Electric side energy efficiency management and loss control method and system under dynamic load characteristics

Abstract

The application provides a method and system for managing and controlling energy efficiency and loss on the power side under dynamic load characteristics, comprising: if the energy flow in the grouped load cluster is unbalanced, calculating the loss value by simulating the energy transmission path, determining the location and severity of the loss key node; for the determined loss key node, obtaining transformer end operation state data including temperature and load rate, and judging whether the transformer end is in an overload state; through the obtained adjusted load distribution scheme, the remote operation and maintenance system issues control instructions to the intelligent interconnected device, determines the reconfiguration of the path in the network structure, and realizes loss management and control; according to the reconfigured network structure, the change of the energy efficiency index on the power side is monitored, the optimized energy balance result is obtained, and it is judged whether the preset management threshold is reached.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a method and system for power-side energy efficiency management and loss control under dynamic load characteristics. Background Technology

[0002] In the operation and management of power systems, power-side energy efficiency management, transformer-end loss control, and intelligent interconnected remote operation and maintenance constitute key areas for ensuring efficient energy utilization and stable system operation. The importance of these areas is self-evident; they directly relate to the conservation of electricity resources, the extension of equipment lifespan, and the reduction of electricity costs for users, serving as core support for the modernization of power systems towards intelligence and green practices.

[0003] However, current methods in these fields often struggle to adapt to complex real-world operating environments, especially when facing scenarios with uneven load distribution and dynamic changes, lacking comprehensive control and precise analysis of energy flow in the power network. This deficiency not only leads to resource waste but may also exacerbate equipment aging and even trigger system failures due to the failure to detect losses in a timely manner.

[0004] A deeper technical challenge lies in the fact that the physical connection methods of loads in the distribution network and the distribution characteristics of energy flow are not yet fully understood. Different load connection locations directly affect energy losses along the distribution path, and these losses are closely related to the operating status of the transformer. Without an accurate understanding of how load distribution affects energy transmission in the network, it is impossible to effectively identify which paths or nodes are the main sources of loss, leading to a lack of targeted mitigation measures. For example, in an industrial park, the improper location of lines connected to certain high-energy-consuming equipment may cause one section of the line to be overloaded, while another section is under low load. This imbalance not only increases the burden on the transformer but may also lead to localized overheating, accelerating equipment wear.

[0005] Therefore, accurately identifying key nodes of energy loss in the power distribution network under dynamically changing load conditions, and balancing the relationship between load distribution and transformer operating status by adjusting the network structure, has become a key issue in power-side energy efficiency management and loss control. Summary of the Invention

[0006] This invention provides a method for power-side energy efficiency management and loss control under dynamic load characteristics, mainly including: By collecting voltage, current and power data of each node in the power distribution network in real time through sensor networks, dynamic load distribution information can be obtained to determine the degree of influence of load connection location on energy flow. Based on the obtained dynamic load distribution information, the k-means clustering algorithm is used to group the network nodes to obtain the grouped load clusters, and it is determined which clusters have uneven energy flow. If the energy flow is uneven in the grouped load cluster, the loss value is calculated by simulating the energy transfer path to determine the location and severity of the critical loss nodes; For the identified critical loss nodes, obtain the transformer terminal operating status data, including temperature and load rate, to determine whether the transformer terminal is in an overload state. If the transformer is in an overload state, a deep reinforcement learning algorithm is used to optimize the network structure adjustment strategy to obtain an adjusted load distribution scheme that balances energy flow and transformer operating status. Based on the adjusted load distribution scheme, the remote operation and maintenance system sends control commands to the intelligent interconnected devices to determine the reconfiguration of paths in the network structure and achieve loss control. Based on the reconfigured network structure, monitor changes in power-side energy efficiency indicators to obtain optimized energy balance results and determine whether the preset governance threshold has been reached.

[0007] This invention provides a power-side energy efficiency management and loss control system under dynamic load characteristics, mainly comprising: The data acquisition module is used to collect voltage, current and power data of each node in the power distribution network in real time through the sensor network, obtain dynamic load distribution information, and determine the degree of influence of load connection location on energy flow. The node grouping module is used to group network nodes using the k-means clustering algorithm based on the acquired dynamic load distribution information, obtain the grouped load clusters, and determine which clusters have uneven energy flow. The loss calculation module is used to calculate the loss value by simulating the energy transfer path if the energy flow in the grouped load cluster is uneven, and to determine the location and severity of the loss critical nodes. The transformer status judgment module is used to acquire transformer terminal operating status data, including temperature and load rate, for identified key loss nodes, and to determine whether the transformer terminal is in an overload state. The network optimization module is used to optimize the network structure adjustment strategy using a deep reinforcement learning algorithm if the transformer is in an overload state, so as to obtain an adjusted load distribution scheme and balance energy flow and transformer operating status. The path configuration module is used to send control commands from the remote operation and maintenance system to the intelligent interconnected devices based on the obtained adjusted load distribution scheme, determine the reconfiguration of paths in the network structure, and realize loss control. The energy efficiency monitoring module is used to monitor changes in power-side energy efficiency indicators based on the reconfigured network structure, obtain optimized energy balance results, and determine whether the preset governance threshold has been reached.

[0008] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses a method for energy efficiency management and loss control on the power side under dynamic load characteristics, proposing a complete solution to the problem of uneven energy flow and losses caused by dynamic load distribution in distribution networks. This invention collects voltage, current, and power data of each node in real time through a sensor network, groups nodes using a k-means clustering algorithm to accurately identify load clusters with uneven energy flow, and calculates loss values ​​by simulating energy transfer paths to locate key loss nodes. Simultaneously, it assesses overload conditions by combining transformer operating status data, employs a deep reinforcement learning algorithm to optimize network structure adjustment strategies, and generates a load distribution scheme that balances energy flow and transformer status. Finally, it achieves path reconfiguration and loss control through remote operation and maintenance control commands. This invention also continuously monitors power side energy efficiency indicators to ensure that the optimization results reach preset management thresholds. Its core lies in intelligent optimization and real-time control under dynamic load characteristics, significantly improving the energy efficiency level and operational stability of distribution networks. Attached Figure Description

[0009] Figure 1 This is a flowchart of the method for power-side energy efficiency management and loss control under dynamic load characteristics according to the present invention.

