Low-voltage treatment method based on big data clustering and analogue simulation

Abstract

The invention discloses a low-voltage treatment method based on big data clustering and analogue simulation, and relates to the technical field of power distribution network voltage treatment, and the method comprises the steps: S1, carrying out the real-time collection and preprocessing of data, and accessing the multi-source data in the operation process of a power distribution network in real time; according to the method, the dynamic clustering model is constructed by accessing the power grid operation data in real time, so that working condition changes such as load abrupt change and equipment fault can be quickly responded, a targeted treatment scheme is synchronously generated, the problems of low-voltage discovery lagging and passive remedy in traditional treatment are solved, real-time response and flexible adaptation to the uncertainty of the power grid are realized, and the reliability of the power grid is improved. According to the low-voltage management system, deep linkage of analogue simulation and actual management is achieved by building an intelligent interaction system, simulation model parameters are dynamically optimized, it is ensured that a simulation result fits the field reality, the problems that in the prior art, simulation and management are disjointed, and the pertinence of a management scheme is insufficient are solved, and the timeliness, flexibility and accuracy of low-voltage management are overall improved.

Description

Technical Field

[0001] This invention relates to the field of distribution network voltage management technology, specifically a low-voltage management method based on big data clustering and simulation. Background Technology

[0002] Low voltage is a common voltage quality problem in power system operation, especially during peak electricity consumption periods such as summer and winter, when it occurs frequently due to a large user base. These problems not only lead to a poor electricity experience for users but also significantly increase the workload of frontline service staff. Implementing low voltage management—that is, developing scientific, effective, and economical analysis and management solutions and evaluation systems based on a full, clear, and comprehensive understanding of low voltage problems in the distribution network—is a key measure for power grid companies to improve distribution network voltage quality, promote refined management of the distribution network, improve power supply service quality, and establish a positive social image.

[0003] With the continuous growth of electricity demand, especially during peak summer and winter periods, low voltage issues are becoming increasingly prominent, causing poor user experience and increasing the workload of grassroots power supply service personnel. Implementing low voltage management and developing scientific and effective analysis and management solutions are crucial for improving the voltage quality of the distribution network, promoting refined management of the distribution network, enhancing power supply service quality, and improving the social image of power grid companies. Currently, comprehensively building a modern, intensive, and digitalized power supply service command system has become an important direction for power grid companies to enhance their core functions and improve their core competitiveness. Integrating service resources from equipment, marketing, and dispatch departments with the power supply service command center as the hub, and eliminating bottlenecks and departmental barriers in power supply services, can inject digital, intelligent, and lean "heartfelt service" into serving a better life and building a new power system. Efficient low voltage management is a crucial link in this system to ensure the quality of power supply services, and it is of great significance for promoting the digital transformation of the power grid and improving the reliability of power grid operation.

[0004] However, existing low-voltage management technologies still have certain shortcomings. Traditional management relies heavily on manual data statistical analysis, which cannot perform real-time attribution analysis and intelligent judgment of user demands. Furthermore, data from multiple systems are difficult to correlate effectively, resulting in delayed low-voltage detection and only passive post-event remediation. In addition, existing clustering analysis models are difficult to adapt to the dynamic changes in the power grid's operating status. They cannot adjust the analysis results in a timely manner when faced with sudden load changes, equipment failures, and other operating conditions. At the same time, simulation is disconnected from actual management, and the fixed parameters of the simulation model make it difficult to optimize based on field feedback. This results in insufficient targeting and effectiveness of management solutions, failing to meet the needs of refined and intelligent management of distribution networks. Therefore, developing a low-voltage management method based on big data clustering and simulation is of great significance. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a low-voltage management method based on big data clustering and simulation. It can construct a dynamic clustering model by accessing real-time power grid operation data, and can quickly respond to changes in operating conditions such as load surges and equipment failures, and simultaneously generate targeted management solutions. It achieves real-time response and flexible adaptation to power grid uncertainties. By building an intelligent interactive system, it realizes deep linkage between simulation and actual management, dynamically optimizes simulation model parameters, and ensures that simulation results are consistent with the actual situation on site, thus improving the timeliness, flexibility and accuracy of low-voltage management.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a low-voltage management method based on big data clustering and simulation, comprising the following steps:

[0007] S1. Real-time data acquisition and preprocessing: Real-time access to multi-source data during the operation of the power distribution network, including voltage data, current data, load data and equipment operating status data. The acquired multi-source data is preprocessed to remove null values ​​and outliers, forming standardized data for subsequent analysis.

