A method and apparatus for constructing an industrial safety inspection system based on configurable components
By using component-based encapsulation and template-based implementation, the business logic and technical implementation are decoupled, enabling rapid deployment and flexible evolution of the industrial safety inspection system. This solves the problems of low development efficiency and poor reusability in existing technologies and is suitable for industrial environments such as steel, mining, and chemical industries.
Patent Information
- Application Number
- CN202611618453.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-19
- Publication Date
- 2026-11-17
AI Technical Summary
Existing industrial safety inspection systems suffer from low development efficiency, poor reusability, and high evolution costs. The high coupling between business logic and technical implementation results in long development cycles, high costs, and difficulty in cross-project reuse.
By adopting component-based encapsulation, template-based accumulation, and configuration-based reuse, and by building a pluggable component library, a configurable engine, and scene templates, it achieves decoupling of business logic and technical implementation, allowing business experts to define complex business rules through a graphical interface by configuring JSON templates.
This industrial safety inspection system enables rapid deployment and flexible evolution, shortening the deployment cycle, reducing development costs, and improving the system's reusability and evolvability. It is suitable for industrial environments such as steel, mining, and chemical industries.
Smart Images

Figure CN123123398A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of industrial safety monitoring technology, and in particular to a method and apparatus for constructing an industrial safety inspection system based on configurable components. Background Technology
[0002] With the development of artificial intelligence and computer vision technologies, industrial safety inspection systems have evolved from traditional manual monitoring to intelligent automated inspection. Currently, the mainstream industrial safety inspection system implementation solutions mainly include the following:
[0003] A solution based on a single model API call: The system developer trains a deep learning model for a single inspection target (such as helmet wearing) and calls the model through an API interface for detection. The drawback of this approach is that each independent inspection task (such as personnel intrusion or vehicle violations) requires the separate deployment, calling, and maintenance of one or more models. When the business scenario becomes complex, requiring the fusion of results from multiple models (such as simultaneously detecting helmets, work clothes, and personnel location) and combining this with equipment PLC signals (such as whether the equipment is running) for comprehensive judgment, the system logic becomes extremely complex, with high code coupling, making it difficult to maintain and extend.
[0004] Customized project development based on hard-coded implementations: For each new industrial site (such as the mill platform in Diqing and the filling area of Wuhan Iron and Steel), the development team needs to write code from scratch. This includes: video stream access, image decoding, calling specific algorithms, writing business logic (e.g., triggering an alarm if a safety helmet is detected in a hazardous area and the equipment is running), and handling various abnormal situations. The pain points of this model are long development cycles, high costs, and the inability to effectively accumulate and reuse project experience. For example, the code logic written for the "Diqing mill platform hoisting safety" project can hardly be directly reused in the "Wuhan Iron and Steel filling area safety" project, resulting in serious duplication of work and waste of resources.
[0005] Systems based on fixed rule engines: Some systems offer simple rule configuration functions, but they are usually limited to a single condition such as "if A is detected, then trigger B". For complex logic that requires multimodal data fusion (vision + sensor + time series), such as "if a person is detected entering area A, and a vehicle is detected in area B within 30 seconds and its speed is below a threshold, then trigger the 'prepare to unload' event", this cannot be accomplished through configuration and still requires secondary development.
[0006] In summary, existing technical solutions generally suffer from the following drawbacks:
[0007] Low development efficiency: It is impossible to quickly replicate across projects, and each new scenario faces the dilemma of "starting from scratch".
[0008] High coupling between business and technology: Complex business logic (such as security procedures) needs to be implemented by software developers through code, and business experts cannot directly participate in the definition of rules, resulting in high communication costs and long delivery cycles.
[0009] Poor reusability: Validated algorithm models, processing flows, and business rules cannot be transformed into reusable assets, making knowledge accumulation difficult.
