Power distribution line power factor real-time monitoring and collaborative optimization system and method

Abstract

The invention relates to a distribution line power factor real-time monitoring and collaborative optimization system and method, and belongs to the technical field of distribution line energy efficiency management systems and methods. According to the technical scheme, the system comprises a multi-source data acquisition and preprocessing module, an edge node calculation and edge analysis module, a power factor real-time analysis and state evaluation module, a multi-subject collaborative optimization decision module and a dynamic reactive compensation control execution module. An edge and cloud data interaction and full life cycle management module; and an abnormity early warning and fault diagnosis module. The method has the beneficial effects that the line network loss and the operation and maintenance cost are reduced at the same time, the fault-tolerant capability is achieved, the modular design supports multi-scene expansion, the method not only adapts to the process requirements of vanadium-titanium steel rolling, smelting and the like, but also can be flexibly applied to other high-load industrial 11kV lines, and the power factor regulation and control effectiveness and the system full-life-cycle benefits are improved.

Description

Technical Field

[0001] This invention relates to a real-time monitoring and collaborative optimization system and method for the power factor of power distribution lines, belonging to the technical field of power distribution line energy efficiency management systems and methods. Background Technology

[0002] In heavy industrial enterprises, 11kV distribution lines supply power to high-power industrial loads such as rolling mills, electric arc furnaces, and fans. These loads are characterized by frequent start-stop cycles, large fluctuations in reactive power demand, and high harmonic content, resulting in a power factor that remains consistently low, between 0.75 and 0.85. Current power factor management for distribution lines generally faces the following technical bottlenecks: Traditional monitoring methods that rely on centralized cloud computing require uploading all data such as line voltage and current to a remote server. Due to the on-site communication conditions of 11kV lines (such as electromagnetic interference in the factory area and the dispersed distribution of towers), the data transmission delay reaches the second level, making it impossible to achieve real-time power factor perception. The existing solution only uses a single reactive power compensation device (such as a fixed capacitor bank) for passive adjustment, without considering the coordinated linkage between the power grid dispatch center, the vanadium-titanium production load side, and the compensation device. This results in a mismatch between the compensation capacity and the actual reactive power demand, and is prone to over-compensation or under-compensation. Equipment such as electric arc furnaces in vanadium-titanium production generate a large amount of harmonic current. Traditional power factor calculation methods do not eliminate harmonic components, resulting in a large deviation between the monitored value and the actual fundamental power factor, which in turn affects the effectiveness of optimization decisions. Dynamic changes in industrial loads (such as sudden load changes caused by the switching of steel rolling processes) require optimization strategies to have rapid response capabilities. However, the control algorithms of existing systems are mostly based on fixed parameters and cannot be dynamically adjusted according to load characteristics, resulting in unstable power factor optimization effects.

[0003] Therefore, there is an urgent need for a real-time power factor monitoring and collaborative optimization system for power distribution lines to improve the energy efficiency and operational stability of 11kV power distribution lines. Summary of the Invention

[0004] The purpose of this invention is to provide a real-time power factor monitoring and collaborative optimization system and method for power distribution lines. It achieves millisecond-level real-time power factor sensing through an edge computing architecture, avoids over- and under-compensation due to single-entity collaborative optimization, eliminates harmonic interference using a fundamental frequency extraction algorithm to ensure accurate monitoring, and adapts to load fluctuations and process changes through hierarchical control and flexible upgrades. Simultaneously, it reduces line network losses and maintenance costs, possesses fault tolerance capabilities, and its modular design supports multi-scenario expansion. It is suitable for processes such as vanadium-titanium steel rolling and smelting, and can also be flexibly applied to other high-load industrial 11kV lines, improving the effectiveness of power factor regulation and the system's overall lifecycle benefits, effectively solving the aforementioned problems in the background technology.

[0005] The technical solution of the present invention is: a power factor real-time monitoring and collaborative optimization system for power distribution lines, comprising a multi-source data acquisition and preprocessing module, an edge node calculation and edge analysis module, a power factor real-time analysis and status assessment module, a multi-subject collaborative optimization decision-making module, a dynamic reactive power compensation control execution module, an edge and cloud data interaction and full life cycle management module, and an anomaly early warning and fault diagnosis module; The multi-source data acquisition and preprocessing module collects 11kV line electrical parameters and equipment status data through distributed sensing terminals, and performs preprocessing using moving average filtering and data normalization. The electrical parameters include voltage, current, active power and reactive power. The edge node computing and edge analysis module is connected to the output of the multi-source data acquisition and preprocessing module. It adopts a distributed architecture of master edge nodes and slave edge nodes and is deployed on the line site to realize data dimensionality reduction, real-time computing scheduling and local decision caching, and reduce data transmission latency. The real-time power factor analysis and status assessment module is connected to the output of the edge node calculation and edge analysis module. It calculates the actual fundamental power factor based on the fundamental extraction algorithm, establishes a four-level status assessment standard, and identifies factors affecting the power factor. The four-level status assessment standard includes optimal, qualified, early warning, and emergency. The multi-subject collaborative optimization decision-making module connects to the output of the power factor real-time analysis and status assessment module and the edge and cloud data interaction and full life cycle management module. The multi-subject collaborative optimization decision-making module is linked with the power grid dispatch center, the vanadium-titanium production load side and the reactive power compensation device to construct an optimization function with the goal of minimizing network loss and maximizing the power factor compliance rate, and generate collaborative optimization strategies. The dynamic reactive power compensation control execution module connects to the output of the multi-subject collaborative optimization decision-making module and the abnormal early warning and fault diagnosis module. Its output is fed back to the edge node calculation and edge analysis module, which transforms the optimization strategy into control commands. It adopts hierarchical control logic with fast response and steady-state adjustment to drive the compensation device and load-side equipment to perform adjustments. The edge and cloud data interaction and full lifecycle management module is interconnected with the edge node computing and edge analysis module to realize encrypted data interaction between the edge and the cloud, and complete the full lifecycle management of system deployment, operation, maintenance and upgrade; The anomaly warning and fault diagnosis module is connected to the output of the power factor real-time analysis and status assessment module. Based on the fusion method of multi-dimensional thresholds and rule bases and machine learning, it monitors anomalies and diagnoses faults, triggering graded warnings.

[0006] The multi-source data acquisition and preprocessing module includes a data acquisition unit and a data preprocessing unit; Data acquisition unit: A distributed deployment approach is adopted, with intelligent sensing terminals installed on 11kV line towers, transformer outlets, and the load side of the vanadium-titanium production workshop. This includes: a three-phase voltage and current combined sensor used to collect the voltage of each phase of the line. and current Real-time data, adapting to the electromagnetic compatibility requirements of industrial environments; active and reactive power sensors: directly collect data on the total active power P, total reactive power Q, and reactive power distribution of each load branch; environmental and equipment status sensors: collect data on the operating temperature and switching status of reactive power compensation devices, as well as the temperature, humidity, and electromagnetic interference intensity of the environment in which the line is located. Data preprocessing unit: Addressing the susceptibility of data to electromagnetic interference and fluctuations caused by sudden load changes in industrial environments, a two-stage preprocessing mechanism is employed: Stage 1: Moving average filtering algorithm to eliminate high-frequency noise interference, the formula is: , in, The filtered data is at time k. The data represents the raw data collected at time i, and N is the length of the sliding window, which is dynamically adjusted according to the load fluctuation frequency. The second level is data synchronization and normalization processing. Based on the clock synchronization signal of the edge node, the data collected by different sensing terminals are unified with a timestamp. At the same time, the data of different dimensions of voltage and current are normalized to the [0,1] interval, which is convenient for subsequent edge computing analysis.