[0008] S2. Dynamic clustering analysis: Establish a dynamic clustering model, input the obtained standardized data into the dynamic clustering model, and continuously monitor the power grid operation status. When a load change or equipment failure is detected, the re-clustering analysis process is automatically triggered to update the clustering results and divide the low voltage impact area and feature categories.

[0009] S3. Preliminary governance plan generation: Based on the obtained clustering results, combined with the distribution network structure and historical governance experience, a preliminary governance plan for the current low voltage situation is generated.

[0010] S4. Simulation and governance interaction optimization: Build an intelligent interaction system, input the generated preliminary governance plan into the simulation module of the intelligent interaction system, and adjust the parameters and assumptions of the simulation model based on the actual power grid feedback operation data during the simulation process. At the same time, new strategies and measures in actual governance are fed back to the simulation model to verify and optimize the new strategies and measures.

[0011] S5. Final resolution plan execution: Based on the simulation optimization results, determine the optimal low voltage resolution plan and execute the optimal low voltage resolution plan.

[0012] Furthermore, the real-time data acquisition and preprocessing process in step S1 includes:

[0013] First, input voltage data, current data, load data, and equipment operating status data. At the same time, supplement the input with user profile data, 95598 work order data, PMS2.0 system data, and external weather data.

[0014] Next, the collected multi-source data is preprocessed. For null values, the RU-Net power missing data reconstruction model is used. After training the model by dividing it into training and test sets, the null values ​​are output by the trained model.

[0015] For outliers, the standard deviation of the data is calculated to determine the outlier threshold. Data exceeding the threshold are marked as outliers and removed to form standardized data.

[0016] Furthermore, the dynamic clustering analysis process in step S2 includes:

[0017] First, input the obtained standardized data into a dynamic clustering model built based on the improved K-means algorithm. The model initializes the weighted cluster centers using the following formula: Among them, C j For the j-th cluster center, x i For the i-th standardized data point, ω ij For data point x i For cluster center C j The initial weights, ω ij The importance of a data point is determined based on the importance of the corresponding distribution area. The importance of a distribution area is comprehensively evaluated by the number of users, load level, and power supply reliability requirements. The higher the importance, the greater the weight of the corresponding data point.

[0018] Then, the Euclidean distance between each data point and the cluster center is calculated, and the data points are assigned to the nearest cluster. The power grid operation data is then continuously tracked through the built-in state monitoring module of the model.

[0019] When the load data changes by more than a preset value within a unit of time, it is determined to be a load change; when the equipment operating status data exceeds the normal threshold, it is determined to be an equipment failure.

[0020] When a sudden load change or equipment failure occurs, re-clustering is triggered. During re-clustering, the cluster centers are first updated using the following formula: Among them, C′ j Let S be the updated j-th cluster center, α be the weight coefficient of historical cluster centers, and S be the weight coefficient of historical cluster centers. j Let |S| be the set of data points contained in the j-th cluster. j |For set S jThe number of data points, α, is determined through historical cluster stability analysis. The smaller the fluctuation of historical cluster results, the larger the value of α, and vice versa. The distance calculation and data allocation steps are repeated to output the updated cluster results and divide the low voltage affected area and feature category.

[0021] Furthermore, the simulation and governance interaction optimization process in step S4 includes:

[0022] First, input the preliminary treatment plan into the simulation module of the intelligent interactive system. The module loads the distribution network topology parameters and initializes the simulation environment.

[0023] Next, the simulation is run to receive real-time voltage and current data from the actual power grid and compare the feedback data with the simulation output data.

[0024] When the deviation exceeds the allowable range, adjust the simulation model parameters using the following formula: Among them, P′ k For the adjusted k-th model parameter, P k Here, β is the k-th model parameter before adjustment, and D is the parameter adjustment coefficient. k β is the deviation between the simulated value and the actual feedback value corresponding to the k-th parameter. It is determined based on parameter sensitivity analysis. The greater the influence of the parameter on the simulation results, the smaller the value of β should be, so as to avoid excessive parameter adjustment.

[0025] Simultaneously, new strategies and measures adopted in actual governance are collected and transformed into simulation parameter input models to simulate the governance effects under different working conditions. Based on the simulation results after parameter adjustment and the simulation effects of new measures, the simulation model is optimized, and the optimized governance scheme parameters are output.

[0026] Furthermore, in step S1, when collecting multi-source data, streaming data real-time computing technology is adopted. The operating data of each substation and each line is synchronously acquired through distributed data acquisition nodes. The acquisition frequency is determined according to the data type. The acquisition frequency of voltage and current data is no less than once every 5 minutes, and the acquisition frequency of equipment operating status data is no less than once every 15 minutes.