[0010] Weak scalability: When the algorithm model is upgraded or the business rules are fine-tuned, the entire project code needs to be modified and redeployed, resulting in high iteration costs. Summary of the Invention
[0011] To address the shortcomings of existing technologies, this invention proposes a method and apparatus for constructing an industrial safety inspection system based on configurable components. This solves the problems of low development efficiency, poor reusability, and high evolution costs caused by the high coupling between business logic and technical implementation in the development of existing industrial safety inspection systems.
[0012] This invention transforms the traditional customized industrial safety project development model into a configurable product model that can be quickly deployed and flexibly evolved through component-based encapsulation, template-based accumulation, and configurable reuse. It is applicable to intelligent safety management in various industrial production environments such as steel, mining, and chemical industries.
[0013] The technical terms used in this invention and their related explanations are as follows:
[0014] Safety building blocks: refers to reusable units formed by standardizing and modularizing the algorithm models, business logic, and processing action elements in industrial safety inspection.
[0015] Logical expression configuration: A method for defining complex business rules by combining logic blocks (AND, OR, NOT, timing) and data blocks (visual inspection results, PLC signals, etc.) through graphical drag-and-drop or structured languages (such as JSON).
[0016] Processing chain orchestration refers to the process of assembling a series of preprocessing and postprocessing nodes (such as dehazing, super-resolution, image stabilization, etc.) for a specific video stream or data stream in a visual manner to form a customized data processing pipeline.
[0017] Scene template: Refers to a validated, reusable JSON configuration file for a specific industrial scene (such as hoisting operations or vehicle parking), containing model combinations, processing chains, and logical rules.
[0018] In a first aspect, the present invention provides a method for constructing an industrial safety inspection system based on configurable components, comprising the following steps:
[0019] S1. Construct a pluggable component library to uniformly encapsulate the algorithm model, data access, image or video processing, logical operation, data block and output linkage functions in industrial safety inspection, so that each component has a component identifier, version, input and output format, configurable parameters and adapter interface.
[0020] S2. Establish a configurable engine, which is deployed on an edge computing device, industrial server, or cloud computing device and communicates with cameras, PLCs / sensors, and output linkage devices; configure the engine to parse the data source, processing chain, rules, and action fields in the structured configuration file to generate an inspection task execution plan.
[0021] S3. Abstract the configuration file that has been verified on-site into a scene template, extract the detection target, area coordinates, threshold, PLC point, processing chain node and output action as variable parameters, and store the scene template into the template library.
[0022] S4. In the new inspection scenario, call the scenario template and modify the variable parameters to generate the target configuration file; the configuration engine loads components, binds data sources and linkage devices, orchestrates processing chains, executes rule expressions and outputs inspection results or linkage actions according to the target configuration file.
[0023] Secondly, the present invention provides an industrial safety inspection system construction device based on configurable components to implement the steps of the method described in the first aspect. The construction device includes a component library construction module, a configuration engine module, a template accumulation module, a deployment execution module, and a runtime monitoring and version management module.
[0024] The component library building module is used to encapsulate and maintain algorithm model components, data access components, data processing components, logic components, data block components, and output components; the configuration engine module is used to parse configuration files and generate processing chain execution graphs and rule expression trees; the template accumulation module is used to parameterize verified configuration files and store them as scene templates; the deployment and execution module is used to schedule components to run collaboratively according to the target configuration file; and the operation monitoring and version management module is used to record the running status, component version, and configuration version.
[0025] In one implementation, the configuration platform and template library of the device are deployed on a central server, and the component running instances are deployed on edge computing devices or industrial servers; the edge computing devices communicate with field cameras, PLCs / sensors, LED screens, audible and visual alarms or other linked devices, thereby forming a closed loop of configuration, deployment, execution and feedback.
[0026] Thirdly, the present invention provides a computer device, including 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 described in the first aspect, or to implement the functions of each module in the device described in the second aspect.
[0027] Fourthly, the present invention provides a computer-readable storage medium or computer program product 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 described in the first aspect, or implements the functions of the modules in the apparatus described in the second aspect.
