Power equipment state management system and work ticket processing method

Abstract

The application provides a power equipment state management system and a work ticket processing method, which first accesses multi-source heterogeneous equipment data through multiple protocols, constructs a real-time equipment state library by using a dynamic weighted fusion algorithm based on the online duration of a signal source, and reduces unstable signal source interference by using a nonlinear weight adjustment mechanism. In the work ticket permission stage, a three-check mechanism including heterogeneous data consistency verification is introduced, the normalized values of physical sensors and the confidence of AI visual analysis are quantitatively compared, and sensor drift or visual recognition abnormalities are effectively identified. At the same time, the system integrates multi-dimensional conflict detection, dynamic deviation monitoring during the operation process, and a hierarchical alarm strategy, and adopts edge computing and a cold and hot data hierarchical storage architecture. The application solves the problems of unreliable state sensing and artificial verification blind spots in traditional operation and maintenance, realizes whole-process management and control from multi-source sensing, intelligent permission to closed-loop monitoring, and significantly improves the safety and robustness of power operation.

Description

Technical Field

[0001] This invention relates to the fields of industrial automation and power system safety technology, and in particular to a power equipment status management system and a work order processing method. Background Technology

[0002] Currently, power equipment condition management mainly relies on a single data source (such as SCADA system), which results in data silos and an inability to effectively integrate multi-source data such as sensors and images for cross-verification, leading to insufficient accuracy in condition judgment. Manual inspection is inefficient and lacks real-time capability, making it difficult to meet the real-time monitoring needs of high-risk operations.

[0003] Among existing technologies, CN119991040A proposes a power plant safety management and control system based on intelligent work permits. Although it achieves data integration and partial linkage, its data processing mainly relies on simple integration and lacks a dynamic weighting mechanism for information source stability. It is prone to misjudgment under sensor drift or environmental interference. Another existing technology, CN116862162A, proposes an intelligent management system for electronic work permits. Although it has certain conflict detection functions, its detection logic is mostly based on static database records and lacks the ability to verify the real-time consistency of heterogeneous data based on "physical sensing + visual perception". When facing complex sites where the physical state of equipment is inconsistent with the system records, security vulnerabilities still exist.

[0004] Therefore, traditional systems still have significant shortcomings in terms of the reliability of multi-source data fusion, dynamic monitoring during operation, and conflict detection in complex scenarios. Summary of the Invention

[0005] The main objective of this invention is to provide a power equipment status management system and a work order processing method, which solves the problems of single and unreliable sources of equipment status data, lack of heterogeneous data consistency verification for work order permits, and inability to dynamically detect spatiotemporal and status conflicts in real time when multiple tasks are running in parallel in the existing power system operation and maintenance.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a work order processing method for a power equipment status management system, the processing method comprising: S1. Access multi-source device data through various industrial protocols, clean the data, and then execute a dynamic weighted fusion algorithm based on the online duration of the data source to fuse physical sensor data and visual analysis data to build a real-time device status database. S2. Create work orders based on the device status library, trigger multi-dimensional conflict detection upon submission, identify conflicts between different work orders in terms of status, time and spatial region, and generate conflict resolution solutions; S3. Before work permit approval, a triple verification mechanism is implemented, which includes heterogeneous data consistency verification of sensor values ​​and AI image analysis results, comparison of real-time status with ticket requirements, and manual audit coverage. S4. During the operation, periodically calculate the deviation between the real-time status value of the equipment and the preset safety value of the work order, and trigger a multi-level alarm strategy based on the deviation range; S5. Before the work order is terminated, retrieve historical status trajectory data to verify whether the device has been restored to its initial physical state.

[0007] In the preferred embodiment, the step of dynamic weighted fusion based on the online duration of the data source in step S1 is as follows: S11. Define the physical sensors that directly contact the equipment as the first data source and obtain the status values ​​of the first data source. Define the camera-based AI visual analysis results as the second data source and obtain the state value of the second data source. ; S12. Real-time statistics of continuous online fault-free time for each data source. The unit is hours; S13. Calculate the weighting coefficient of the first data source using a non-linear growth function. The calculation formula is: Simultaneously, a weight saturation zone is set to limit... The value range is [0.5, 0.8], ensuring that the weight of a single information source does not exceed 80%; Weighting coefficients of the second data source Set as ; S14.Use the formula Calculate the final device fusion state value This reduces the interference of frequent disconnections or new incoming information sources on system decision-making.