[0027] Furthermore, after dividing the low voltage impact area and feature category in step S2, a visualization display interface is constructed based on a distribution network map. The low voltage impact area is marked on the power grid topology map of the interface with different colors, and the color depth corresponds to the severity of low voltage. Feature categories are associated with the corresponding areas in the form of labels.

[0028] Furthermore, in step S3, when generating a preliminary governance plan based on the distribution network structure, the line connection relationships, transformer capacity, and conductor cross-sectional parameters of the distribution network are first obtained. The governance priority of the low-voltage area is calculated using the following formula: P = γ1L + γ2U + γ3C, where P is the governance priority, L is the standardized value of the duration of low voltage, U is the standardized value of the number of affected users, C is the standardized value of the transformer capacity, and γ1, γ2, and γ3 are weight coefficients. γ1, γ2, and γ3 are determined by the analytic hierarchy process (AHP). Experts in power grid operation and maintenance, planning, and marketing are invited to construct a judgment matrix. After passing the matrix consistency test, the final weights are obtained. The correlation between the current low-voltage area and the key nodes of the network is analyzed. Based on the correlation and governance priority, governance measures that are close to the key nodes and have low implementation difficulty are selected and incorporated into the preliminary governance plan.

[0029] Furthermore, in step S4, when verifying the new strategy and measures, multiple sets of comparative simulation experiments are set up. One set adopts the new strategy and measures, while the other set adopts traditional governance measures. The simulation is run under the same initial operating conditions, and the voltage recovery time and voltage stability index in the simulation results of the two sets are compared.

[0030] Furthermore, when implementing the optimal low voltage mitigation scheme in step S5, a phased implementation plan is formulated. The first phase addresses areas with high low voltage severity and affecting many users, the second phase addresses areas with moderate severity, and the third phase addresses the remaining areas. After each phase, the effectiveness is verified.

[0031] Furthermore, throughout the entire governance process, a data log recording system is established to record data collection results, cluster analysis process, scheme generation content, simulation optimization parameters, and governance execution status in real time, forming a complete governance data archive.

[0032] Compared with existing technologies, this low-voltage management method based on big data clustering and simulation has the following advantages:

[0033] This invention constructs a dynamic clustering model by accessing real-time power grid operation data, enabling rapid response to changes in operating conditions such as load surges and equipment failures. It simultaneously generates targeted mitigation solutions, solving the problems of delayed low-voltage detection and passive remediation in traditional mitigation methods. This achieves real-time response and flexible adaptation to power grid uncertainties. By building an intelligent interactive system, it achieves deep linkage between simulation and actual mitigation, dynamically optimizing simulation model parameters to ensure simulation results closely match real-world conditions. This addresses the issues of disconnect between simulation and mitigation, and insufficient targeting of mitigation solutions in existing technologies. Overall, it improves the timeliness, flexibility, and accuracy of low-voltage mitigation, promoting a shift in distribution network management from process-driven to data-driven approaches, reducing operation and maintenance costs, and increasing user satisfaction with electricity usage.

[0034] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from an examination of the following, or may be learned from the practice of the invention. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0036] Figure 1 This is a flowchart of a low-voltage management method based on big data clustering and simulation.

[0037] Figure 2 This is a flowchart of a low-voltage management method based on big data clustering and simulation. Detailed Implementation

[0038] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0039] This invention patent provides a flowchart of a low-voltage management method based on big data clustering and simulation, which clarifies the complete technical solution. (See attached document.) Figure 1 and Figure 2 The core content is as follows:

[0040] Real-time data acquisition and preprocessing: Real-time access to multi-source data such as voltage, current, load, and equipment operating status of the power distribution network, supplementing user profiles, 95598 work orders, PMS2.0 system, and external weather data. A specific model is used to fill in missing values, and outliers are removed by determining thresholds based on standard deviation, forming standardized data. Data acquisition frequency is set according to data type to ensure real-time performance.

[0041] Dynamic clustering analysis: A dynamic clustering model based on an improved algorithm is constructed, and standardized data is continuously input to monitor the power grid status. When a sudden load change or equipment failure is detected, re-clustering is automatically triggered, the clustering results are updated, the low-voltage impact area and feature categories are accurately divided, and the area and features are labeled through a visual interface.