[0028] This invention proposes a method and apparatus for constructing an industrial safety inspection system based on configurable components. By modularizing and templating algorithms, logic, and processing flows, it forms "safety building blocks." Business experts can directly "translate" safety procedures into executable digital rules by configuring JSON templates through a graphical interface without coding.
[0029] This invention addresses the pain point of existing technologies that require "customization from scratch for each project," achieving a productized model of "one-time development, reusable everywhere." Through this method, the deployment cycle for a new scenario can be compressed from the traditional week or even several weeks to just a few hours; and when the algorithm model is upgraded, the business side can seamlessly acquire the latest capabilities without modifying any business configurations. This significantly improves the productivity of intelligent industrial safety, drastically reduces project delivery costs, and gives the system the flexibility for continuous evolution, providing an effective solution for the standardization and intelligentization of industrial safety management. Attached Figure Description
[0030] Figure 1 This is a schematic diagram of the overall architecture of a method for constructing an industrial safety inspection system based on configurable components, as disclosed in this invention.
[0031] Figure 2 This is a flowchart illustrating the process of achieving rapid reuse across scenarios through a configuration paradigm.
[0032] Figure 3 This is a schematic diagram of the configuration interface for implementing compliant vehicle parking rules through logical expressions. Detailed Implementation
[0033] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Steps S1, S2… in the described embodiments of the present invention do not limit the scope of execution of the present invention; the various models, simulation environments, and software described in the present invention are not considered as the only limiting methods of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0034] In this invention, computer device / equipment / system refers to a related entity applied to a computer, such as hardware, a combination of hardware and software, software, or software in execution. More specifically, for example, software includes, but is not limited to, a process running on a processor, a processor, an object, executable software, an execution thread, a program, and / or a computer. Furthermore, an application program or script running on a server, and the server itself, can also be software. One or more software programs may be in an execution process and / or thread, and the software may be localized on one computer and / or distributed across two or more computers, and may be run on various computer-readable media.
[0035] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.
[0036] In a first aspect, the present invention provides an embodiment of a method for constructing an industrial safety inspection system based on configurable components, such as... Figure 1 As shown, the specific process can be as follows:
[0037] S1. Build a pluggable component library;
[0038] Standardize and encapsulate various algorithm models (such as safety helmet detection and vehicle detection), data processing nodes (such as defogging, super-resolution, image stabilization, and multi-frame fusion), and business logic (such as AND, OR, NOT, and timing) to form independent, pluggable "safety building block" components, and store them in the component library.
[0039] S2, Establish a configurable engine;
[0040] A configuration engine is provided that can parse structured configuration files (such as JSON). The configuration file defines the complete workflow of the inspection task, including:
[0041] Data source configuration: Specify the input source such as video stream address and PLC signal point.
[0042] Processing chain orchestration: Select required processing nodes from the component library, and assemble a customized preprocessing pipeline for a specific data source (for example: video stream → dehazing → image stabilization → model inference).
[0043] Rule definition: Combine data blocks (model inference results, PLC signal values) and logic blocks through logical expressions to define complex business rules. For example, the rule [vehicle detected in area A] AND [vehicle angle within threshold] AND [parking duration > 30 seconds] is configured as a logical expression.
[0044] In one implementation, the configuration engine, as the operating core of the inspection software, is deployed on edge computing devices, on-site industrial servers or cloud computing devices. The computing device may be configured with a CPU, GPU or AI acceleration card, access camera video streams through a network interface, access PLC / sensor signals through an industrial communication interface or gateway, and output results to an alarm platform, LED screen, audible and visual alarm or PLC linkage device through message queue, HTTP interface, OPC UA, Modbus or other industrial protocols.
