embodiment, the synchronous configuration of the digital twin model of the customized industrial application system in step S3 specifically includes:

[0087] S34 receives CAD drawings or 3D models of physical equipment uploaded by users and performs lightweight processing and format conversion using WebGL rendering technology;

[0088] S35: Import the converted 3D model into the visual scene editor and configure the real-time data binding channel between the model and the physical device.

[0089] S36 overlays key indicator charts in a visualization scenario to generate a real-time monitoring panel.

[0090] In one embodiment, the synchronous configuration of the digital twin model of the customized industrial application system in step S3 specifically includes:

[0091] The S34 digital twin engine allows users to upload CAD drawings or 3D model files of physical devices. Based on WebGL and Three.js technologies, the engine achieves 3D visualization rendering. After a user uploads a model, the engine automatically performs lightweight processing and format conversion to reduce the model's file size and rendering resource consumption, enabling efficient loading and display on the web. Lightweight processing includes model mesh simplification, texture compression, and removal of redundant data, while format conversion ensures that model files from different sources (such as STEP, IGES, OBJ, FBX, etc.) can be uniformly converted to rendering formats supported by the platform.

[0092] S35. After conversion, the 3D model is imported into a visual scene editor, where users can position and adjust its location, orientation, and viewpoint. Subsequently, by configuring a real-time data binding channel between the model and physical devices, a mapping relationship is established between 3D model components and IoT device measurement points. By binding IoT data, real-time synchronization between the 3D model and the actual device status is achieved. For example, the spindle speed display component of the CNC machine tool model is bound to the machine tool spindle speed measurement points collected by the IoT engine, and the position of the AGV model is bound to the real-time coordinate data of the AGV. After configuration, when the status of the physical device changes, the digital twin model automatically updates its display.

[0093] The S36 digital twin engine supports overlaying key performance indicator (KPI) charts within 3D visualization scenes. Key indicators can be displayed as charts overlaid on the 3D scene. Users can customize the configuration through the data panel, selecting equipment parameters to monitor, such as temperature trend charts, equipment OEE (Overall Equipment Efficiency) dashboards, production progress bars, alarm lists, etc., which are overlaid on the 3D scene as floating panels or in fixed positions. Through this real-time monitoring panel, operators can simultaneously view the 3D spatial layout and key operating indicators of workshop equipment on a single interface, achieving comprehensive digital management and visual monitoring of physical equipment.

[0094] In one embodiment, step S4, which utilizes the industrial vertical large model engine for analysis, specifically includes:

[0095] S41 cleans and extracts features from the real-time collected production data to identify key quality data and equipment status data.

[0096] S42, compare the extracted data with historical normal operating condition data, and use the preset anomaly detection algorithm to identify potential equipment failures or quality anomalies;

[0097] S43, when an anomaly or area for optimization is detected, generates an alert message containing processing suggestions and pushes it to the user display layer.

[0098] In one embodiment, step S4, which utilizes the industrial vertical large model engine for analysis, specifically includes:

[0099] S41, the industrial vertical large-scale model engine, acquires real-time production data from physical equipment collected by the IoT engine, as well as business data from customized industrial application systems. The raw data may contain noise, missing values, or inconsistent formats. The engine first cleans the data, including removing anomalous jump values, filling in missing data, and standardizing data formats. Subsequently, the engine extracts features from the cleaned data, identifying key quality data (such as product dimensional accuracy, surface defect rate, and key process parameters) and equipment status data (such as spindle vibration values, motor temperature, and equipment operating time) from the massive dataset, laying the foundation for subsequent anomaly detection and optimization analysis.

[0100] In step S42, the engine compares the key data extracted in step S41 with historical normal operating condition data stored in the industry knowledge base. The AI ​​large model engine can intelligently analyze the business data accumulated on the platform, providing intelligent services such as equipment maintenance early warning and quality anomaly detection. The engine uses preset anomaly detection algorithms to identify situations where data deviates from the normal range. For example, when the equipment vibration value continuously exceeds the historical normal threshold, it is determined that there is a potential risk of equipment failure; when key process parameters exceed the standard control limits, it is determined that there is a trend of quality anomalies. The anomaly detection algorithm can be implemented based on statistical analysis methods (such as the 3σ principle), machine learning models (such as isolated forests, autoencoders), or rule engines, and can be flexibly configured according to different data characteristics and application scenarios.

[0101] S43: When the engine identifies potential equipment malfunctions, quality anomalies, or areas for optimization, it automatically generates warning messages. These messages include the anomaly type, location, severity, and suggested handling. For example, if an abnormally high spindle temperature is detected on a CNC machine tool, the warning message might state, "Equipment ID: MC-03, spindle temperature has reached 85℃, exceeding the normal threshold of 75℃. It is recommended to immediately check the cooling system or reduce the machining load." The engine can assist in layout optimization and parameter verification, providing optimization suggestions. Warning messages are pushed to the user display layer via the interface service layer and displayed on the monitoring interface of the customized industrial application system, such as in pop-ups, alarm lists, or highlighted flashing elements in a digital twin scenario, reminding operators to handle the anomaly promptly.