[0042] Preliminary governance plan generation: Combining clustering results, distribution network structure parameters, and historical governance experience, governance priorities are calculated, and governance measures that are close to key nodes and have low implementation difficulty are selected to generate a targeted preliminary plan.

[0043] Simulation and governance interaction optimization: An intelligent interactive system is built, and the preliminary plan is input into the simulation module to load the network topology parameters and initialize the environment. During simulation, the model parameters are adjusted based on actual power grid feedback data, and new strategies and measures from actual governance are input into the model. The optimization is verified by comparing with simulation experiments.

[0044] Final governance plan implementation: The optimal solution is determined based on the simulation optimization results, and implemented in stages according to the severity and scope of the low voltage. The effect is verified after each stage.

[0045] Throughout the governance process, a data log recording system was established to record data collection, cluster analysis, scheme generation, simulation optimization, and execution in real time, forming a complete governance archive. This scheme achieves real-time response to changes in power grid operating conditions through dynamic clustering, and enhances the scheme's pertinence through deep integration of simulation and governance. It solves problems such as delayed detection and insufficient scheme adaptability in traditional governance, and improves the timeliness, flexibility, and accuracy of low-voltage governance as a whole.

[0046] Example 1

[0047] This embodiment is applied to the low-voltage management scenario of urban power distribution networks. Given the characteristics of this area—dense user populations, large load fluctuations, and pronounced seasonal peaks—low-voltage problems frequently occur during peak electricity consumption periods such as summer and winter. Traditional management methods suffer from slow response times and insufficient targeted solutions, severely impacting user experience and grid stability. (See also...) Figure 1 and Figure 2 This embodiment uses a low-voltage management method based on big data clustering and simulation to achieve rapid identification, accurate analysis and efficient management of low-voltage problems in the distribution network, thereby comprehensively improving the voltage quality and power supply reliability of the distribution network.

[0048] Real-time data acquisition and preprocessing employs streaming real-time computing technology, synchronously acquiring operational data from each transformer substation and line of the city's power distribution network through distributed data acquisition nodes. Core acquired data includes voltage, current, load, and equipment operating status data. Supplementary data includes user profile data, 95598 work order data, PMS2.0 system data, and external weather data, ensuring comprehensiveness and diversity of data sources.

[0049] Data acquisition frequency is strictly set according to the differences in data types. Voltage and current data are acquired at least once every 5 minutes, and equipment operating status data are acquired at least once every 15 minutes to ensure data real-time performance. When preprocessing the acquired multi-source data, the RU-Net power missing data reconstruction model is used to address missing values. The model is first trained by dividing the data into training and testing sets, and then the trained model outputs the corresponding filler values ​​for the missing values ​​to ensure data integrity.

[0050] For outliers, an outlier threshold is determined by calculating the standard deviation of the data. Data exceeding the threshold are marked as outliers and removed, ultimately forming standardized data for subsequent analysis, laying the data foundation for subsequent cluster analysis and scheme generation.

[0051] Dynamic clustering analysis was performed by establishing a dynamic clustering model based on an improved K-means algorithm. Preprocessed, standardized data was input into the model. The model first uses the formula... Initialize weighted cluster centers, where C j x is the γth cluster center. i For the i-th standardized data point, ω ij For data point x i For cluster center C j The initial weights, ω ij The importance of a data point is determined based on the importance of the corresponding distribution area. The importance of a distribution area is comprehensively evaluated by the number of users, load level, and power supply reliability requirements. The higher the importance, the greater the weight of the corresponding data point.

[0052] The Euclidean distance between each data point and the cluster center is then calculated, and the data points are assigned to the nearest cluster. The model's built-in status monitoring module continuously tracks power grid operation data. When the load data changes by more than a preset value within a unit of time, it is determined to be a load surge; when the equipment operation status data exceeds the normal threshold, it is determined to be an equipment failure.

[0053] When a sudden load change or equipment failure occurs, the re-clustering process is automatically triggered. During re-clustering, the process is first performed using a formula... Update the cluster centers, where C′ j Let S be the updated j-th cluster center, α be the weight coefficient of historical cluster centers, and S be the weight coefficient of historical cluster centers. j Let |S| be the set of data points contained in the j-th cluster. j |For set S j The number of data points, α, is determined through historical cluster stability analysis. The smaller the fluctuation in historical cluster results, the larger the value of α, and vice versa. The distance calculation and data allocation steps are then repeated to output the updated clustering results, accurately dividing the low-voltage affected areas and feature categories.