[0045] The functional layout of the configuration engine includes: a configuration file parser, a component registration and loader, a processing chain orchestrator, a rule executor, a task scheduler and an action executor. The configuration file parser reads the data source, pipeline, rules and actions fields in the JSON configuration file; the component registration and loader locates the component package from the component library according to the component identifier and version number and binds parameters; the processing chain orchestrator converts the pipeline field into a serial processing chain or a directed acyclic graph; the rule executor converts the rules field into a logical expression tree and integrates visual detection results, timing results and PLC signals; the action executor triggers an alarm, display or linkage control according to the rule result.
[0046] S3. Precipitate and reuse scenario templates;
[0047] As Figure 2 shown, a configuration file that has been verified on site and contains complete data sources, processing chains and rule definitions is solidified into a "scenario template" and stored in the template library. When a new scenario arises, a configuration file for the new scenario can be quickly generated by calling a similar scenario template in the template library and modifying some of its parameters (such as detection target, area coordinates, linkage signal), so as to realize rapid reuse with "zero code modification, only adjustment of configuration parameters".
[0048] S4. Automated deployment and execution;
[0049] The configuration engine loads the final generated target configuration file. Based on the data source fields, processing chain fields, rule fields, and action fields defined in the target configuration file, it automatically orchestrates and executes the entire inspection task process, driving the collaborative work of various components, and ultimately outputting inspection results or triggering linked actions. The automatic orchestration includes verifying component dependencies, loading component packages by component identifier and version, binding video streams and PLC points, generating a processing chain execution graph, and starting the rule executor.
[0050] Secondly, the present invention provides an embodiment of an industrial safety inspection system construction device based on configurable components to implement the steps of the method described in the first aspect.
[0051] The construction device includes: a component library construction module for registering and maintaining various safety building block components; a configuration engine module for parsing configuration files, loading components, generating processing chains and rule execution plans; a template library management module for storing, querying, and versioning scenario templates; a configuration editing module for providing a graphical drag-and-drop interface and a parameter editing interface; a deployment execution module for publishing the target configuration file to an edge computing device or industrial server and initiating inspection tasks; and a result output module for outputting alarms, compliance records, on-site displays, or PLC linkage signals.
[0052] The aforementioned modules can be centrally deployed on a single computer device or distributed across multiple devices. For example, the configuration editing module, template library management module, and component library building module can be deployed on a central server, while the configuration engine module and deployment execution module can be deployed on edge computing devices located near the field cameras and PLCs. The result output module can be connected to the alarm platform, LED screen, and PLC linkage interface.
[0053] Example 1: Configuring inspection rules for compliant vehicle parking in a factory's filling area;
[0054] like Figure 3 As shown in the figure, this embodiment takes the deployment of a "vehicle compliance parking" inspection task in a filling area of a factory as an example to illustrate the implementation process of the method of the present invention in detail.
[0055] This inspection task uses cameras to capture video streams from the filling platform and reads gas emission status, operational status, or other on-site interlocking signals through the PLC interface. The system identifies vehicles, guardrails, parking lines, and area boundaries, and makes a comprehensive judgment based on whether the vehicle has entered the designated area, the vehicle's parking angle, the duration of parking, and the PLC signal status. When the vehicle's parking position and angle meet preset rules, a compliance record is output; when the vehicle is parked at an angle, crosses the boundary, has insufficient parking time, or does not match the on-site process signals, an alarm is pushed to the platform, a prompt is displayed on the LED screen, and a prompt or control signal can be sent to the PLC linkage interface, thereby realizing automatic inspection and real-time handling of vehicle parking status in the filling area.
[0056] S1. Build the component library (prepare "safe building blocks");
[0057] The system has pre-packaged various components and stored them in the component library:
[0058] Visual algorithm components: "Vehicle Detection v2.0" and "Parking Line Detection v1.0".
[0059] Data processing components: "Video Stream Access Node", "Defogging Node", "Image Stabilization Node".
[0060] Logic components: "AND", "OR", "timing (A then B)", "counter".
[0061] Data block components: "Video stream detection result area A", "PLC signal value".
[0062] Output components: "Alarm push notification", "LED screen display".