[0008] In the preferred embodiment, the detailed execution method of the triple verification mechanism in step S3 includes: S31. Receive real-time RTSP video stream from a high-definition camera, extract image frames at a preset frequency, and input them into a YOLO object detection model based on TensorRT acceleration deployed in an edge computing node; output the device appearance status and corresponding confidence probability through model inference; if the confidence probability is lower than 80%, mark it as a state to be reviewed; if the confidence probability is higher than 80%, extract the state confidence value output by the AI ​​model. Simultaneously, analog data from physical sensors is extracted and normalized to obtain numerical values. Calculate the absolute difference between the two. Only when If the verification is successful, it will be determined that the sensor is drifting or visual recognition is blocked, and the authorization process will be automatically blocked. S32. Retrieve the current fusion status value of the device from the real-time device status database and perform a logical comparison with the "safety measure requirement status" field defined in the work order data structure to ensure that the physical status is completely consistent with the management requirements; S33. When an anomaly or data inconsistency occurs in the automatic verification process, a manual intervention process is triggered. Only work permit holders with encrypted digital signatures are allowed to perform mandatory confirmation or correction of the abnormal status after passing two-factor authentication, and an audit log is recorded.

[0009] In the preferred embodiment, in step S1, data cleaning includes step S15: S15. Calculate two consecutive sampling times. and If the rate of change of the data exceeds a preset physical limit threshold, it is determined to be an abnormal jump, and the system automatically retains the previous time frame. The valid state value is determined, and the data source is marked as unstable.

[0010] 5. The work order processing method for a power equipment status management system according to claim 1, characterized in that: in step S4, the specific steps of the multi-level alarm strategy include: S41. When the calculated deviation When this occurs, an alarm flashing on the HMI interface is triggered; S42. When the calculated deviation When this happens, a pop-up blocking alarm is triggered; S43. When the calculated deviation When this happens, an SMS push alarm is triggered and associated operation permissions are locked.

[0011] In the preferred embodiment, step S16, which involves constructing the real-time device state database, includes the following storage steps: S16. Use an in-memory database cluster as the primary storage to store real-time status data for the most recent hour; use a time-series database as the secondary storage to store historical time-series data for 3 months to 5 years; and use an unstructured object storage system as the tertiary storage to store image snapshots of device status changes along the time path.

[0012] In the preferred embodiment, step S21 of the region conflict detection in multidimensional conflict detection includes: S21. Obtain the set of work space coordinates for the new work order and the set of work space coordinates for all ongoing work orders; calculate the Euclidean distance between the two sets of coordinates; if the distance is less than the preset safe physical isolation distance, it is determined to be an area conflict, and physical isolation measures are recommended.

[0013] In the preferred embodiment, step S10, which involves accessing data from multiple devices, includes: S10. Access monitoring system data via OPCUA protocol; access miniature contact sensor data via Modbus-RTU to MQTT protocol; access displacement sensor data via Modbus-TCP protocol; and access high-definition video stream data via RTSP protocol.

[0014] A power equipment status management system includes a multi-source data acquisition module for accessing physical sensor data and visual image data of the equipment via various industrial fieldbus protocols. The edge intelligence processing module is equipped with a heterogeneous data fusion unit. The heterogeneous data fusion unit is configured to execute a dynamic weighted fusion algorithm based on the online duration of the data source, dynamically adjust the weight of each data source according to the online stability of the information source to calculate the final state value, and perform AI image recognition. A tiered storage module is used to store real-time data, historical time-series data, and unstructured image data according to data access frequency and type. The intelligent service module includes a conflict detection service unit and a process engine service unit, which are used to perform multi-dimensional conflict calculations and triple verification mechanisms. The application module provides a human-computer interaction interface to display work order status and alarm information; The multi-source data acquisition module connects to the edge intelligent processing module, and the processed data is written to the hierarchical storage module. The intelligent service module reads the data from the hierarchical storage module, performs logical operations, and then drives the application module.

[0015] In the preferred embodiment, the multi-source data acquisition module includes an OPCUA interface, an MQTT gateway interface, and an RTSP video stream interface; The edge intelligence processing module integrates the TensorRT inference engine for running the YOLO object detection model; The tiered storage module consists of a RedisCluster cluster, an InfluxDB time-series database, and a MinIO object storage service.