[0054] A visualization interface is built based on a distribution network map. The areas affected by low voltage are marked with different colors on the power grid topology map of the interface. The color depth corresponds to the severity of low voltage. Feature categories are associated with the corresponding areas in the form of labels, which makes it easy for staff to intuitively understand the low voltage situation.

[0055] A preliminary governance plan was generated based on the results of dynamic clustering analysis, combined with the city's power distribution network structure and historical governance experience. First, key information such as the line connections, transformer capacity, and conductor cross-sectional parameters of the power distribution network were obtained. The governance priority for low-voltage areas was calculated using the formula P = γ1L + γ2U + γ3C, where P is the governance priority, L is the standardized value of low-voltage duration, U is the standardized value of the number of affected users, C is the standardized value of transformer capacity, and γ1, γ2, and γ3 are weighting coefficients determined using the analytic hierarchy process (AHP). Experts in power grid operation and maintenance, planning, and marketing were invited to construct a judgment matrix, and the final weights were obtained after matrix consistency verification.

[0056] A thorough analysis of the relationship between the current low-voltage areas and key nodes of the power grid is conducted. Based on the relationship and governance priorities, governance measures that are close to the key nodes and easy to implement, such as adjusting transformer tap changes and optimizing line reactive power compensation configurations, are selected and incorporated into the preliminary governance plan to ensure the feasibility and relevance of the plan.

[0057] Simulation and governance are interactively optimized by building an intelligent interactive system. The generated preliminary governance plan is input into the simulation module of this system. The module loads the distribution network topology parameters, completes the simulation environment initialization, and starts the simulation. During the simulation, real-time voltage and current data from the actual power grid are received, and the feedback data is compared with the simulation output data to calculate the deviation between the two.

[0058] When the deviation exceeds the allowable range, it is determined by the formula. Adjust the simulation model parameters, where P′ k For the adjusted k-th model parameter, P k Here, β is the k-th model parameter before adjustment, and D is the parameter adjustment coefficient. k The deviation between the simulated value and the actual feedback value corresponding to the k-th parameter is β, which is determined based on parameter sensitivity analysis. The greater the influence of the parameter on the simulation results, the smaller the value of β should be, so as to avoid excessive parameter adjustment.

[0059] Simultaneously, new strategies and measures adopted in actual governance are collected and transformed into simulation parameter input models to simulate the governance effects under different operating conditions. Multiple sets of comparative simulation experiments are set up, one using new strategies and measures and the other using traditional governance measures. The simulations are run under the same initial operating conditions, and the voltage recovery time and voltage stability indicators in the simulation results of the two sets are compared to verify and optimize the new strategies and measures, and the optimized governance scheme parameters are output.

[0060] The final remediation plan will be implemented by determining the optimal low-voltage remediation scheme based on the simulation optimization results and formulating a phased implementation plan. The first phase will address areas with high low-voltage severity and a large number of affected users, prioritizing the normal power supply for the majority of users. The second phase will address areas with moderate severity, gradually expanding the remediation coverage. The third phase will address the remaining areas, achieving comprehensive remediation of the low-voltage problem.

[0061] After each phase, effectiveness verification is conducted. This involves collecting power grid operation data and gathering user feedback to assess whether the governance effect has met expectations. Throughout the governance process, a data log recording system is established to record data collection results, cluster analysis processes, scheme generation content, simulation optimization parameters, and governance implementation status in real time, forming a complete governance data archive to provide a reference for subsequent low-voltage governance.

[0062] In summary, this embodiment achieves precise management of low-voltage issues in urban power distribution networks through a complete technical process. The dynamic clustering model can quickly respond to changes in operating conditions such as load surges and equipment failures, solving the problem of delayed low-voltage detection in traditional management methods. The deep integration of simulation and management ensures the targeted nature and effectiveness of the management solution, avoiding the drawbacks of a disconnect between simulation and management. After implementation, the response time to low-voltage issues in the urban power distribution network has been significantly shortened, voltage quality stability has been greatly improved, and user satisfaction with electricity usage has increased significantly. Simultaneously, it has reduced power grid operation and maintenance costs, promoting the transformation of power distribution network management from process-driven to data-driven approaches, and providing a feasible practical path for the refined and intelligent management of power distribution networks.