[0007] The edge node computing and edge analysis module includes an edge node deployment unit and an edge computing processing unit; Edge node deployment unit: adopts a distributed architecture of master edge nodes and slave edge nodes. Main edge node: Deployed in the substation of the Chengde Vanadium Titanium 11kV line, it adopts an industrial-grade edge computing gateway, integrating CPU and FPGA chips, and is responsible for global data aggregation and complex algorithm calculation. From the edge node: Deployed near the towers and load side in the middle section of the line, using lightweight edge terminals, responsible for local data preprocessing and simple calculations, and uploading the processing results to the main edge node; Edge computing processing unit: Enables real-time data analysis and preliminary decision-making, including: Data dimensionality reduction: Principal component analysis (PCA) algorithm is used to extract core features from multi-dimensional data to reduce data redundancy; Real-time computing scheduling: Based on task priority, dynamically allocate computing resources to edge nodes to ensure the response speed of critical tasks; Local decision caching: Frequently used optimization instructions are temporarily stored locally on edge nodes to avoid latency caused by relying on the cloud, and only periodically summarized data is uploaded to the cloud.

[0008] The real-time power factor analysis and condition assessment module includes a power factor accurate calculation unit, a condition assessment unit, and an influencing factor analysis unit. Power factor precision calculation unit: Addressing the harmonic interference problem in vanadium-titanium production loads, it employs a fundamental frequency extraction and power factor correction algorithm. The calculation formula is as follows: , Where λ is the actual fundamental power factor of the line; P base Q represents the fundamental active power of the line. base Q represents the fundamental reactive power of the line. harm This refers to harmonic reactive power; Condition Assessment Unit: Establish a power factor condition classification standard. Based on the operational requirements of the Chengde Vanadium Titanium 11kV line, with a target power factor ≥ 0.92, the conditions are divided into: Optimal state λ≥0.95: No optimization required, maintain the existing operating mode; Acceptable state 0.92≤λ<0.95: Slight optimization, achieved only through load-side fine-tuning; Warning status 0.85≤λ<0.92: Moderate optimization, activate reactive power compensation device; Emergency state λ < 0.85: Deep optimization, coordinated control of power grid dispatch and load side; Influencing Factor Analysis Unit: Through correlation analysis algorithms, it identifies key factors that lead to a decrease in power factor, including load-side factors, equipment-side factors, and grid-side factors.

[0009] The multi-agent collaborative optimization decision-making module includes a multi-agent interaction unit, an optimization objective function construction unit, and an optimization strategy generation unit; Multi-entity interactive unit: Establishes real-time communication with various entities through edge nodes: Power grid dispatch center: Receives the reactive power control indicators of the 11kV line issued by it; Vanadium-titanium production load side: Obtains the production plan and adjustable load list; Reactive power compensation device: Collects the available capacity and response speed parameters of SVG and capacitor bank; Optimization objective function building block: To minimize network loss and maximize power factor compliance, the optimization objective function is established as follows: , Where F is the overall cost of the optimization target; ΔP is the 11kV line network loss; η is the power factor compliance rate; α and β are weighting coefficients, which are dynamically adjusted according to the grid dispatch priority and vanadium-titanium production demand, and α+β=1; Optimization Strategy Generation Unit: Based on the objective function and the constraints of each subject, the Particle Swarm Optimization (PSO) algorithm is used to generate strategies: Load-side optimization: When production processes allow, the operating sequence of adjustable loads is adjusted to reduce the impact of reactive power superposition; Reactive power compensation device optimization: The compensation capacity of SVG and capacitor banks is dynamically allocated according to reactive power demand, with priority given to using SVG to cope with high-frequency fluctuations and capacitor banks to cope with steady-state reactive power deficits; Grid-side coordination: When the reactive power deficit of the line exceeds the local compensation capacity, reactive power support is requested from the grid dispatch center.

[0010] The dynamic reactive power compensation control execution module includes a control command parsing unit, a hierarchical control execution unit, and an execution status feedback unit. Control instruction parsing unit: converts optimization strategies into executable control instructions, including: Reactive power compensation device instructions: target reactive power output value of SVG and number of capacitor banks to be switched on and off, with a fixed capacity for each bank; Load side instructions: adjustable load start and stop time and operating power adjustment range; Hierarchical control execution unit: Employs hierarchical control logic with fast response and steady-state adjustment. Fast Response Layer: For sudden drops in power factor caused by load fluctuations, millisecond-level reactive power compensation is achieved through the IGBT module of the SVG. The control logic is as follows: when the power factor is detected to be <0.88, the SVG is immediately triggered to output at full capacity until it recovers to above 0.90. Steady-State Regulation Layer: When the power factor stabilizes in the 0.90-0.95 range, it is optimized to the optimal range of 0.92-0.95 by switching capacitor banks and adjusting the load operation mode. The control cycle is dynamically set according to the load fluctuation frequency. Execution status feedback unit: collects the execution status of the compensation device and the load-side equipment in real time, feeds it back to the edge node to form a control closed loop, and triggers re-optimization if execution deviation occurs.

[0011] The edge and cloud data interaction and full lifecycle management module includes a data interaction unit and a full lifecycle management unit; Data interaction unit: Employs a communication mode of proactive edge push and on-demand cloud retrieval. Edge proactive push: periodic and abnormal data are encrypted and pushed to the cloud; On-demand data retrieval from the cloud: The cloud platform sends data retrieval commands to edge nodes based on analysis needs; The full lifecycle management unit includes the deployment phase, operation phase, maintenance phase, and upgrade phase; Deployment phase: The cloud platform assigns unique identifiers to edge nodes and sensor terminals and completes parameter configuration; Operation phase: The cloud remotely monitors the computing resources of edge nodes and the operating status of sensor terminals. When the load on edge nodes is too high, some non-real-time tasks are dynamically migrated to the cloud. Maintenance phase: Based on the device operation data uploaded by edge nodes, maintenance work orders are generated and pushed to maintenance personnel; Upgrade phase: The cloud pushes the updated algorithm model package to the edge nodes, supporting remote OTA upgrades without requiring on-site downtime.

[0012] The anomaly warning and fault diagnosis module includes an anomaly monitoring unit, a fault diagnosis unit, and an early warning and handling unit; Anomaly monitoring unit: Sets multi-dimensional anomaly detection thresholds, including: Abnormal power factor: <0.85 or >0.98; Data anomalies: The fluctuation range of the collected data exceeds the normal range and the data interruption time exceeds the set duration; Equipment malfunctions: The operating temperature of the compensation device exceeds the threshold, the switch fails to operate, and the reactive power demand of the load-side equipment changes more rapidly than expected; Fault diagnosis unit: Employs a diagnostic method that combines rule base and machine learning. Rule base diagnosis: Establishing fault rules based on expert experience; Machine learning diagnostics: By using historical fault data accumulated at edge nodes, a fault diagnosis model is trained to identify complex faults; Early warning and response unit: Different early warning methods are triggered according to the level of abnormality. The levels of abnormality are divided into general abnormality, severe abnormality and emergency abnormality. General anomalies: Local audible and visual alarms are triggered at edge nodes, and logs are recorded on the cloud platform; Serious anomaly: A warning message will be pushed to the mobile app of the maintenance personnel, along with preliminary diagnostic results; Emergency Anomaly: In addition to push notifications, backup optimization strategies will be automatically activated to prevent the fault from escalating.

[0013] A method for real-time monitoring and collaborative optimization of power factor in power distribution lines, characterized by comprising the following steps: Step 1: System deployment and initialization, completing the deployment of edge nodes and sensor terminals, as well as cloud parameter configuration and interface debugging; Step 2: Multi-source data acquisition and preprocessing. The sensor terminal acquires data, and the edge nodes process the data through moving average filtering and normalization. Step 3: Edge computing and real-time power factor analysis. The main edge node calculates the fundamental power factor, assesses the operating status, and identifies influencing factors. Step 4: Multi-agent collaborative optimization decision-making, acquire multi-agent data, construct optimization functions and generate collaborative optimization strategies; Step 5: Dynamic reactive power compensation control is executed by parsing control commands and driving the equipment to perform adjustments through hierarchical control. Step 6: Edge and cloud data interaction and status feedback. Edge nodes push data to the cloud, and the cloud generates an operation report and issues optimization suggestions. Step 7: Full lifecycle management and optimization iteration, cloud monitoring of equipment status, regular updates to algorithm models, and adaptation to production process adjustments.