[0016] This invention provides a power equipment status management system and a work order processing method. This application integrates physical sensing and visual analysis data through multiple industrial protocols, employing a dynamic weighting algorithm based on the online duration of the signal source. This effectively reduces interference from frequent disconnections or unstable signal sources, significantly improving the accuracy and robustness of equipment status data. The system introduces a triple verification mechanism, including heterogeneous data consistency verification, during the work order authorization stage. By quantitatively comparing sensor values ​​with AI image analysis results, it can automatically identify and block the risk of misoperation caused by sensor drift or visual recognition errors, ensuring the authenticity of the authorization basis. Simultaneously, through multi-dimensional conflict detection and dynamic deviation monitoring during the operation process, it achieves automatic identification and real-time early warning of status conflicts, time overlaps, and regional intersections. This constructs a closed-loop safety management system covering the entire process from status perception and authorization verification to process monitoring, greatly improving the intelligence level and on-site safety of power operations. Attached Figure Description

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a diagram of the power equipment status management system of the present invention; Figure 2 This is a system diagram of the work order processing method of the present invention. Detailed Implementation

[0018] Example 1 like Figure 1-2 As shown, a work order processing method for a power equipment status management system includes the following steps: S1. Access multi-source device data through various industrial protocols, clean the data, and then execute a dynamic weighted fusion algorithm based on the online duration of the data source to fuse physical sensor data and visual analysis data to build a real-time device status database. S2. Create work orders based on the device status library, trigger multi-dimensional conflict detection upon submission, identify conflicts between different work orders in terms of status, time and spatial region, and generate conflict resolution solutions; S3. Before work permit approval, a triple verification mechanism is implemented, which includes heterogeneous data consistency verification of sensor values ​​and AI image analysis results, comparison of real-time status with ticket requirements, and manual audit coverage. S4. During the operation, periodically calculate the deviation between the real-time status value of the equipment and the preset safety value of the work order, and trigger a multi-level alarm strategy based on the deviation range; S5. Before the work order is terminated, retrieve historical status trajectory data to verify whether the device has been restored to its initial physical state.

[0019] Step S1 addresses the unreliability of single-source data by accessing and cleaning multi-source device data and employing a dynamic weighted fusion algorithm based on the online duration of the data sources. It leverages the complementarity of physical sensor data and visual analysis data to construct a real-time device status database, ensuring the accuracy of status perception. Step S2 triggers multi-dimensional conflict detection when creating a work order. The system automatically identifies status conflicts, temporal conflicts, and spatial area conflicts, proactively mitigating safety hazards during multi-task parallel processing. Step S3 introduces a triple verification mechanism, particularly heterogeneous data consistency verification. Technical means forcefully compare sensor values ​​with AI image analysis results, preventing human error caused by single system failures. Step S4 performs periodic deviation calculations and alarms during work execution, achieving dynamic closed-loop monitoring of the field status. Step S5 verifies the equipment status recovery before termination, ensuring the safety of the final step in closed-loop management. The beneficial effects of this claim are that through full-process digital and intelligent control, it eliminates the blind spots of traditional manual verification, significantly improves the robustness of power equipment status perception using multi-source fusion technology, and ensures inherent safety in complex operating environments.

[0020] In the preferred embodiment, the step of dynamic weighted fusion based on the online duration of the data source in step S1 is as follows: S11. Define the physical sensors that directly contact the equipment as the first data source and obtain the status values ​​of the first data source. Define the camera-based AI visual analysis results as the second data source and obtain the state value of the second data source. ; S12. Real-time statistics of continuous online fault-free time for each data source. The unit is hours; S13. Calculate the weighting coefficient of the first data source using a non-linear growth function. The calculation formula is: Simultaneously, a weight saturation zone is set to limit... The value range is [0.5, 0.8], ensuring that the weight of a single information source does not exceed 80%; Weighting coefficients of the second data source Set as ; S14.Use the formula Calculate the final device fusion state value This reduces the interference of frequent disconnections or new incoming information sources on system decision-making.

[0021] Step S11 clarifies that physical sensors are the first data source and AI visual analysis is the second data source, thus establishing the foundation for heterogeneous data.