[0063] S2, Call the scene template (reuse "safety blocks");
[0064] Since the "Dynamic Dangerous Area Protection" rule had already been deployed for the "Lifting Safety at a Certain Factory" project, the corresponding scenario template (Template_DynamicDangerZone.json) has been stored in the template library. When business experts (non-developers) create a new task for the "Filling Area" project in the configuration platform, they can directly call this template.
[0065] The dynamic hazardous area protection template is used to determine whether personnel, vehicles, or loads have entered a hazardous area during operations such as hoisting, loading and unloading, or equipment operation, based on target detection results, area boundaries, duration, and PLC operation signals. It triggers an alarm or on-site linkage when violation conditions are met. This template, like the vehicle compliance parking task in this embodiment, requires video stream access, target-area relationship determination, timing, PLC signal fusion, and alarm / display / linkage output. Therefore, its data source structure, area parameter structure, timing logic, and output action structure can be directly reused; only the detection target, area coordinates, threshold, and linkage points need to be replaced.
[0066] S3. Modify configuration parameters (to achieve rapid adaptation);
[0067] Business experts should open the JSON configuration file for this template (or edit it via the graphical interface) and modify only the following key parameters:
[0068] Modify the detection target: Change "target_object": "lifted object" in the configuration file to "target_object": "vehicle" and "target_object": "protective post".
[0069] Modify zone coordinates: Change the coordinates of the danger zone "zone_coordinates": [...] from the mill platform area to the filling platform area.
[0070] Modify the linkage signal: Change the associated signal that triggers the alarm from "trigger_signal": "PLC_weight" to "trigger_signal": "PLC_gas emission".
[0071] Optimized processing chain: Due to the high dust levels in the filling area, business experts added an extra "dehazing node" to the video stream processing chain, forming a new processing chain: Video Stream → Dehazing → Image Stabilization → Vehicle Detection. This was achieved by modifying the "pipeline" field in the configuration file.
[0072] Define the rules: Business experts combine logic blocks and data blocks to form logical expressions, such as... Figure 3 As shown: [Vehicle detected in area A] AND [Vehicle angle within threshold] AND [Parking duration > 30 seconds]. If the conditions are met, "Output [Compliance]" is triggered; otherwise, "Output [Illegal angled parking] Alarm" is triggered. This operation automatically generates the "rules" field in the configuration file.
[0073] In this embodiment, the configuration file adopts a hierarchical organizational structure, including fields such as datasources, components, parameters, pipeline, rules, and actions. The datasources field records the video stream address, frame rate, PLC connection parameters, and PLC location; the components field records the component identifiers and versions of vehicle detection, parking line detection, defogging, image stabilization, timers, logic operations, and output components; the parameters field records field parameters such as area coordinates, angle thresholds, parking duration thresholds, and alarm levels.
[0074] The `pipeline` field defines the processing chain as a list of nodes or a directed graph. Each node includes `node_id`, `component_id`, `input`, `output`, and `params`, describing the execution order and input / output mapping of "video stream inlet → defogging node → image stabilization node → vehicle detection component → region / angle post-processing component". The `rules` field defines rules as a logical expression tree. Each rule includes `rule_id`, `operator`, `conditions`, `true_action`, and `false_action`, describing the combined conditions of "vehicle located in region A", "angle within the threshold", and "parking time greater than 30 seconds". The `actions` field defines output actions such as compliance recording, violation alarms, LED screen display, and PLC linkage.
[0075] S4, one-click deployment and execution;
[0076] The business expert saves the modified configuration file and clicks "Publish". The configuration engine automatically parses the JSON file and generates component runtime instances based on the component identifier, version number, and parameters in the configuration file. Specifically, at runtime, it creates a corresponding memory object, process object, or service call object for each selected component and binds the video stream, PLC points, area coordinates, angle thresholds, and parking duration thresholds to the corresponding runtime instances. Subsequently, it loads the required visual algorithm components (vehicle detection v2.0), data processing components (defogging node, image stabilization node), and logic components, and creates a rule executor based on the logical expressions. After the system starts running, it analyzes the video stream and PLC signals in real time, triggering corresponding actions once "compliant" or "non-compliant" conditions are met.