[0014] The beneficial effects of this invention are: it achieves millisecond-level real-time power factor sensing through edge computing architecture; multi-entity collaborative optimization avoids over- and under-compensation due to single compensation; fundamental frequency extraction algorithm eliminates harmonic interference to ensure accurate monitoring; and hierarchical control and flexible upgrades adapt to load fluctuations and process changes. At the same time, it reduces line network losses and operation and maintenance costs, has fault tolerance capabilities, and its modular design supports multi-scenario expansion. It is suitable for the needs of processes such as vanadium-titanium steel rolling and smelting, and can also be flexibly applied to other high-load industrial 11kV lines, improving the effectiveness of power factor regulation and the benefits of the system throughout its entire life cycle. Attached Figure Description

[0015] Figure 1 This is a system principle block diagram of the present invention; Figure 2 This is a block diagram illustrating the principle of the edge node calculation and edge analysis module of the present invention; Figure 3 This is a block diagram of the dynamic reactive power compensation control execution module of the present invention; Figure 4 This is a block diagram illustrating the principle of the edge and cloud data interaction and full lifecycle management module of this invention; Figure 5 This is a flowchart of the method of the present invention. Detailed Implementation

[0016] To make the purpose, technical solutions, and advantages of the invention's embodiments clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only a small part of the embodiments of the present invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0017] A power factor real-time monitoring and collaborative optimization system for power distribution lines includes a multi-source data acquisition and preprocessing module, an edge node calculation and edge analysis module, a power factor real-time analysis and status assessment module, a multi-subject collaborative optimization decision-making module, a dynamic reactive power compensation control execution module, an edge and cloud data interaction and full lifecycle management module, and an anomaly early warning and fault diagnosis module. The multi-source data acquisition and preprocessing module collects 11kV line electrical parameters and equipment status data through distributed sensing terminals, and performs preprocessing using moving average filtering and data normalization. The electrical parameters include voltage, current, active power and reactive power. The edge node computing and edge analysis module is connected to the output of the multi-source data acquisition and preprocessing module. It adopts a distributed architecture of master edge nodes and slave edge nodes and is deployed on the line site to realize data dimensionality reduction, real-time computing scheduling and local decision caching, and reduce data transmission latency. The real-time power factor analysis and status assessment module is connected to the output of the edge node calculation and edge analysis module. It calculates the actual fundamental power factor based on the fundamental extraction algorithm, establishes a four-level status assessment standard, and identifies factors affecting the power factor. The four-level status assessment standard includes optimal, qualified, early warning, and emergency. The multi-subject collaborative optimization decision-making module connects to the output of the power factor real-time analysis and status assessment module and the edge and cloud data interaction and full life cycle management module. The multi-subject collaborative optimization decision-making module is linked with the power grid dispatch center, the vanadium-titanium production load side and the reactive power compensation device to construct an optimization function with the goal of minimizing network loss and maximizing the power factor compliance rate, and generate collaborative optimization strategies. The dynamic reactive power compensation control execution module connects to the output of the multi-subject collaborative optimization decision-making module and the abnormal early warning and fault diagnosis module. Its output is fed back to the edge node calculation and edge analysis module, which transforms the optimization strategy into control commands. It adopts hierarchical control logic with fast response and steady-state adjustment to drive the compensation device and load-side equipment to perform adjustments. The edge and cloud data interaction and full lifecycle management module is interconnected with the edge node computing and edge analysis module to realize encrypted data interaction between the edge and the cloud, and complete the full lifecycle management of system deployment, operation, maintenance and upgrade; The anomaly warning and fault diagnosis module is connected to the output of the power factor real-time analysis and status assessment module. Based on the fusion method of multi-dimensional thresholds and rule bases and machine learning, it monitors anomalies and diagnoses faults, triggering graded warnings.

[0018] The multi-source data acquisition and preprocessing module includes a data acquisition unit and a data preprocessing unit; Data acquisition unit: A distributed deployment approach is adopted, with intelligent sensing terminals installed on 11kV line towers, transformer outlets, and the load side of the vanadium-titanium production workshop. This includes: a three-phase voltage and current combined sensor used to collect the voltage of each phase of the line. and current Real-time data, adapting to the electromagnetic compatibility requirements of industrial environments; active and reactive power sensors: directly collect data on the total active power P, total reactive power Q, and reactive power distribution of each load branch; environmental and equipment status sensors: collect data on the operating temperature and switching status of reactive power compensation devices, as well as the temperature, humidity, and electromagnetic interference intensity of the environment in which the line is located. Data preprocessing unit: Addressing the susceptibility of data to electromagnetic interference and fluctuations caused by sudden load changes in industrial environments, a two-stage preprocessing mechanism is employed: Stage 1: Moving average filtering algorithm to eliminate high-frequency noise interference, the formula is: , in, The filtered data is at time k. The data represents the raw data collected at time i, and N is the length of the sliding window, which is dynamically adjusted according to the load fluctuation frequency. The second level is data synchronization and normalization processing. Based on the clock synchronization signal of the edge node, the data collected by different sensing terminals are unified with a timestamp. At the same time, the data of different dimensions of voltage and current are normalized to the [0,1] interval, which is convenient for subsequent edge computing analysis.

[0019] The edge node computing and edge analysis module includes an edge node deployment unit and an edge computing processing unit; Edge node deployment unit: adopts a distributed architecture of master edge nodes and slave edge nodes. Main edge node: Deployed in the substation of the Chengde Vanadium Titanium 11kV line, it adopts an industrial-grade edge computing gateway, integrating CPU and FPGA chips, and is responsible for global data aggregation and complex algorithm calculation. From the edge node: Deployed near the towers and load side in the middle section of the line, using lightweight edge terminals, responsible for local data preprocessing and simple calculations, and uploading the processing results to the main edge node; Edge computing processing unit: Enables real-time data analysis and preliminary decision-making, including: Data dimensionality reduction: Principal component analysis (PCA) algorithm is used to extract core features from multi-dimensional data to reduce data redundancy; Real-time computing scheduling: Based on task priority, dynamically allocate computing resources to edge nodes to ensure the response speed of critical tasks; Local decision caching: Frequently used optimization instructions are temporarily stored locally on edge nodes to avoid latency caused by relying on the cloud, and only periodically summarized data is uploaded to the cloud.

[0020] The real-time power factor analysis and condition assessment module includes a power factor accurate calculation unit, a condition assessment unit, and an influencing factor analysis unit. Power factor precision calculation unit: Addressing the harmonic interference problem in vanadium-titanium production loads, it employs a fundamental frequency extraction and power factor correction algorithm. The calculation formula is as follows: , Where λ is the actual fundamental power factor of the line; P base Q represents the fundamental active power of the line. base Q represents the fundamental reactive power of the line. harm This refers to harmonic reactive power; Condition Assessment Unit: Establish a power factor condition classification standard. Based on the operational requirements of the Chengde Vanadium Titanium 11kV line, with a target power factor ≥ 0.92, the conditions are divided into: Optimal state λ≥0.95: No optimization required, maintain the existing operating mode; Acceptable state 0.92≤λ<0.95: Slight optimization, achieved only through load-side fine-tuning; Warning status 0.85≤λ<0.92: Moderate optimization, activate reactive power compensation device; Emergency state λ < 0.85: Deep optimization, coordinated control of power grid dispatch and load side; Influencing Factor Analysis Unit: Through correlation analysis algorithms, it identifies key factors that lead to a decrease in power factor, including load-side factors, equipment-side factors, and grid-side factors.