[0022] Steps S12 and S13 specify the calculation logic for the weights, including the formula... The weighting coefficients were clarified. With online time The quantitative relationship between them. The formula means that the initial weight of a newly connected data source is 0.5. As the continuous online fault-free operation time of the device increases by 1000 hours, its weight increases by 0.2 until it reaches the saturation upper limit of 0.8.

[0023] This design uses a mathematical model to give higher confidence to long-term stable signal sources, while using saturation region limitations to prevent excessive weighting of a single signal source from causing the system to be slow to respond to new faults. Step S14 uses the formula... Calculate the final fusion state This achieves weighted output based on source reliability. The beneficial effect of this claim is that by dynamically adjusting the fusion ratio of heterogeneous data through a quantitative mathematical model, it effectively suppresses data noise caused by the instability of newly accessed devices or sensor aging and drift, ensuring that the generated device status database data can truly reflect the on-site physical conditions, and providing a highly reliable data foundation for subsequent security verification.

[0024] In the preferred embodiment, the detailed execution method of the triple verification mechanism in step S3 includes: S31. Receive real-time RTSP video stream from a high-definition camera, extract image frames at a preset frequency, and input them into a YOLO object detection model based on TensorRT acceleration deployed in an edge computing node; output the device appearance status and corresponding confidence probability through model inference; if the confidence probability is lower than 80%, mark it as a state to be reviewed; if the confidence probability is higher than 80%, extract the state confidence value output by the AI ​​model. Simultaneously, analog data from physical sensors is extracted and normalized to obtain numerical values. Calculate the absolute difference between the two. Only when If the verification is successful, it will be determined that the sensor is drifting or visual recognition is blocked, and the authorization process will be automatically blocked. S32. Retrieve the current fusion status value of the device from the real-time device status database and perform a logical comparison with the "safety measure requirement status" field defined in the work order data structure to ensure that the physical status is completely consistent with the management requirements; S33. When an anomaly or data inconsistency occurs in the automatic verification process, a manual intervention process is triggered. Only work permit holders with encrypted digital signatures are allowed to perform mandatory confirmation or correction of the abnormal status after passing two-factor authentication, and an audit log is recorded.

[0025] The detailed explanation of the triple verification mechanism in step S3 focuses on the heterogeneous data consistency verification in step S31. This step specifies the use of edge computing nodes to deploy a TensorRT-accelerated YOLO object detection model to process RTSP video streams, and the use of deep learning technology to extract the device appearance status and confidence level.

[0026] The technical solution explicitly sets a confidence threshold of 80%. Only values ​​exceeding this threshold qualify for comparison, thus avoiding false alarms from AI with low confidence levels. The core comparison logic calculates normalized values ​​from physical sensors. AI confidence score absolute difference value And set the difference threshold to 15 percent.

[0027] When the difference is less than or equal to the threshold, it indicates that the physical signal is consistent with the visual representation and the verification is successful; otherwise, the system determines that the sensor may be drifting or the camera may be blocked, and automatically blocks the permission.

[0028] Step S33 further introduces a two-factor authentication manual audit based on encrypted digital signatures, ensuring a final line of defense in case of automatic verification failure. The beneficial effect of this claim is that by forcibly comparing physical and visual quantities across modalities, it solves the problem of a single system being unable to self-check when hacked or physically damaged, significantly improving the rigor and anti-interference capability of high-risk operation permits.

[0029] In the preferred embodiment, in step S1, data cleaning includes step S15: S15. Calculate two consecutive sampling times. and If the rate of change of the data exceeds a preset physical limit threshold, it is determined to be an abnormal jump, and the system automatically retains the previous time frame. The valid state value is determined, and the data source is marked as unstable.

[0030] This step involves calculating two consecutive sampling times. and The data change rate is compared with the physical limit threshold to filter out abnormal and sudden data. In power equipment, the actions of mechanical switches or valves have physical inertia, making it impossible for them to produce state jumps exceeding physical limits in a very short time. The beneficial effect of this claim is that by constraining the validity of data through physical laws, it can automatically identify and eliminate instantaneous spike data caused by electromagnetic interference or communication errors, prevent false alarms from interfering with the judgment of operation and maintenance personnel, and further purify the data quality of the equipment status database.

[0031] In the preferred embodiment, the specific steps of the multi-level alarm strategy in step S4 include: S41. When the calculated deviation When this occurs, an alarm flashing on the HMI interface is triggered; S42. When the calculated deviation When this happens, a pop-up blocking alarm is triggered; S43. When the calculated deviation When this happens, an SMS push alarm is triggered and associated operation permissions are locked.