[0102] The following describes the platform architecture for implementing the method of this application.

[0103] A low-code platform based on agile development for industrial applications includes the following core components:

[0104] The platform adopts a four-layer service architecture design, including:

[0105] Engine Service Layer: Used for building the underlying program and framework, enabling interaction with the database, and supporting unified management and efficient processing of mainstream relational databases such as MySQL, SQL Server, Oracle, and DM.

[0106] Module Service Layer: Used for business logic processing, AI large model algorithm processing, etc., enabling the rapid construction of modules such as production management, process management, supply chain management, raw material management, planning management, and equipment management in industrial software.

[0107] Interface Service Layer: This layer exposes API interfaces to the outside world for use by the user layer or other systems, enabling data exchange between systems.

[0108] User Presentation Layer: Used for building and operating user function modules. Users can use this layer to quickly build software systems for different enterprises and industries.

[0109] The platform integrates six core engine tools, achieving high configurability and flexibility:

[0110] (1) Data Engine: Supports configuration and connection of mainstream relational databases, realizes unified management and efficient processing of multiple types of databases, and can be flexibly configured and compatible.

[0111] (2) Form Engine: A visual form design tool that provides a rich variety of field types. It allows users to customize form styles and quickly complete form designs by using a drag-and-drop interface. It includes multiple modules such as operation buttons, form lists, table lists, and linked tables.

[0112] (3) Process Engine: The engine automates business processes through configuration, supports quick configuration of production processes and procedures via drag-and-drop, and synchronizes them to the production workshop for execution. It supports the identification and handling logic of abnormal nodes and can perform logical processing of abnormal processes according to the set rules.

[0113] (4) AI Large Model Engine: Based on the structure and knowledge of the basic large model, and integrating data and expert experience from various industrial sub-sectors, the standard business functions of different industries in the industrial field are automatically generated through natural language parsing in a modular manner to adapt to different industries. It can realize the rapid generation of engineering drawings and design schemes, and assist in layout optimization and parameter verification.

[0114] (5) IoT Engine: Supports access configuration for IoT devices, including gateway management, status monitoring, control command issuance, data collection point management, device configuration and management, monitoring point management, alarm configuration, and alarm linkage. Supports data collection, data calculation, and manual data entry by IoT devices, and realizes the calculation and storage of measurement point data through configuration of calculation logic.

[0115] (6) Digital Twin Engine: Modeling and designing through entity prototype diagrams, business data panels, etc., to replicate the static and dynamic characteristics of the entity and realize comprehensive digital management of the entity itself, its surrounding environment and application system.

[0116] The platform's core innovation lies in its deep integration of large AI models with a low-code platform, enabling the following functionalities:

[0117] (1) Business requirements are described in natural language, and the AI ​​engine automatically analyzes and generates system templates for the corresponding industries;

[0118] (2) Based on data and expert experience from subdivided industrial sectors, form vertical, scenario-based, and professional application models;

[0119] (3) Analyze industrial and industry data to help manufacturing enterprises optimize production data, equipment data, quality data and business data.

[0120] Compared with the prior art, the present invention has the following beneficial effects:

[0121] (1) Significantly improved development efficiency: Through the "drag, drop, and drag" visual configuration method, industrial application software can be quickly generated without writing a lot of code, shortening the development cycle by more than 70%.

[0122] (2) AI intelligent assisted generation: Combining AI large model technology, business function modules adapted to different industries can be automatically generated, further reducing the workload of manual configuration and improving the overall R&D efficiency.

[0123] (3) Deep adaptation to industrial scenarios: It is specially designed for the characteristics of the manufacturing industry and supports complex production process configuration, equipment management, quality traceability and other industrial-specific scenarios.

[0124] (4) Seamless integration with the Internet of Things: Built-in Internet of Things engine supports rapid access, data collection and remote control of industrial equipment, promoting the intelligent development of industry.

[0125] (5) Digital twin capability: Supports digital twin modeling of physical devices to achieve comprehensive digital management and visual monitoring.

[0126] (6) Flexible deployment methods: Supports single-machine deployment, cluster deployment, distributed deployment and other methods, as well as private deployment and cloud deployment, to meet the IT architecture needs of different enterprises.

[0127] (7) High availability and high performance: Supports the needs of a large number of users, with fast response speed and low latency, supports the management of multiple service clusters, automatic information backup, and ensures stable system operation.