[0021] The multi-agent collaborative optimization decision-making module includes a multi-agent interaction unit, an optimization objective function construction unit, and an optimization strategy generation unit; Multi-entity interactive unit: Establishes real-time communication with various entities through edge nodes: Power grid dispatch center: Receives the reactive power control indicators of the 11kV line issued by it; Vanadium-titanium production load side: Obtains the production plan and adjustable load list; Reactive power compensation device: Collects the available capacity and response speed parameters of SVG and capacitor bank; Optimization objective function building block: To minimize network loss and maximize power factor compliance, the optimization objective function is established as follows: , Where F is the overall cost of the optimization target; ΔP is the 11kV line network loss; η is the power factor compliance rate; α and β are weighting coefficients, which are dynamically adjusted according to the grid dispatch priority and vanadium-titanium production demand, and α+β=1; Optimization Strategy Generation Unit: Based on the objective function and the constraints of each subject, the Particle Swarm Optimization (PSO) algorithm is used to generate strategies: Load-side optimization: When production processes allow, the operating sequence of adjustable loads is adjusted to reduce the impact of reactive power superposition; Reactive power compensation device optimization: The compensation capacity of SVG and capacitor banks is dynamically allocated according to reactive power demand, with priority given to using SVG to cope with high-frequency fluctuations and capacitor banks to cope with steady-state reactive power deficits; Grid-side coordination: When the reactive power deficit of the line exceeds the local compensation capacity, reactive power support is requested from the grid dispatch center.

[0022] The dynamic reactive power compensation control execution module includes a control command parsing unit, a hierarchical control execution unit, and an execution status feedback unit. Control instruction parsing unit: converts optimization strategies into executable control instructions, including: Reactive power compensation device instructions: target reactive power output value of SVG and number of capacitor banks to be switched on and off, with a fixed capacity for each bank; Load side instructions: adjustable load start and stop time and operating power adjustment range; Hierarchical control execution unit: Employs hierarchical control logic with fast response and steady-state adjustment. Fast Response Layer: For sudden drops in power factor caused by load fluctuations, millisecond-level reactive power compensation is achieved through the IGBT module of the SVG. The control logic is as follows: when the power factor is detected to be <0.88, the SVG is immediately triggered to output at full capacity until it recovers to above 0.90. Steady-State Regulation Layer: When the power factor stabilizes in the 0.90-0.95 range, it is optimized to the optimal range of 0.92-0.95 by switching capacitor banks and adjusting the load operation mode. The control cycle is dynamically set according to the load fluctuation frequency. Execution status feedback unit: collects the execution status of the compensation device and the load-side equipment in real time, feeds it back to the edge node to form a control closed loop, and triggers re-optimization if execution deviation occurs.

[0023] The edge and cloud data interaction and full lifecycle management module includes a data interaction unit and a full lifecycle management unit; Data interaction unit: Employs a communication mode of proactive edge push and on-demand cloud retrieval. Edge proactive push: periodic and abnormal data are encrypted and pushed to the cloud; On-demand data retrieval from the cloud: The cloud platform sends data retrieval commands to edge nodes based on analysis needs; The full lifecycle management unit includes the deployment phase, operation phase, maintenance phase, and upgrade phase; Deployment phase: The cloud platform assigns unique identifiers to edge nodes and sensor terminals and completes parameter configuration; Operation phase: The cloud remotely monitors the computing resources of edge nodes and the operating status of sensor terminals. When the load on edge nodes is too high, some non-real-time tasks are dynamically migrated to the cloud. Maintenance phase: Based on the device operation data uploaded by edge nodes, maintenance work orders are generated and pushed to maintenance personnel; Upgrade phase: The cloud pushes the updated algorithm model package to the edge nodes, supporting remote OTA upgrades without requiring on-site downtime.

[0024] The anomaly warning and fault diagnosis module includes an anomaly monitoring unit, a fault diagnosis unit, and an early warning and handling unit; Anomaly monitoring unit: Sets multi-dimensional anomaly detection thresholds, including: Abnormal power factor: <0.85 or >0.98; Data anomalies: The fluctuation range of the collected data exceeds the normal range and the data interruption time exceeds the set duration; Equipment malfunctions: The operating temperature of the compensation device exceeds the threshold, the switch fails to operate, and the reactive power demand of the load-side equipment changes more rapidly than expected; Fault diagnosis unit: Employs a diagnostic method that combines rule base and machine learning. Rule base diagnosis: Establishing fault rules based on expert experience; Machine learning diagnostics: By using historical fault data accumulated at edge nodes, a fault diagnosis model is trained to identify complex faults; Early warning and response unit: Different early warning methods are triggered according to the level of abnormality. The levels of abnormality are divided into general abnormality, severe abnormality and emergency abnormality. General anomalies: Local audible and visual alarms are triggered at edge nodes, and logs are recorded on the cloud platform; Serious anomaly: A warning message will be pushed to the mobile app of the maintenance personnel, along with preliminary diagnostic results; Emergency Anomaly: In addition to push notifications, backup optimization strategies will be automatically activated to prevent the fault from escalating.

[0025] A method for real-time monitoring and collaborative optimization of power factor in power distribution lines, characterized by comprising the following steps: Step 1: System deployment and initialization, completing the deployment of edge nodes and sensor terminals, as well as cloud parameter configuration and interface debugging; Step 2: Multi-source data acquisition and preprocessing. The sensor terminal acquires data, and the edge nodes process the data through moving average filtering and normalization. Step 3: Edge computing and real-time power factor analysis. The main edge node calculates the fundamental power factor, assesses the operating status, and identifies influencing factors. Step 4: Multi-agent collaborative optimization decision-making, acquire multi-agent data, construct optimization functions and generate collaborative optimization strategies; Step 5: Dynamic reactive power compensation control is executed by parsing control commands and driving the equipment to perform adjustments through hierarchical control. Step 6: Edge and cloud data interaction and status feedback. Edge nodes push data to the cloud, and the cloud generates an operation report and issues optimization suggestions. Step 7: Full lifecycle management and optimization iteration, cloud monitoring of equipment status, regular updates to algorithm models, and adaptation to production process adjustments.

[0026] In practical applications, this invention provides a real-time monitoring and collaborative optimization system for power factor of power distribution lines, including a multi-source data acquisition and preprocessing module, an edge node calculation and edge analysis module, a real-time power factor analysis and status assessment module, a multi-subject collaborative optimization decision-making module, a dynamic reactive power compensation control execution module, an edge and cloud data interaction and full life cycle management module, and an anomaly warning and fault diagnosis module. The multi-source data acquisition and preprocessing module collects 11kV line electrical parameters and equipment status data through distributed sensing terminals, and performs preprocessing using moving average filtering and data normalization. The electrical parameters include voltage, current, active power, and reactive power. The edge node computing and edge analysis module adopts a distributed architecture of master edge nodes and slave edge nodes, deployed on the line site, to realize data dimensionality reduction, real-time computing scheduling and local decision caching, and reduce data transmission latency; The real-time power factor analysis and status assessment module calculates the actual fundamental power factor based on the fundamental frequency extraction algorithm, establishes a four-level status assessment standard, and identifies factors affecting the power factor. The four-level status assessment standard includes optimal, qualified, early warning, and emergency. The multi-entity collaborative optimization decision-making module links the power grid dispatch center, the vanadium-titanium production load side, and the reactive power compensation device to construct an optimization function with the objectives of minimizing network losses and maximizing the power factor compliance rate, and generates a collaborative optimization strategy. The dynamic reactive power compensation control execution module transforms the optimization strategy into control commands and uses hierarchical control logic with fast response and steady-state adjustment to drive the compensation device and load-side equipment to perform adjustments. The edge and cloud data interaction and full lifecycle management module enables encrypted data interaction between the edge and the cloud, and completes full lifecycle management of system deployment, operation, maintenance and upgrade; The anomaly warning and fault diagnosis module is based on a fusion method of multi-dimensional thresholds, rule bases and machine learning to monitor anomalies and diagnose faults, triggering tiered warnings.