[0032] Step S41: Set the deviation A level 1 HMI flashing alarm is triggered when the deviation is between 0% and 10% to indicate a minor anomaly; step S42 sets a level 2 pop-up blocking alarm to be triggered when the deviation is between 10% and 30%, forcibly pausing the current operation; step S43 sets a level 3 SMS push and permission lock to be triggered when the deviation is greater than 30%. This tiered strategy adopts progressive intervention measures based on the severity of the state deviation. The beneficial effect of this claim is that it avoids alarm fatigue caused by overreaction due to minor fluctuations, while ensuring that the system can automatically take strong blocking measures when a serious state deviation occurs, realizing refined management of security monitoring and improving the efficiency and accuracy of emergency response.

[0033] In the preferred embodiment, step S16, which involves constructing the real-time device state database, includes the following storage steps: S16. Use an in-memory database cluster as the primary storage to store real-time status data for the most recent hour; use a time-series database as the secondary storage to store historical time-series data for 3 months to 5 years; and use an unstructured object storage system as the tertiary storage to store image snapshots of device status changes along the time path.

[0034] Step S16 proposes a three-tiered storage strategy for separating hot and cold data: a memory database cluster is used as the primary storage to handle high-frequency read / write real-time data, ensuring millisecond-level response; a time-series database is used as the secondary storage to retain historical trend data over long periods, supporting accident tracing; and an unstructured object storage system is used as the tertiary storage to manage image snapshots. The beneficial effect of this claim is that, considering the characteristics of massive, multimodal data in power systems, it optimizes storage resource configuration, meeting the high-performance requirements of real-time conflict detection while solving the problem of low-cost long-term archiving of massive historical data, providing a solid underlying data support for the efficient operation and long-term maintenance of the system.

[0035] In the preferred embodiment, step S21 of the region conflict detection in multidimensional conflict detection includes: S21. Obtain the set of work space coordinates for the new work order and the set of work space coordinates for all ongoing work orders; calculate the Euclidean distance between the two sets of coordinates; if the distance is less than the preset safe physical isolation distance, it is determined to be an area conflict, and physical isolation measures are recommended.

[0036] Step S21 transforms abstract safety management requirements into a calculable geometric problem by acquiring the set of work space coordinates for different work orders and calculating the Euclidean distance. When the calculated distance is less than the preset safe physical isolation distance, the system automatically determines that there is a regional conflict. The beneficial effect of this claim is that it uses digital means to solve the problem of inaccurate judgment of work distances based on traditional human experience, effectively preventing cross-injury accidents caused by insufficient safe distance between adjacent work areas. It has significant safety protection value, especially for three-dimensional cross-operation scenarios in large and complex substations or hydropower stations.

[0037] In the preferred embodiment, step S10, which involves accessing data from multiple devices, includes: S10. Access monitoring system data via OPCUA protocol; access miniature contact sensor data via Modbus-RTU to MQTT protocol; access displacement sensor data via Modbus-TCP protocol; and access high-definition video stream data via RTSP protocol.

[0038] Step S10 lists various standard industrial protocols such as OPC UA, Modbus to MQTT, Modbus TCP, and RTSP. These protocols cover mainstream communication methods from underlying sensors and PLC controllers to video surveillance. The beneficial effect of this claim is that it establishes the system's universality and compatibility, enabling the method to adapt to existing equipment from different manufacturers and eras, breaking down common equipment interface barriers in industrial sites, and providing a technical implementation path for achieving unified acquisition and fusion of data across the entire site.

[0039] Example 2 Further explanation in conjunction with Example 1, such as Figure 1-2 As shown, a power equipment status management system includes a multi-source data acquisition module for accessing physical sensor data and visual image data of the equipment through various industrial fieldbus protocols. The edge intelligence processing module is equipped with a heterogeneous data fusion unit. The heterogeneous data fusion unit is configured to execute a dynamic weighted fusion algorithm based on the online duration of the data source, dynamically adjust the weight of each data source according to the online stability of the information source to calculate the final state value, and perform AI image recognition. A tiered storage module is used to store real-time data, historical time-series data, and unstructured image data according to data access frequency and type. The intelligent service module includes a conflict detection service unit and a process engine service unit, which are used to perform multi-dimensional conflict calculations and triple verification mechanisms. The application module provides a human-computer interaction interface to display work order status and alarm information; The multi-source data acquisition module connects to the edge intelligent processing module, and the processed data is written to the hierarchical storage module. The intelligent service module reads the data from the hierarchical storage module, performs logical operations, and then drives the application module.