[0063] Example 2

[0064] This embodiment applies to the low-voltage management scenario of rural power distribution networks. Rural power distribution networks are characterized by wide line distribution, dispersed transformer areas, and significant seasonal load variations. Especially during busy irrigation seasons and peak electricity consumption periods like the Spring Festival travel rush, a large number of agricultural devices and household appliances operate simultaneously, easily leading to low-voltage problems. Traditional management methods often suffer from incomplete data collection and analysis models that cannot adapt to dynamic operating conditions, resulting in delayed solutions and unsatisfactory results, impacting agricultural production and villagers' daily electricity use. (See also...) Figure 1 and Figure 2This embodiment adopts a low-voltage management method based on big data clustering and simulation, which is adapted to the unique operating conditions of rural power distribution networks, to achieve efficient identification and precise management of low-voltage problems and ensure reliable power supply in rural areas.

[0065] Real-time data acquisition and preprocessing employs streaming data real-time computing technology, deploying distributed data acquisition nodes at key nodes of transformers and branch lines in various rural power distribution networks to simultaneously acquire multi-source operational data. Core acquired data includes voltage, current, load, and equipment operating status data. Supplementary data includes user profiles recording villagers' electricity consumption characteristics, 95598 work order data tracking user requests, PMS2.0 system data on line and equipment ledgers, and external weather data linking load-influencing factors such as rainfall, irrigation, and low-temperature heating.

[0066] Data acquisition frequencies are set differently based on data type. Voltage and current data are acquired at least once every 5 minutes to ensure the capture of rapid fluctuations in power load. Equipment operating status data is acquired at least once every 15 minutes to promptly monitor the operating status of equipment such as line switches and transformers. Preprocessing is performed on the acquired multi-source data. For null values, the RU-Net power missing data reconstruction model is used. After training and testing sets are divided, the model is trained and the corresponding filler values ​​are output. For outliers, the standard deviation of the data is calculated to determine the outlier threshold. Data exceeding the threshold is marked as outliers and removed, ultimately forming standardized data to provide reliable data support for subsequent analysis.

[0067] Dynamic clustering analysis is performed by constructing a dynamic clustering model based on an improved K-means algorithm, inputting standardized data into the model. The model first uses formulas... Initialize weighted cluster centers, then calculate the Euclidean distance between each data point and the cluster center, and assign the data points to the nearest cluster. The model's built-in status monitoring module continuously tracks power grid operation data. When the load data changes by more than a preset value within a unit of time, it is determined to be a load surge, such as the simultaneous start of multiple irrigation pumps during busy farming seasons. When the equipment operation status data exceeds normal thresholds, it is determined to be an equipment fault, such as line overload causing switch malfunction.

[0068] When a sudden load change or equipment failure occurs, the re-clustering process is triggered. During re-clustering, the process is first performed using a formula... The cluster centers are updated, and the distance calculation and data allocation steps are repeated to output the updated clustering results, accurately classifying the low-voltage affected areas and feature categories, such as "low-voltage areas with concentrated irrigation loads" and "low-voltage areas with a surge in returning migrant workers." A visualization interface is built based on a distribution network map, marking the low-voltage affected areas on the power grid topology map with different colors. The color depth corresponds to the severity of the low voltage, and feature categories are associated with corresponding areas in the form of labels, making it convenient for operation and maintenance personnel to quickly locate the problem area and its cause.

[0069] A preliminary governance plan is generated by combining dynamic clustering results, rural power distribution network structure, and historical governance experience. First, the line connection relationships, transformer capacity, and conductor cross-sectional parameters of the power distribution network are obtained to clarify the upper limit of power supply capacity in each area. Then, the governance priority for low-voltage areas is calculated using the formula P = γ1L + γ2U + γ3C, prioritizing areas with long durations of low voltage, a large number of affected users, and critical transformer capacity.

[0070] Analyze the relationship between low-voltage areas and key nodes of the power grid. For example, transformer substations near main lines can be prioritized for optimization through reactive power compensation. For remote substations, transformer capacity expansion or tap changer adjustment can be considered. Select management measures that are easy to implement and quick to take effect and incorporate them into the preliminary plan. For example, formulate a "peak-shifting irrigation guidance + temporary reactive power compensation device access" plan for areas with concentrated irrigation loads, and formulate a "transformer tap changer adjustment + line load balancing distribution" plan for areas with a surge in returning migrant workers.

[0071] Simulation and governance interaction optimization is implemented by building an intelligent interactive system. Preliminary governance plans are input into the system's simulation module. The module loads rural power distribution network topology parameters, including line lengths, conductor types, and transformer parameters, and starts the simulation after initializing the simulation environment. During the simulation, voltage and current data from the actual power grid are received in real time. The feedback data is compared with the simulation output data, and the deviation between the two is calculated.