[0077] Through the above steps, a project that originally required a week of development was deployed and launched within hours. The entire process involved no coding, achieving complete decoupling between business logic and technical implementation. When the algorithm team released the more powerful "Vehicle Detection v3.0" model, the system administrator only needed to upgrade the "Vehicle Detection" component from v2.0 to v3.0 in the component library. All inspection tasks that referenced this component (including the Wuhan Iron and Steel filling area task in this example) would automatically acquire the latest capabilities without any further operation.
[0078] The following is a comparative analysis of the embodiments of the present invention with other embodiments, in order to further illustrate the technical effects of the present invention.
[0079] Comparison with Implementation Method 1: Traditional development based on a low-code platform;
[0080] This comparative approach provides a low-code development platform, allowing developers to create applications through a graphical interface and with minimal code. However, for complex industrial safety inspection logic, developers still need to write a significant amount of code to handle multimodal data fusion, timing logic, and other complex tasks.
[0081] Compared to this solution, the advantages of the present invention are as follows:
[0082] Business experts can participate directly: This invention is aimed at business experts, not professional developers. Through "logic expression configuration" and "scenario template reuse," business personnel can directly convert security procedures into system configurations, while low-code platforms still require developers with programming mindset.
[0083] Finer-grained asset accumulation: The "scenario templates" accumulated in this invention are encapsulations of complete business scenarios, with a larger granularity for reuse and higher efficiency. Low-code platforms typically accumulate code snippets or modules, which require more understanding and assembly work for reuse.
[0084] True zero-code reuse: The "template parameter tuning" mode of this invention achieves "zero code modification" when reusing scenarios. Low-code platforms often cannot avoid fine-tuning and adapting the code at the new scenario when reusing complex logic.
[0085] Comparative Implementation Method Two: Customized Development Based on Microservice Architecture;
[0086] This approach breaks down different algorithms and functions into independent microservices (such as safety helmet detection service and PLC data service), and then orchestrates these services by writing code at the business layer.
[0087] Compared to this solution, the advantages of the present invention are as follows:
[0088] Configuration-driven rather than code-driven: This invention drives the entire business process through configuration files. Changes in business logic no longer require recompiling, building, and deploying microservices. Instead, only the configuration needs to be modified and the service restarted or hot-loaded, which greatly improves iteration speed and operational convenience.
[0089] Standardized rule definitions: This invention standardizes complex business rules into unified logical expressions, making them easy to understand, share, and verify. In contrast, the code logic of microservice orchestration varies from person to person, lacks a unified standard, and is difficult to form reusable knowledge assets.
[0090] Explicit Management of Processing Chains: This invention explicitly incorporates image preprocessing techniques such as "dehazing" and "image stabilization" as programmable components into the configuration, making optimization solutions for specific environmental quality issues configurable and reusable. In contrast, in microservice architectures, these processes are typically hard-coded into the call flow, making them difficult to discover and reuse.
[0091] In summary, compared to other alternatives, this invention, through its core principles of "component-based encapsulation, template-based accumulation, and configurable reuse," achieves complete decoupling between business logic and technical implementation, shifting the construction of industrial safety inspection systems from a "customized project" model to a "configurable product" model. This invention offers significant advantages in reducing development costs, shortening deployment cycles, accumulating knowledge assets, and enhancing system evolvability, bringing substantial improvements to intelligent industrial safety.