[0040] The system comprises a multi-source data acquisition module, an edge intelligent processing module, a hierarchical storage module, an intelligent service module, and an application module, clearly defining the connections and data flow relationships between these modules. The beneficial effect of this claim is that it solidifies the methodological process into specific system functional modules, clarifies the hardware and software architecture layout, makes the technical solution feasible, provides a clear system framework for the development and deployment of actual products, and ensures smooth data flow throughout the entire chain from acquisition to application.

[0041] In the preferred embodiment, the multi-source data acquisition module includes an OPCUA interface, an MQTT gateway interface, and an RTSP video stream interface; The edge intelligence processing module integrates the TensorRT inference engine for running the YOLO object detection model; The tiered storage module consists of a RedisCluster cluster, an InfluxDB time-series database, and a MinIO object storage service.

[0042] The specific interface type of the acquisition module, the TensorRT inference engine integrated in the edge processing module, and the Redis, InfluxDB, and MinIO components specifically selected in the hierarchical storage module are clearly defined. The beneficial effect of this claim is that it specifies the specific technical choices required to implement the invention. The combined application of these components can maximize the system's performance advantages. For example, TensorRT can significantly reduce AI inference latency, and Redis can support high-concurrency state queries, ensuring high performance and high reliability of the system in practical industrial applications, and providing sufficient hardware and software environment support for the technical solution.

[0043] Example 3 Further explanation in conjunction with Example 1, such as Figure 1-2As shown, this embodiment provides a power equipment status management system and a work order intelligent processing method applied to a hydroelectric power plant. The system's hardware architecture deploys a multi-source data acquisition module, an edge intelligent processing module, a hierarchical storage module, an intelligent service module, and an application module. The multi-source data acquisition module accesses circuit breaker and disconnector status data from the power plant monitoring system via the OPC UA protocol interface, accesses data from miniature contact sensors installed on the protection pressure plate via the Modbus-RTU to MQTT protocol interface, accesses data from turbine guide vane displacement sensors via the Modbus-TCP interface, and accesses high-definition camera video streams covering key equipment areas via the RTSP protocol. The edge intelligent processing module integrates a computing unit based on the TensorRT inference engine for running the YOLO object detection model, and is also configured with a processing unit to execute data fusion algorithms. The hierarchical storage module consists of a RedisCluster in-memory database cluster, an InfluxDB time-series database, and a MinIO unstructured object storage system, responsible for storing real-time hot data, historical time-series data, and on-site image snapshots, respectively. The application module provides a human-machine interface on the central control room workstation to display the work order circulation status and equipment alarm information.

[0044] During system operation, the first step is to perform multi-source heterogeneous data fusion. Taking a high-voltage circuit breaker as an example, the system defines its physical auxiliary contact signal as the first data source, acquiring the state value S1. Simultaneously, the AI ​​recognition result of the high-definition camera pointing at the circuit breaker's open / close indicator is defined as the second data source, acquiring the state value S2. The edge intelligent processing module calculates the continuous online fault-free duration of each data source in real time. Assuming the physical contact sensor has been running continuously and stably for 2000 hours, the system uses a non-linear growth function to calculate its weighting coefficient alpha. The calculation process is 0.5 plus 0.2 multiplied by 2000 divided by 1000, resulting in an alpha value of 0.9. Since the system sets a saturation zone of 0.5 to 0.8, the final alpha value is limited to 0.8, and the weighting coefficient beta of the second data source is set to 0.2. Subsequently, the system uses the formula W equal to 0.8 multiplied by S1 plus 0.2 multiplied by S2 to calculate the final device fusion state value W. Meanwhile, during the data cleaning process, the system calculates the rate of change of data between two consecutive sampling times. If it finds that the value of a displacement sensor changes abruptly from zero to 100% within milliseconds, exceeding the physical limit threshold, the system determines it as an abnormal jump, automatically maintains the valid value of the previous moment, and marks the source as unstable.