[0027] The multi-source data acquisition and preprocessing module includes a data acquisition unit and a data preprocessing unit; Data acquisition unit: A distributed deployment approach is adopted, with intelligent sensing terminals installed at 11kV line towers, transformer outlets, and the load side of the vanadium-titanium production workshop. This includes: a three-phase voltage and current combined sensor used to collect the voltage of each phase of the line. and current Real-time data, adapting to the electromagnetic compatibility requirements of industrial environments; active and reactive power sensors: directly collect data on the total active power P, total reactive power Q, and reactive power distribution of each load branch; environmental and equipment status sensors: collect data on the operating temperature and switching status of reactive power compensation devices (such as SVG and capacitor banks), as well as the temperature, humidity, and electromagnetic interference intensity of the environment in which the line is located. Data preprocessing unit: Addressing the susceptibility of data to electromagnetic interference and fluctuations caused by sudden load changes in industrial environments, a two-stage preprocessing mechanism is employed: Stage 1: Moving average filtering algorithm to eliminate high-frequency noise interference, the formula is: , in, The filtered data is at time k. The first level is the original data collected at time i, and N is the length of the sliding window, which is dynamically adjusted according to the load fluctuation frequency. The second level is data synchronization and normalization processing. Based on the clock synchronization signal of the edge node, the data collected by different sensing terminals are unified with a timestamp. At the same time, the data of different dimensions of voltage and current are normalized to the [0,1] interval, which is convenient for subsequent edge computing analysis.

[0028] The edge node computing and edge analysis module includes an edge node deployment unit and an edge computing processing unit; Edge node deployment unit: adopts a distributed architecture of master edge nodes and slave edge nodes. Main edge node: Deployed in the substation of Chengde Vanadium Titanium 11kV line, it adopts an industrial-grade edge computing gateway, integrates CPU and FPGA chips, and is responsible for global data aggregation and complex algorithm calculation (such as power factor deep analysis and collaborative optimization decision-making). From the edge node: Deployed near the towers and load side in the middle of the line, using lightweight edge terminals, responsible for local data preprocessing, simple calculations (such as preliminary calculation of real-time power factor and screening of abnormal data), and uploading the processing results to the main edge node; Edge computing processing unit: Enables real-time data analysis and preliminary decision-making, including: Data dimensionality reduction: Principal component analysis (PCA) algorithm is used to extract core features (such as reactive power fluctuation amplitude and voltage deviation rate) from multi-dimensional data (voltage, current, power, environmental parameters, etc.) to reduce data redundancy; Real-time computing scheduling: Based on task priority (e.g., the priority of analysis tasks when the power factor is abnormal is higher than that of regular monitoring tasks), computing resources of edge nodes are dynamically allocated to ensure the response speed of critical tasks. Local decision caching: Frequently used optimization instructions (such as fine-tuning instructions for reactive power compensation devices) are temporarily stored locally on the edge nodes to avoid latency caused by relying on the cloud. Only periodically summarized data (such as daily / weekly power factor statistics) are uploaded to the cloud.

[0029] The real-time power factor analysis and condition assessment module includes a power factor accurate calculation unit, a condition assessment unit, and an influencing factor analysis unit; Power factor precision calculation unit: Addressing the harmonic interference problem in vanadium-titanium production loads, it employs a fundamental frequency extraction and power factor correction algorithm. The calculation formula is as follows: , Where λ is the actual fundamental power factor of the line; P base Q represents the fundamental active power of the line. base Q represents the fundamental reactive power of the line. harm This refers to harmonic reactive power; Condition Assessment Unit: Establish a power factor condition classification standard. Based on the operational requirements of the Chengde Vanadium Titanium 11kV line, with a target power factor ≥ 0.92, the conditions are divided into: Optimal state λ≥0.95: No optimization required, maintain the existing operating mode; Acceptable state 0.92≤λ<0.95: Slight optimization, achieved only through load-side fine-tuning; Warning status 0.85≤λ<0.92: Moderate optimization, activate reactive power compensation device; Emergency state λ < 0.85: Deep optimization, coordinated control of power grid dispatch and load side; Influencing Factor Analysis Unit: Through correlation analysis algorithms, it identifies key factors that lead to a decrease in power factor, including load-side factors, equipment-side factors, and grid-side factors. Load-side factors: such as reactive power impact caused by the start-up and shutdown of the rolling mill, and excessive harmonic content of the electric arc furnace; Equipment-side factors: such as failure of reactive power compensation device, excessive reactive power of transformer excitation; Grid-side factors: such as 11kV line voltage deviation and reactive power flow fluctuations in the upstream grid.

[0030] The multi-agent collaborative optimization decision-making module includes a multi-agent interaction unit, an optimization objective function construction unit, and an optimization strategy generation unit; Multi-entity interactive unit: Establishes real-time communication with various entities through edge nodes: Power grid dispatch center: Receives the reactive power control indicators of 11kV lines (such as regional reactive power balance requirements) issued by the power grid dispatch center; Vanadium-titanium production load side: Obtains production plans (such as steel rolling process arrangements) and adjustable load lists (such as start-up and shutdown time windows of fans and water pumps); Reactive power compensation device: Collects available capacity and response speed parameters of SVG and capacitor banks; Optimization objective function building block: To minimize network loss and maximize power factor compliance, the optimization objective function is established as follows: , Where F is the overall cost of the optimization target; ΔP is the 11kV line network loss; η is the power factor compliance rate; α and β are weighting coefficients, which are dynamically adjusted according to the grid dispatch priority and vanadium-titanium production demand, and α+β=1; Optimization Strategy Generation Unit: Based on the objective function and constraints of each subject (such as uninterrupted production process on the load side and capacity limitations of compensation devices), the Particle Swarm Optimization (PSO) algorithm is used to generate strategies: Load-side optimization: When production processes allow, the operating sequence of adjustable loads is adjusted (such as staggering the start-up and shutdown times of high-power equipment) to reduce the impact of reactive power superposition; Reactive power compensation device optimization: The compensation capacity of SVG (continuous regulation) and capacitor banks (tiered regulation) is dynamically allocated according to reactive power demand, with priority given to using SVG to cope with high-frequency fluctuations and capacitor banks to cope with steady-state reactive power deficits; Grid-side coordination: When the reactive power deficit of the line exceeds the local compensation capacity, reactive power support is requested from the grid dispatch center (such as adjusting the main transformer taps of the upstream substation).

[0031] The dynamic reactive power compensation control execution module includes a control command parsing unit, a hierarchical control execution unit, and an execution status feedback unit; Control instruction parsing unit: converts optimization strategies into executable control instructions, including: Reactive power compensation device instruction: Target reactive power output value Q of SVG SVG The number of capacitor banks to be switched (n) is fixed, and the capacity of each bank is fixed; load-side command: adjustable load start / stop time (t). start t stop Operating power adjustment range ΔP load ; Hierarchical control execution unit: Employs hierarchical control logic with fast response and steady-state adjustment. Fast Response Layer: For sudden drops in power factor caused by load changes (such as electric arc furnace startup), millisecond-level reactive power compensation is achieved through the IGBT module of the SVG. The control logic is as follows: when λ < 0.88 is detected, the SVG is immediately triggered to output at full capacity until λ recovers to above 0.90. Steady-State Regulation Layer: When the power factor is stable in the 0.90-0.95 range, λ is optimized to the optimal range of 0.92-0.95 by switching capacitor banks and adjusting the load operation mode. The control cycle is dynamically set according to the load fluctuation frequency. Execution status feedback unit: Real-time acquisition of the execution status of the compensation device and load-side equipment (such as the actual reactive power output of SVG, capacitor bank switching status, and load operating power), and feedback to the edge node to form a control closed loop. If an execution deviation occurs (such as λ not meeting expectations after command execution), re-optimization is triggered.

[0032] The edge and cloud data interaction and full lifecycle management module includes a data interaction unit and a full lifecycle management unit; Data interaction unit: Employs a communication mode of proactive edge push and on-demand cloud retrieval. Edge proactive push: Periodic data (such as power factor statistics every 15 minutes and daily optimization effect reports) and abnormal data (such as power factor emergency alarms) are encrypted and pushed to the cloud; Cloud-based on-demand data retrieval: The cloud platform sends data retrieval commands to edge nodes based on analysis needs (such as raw data collected during a specific time period or the operation logs of a compensation device). The full lifecycle management unit includes the deployment phase, operation phase, maintenance phase, and upgrade phase; Deployment phase: The cloud platform assigns unique identifiers to edge nodes and sensing terminals and completes parameter configuration (such as sampling frequency and communication protocol). Operation phase: The cloud remotely monitors the computing resources of edge nodes and the operating status of sensor terminals. When the load on edge nodes is too high, some non-real-time tasks are dynamically migrated to the cloud. Maintenance phase: Based on the equipment operation data uploaded by the edge nodes (such as the failure rate of the compensation device and the drift of the sensor), a maintenance work order is generated and pushed to the operation and maintenance personnel; Upgrade phase: The cloud pushes update packages of algorithm models (such as power factor calculation models and optimization decision algorithms) to edge nodes, supporting remote OTA upgrades without requiring on-site downtime.