Claims
1. A method for constructing an industrial safety inspection system based on configurable components, characterized in that, Includes the following steps: S1. Construct a pluggable component library to standardize and encapsulate the algorithm model, data access, data processing, logical operation, data block and output linkage functions for industrial safety inspection, and generate components with unified description information. S2. Establish a configuration engine to communicate with video acquisition equipment, PLC / sensors, and output linkage devices; parse structured configuration files to define the execution plan for inspection tasks; among which, The inspection task process includes: configuring data sources, orchestrating processing chains, rule logic transformation, and action output; S3. Abstract the variable parameters from the configuration file that has been verified on-site and solidify them into scene templates, then store them in the template library; S4. In the new inspection scenario, the scenario template is called and the parameters are modified to generate the target configuration file. The configuration engine loads components, arranges data processing flow, executes inspection rules and outputs inspection results or linkage actions according to the target configuration file.
2. The method according to claim 1, characterized in that, The unified description information includes component identifier, component version, function type, input data format, output data format, configurable parameters, parameter constraints, dependent resources, operating environment, and adapter interface; the components, according to their function types, include at least visual algorithm components, data access components, image or video processing components, multi-frame fusion components, logic components, data block components, and output components.
3. The method according to claim 1, characterized in that, The configuration engine includes a configuration parsing module, a component adaptation module, a process orchestration module, a rule execution module, a task scheduling module, and an action execution module. The configuration parsing module parses structured configuration files; the component adaptation module loads corresponding components based on component identifiers and versions; the process orchestration module converts processing chain configurations into executable flowcharts; the rule execution module converts rule configurations into logical expression trees and evaluates the rules; the task scheduling module allocates computing resources; and the action execution module outputs alarm, display, or PLC linkage instructions.
4. The method according to claim 1, characterized in that, The structured configuration file includes a data source field, a component field, a parameter field, a processing chain field, a rule field, and an action field. The processing chain field is used to define the execution order or directed graph relationship of processing nodes and the input-output mapping between processing nodes. The rule field is used to define the combination relationship of region judgment, threshold comparison, duration, count, timing, and AND, OR, and NOT logical operations. The action field is used to define the output action triggered when the rule is true or false.
5. The method according to claim 1, characterized in that, The scenario template is generated by parameterizing a configuration file that has been verified in the field. The variable parameters include the detection target, area coordinates, threshold, PLC point, output action, component version, and processing chain node. The scenario template also records the applicable scenario, parameter value constraints, template version, and reuse instructions.
6. The method according to claim 1, characterized in that, Calling scene templates and modifying parameters in a new inspection scenario includes: selecting a scene template that matches the new inspection scenario from the template library, replacing or adding detection targets, monitoring area coordinates, duration thresholds, alarm thresholds, PLC signal points, and output linkage devices, and having the configuration engine perform field validation, dependency validation, and component version validation on the modified target configuration file.
7. The method according to claim 1, characterized in that, When the configuration engine executes the target configuration file, it receives video streams and PLC / sensor data, and sequentially performs data access, preprocessing, model inference, and postprocessing according to the processing chain fields. It also performs fusion judgment on the model inference results, area parameters, timing results, and PLC / sensor data according to the rule fields. When the component version in the component library is updated, the component capability is upgraded by modifying the component version mapping or the component version field in the target configuration file, without modifying the business rule code.
8. A device for constructing an industrial safety inspection system based on configurable components, characterized in that, include: The component library building module is used to standardize and encapsulate the algorithm models, data access, data processing, logical operations, data blocks and output linkage functions required for industrial safety inspections and store them in the component library; the configuration engine module is used to parse structured configuration files and generate inspection task execution plans. The template storage module is used to abstract the field-verified configuration files into scene templates and store them in the template library; The deployment and execution module is used to call the scene template, generate the target configuration file, and drive the components to collaboratively execute the inspection task in the new inspection scenario; the operation monitoring and version management module is used to monitor the running status of the inspection task and maintain the component version.
9. 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 1-7, or to implement the functions of each module in the apparatus according to claim 8.
10. A computer-readable storage medium or computer program product having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method described in any one of claims 1-7, or implements the functions of the modules in the apparatus described in claim 8.