[0128] Platform Overall Architecture Implementation: The low-code platform of this invention adopts a cloud-native architecture design and is built based on the microservice concept. The platform front-end adopts a responsive design, supporting access from both web and mobile devices; the back-end adopts a distributed service architecture, where each engine service can be deployed and scaled independently.

[0129] Data Engine Implementation: The data engine achieves unified management of various relational databases by abstracting the database access layer. It employs connection pooling technology to optimize database connection performance, supports read / write separation and database sharding, and meets the needs of large-scale data processing. Users configure database connection parameters through a visual interface, and the system automatically generates the corresponding data access interface.

[0130] Form Engine Implementation: The form engine provides a rich library of field components, including basic components such as text, number, date, dropdown selection, and file upload, as well as specialized components specific to industrial scenarios such as equipment selection, process parameters, and quality inspection. Users can add components to the form design area by dragging and dropping, and then generate a complete form interface after configuring the field attributes. The form engine supports data linkage between forms and master-detail table relationships.

[0131] Workflow Engine Implementation: The workflow engine is implemented based on the BPMN 2.0 specification and supports various workflow elements such as sequential flow, parallel gateways, and exclusive gateways. For the specific needs of industrial scenarios, the workflow engine adds extended elements such as process nodes, exception handling nodes, and equipment linkage nodes. Users can construct production process flow diagrams through a visual workflow designer using drag-and-drop, configuring the handlers, processing rules, and trigger conditions for each node. The workflow engine supports real-time monitoring and exception alerts for the workflow.

[0132] AI Large-Scale Model Engine Implementation: The AI ​​Large-Scale Model Engine uses a pre-trained large language model as its foundation, fine-tuned through an industrial domain knowledge base to form specialized models for vertical industrial sectors. The engine receives business requirements described by users in natural language, and through semantic understanding, intent recognition, and knowledge retrieval, automatically generates corresponding industry template configurations. Simultaneously, the engine can intelligently analyze the business data accumulated on the platform, providing intelligent services such as production optimization suggestions, equipment maintenance warnings, and quality anomaly detection.

[0133] IoT Engine Implementation: The IoT engine supports multiple IoT communication protocols such as MQTT, CoAP, and HTTP, and can connect to various industrial sensors, PLCs, SCADA systems, and other devices. The engine provides device modeling capabilities, allowing users to define device types, attributes, commands, and alarm rules. Collected device data is processed through edge computing and then stored in a time-series database for querying and analysis by upper-layer applications. The engine supports remote command issuance, enabling remote control of industrial equipment.

[0134] Digital Twin Engine Implementation: The digital twin engine utilizes WebGL and Three.js technologies to achieve 3D visualization rendering. Users upload CAD drawings or 3D models of physical devices, and the engine automatically performs lightweight processing and format conversion. By binding IoT data, real-time synchronization between the 3D model and the actual device status is achieved. The engine supports customizable data dashboard configurations, allowing key indicators to be overlaid and displayed as charts in the 3D scene.

[0135] Taking the implementation of an MES system by an automotive parts manufacturing company as an example, the specific application process of this invention is as follows:

[0136] Enterprises input their requirements for an "Automotive Parts Manufacturing MES System" into the AI ​​large model engine, and the engine automatically generates industry templates that include modules such as production planning, process management, quality management, and equipment management.

[0137] Technical staff use the form engine to drag and drop and configure various business forms according to the company's actual needs for work order formats, quality inspection forms, etc.

[0138] Configure production processes using the process engine, including process definitions for each step such as stamping, welding, painting, and final assembly, and set flow conditions and exception handling rules between steps.

[0139] By connecting to CNC machine tools, robots, AGVs and other equipment in the workshop through the Internet of Things engine, equipment status monitoring and data collection can be achieved.

[0140] A digital twin engine is used to build a 3D visualization scene of the workshop, enabling real-time monitoring and digital twin display of the production process.

[0141] Once the system is deployed and launched, it will integrate with the enterprise's existing ERP, WMS, and other systems through the interface service layer to achieve data interoperability.

[0142] In one embodiment, a low-code application building system for industrial scenarios is provided, for implementing the method of the embodiment described above, characterized in that it includes:

[0143] The requirements parsing and generation module is used to receive natural language input and generate basic system templates.

[0144] The visual configuration module is used to configure forms and processes in the basic system template by dragging and dropping.

[0145] The IoT engine module is used to configure device access protocols and establish communication with physical devices;

[0146] The digital twin building block is used to create virtual models of physical devices and perform data mapping.

[0147] The intelligent monitoring module is used to collect data and generate optimization suggestions based on AI models;

[0148] The output of the requirement parsing and generation module is connected to the visualization configuration module. The configuration information generated by the visualization configuration module is transmitted to the IoT engine module and the digital twin construction module. The intelligent monitoring module obtains data from the IoT engine module and the digital twin construction module respectively. The modules work together through a distributed service architecture to realize the full lifecycle management of industrial applications.