[0033] The anomaly warning and fault diagnosis module includes an anomaly monitoring unit, a fault diagnosis unit, and an early warning and handling unit; Anomaly monitoring unit: Sets multi-dimensional anomaly detection thresholds, including: Power factor anomaly: λ < 0.85 (emergency threshold) or λ > 0.98 (overcompensation threshold); Data anomalies: The fluctuation range of the collected data exceeds the normal range, and the data interruption time exceeds the set duration; Equipment malfunctions: The operating temperature of the compensation device exceeds the threshold, the switch fails to operate, and the reactive power demand of the load-side equipment changes more suddenly than expected; Fault diagnosis unit: Employs a diagnostic method that combines rule base and machine learning. Rule base diagnosis: Fault rules are established based on expert experience (e.g., "SVG outputs zero reactive power and temperature exceeds the limit → SVG overheating fault"). Machine learning diagnostics: By using historical fault data accumulated at edge nodes, a fault diagnosis model (such as a decision tree model) is trained to identify complex faults (such as falsely low power factor caused by harmonic interference). Early warning and response unit: Different early warning methods are triggered according to the level of abnormality (general, severe, emergency). The levels of abnormality are divided into general abnormality, severe abnormality and emergency abnormality. General anomalies: Local audible and visual alarms are triggered at edge nodes, and logs are recorded on the cloud platform; Serious anomaly: A warning message will be pushed to the mobile app of the maintenance personnel, along with preliminary diagnostic results; Emergency Anomaly: In addition to push notifications, automatically activate backup optimization strategies (such as deploying backup capacitor banks) to prevent the fault from escalating.

[0034] The optimization method for a real-time monitoring and collaborative optimization system for power factor of distribution lines includes the following steps: Step 1: System Deployment and Initialization 1.1 Deploy main edge nodes in the substation of the Chengde Vanadium Titanium 11kV distribution line, and deploy slave edge nodes in the middle towers and load-side workshops to complete the communication network between edge nodes (using industrial Ethernet + wireless LoRa backup mode). 1.2 Install voltage and current sensors, power sensors, and equipment status sensors on 11kV line towers, transformer outlets, and load sides, and connect the sensing terminals to the nearest edge nodes; 1.3 The cloud platform assigns a unique ID to each edge node and sensor terminal, and configures the data acquisition frequency (set according to the load fluctuation characteristics, such as the acquisition frequency on the load side of the rolling mill is higher than that of the line tower) and communication protocol (MQTT protocol, adapted for edge and cloud data interaction). 1.4 Load the initial algorithm model (power factor calculation model, optimization decision algorithm model) at the edge node, complete the interface debugging with the power grid dispatch center, vanadium-titanium production dispatch system, and reactive power compensation device control system, and the system enters standby state; Step 2: Multi-source data acquisition and preprocessing. 2.1 The sensing terminal collects the voltage (U) of the 11kV line in real time according to the configured acquisition frequency. a U b U c ), current ( I a , I b , I c Data such as active power P, reactive power Q, and temperature and load operating status of the compensation device; 2.2 First-level preprocessing of locally acquired data from edge nodes: The moving average filtering algorithm is used to eliminate high-frequency noise, filter out invalid data (such as missing values ​​caused by sensor offline) and mark them; 2.3 Synchronize the preprocessed local data from the edge nodes to the main edge nodes. The main edge nodes perform a second-level preprocessing on the data of the entire line: unify the timestamp, normalize data of different dimensions, extract core features (such as reactive power fluctuation amplitude), and form a standardized dataset. Step 3: Edge computing and real-time power factor analysis. 3.1 The main edge node invokes the power factor accurate calculation model, based on a standardized dataset, and extracts the fundamental active power P through Fourier transform. base Fundamental reactive power Q base Calculate the harmonic reactive power Q harm Substitute into the formula Calculate the actual fundamental power factor λ of the line; 3.2 The main edge node assesses the current status (optimal, qualified, warning, and emergency) based on the power factor status classification standard, and identifies the key influencing factors that lead to the current status (such as excessive harmonics of the electric arc furnace on the load side and insufficient capacity of the compensation device) through correlation analysis algorithm. 3.3 The main edge node caches the power factor calculation results, status assessment conclusions, and influencing factor analysis reports locally, and pushes them to the cloud platform at a set period (e.g., every 5 minutes). If it is in an early warning or emergency state, the abnormal early warning module is triggered immediately. Step 4: Multi-agent collaborative optimization decision-making. 4.1 The main edge node initiates data requests to the power grid dispatch center, vanadium-titanium production dispatch system, and reactive power compensation device control system to obtain data such as power grid reactive power control indicators, production plans (such as steel rolling process arrangements), adjustable load lists, and available capacity of compensation devices; 4.2 Optimization Objective Function for Main Edge Node Construction Set constraints (such as uninterrupted production process on the load side and maximum output capacity limit of the compensation device), and use the particle swarm optimization (PSO) algorithm to find the optimal solution. 4.3 Generate a collaborative optimization strategy based on the optimized solution: Clarify the adjustment sequence of load-side adjustable equipment, the compensation capacity allocation of reactive power compensation devices (SVG and capacitor banks), and whether it is necessary to apply for reactive power support from the power grid dispatch center, and form a standardized control instruction set; Step 5: Execution of dynamic reactive power compensation control. 5.1 The main edge node sends the control command set to the dynamic reactive power compensation control execution module, and the command parsing unit converts it into specific parameters that the device can execute (such as the SVG target reactive power output value Q). SVG =200kvar, number of capacitor banks switched (n=2). 5.2 The hierarchical control execution unit executes instructions based on the power factor status: if in an emergency state, it first activates the SVG fast response (millisecond level) to raise λ back to above 0.90; if in a warning or qualified state, it switches capacitor banks and adjusts the load operation mode through the steady-state regulation layer. 5.3 The execution status feedback unit collects the execution results of the equipment in real time (such as the actual reactive power output of SVG and the load operating power) and uploads them to the main edge node. The main edge node compares the execution results with the optimization target. If the deviation exceeds 5%, the strategy is triggered to be re-optimized. Step 6: Edge and cloud data interaction and status feedback. 6.1 The main edge node pushes the following data to the cloud platform at a set period: real-time power factor monitoring curve, optimization strategy execution log, device operating status data, and abnormal warning records; 6.2 The cloud platform summarizes and analyzes the received data to generate the "Daily / Weekly / Monthly Operation Report of Power Factor of Chengde Vanadium Titanium 11kV Line", which includes indicators such as power factor compliance rate, reduction in network loss, and operating efficiency of compensation devices; 6.3 Based on the operation report, the cloud platform sends optimization suggestions (such as adjusting algorithm model parameters and updating the fault diagnosis rule base) to the main edge node, and the main edge node completes the local configuration update after receiving the suggestions. Step 7: Full lifecycle management and optimization iteration. 7.1 The cloud platform remotely monitors the operating status of edge nodes and sensor terminals. When a device malfunction is detected (such as sensor drift or excessive computing load on edge nodes), a maintenance work order is generated and pushed to the maintenance personnel. 7.2 Regularly (e.g., quarterly) optimize the power factor calculation model and collaborative optimization algorithm model based on historical data accumulated in the cloud, and push update packages to edge nodes via OTA. Edge nodes complete the upgrade during non-peak production periods (e.g., early morning). 7.3 Based on the adjustments to the vanadium-titanium production process in Chengde (such as the addition of smelting equipment), expand the deployment of sensor terminals and update the multi-entity collaboration list through the cloud platform to ensure that the system continuously adapts to the line operation requirements.