[0045] When the work supervisor creates and submits a maintenance work order for the circuit breaker through the application module, the intelligent service module immediately triggers multi-dimensional conflict detection. The system iterates through all permitted and ongoing work orders. If it finds another work order requiring the circuit breaker to be in a closed operating state, while the new work order requires it to be in an open state for maintenance, it identifies this as a state conflict. Simultaneously, the system obtains the work space coordinate set of the new work order and the coordinate set of the existing work order, calculating the Euclidean distance between them. If the calculated result is less than the preset safe physical isolation distance, the system determines it as a zone conflict and automatically generates a recommended solution to add a physical isolation barrier.

[0046] Before the work permit enters the licensing stage, the system executes a strict triple verification mechanism. Edge computing nodes extract video stream image frames at a preset frequency and input them into the YOLO model for inference. If the device status confidence probability output by the model is 85%, which is higher than the 80% verification threshold, the system extracts the confidence value VAI. Simultaneously, the system acquires analog data from physical sensors and normalizes it to obtain the value Vsensor. The system calculates the absolute difference between the two; assuming Vsensor is 1.0 and VAI is 0.85, the difference is 0.15. Since this difference is less than or equal to the 15% judgment threshold, the system passes the heterogeneous data consistency verification. Subsequently, the system compares the fused status value in the real-time status database with the security measure status required on the work permit. If they are consistent, the system allows licensing. If data inconsistency occurs in the automatic verification process, the system will block the process, allowing only the work permit holder with the encrypted digital signature to manually force correction after two-factor authentication, and recording the audit log.

[0047] After the work begins, the system enters the dynamic deviation monitoring phase. The system periodically calculates the deviation between the real-time status value of the equipment and the preset safety value of the work order. If a deviation of 5% is detected at any time, falling within the range of 0% to 10%, the application module will trigger a red flashing alarm on the device icon in the HMI interface. If the deviation increases to 20%, falling within the range of 10% to 30%, the system will automatically pop up a blocking window to suspend the operation. If the deviation exceeds 30%, the system will immediately trigger the highest level alarm, send SMS messages to relevant personnel, and lock associated operation permissions. In the work completion phase, the system retrieves historical status trajectory data from InfluxDB to verify whether the equipment has been fully restored to its initial physical state before the work. After confirmation, the work order is archived, completing closed-loop management.

[0048] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A work order processing method for a power equipment status management system, characterized by: The processing methods include: S1. Access multi-source device data through various industrial protocols, clean the data, and then execute a dynamic weighted fusion algorithm based on the online duration of the data source to fuse physical sensor data and visual analysis data to build a real-time device status database. S2. Create work orders based on the device status library, trigger multi-dimensional conflict detection upon submission, identify conflicts between different work orders in terms of status, time and spatial region, and generate conflict resolution solutions; S3. Before work permit approval, a triple verification mechanism is implemented, which includes heterogeneous data consistency verification of sensor values ​​and AI image analysis results, comparison of real-time status with ticket requirements, and manual audit coverage. S4. During the operation, periodically calculate the deviation between the real-time status value of the equipment and the preset safety value of the work order, and trigger a multi-level alarm strategy based on the deviation range; S5. Before the work order is terminated, retrieve historical status trajectory data to verify whether the device has been restored to its initial physical state.

2. The work order processing method for a power equipment status management system according to claim 1, characterized in that: in In step S1, the dynamic weighted fusion based on the online duration of the data source is as follows: S11. Define the physical sensors that directly contact the equipment as the first data source and obtain the status values ​​of the first data source. Define the camera-based AI visual analysis results as the second data source and obtain the state value of the second data source. ; S12. Real-time statistics of continuous online fault-free time for each data source. The unit is hours; S13. Calculate the weighting coefficient of the first data source using a non-linear growth function. The calculation formula is: Simultaneously, a weight saturation zone is set to limit... The value range is [0.5, 0.8], ensuring that the weight of a single information source does not exceed 80%; Weighting coefficients of the second data source Set as ; S14.Use the formula Calculate the final device fusion state value This reduces the interference of frequent disconnections or new incoming information sources on system decision-making.