[0072] When the deviation exceeds the allowable range, it is determined by the formula. Adjust the simulation model parameters to ensure the simulation results closely match the actual site conditions. Simultaneously, collect new strategies and measures explored in actual governance, such as "temporary grid connection of transformer substations during busy farming seasons" and "orderly charging and discharging of household energy storage devices to assist voltage regulation," and convert them into simulation parameter input models. Set up multiple sets of comparative simulation experiments: one set using the new strategies and measures, and the other using traditional governance measures. Run the simulations under the same initial operating conditions, compare the voltage recovery time and voltage stability indicators of the two sets of simulation results, complete the verification and optimization of the new strategies and measures, and output the optimized governance scheme parameters.

[0073] The final governance plan will be implemented by determining the optimal low-voltage governance scheme based on simulation optimization results and formulating a phased implementation plan. The first phase will address areas with high low-voltage severity that affect agricultural production and villagers' lives, such as core irrigation areas and densely populated villages; the second phase will address areas with moderate severity, such as remote villages and peripheral transformer areas; and the third phase will address the remaining areas with minor low voltage, achieving full coverage.

[0074] After each phase, effectiveness verification is conducted. Indicators such as the voltage compliance rate of the distribution area and user feedback satisfaction are collected to assess whether the governance effect meets the standards. Throughout the governance process, a data log recording system is established to record data collection results, cluster analysis process, scheme generation content, simulation optimization parameters, and governance implementation status in real time, forming a complete governance data archive to provide experience and reference for subsequent low-voltage governance of rural power distribution networks.

[0075] In summary, this embodiment, tailored to the specific operating conditions of rural power distribution networks, achieves precise management of low-voltage issues through a complete technical process. The dynamic clustering model effectively adapts to dynamic changes such as agricultural load fluctuations and population movement, solving the problem of delayed low-voltage identification in traditional management methods. Deep integration of simulation and management ensures the solution aligns with the realities of rural power grids, avoiding the drawbacks of a "one-size-fits-all" approach. After implementation, the response time to low-voltage issues in rural power distribution networks is shortened, and the voltage qualification rate is significantly improved. This not only ensures the electricity needs of agricultural production during busy seasons but also improves the daily electricity experience for villagers, while reducing power grid operation and maintenance costs. It provides a practical and feasible technical path for the intelligent upgrading of rural power distribution networks.

[0076] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A low-voltage management method based on big data clustering and simulation, characterized in that, The method includes the following steps: S1. Real-time data acquisition and preprocessing: Real-time access to multi-source data during the operation of the power distribution network, including voltage data, current data, load data and equipment operating status data. The acquired multi-source data is preprocessed to remove null values ​​and outliers, forming standardized data for subsequent analysis. S2. Dynamic clustering analysis: Establish a dynamic clustering model, input the obtained standardized data into the dynamic clustering model, and continuously monitor the power grid operation status. When a load change or equipment failure is detected, the re-clustering analysis process is automatically triggered to update the clustering results and divide the low voltage impact area and feature categories. S3. Preliminary governance plan generation: Based on the obtained clustering results, combined with the distribution network structure and historical governance experience, a preliminary governance plan for the current low voltage situation is generated. S4. Simulation and governance interaction optimization: Build an intelligent interaction system, input the generated preliminary governance plan into the simulation module of the intelligent interaction system, and adjust the parameters and assumptions of the simulation model based on the actual power grid feedback operation data during the simulation process. At the same time, new strategies and measures in actual governance are fed back to the simulation model to verify and optimize the new strategies and measures. S5. Final resolution plan execution: Based on the simulation optimization results, determine the optimal low voltage resolution plan and execute the optimal low voltage resolution plan.

2. The low-voltage management method based on big data clustering and simulation according to claim 1, characterized in that, The real-time data acquisition and preprocessing process in step S1 includes: First, input voltage data, current data, load data, and equipment operating status data. At the same time, supplement the input with user profile data, 95598 work order data, PMS2.0 system data, and external weather data. Next, the collected multi-source data is preprocessed. For null values, the RU-Net power missing data reconstruction model is used. After training the model by dividing it into training and test sets, the null values ​​are output by the trained model. For outliers, the standard deviation of the data is calculated to determine the outlier threshold. Data exceeding the threshold are marked as outliers and removed to form standardized data.