[0035] Example: System edge node deployment: The main edge node is deployed at the edge cabinet of the substation in the rolling mill workshop, and the secondary edge nodes are deployed on the poles near rolling mills #1 and #2 in the workshop; Sensing terminals: Three-phase voltage and current sensors are installed on the 11kV incoming line towers, active and reactive power sensors are installed at the outlet of the rolling mill transformer, and temperature and status sensors are installed next to the SVG device and capacitor bank. Communication network: Master and slave edge nodes are connected via industrial Ethernet, sensor terminals and slave edge nodes communicate wirelessly via LoRa (anti-electromagnetic interference), and edge nodes communicate with the cloud via 5G industrial modules.

[0036] Data acquisition: The sensor terminal collects real-time data on voltage (10.5-11.5kV), current (800-1200A), active power (10-15MW), and reactive power (6-10Mvar) during the operation of the rolling mill. The acquisition frequency is set to 10 times / second to adapt to the fluctuation of the rolling mill load. Edge computing: Data is processed from edge nodes using a moving average filter (N=5), and the fundamental component is extracted from the main edge nodes. The power factor λ=0.83 (emergency state) is calculated. Analysis reveals that the key influencing factor is the excessive harmonic Q of the electric arc furnace of No. 1 rolling mill. harm =1.2Mvar; Collaborative optimization: The main edge node obtains the power grid dispatch indicators (regional reactive power balance requirements), production plan (No. 1 rolling mill needs to run for another 2 hours), and SVG available capacity (3Mvar), constructs the optimization function α=0.4, β=0.6, and generates the strategy: SVG outputs 2.5Mvar reactive power, and adjusts the operation sequence of the No. 2 rolling mill fan (staggering the high load period of the No. 1 rolling mill). Control Execution: The SVG responded within 50ms, outputting 2.5Mvar reactive power, and λ rose to 0.91; 10 minutes later, by switching one capacitor bank (capacity 1Mvar), λ was optimized to 0.93; Feedback and Maintenance: The edge node improved the optimization result λ from 0.83 to 0.93, reducing network loss by 8%, and pushed it to the cloud. The cloud generated an operation report and simultaneously detected that the SVG temperature was slightly high (not reaching the threshold), recorded the log, and prompted the maintenance personnel to focus on checking it during the next inspection.

[0037] This invention specifically addresses the problems of monitoring lag, insufficient strategy coordination, harmonic interference affecting accuracy, and poor system adaptability in power factor management of 11kV distribution lines. It achieves millisecond-level real-time power factor sensing through an edge computing architecture, avoids over- and under-compensation due to single-entity collaborative optimization, eliminates harmonic interference to ensure accurate monitoring through a fundamental frequency extraction algorithm, and adapts to load fluctuations and process changes through hierarchical control and flexible upgrades. Simultaneously, it reduces line network losses and maintenance costs, possesses fault tolerance capabilities, and its modular design supports multi-scenario expansion. It is suitable for processes such as vanadium-titanium steel rolling and smelting, and can also be flexibly applied to other high-load industrial 11kV lines, improving the effectiveness of power factor regulation and the overall lifecycle benefits of the system.

[0038] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A real-time monitoring and collaborative optimization system for power factor of power distribution lines, characterized in that: It includes a multi-source data acquisition and preprocessing module, an edge node computing and edge analysis module, a power factor real-time analysis and status assessment module, a multi-subject collaborative optimization decision-making module, a dynamic reactive power compensation control execution module, an edge and cloud data interaction and full lifecycle management module, and an anomaly early warning and fault diagnosis module; The multi-source data acquisition and preprocessing module collects 11kV line electrical parameters and equipment status data through distributed sensing terminals, and performs preprocessing using moving average filtering and data normalization. The electrical parameters include voltage, current, active power and reactive power. The edge node computing and edge analysis module is connected to the output of the multi-source data acquisition and preprocessing module. It adopts a distributed architecture of master edge nodes and slave edge nodes and is deployed on the line site to realize data dimensionality reduction, real-time computing scheduling and local decision caching, and reduce data transmission latency. The real-time power factor analysis and status assessment module is connected to the output of the edge node calculation and edge analysis module. It calculates the actual fundamental power factor based on the fundamental extraction algorithm, establishes a four-level status assessment standard, and identifies factors affecting the power factor. The four-level status assessment standard includes optimal, qualified, early warning, and emergency. The multi-subject collaborative optimization decision-making module connects to the output of the power factor real-time analysis and status assessment module and the edge and cloud data interaction and full life cycle management module. The multi-subject collaborative optimization decision-making module is linked with the power grid dispatch center, the vanadium-titanium production load side and the reactive power compensation device to construct an optimization function with the goal of minimizing network loss and maximizing the power factor compliance rate, and generate collaborative optimization strategies. The dynamic reactive power compensation control execution module connects to the output of the multi-subject collaborative optimization decision-making module and the abnormal early warning and fault diagnosis module. Its output is fed back to the edge node calculation and edge analysis module, which transforms the optimization strategy into control commands. It adopts hierarchical control logic with fast response and steady-state adjustment to drive the compensation device and load-side equipment to perform adjustments. The edge and cloud data interaction and full lifecycle management module is interconnected with the edge node computing and edge analysis module to realize encrypted data interaction between the edge and the cloud, and complete the full lifecycle management of system deployment, operation, maintenance and upgrade; The anomaly warning and fault diagnosis module is connected to the output of the power factor real-time analysis and status assessment module. Based on the fusion method of multi-dimensional thresholds and rule bases and machine learning, it monitors anomalies and diagnoses faults, triggering graded warnings.

2. The power factor real-time monitoring and collaborative optimization system for power distribution lines according to claim 1, characterized in that: The multi-source data acquisition and preprocessing module includes a data acquisition unit and a data preprocessing unit; Data acquisition unit: A distributed deployment approach is adopted, with intelligent sensing terminals installed on 11kV line towers, transformer outlets, and the load side of the vanadium-titanium production workshop. This includes: a three-phase voltage and current combined sensor used to collect the voltage of each phase of the line. and current Real-time data, adapting to the electromagnetic compatibility requirements of industrial environments; active and reactive power sensors: directly collect data on the total active power P, total reactive power Q, and reactive power distribution of each load branch; environmental and equipment status sensors: collect data on the operating temperature and switching status of reactive power compensation devices, as well as the temperature, humidity, and electromagnetic interference intensity of the environment in which the line is located. Data preprocessing unit: Addressing the susceptibility of data to electromagnetic interference and fluctuations caused by sudden load changes in industrial environments, a two-stage preprocessing mechanism is employed: Stage 1: Moving average filtering algorithm to eliminate high-frequency noise interference, the formula is: , in, The filtered data is at time k. The data represents the raw data collected at time i, and N is the length of the sliding window, which is dynamically adjusted according to the load fluctuation frequency. The second level is data synchronization and normalization processing. Based on the clock synchronization signal of the edge node, the data collected by different sensing terminals are unified with a timestamp. At the same time, the data of different dimensions of voltage and current are normalized to the [0,1] interval, which is convenient for subsequent edge computing analysis.

3. The power factor real-time monitoring and collaborative optimization system for power distribution lines according to claim 1, characterized in that: The edge node computing and edge analysis module includes an edge node deployment unit and an edge computing processing unit; Edge node deployment unit: adopts a distributed architecture of master edge nodes and slave edge nodes. Main edge node: Deployed in the substation of the Chengde Vanadium Titanium 11kV line, it adopts an industrial-grade edge computing gateway, integrating CPU and FPGA chips, and is responsible for global data aggregation and complex algorithm calculation. From the edge node: Deployed near the towers and load side in the middle section of the line, using lightweight edge terminals, responsible for local data preprocessing and simple calculations, and uploading the processing results to the main edge node; Edge computing processing unit: Enables real-time data analysis and preliminary decision-making, including: Data dimensionality reduction: Principal component analysis (PCA) algorithm is used to extract core features from multi-dimensional data to reduce data redundancy; Real-time computing scheduling: Based on task priority, dynamically allocate computing resources to edge nodes to ensure the response speed of critical tasks; Local decision caching: Frequently used optimization instructions are temporarily stored locally on edge nodes to avoid latency caused by relying on the cloud, and only periodically summarized data is uploaded to the cloud.