3. The work order processing method for a power equipment status management system according to claim 1, characterized in that: in In step S3, the detailed execution method of the triple verification mechanism includes: S31. Receive real-time RTSP video stream from a high-definition camera, extract image frames at a preset frequency, and input them into a YOLO object detection model based on TensorRT acceleration deployed in an edge computing node; output the device appearance status and corresponding confidence probability through model inference; if the confidence probability is lower than 80%, mark it as a state to be reviewed; if the confidence probability is higher than 80%, extract the state confidence value output by the AI ​​model. Simultaneously, analog data from physical sensors is extracted and normalized to obtain numerical values. Calculate the absolute difference between the two. Only when If the verification is successful, it will be determined that the sensor is drifting or visual recognition is blocked, and the authorization process will be automatically blocked. S32. Retrieve the current fusion status value of the device from the real-time device status database and perform a logical comparison with the "safety measure requirement status" field defined in the work order data structure to ensure that the physical status is completely consistent with the management requirements; S33. When an anomaly or data inconsistency occurs in the automatic verification process, a manual intervention process is triggered. Only work permit holders with encrypted digital signatures are allowed to perform mandatory confirmation or correction of the abnormal status after passing two-factor authentication, and an audit log is recorded.

4. The work order processing method for a power equipment status management system according to claim 1, characterized in that: in In step S1, data cleaning includes step S15: S15. Calculate two consecutive sampling times. and If the rate of change of the data exceeds a preset physical limit threshold, it is determined to be an abnormal jump, and the system automatically retains the previous time frame. The valid state value is determined, and the data source is marked as unstable.

5. The work order processing method for a power equipment status management system according to claim 1, characterized in that: in In step S4, the specific steps of the multi-level alarm strategy include: S41. When the calculated deviation When this occurs, an alarm flashing on the HMI interface is triggered; S42. When the calculated deviation When this happens, a pop-up blocking alarm is triggered; S43. When the calculated deviation When this happens, an SMS push alarm is triggered and associated operation permissions are locked.

6. The work order processing method for a power equipment status management system according to claim 1, characterized in that: in In step S1, the storage step S16 involved in building the real-time device state database includes: S16. Use an in-memory database cluster as the primary storage to store real-time status data for the most recent hour; use a time-series database as the secondary storage to store historical time-series data for 3 months to 5 years; and use an unstructured object storage system as the tertiary storage to store image snapshots of device status changes along the time path.

7. The work order processing method for a power equipment status management system according to claim 1, characterized in that: in In step S2, the region collision detection step S21 in multidimensional collision detection includes: S21. Obtain the set of work space coordinates for the new work order and the set of work space coordinates for all ongoing work orders; calculate the Euclidean distance between the two sets of coordinates; if the distance is less than the preset safe physical isolation distance, it is determined to be an area conflict, and physical isolation measures are recommended.

8. The work order processing method for a power equipment status management system according to claim 1, characterized in that: in In step S1, step S10, which involves accessing data from multiple sources, includes: S10. Access monitoring system data via OPCUA protocol; access miniature contact sensor data via Modbus-RTU to MQTT protocol; access displacement sensor data via Modbus-TCP protocol; and access high-definition video stream data via RTSP protocol.

9. A power equipment status management system, used to execute the work order processing method of the power equipment status management system according to any one of claims 1-8, characterized in that: include: The multi-source data acquisition module is used to access physical sensor data and visual image data from equipment through various industrial fieldbus protocols. The edge intelligence processing module is equipped with a heterogeneous data fusion unit. The heterogeneous data fusion unit is configured to execute a dynamic weighted fusion algorithm based on the online duration of the data source, dynamically adjust the weight of each data source according to the online stability of the information source to calculate the final state value, and perform AI image recognition. A tiered storage module is used to store real-time data, historical time-series data, and unstructured image data according to data access frequency and type. The intelligent service module includes a conflict detection service unit and a process engine service unit, which are used to perform multi-dimensional conflict calculations and triple verification mechanisms. The application module provides a human-computer interaction interface to display work order status and alarm information; The multi-source data acquisition module connects to the edge intelligent processing module, and the processed data is written to the hierarchical storage module. The intelligent service module reads the data from the hierarchical storage module, performs logical operations, and then drives the application module.

10. The system according to claim 9, characterized in that: The multi-source data acquisition module includes an OPCUA interface, an MQTT gateway interface, and an RTSP video stream interface; The edge intelligence processing module integrates the TensorRT inference engine for running the YOLO object detection model; The tiered storage module consists of a RedisCluster cluster, an InfluxDB time-series database, and a MinIO object storage service.

Images