3. The low-voltage management method based on big data clustering and simulation according to claim 1, characterized in that, The dynamic clustering analysis process in step S2 includes: First, input the obtained standardized data into a dynamic clustering model built based on the improved K-means algorithm. The model initializes the weighted cluster centers using the following formula: Among them, C j For the j-th cluster center, x i For the i-th standardized data point, ω ij For data point x i For cluster center C j The initial weights; Then, the Euclidean distance between each data point and the cluster center is calculated, and the data points are assigned to the nearest cluster. The power grid operation data is then continuously tracked through the built-in state monitoring module of the model. When the load data changes by more than a preset value within a unit of time, it is determined to be a load change; when the equipment operating status data exceeds the normal threshold, it is determined to be an equipment failure. When a sudden load change or equipment failure occurs, re-clustering is triggered. During re-clustering, the cluster centers are first updated using the following formula: Among them, C′ j Let S be the updated j-th cluster center, α be the weight coefficient of historical cluster centers, and S be the weight coefficient of historical cluster centers. j Let |S| be the set of data points contained in the j-th cluster. j |For set S j The number of data points is counted, and the distance calculation and data allocation steps are repeated to output the updated clustering results, dividing the low voltage affected area and feature categories.

4. The low-voltage management method based on big data clustering and simulation according to claim 1, characterized in that, The simulation and governance interaction optimization process in step S4 includes: First, input the preliminary treatment plan into the simulation module of the intelligent interactive system. The module loads the distribution network topology parameters and initializes the simulation environment. Next, the simulation is run to receive real-time voltage and current data from the actual power grid and compare the feedback data with the simulation output data. When the deviation exceeds the allowable range, adjust the simulation model parameters using the following formula: Among them, P′ k For the adjusted k-th model parameter, P k Here, γ represents the k-th model parameter before adjustment, and D is the parameter adjustment coefficient. k This represents the deviation between the simulated value and the actual feedback value corresponding to the k-th parameter. Simultaneously, new strategies and measures adopted in actual governance are collected and transformed into simulation parameter input models to simulate the governance effects under different working conditions. Based on the simulation results after parameter adjustment and the simulation effects of new measures, the simulation model is optimized, and the optimized governance scheme parameters are output.

5. The low-voltage management method based on big data clustering and simulation according to claim 1, characterized in that, When collecting multi-source data in step S1, streaming data real-time computing technology is used. The operating data of each transformer area and each line are synchronously acquired through distributed data acquisition nodes. The acquisition frequency is determined according to the data type. The acquisition frequency of voltage and current data is no less than once every 5 minutes, and the acquisition frequency of equipment operating status data is no less than once every 15 minutes.

6. The low-voltage management method based on big data clustering and simulation according to claim 1, characterized in that, After dividing the low voltage impact area and feature category in step S2, a visualization display interface is constructed based on a distribution network map. The low voltage impact area is marked on the power grid topology map of the interface with different colors. The color depth corresponds to the severity of low voltage. Feature categories are associated with the corresponding areas in the form of labels.

7. The low-voltage management method based on big data clustering and simulation according to claim 1, characterized in that, In step S3, when generating a preliminary governance plan based on the distribution network structure, the line connection relationship, transformer capacity, and conductor cross-sectional parameters of the distribution network are first obtained. The governance priority of the low-voltage area is calculated using the following formula: P = γ1L + γ2U + γ3C, where P is the governance priority, L is the standardized value of the duration of low voltage, U is the standardized value of the number of affected users, C is the standardized value of the transformer area capacity, and γ1, γ2, and γ3 are weighting coefficients. The correlation between the current low-voltage area and the key nodes of the network is analyzed. Based on the correlation and governance priority, governance measures that are close to the key nodes and have low implementation difficulty are selected and incorporated into the preliminary governance plan.

8. The low-voltage management method based on big data clustering and simulation according to claim 1, characterized in that, In step S4, when verifying the new strategy and measures, multiple sets of comparative simulation experiments are set up. One set adopts the new strategy and measures, and the other set adopts traditional governance measures. The simulation is run under the same initial working conditions, and the voltage recovery time and voltage stability index in the simulation results of the two sets are compared.

9. The low-voltage management method based on big data clustering and simulation according to claim 1, characterized in that, When implementing the optimal low voltage mitigation scheme in step S5, a phased implementation plan is formulated. The first phase addresses areas with high low voltage severity and many affected users, the second phase addresses areas with moderate severity, and the third phase addresses the remaining areas. After each phase, the effectiveness is verified.

10. The low-voltage management method based on big data clustering and simulation according to claim 1, characterized in that, Throughout the entire governance process, a data log recording system is established to record data collection results, cluster analysis process, scheme generation content, simulation optimization parameters, and governance execution status in real time, forming a complete governance data archive.

Images