4. The power factor real-time monitoring and collaborative optimization system for power distribution lines according to claim 1, characterized in that: The real-time power factor analysis and condition assessment module includes a power factor accurate calculation unit, a condition assessment unit, and an influencing factor analysis unit. Power factor precision calculation unit: Addressing the harmonic interference problem in vanadium-titanium production loads, it employs a fundamental frequency extraction and power factor correction algorithm. The calculation formula is as follows: Where λ is the actual fundamental power factor of the line; P base Q represents the fundamental active power of the line. base Q represents the fundamental reactive power of the line. harm This refers to harmonic reactive power; Condition Assessment Unit: Establish a power factor condition classification standard. Based on the operational requirements of the Chengde Vanadium Titanium 11kV line, with a target power factor ≥ 0.92, the conditions are divided into: Optimal state λ≥0.95: No optimization required, maintain the existing operating mode; Acceptable state 0.92≤λ<0.95: Slight optimization, achieved only through load-side fine-tuning; Warning status 0.85≤λ<0.92: Moderate optimization, activate reactive power compensation device; Emergency state λ < 0.85: Deep optimization, coordinated control of power grid dispatch and load side; Influencing Factor Analysis Unit: Through correlation analysis algorithms, it identifies key factors that lead to a decrease in power factor, including load-side factors, equipment-side factors, and grid-side factors.

5. The power factor real-time monitoring and collaborative optimization system for power distribution lines according to claim 1, characterized in that: The multi-agent collaborative optimization decision-making module includes a multi-agent interaction unit, an optimization objective function construction unit, and an optimization strategy generation unit; Multi-entity interactive unit: Establishes real-time communication with various entities through edge nodes: Power grid dispatch center: Receives the reactive power control indicators of the 11kV line issued by it; Vanadium-titanium production load side: Obtains the production plan and adjustable load list; Reactive power compensation device: Collects the available capacity and response speed parameters of SVG and capacitor bank; Optimization objective function building block: To minimize network loss and maximize power factor compliance, the optimization objective function is established as follows: , Where F represents the overall cost of the optimization target; ΔP represents the network loss of the 11kV line; η is the power factor compliance rate; α and β are weighting coefficients, which are dynamically adjusted according to the grid dispatch priority and vanadium-titanium production demand, and α+β=1; Optimization Strategy Generation Unit: Based on the objective function and the constraints of each subject, the Particle Swarm Optimization (PSO) algorithm is used to generate strategies: Load-side optimization: When production processes allow, the operating sequence of adjustable loads is adjusted to reduce the impact of reactive power superposition; Reactive power compensation device optimization: The compensation capacity of SVG and capacitor banks is dynamically allocated according to reactive power demand, with priority given to using SVG to cope with high-frequency fluctuations and capacitor banks to cope with steady-state reactive power deficits; Grid-side coordination: When the reactive power deficit of the line exceeds the local compensation capacity, reactive power support is requested from the grid dispatch center.

6. The power factor real-time monitoring and collaborative optimization system for power distribution lines according to claim 1, characterized in that: The dynamic reactive power compensation control execution module includes a control instruction parsing unit, a hierarchical control execution unit, and an execution status feedback unit. Control instruction parsing unit: converts optimization strategies into executable control instructions, including: Reactive power compensation device instructions: SVG target reactive power output value and number of capacitor banks to be switched, with a fixed capacity for each bank; Load-side commands: Adjustable load start / stop times and operating power adjustment range; Hierarchical control execution unit: Employs hierarchical control logic with fast response and steady-state adjustment. Fast Response Layer: For sudden drops in power factor caused by load fluctuations, millisecond-level reactive power compensation is achieved through the IGBT module of the SVG. The control logic is as follows: when the power factor is detected to be <0.88, the SVG is immediately triggered to output at full capacity until it recovers to above 0.

90. Steady-State Regulation Layer: When the power factor stabilizes in the 0.90-0.95 range, it is optimized to the optimal range of 0.92-0.95 by switching capacitor banks and adjusting the load operation mode. The control cycle is dynamically set according to the load fluctuation frequency. Execution status feedback unit: collects the execution status of the compensation device and the load-side equipment in real time, feeds it back to the edge node to form a control closed loop, and triggers re-optimization if execution deviation occurs.

7. The power factor real-time monitoring and collaborative optimization system for power distribution lines according to claim 1, characterized in that: The edge and cloud data interaction and full lifecycle management module includes a data interaction unit and a full lifecycle management unit; Data interaction unit: Employs a communication mode of proactive edge push and on-demand cloud retrieval. Edge proactive push: periodic and abnormal data are encrypted and pushed to the cloud; On-demand data retrieval from the cloud: The cloud platform sends data retrieval commands to edge nodes based on analysis needs; The full lifecycle management unit includes the deployment phase, operation phase, maintenance phase, and upgrade phase; Deployment phase: The cloud platform assigns unique identifiers to edge nodes and sensor terminals and completes parameter configuration; Operation phase: The cloud remotely monitors the computing resources of edge nodes and the operating status of sensor terminals. When the load on edge nodes is too high, some non-real-time tasks are dynamically migrated to the cloud. Maintenance phase: Based on the device operation data uploaded by edge nodes, maintenance work orders are generated and pushed to maintenance personnel; Upgrade phase: The cloud pushes the updated algorithm model package to the edge nodes, supporting remote OTA upgrades without requiring on-site downtime.

8. The power factor real-time monitoring and collaborative optimization system for power distribution lines according to claim 1, characterized in that: The anomaly warning and fault diagnosis module includes an anomaly monitoring unit, a fault diagnosis unit, and an early warning and handling unit; Anomaly monitoring unit: Sets multi-dimensional anomaly detection thresholds, including: Abnormal power factor: <0.85 or >0.98; Data anomalies: The fluctuation range of the collected data exceeds the normal range and the data interruption time exceeds the set duration; Equipment malfunctions: The operating temperature of the compensation device exceeds the threshold, the switch fails to operate, and the reactive power demand of the load-side equipment changes more rapidly than expected; Fault diagnosis unit: Employs a diagnostic method that combines rule base and machine learning. Rule base diagnosis: Establishing fault rules based on expert experience; Machine learning diagnostics: By using historical fault data accumulated at edge nodes, a fault diagnosis model is trained to identify complex faults; Early warning and response unit: Different early warning methods are triggered according to the level of abnormality. The levels of abnormality are divided into general abnormality, severe abnormality and emergency abnormality. General anomalies: Local audible and visual alarms are triggered at edge nodes, and logs are recorded on the cloud platform; Serious anomaly: A warning message will be pushed to the mobile app of the maintenance personnel, along with preliminary diagnostic results; Emergency Anomaly: In addition to push notifications, backup optimization strategies will be automatically activated to prevent the fault from escalating.

9. A method for real-time monitoring and collaborative optimization of power factor in power distribution lines, characterized in that... Includes the following steps: Step 1: System deployment and initialization, completing the deployment of edge nodes and sensor terminals, as well as cloud parameter configuration and interface debugging; Step 2: Multi-source data acquisition and preprocessing. The sensor terminal acquires data, and the edge nodes process the data through moving average filtering and normalization. Step 3: Edge computing and real-time power factor analysis. The main edge node calculates the fundamental power factor, assesses the operating status, and identifies influencing factors. Step 4: Multi-agent collaborative optimization decision-making, acquire multi-agent data, construct optimization functions and generate collaborative optimization strategies; Step 5: Dynamic reactive power compensation control is executed by parsing control commands and driving the equipment to perform adjustments through hierarchical control. Step 6: Edge and cloud data interaction and status feedback. Edge nodes push data to the cloud, and the cloud generates an operation report and issues optimization suggestions. Step 7: Full lifecycle management and optimization iteration, cloud monitoring of equipment status, regular updates to algorithm models, and adaptation to production process adjustments.

Images