[0149] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory, the processor executing the computer program to implement the steps of the method described in the embodiments above.

[0150] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the steps of the method described in the embodiments above.

[0151] In one embodiment, a computer program product includes a computer program / instructions that, when executed by a processor, implement the steps of the method described in the embodiments above.

[0152] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for building low-code applications for industrial scenarios, characterized in that, include: S1 receives a natural language business requirement description input by the user, and uses the built-in industrial vertical big model engine to perform semantic parsing and intent recognition on the business requirement description. Based on the recognition results, it matches the standard business functions of the corresponding industrial sub-fields from the industry knowledge base and automatically generates an adapted basic system template. S2, the basic system template is displayed in the visual configuration interface, responds to the user's "drag, drop, drag" operation, uses the form engine to dynamically adjust the business forms in the basic system template, and uses the process engine to configure the process nodes and flow logic of the production process to generate a customized industrial application system. S3. Based on the data interaction requirements of the customized industrial application system, configure the industrial equipment access protocol through the Internet of Things engine, establish a communication connection with the physical equipment in the workshop, and construct a corresponding digital twin model according to the type of the physical equipment, and map the real-time operating data of the physical equipment into the digital twin model. S4. Start the customized industrial application system, collect the production data of the physical equipment and the business data of the customized industrial application system in real time, analyze the production data and the business data using the industrial vertical large model engine, generate production optimization suggestions or abnormal early warning information, and display them on the monitoring interface of the customized industrial application system.

2. The method according to claim 1, characterized in that, Step S1, which utilizes the built-in industrial vertical large model engine to generate a basic system template, specifically includes: S11, perform word segmentation and entity extraction on the natural language business requirement description input by the user, and extract key business entities and logical relationships; S12, Match the extracted key business entities with standard business objects in the pre-stored industry knowledge base, which integrates expert experience data from subdivided industrial sectors; S13. Based on the matching results, the corresponding functional module interfaces are automatically called to generate an initial system architecture that includes production management, process management, quality management and equipment management modules.

3. The method according to claim 1, characterized in that, The configuration of the form engine and process engine in step S2 specifically includes: S21 responds to the user's component drag-and-drop operation in the form designer, dynamically generates business forms containing text, numbers, dates and industry-specific components, and establishes data linkage relationships between forms; S22 responds to the user's node connection operation in the process designer and constructs a production process flow diagram containing process nodes, exception handling nodes and parallel gateways based on the BPMN2.0 specification. S23, Based on the user-configured flow logic, generate configuration information for controlling workshop execution.

4. The method according to claim 1, characterized in that, Step S3, which configures the industrial equipment access protocol through the IoT engine, specifically includes: S31. Select the corresponding communication protocol adapter according to the type of physical equipment in the workshop. The communication protocol includes MQTT, CoAP or HTTP. S32, establish the connection between the edge computing node and the physical device, and perform edge-side computing processing on the collected device data; S33 stores the processed time-series data in a time-series database and establishes a binding relationship between physical device attributes and digital twin model parameters.

5. The method according to claim 1, characterized in that, The step S3, which involves synchronously configuring the digital twin model of the customized industrial application system, specifically includes: S34 receives CAD drawings or 3D models of physical equipment uploaded by users and performs lightweight processing and format conversion using WebGL rendering technology; S35: Import the converted 3D model into the visual scene editor and configure the real-time data binding channel between the model and the physical device. S36 overlays key indicator charts in a visualization scenario to generate a real-time monitoring panel.

6. The method according to claim 1, characterized in that, Step S4 utilizes the industrial vertical large model engine for analysis, specifically including: S41 cleans and extracts features from the real-time collected production data to identify key quality data and equipment status data. S42, compare the extracted data with historical normal operating condition data, and use the preset anomaly detection algorithm to identify potential equipment failures or quality anomalies; S43, when an anomaly or area for optimization is detected, generates an alert message containing processing suggestions and pushes it to the user display layer.

7. A low-code application building system for industrial scenarios, used to implement the method described in any one of claims 1-6, characterized in that, include: The requirements parsing and generation module is used to receive natural language input and generate basic system templates. The visual configuration module is used to configure forms and processes in the basic system template by dragging and dropping. The IoT engine module is used to configure device access protocols and establish communication with physical devices; The digital twin building block is used to create virtual models of physical devices and perform data mapping. The intelligent monitoring module is used to collect data and generate optimization suggestions based on AI models; The output of the requirement parsing and generation module is connected to the visualization configuration module. The configuration information generated by the visualization configuration module is transmitted to the IoT engine module and the digital twin construction module. The intelligent monitoring module obtains data from the IoT engine module and the digital twin construction module respectively. The modules work together through a distributed service architecture to realize the full lifecycle management of industrial applications.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-6.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1-6.

Images