[0010] Figure 2 This is a schematic diagram of the method for power-side energy efficiency management and loss control under dynamic load characteristics according to the present invention.

[0011] Figure 3 This is another schematic diagram of the method for power-side energy efficiency management and loss control under dynamic load characteristics according to the present invention.

[0012] Figure 4 This is a schematic diagram of the structure of the power-side energy efficiency management and loss control system under dynamic load characteristics according to the present invention. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. The described embodiments are merely some embodiments of the present invention.

[0014] like Figures 1-4 The specific methods and systems for power-side energy efficiency management and loss control under dynamic load characteristics in this embodiment may include: S101. Real-time collection of voltage, current and power data of each node in the power distribution network through sensor network to obtain dynamic load distribution information and determine the degree of influence of load connection location on energy flow.

[0015] A sensor network is used to collect raw data streams of node voltage, current, and power from each node in the power distribution network in real time. These data are stored in a pre-established data warehouse to obtain the initial dataset. For this initial dataset, data cleaning methods are used to remove outliers and missing values. If the voltage or current of a node exceeds a preset threshold, it is marked as abnormal data and removed, resulting in a cleaned node parameter dataset. Based on this cleaned dataset, the dynamic load value of each node is calculated, and time series analysis is used to extract the load distribution over time, determining the load distribution characteristic set of each node.

[0016] ; This formula is used to calculate the dynamic load value of each node; where Represents a node At any moment The dynamic load value, Represents a node At any moment Power data, Represents a node At any moment voltage value, Represents a node At any moment The current value, Indicates the length of the time series. This represents the time weighting coefficient. Represents a node At any moment The formula extracts the load distribution over time using time series analysis to determine the load distribution characteristics of each node.

[0017] For the load distribution feature set, the correlation between the load access location and each node is analyzed. If the load value change at a certain access location causes the power fluctuation of adjacent nodes to exceed a preset threshold, then the location is recorded as a key access point, and a set of key access points is obtained. Through the set of key access points, the energy flow path is modeled using the support vector machine algorithm, and the effect of each key access point on the direction and intensity of energy flow is analyzed to determine the degree of influence of the load access location on energy flow. ; This formula is used to calculate the degree to which energy flow is affected by the location of load access; where Indicates access location The degree of influence on energy flow. This indicates the total number of critical access points. Indicates the first The weight coefficients of key access points Indicates the first Energy flow intensity at key access points Indicates access location The load capacity, The penalty coefficient represents the energy deviation. This represents the reference energy flow value at the j-th critical access point. This formula, combined with a support vector machine algorithm, models the energy flow path and analyzes the influence of each critical access point on the direction and intensity of energy flow.

[0018] Based on the impact value of energy flow, a priority ranking of each key access point is generated. If the impact value of a certain access point is higher than the preset threshold, it is marked as a high priority point, thus obtaining the basis for optimizing and adjusting the load access points in the power distribution network.

[0019] Based on the optimization and adjustment criteria, a load balancing distribution scheme for each node in the power distribution network is generated. By simulating changes in energy flow paths, the final load access location adjustment strategy is determined. A monitoring system covering the entire power distribution network can be constructed by real-time collection of voltage, current, and power data from each node in the power distribution network using a sensor network. For example, 100 sensor nodes can be set up in a certain area, each collecting data once per second. The voltage range is 220V to 240V, and the current range is 0A to 50A. The power calculation formula is P=U*I*cosφ, where cosφ is set to 0.9 as the power factor. Assuming a node collects a voltage of 230V and a current of 20A, the power is P=230*20*0.9=4140W. This data is transmitted in real-time to a cloud server via a wireless communication module. Big data processing technology is used to clean and store the collected data to ensure data integrity. Next, dynamic load distribution information is obtained. Algorithms can be used to analyze the spatiotemporal changes in power data for each node. For example, using the K-means clustering algorithm, the power data of 100 nodes are divided into 5 load clusters based on time period and geographical location. The calculation shows that the average power of a certain cluster is 4000W during peak hours and 1500W during off-peak hours, reflecting the unevenness of load distribution. Subsequently, the impact of load connection locations on energy flow is determined. Based on the above load distribution data, an energy flow model is constructed. Power flow calculations are used to analyze the impact of a new load connection point (e.g., with a power of 2000W) on the network. The calculation results show that the voltage of nodes near the connection point drops by approximately 2.5V, and the current increases by approximately 10A. By comparing network losses before and after the connection, it is found that the loss increases from 500W to 650W, indicating that the connection location has a significant impact on energy flow. To further optimize, historical data can be combined to predict future load trends, and machine learning models such as LSTM can be used to predict load changes in the next hour with an error controlled within 5%. This provides a basis for the dynamic adjustment of the distribution network, forming a complete logical chain from data acquisition to impact analysis to prediction and optimization, ensuring the efficiency and stability of the distribution network operation.

[0020] S102. Based on the obtained dynamic load distribution information, the network nodes are grouped using the k-means clustering algorithm to obtain the grouped load clusters, and it is determined which clusters have uneven energy flow.

[0021] A pre-established data acquisition system is used to obtain dynamic load and distribution information of each node in the power distribution network and store it in a data warehouse to obtain an initial load dataset. For this initial load dataset, a data filtering method is used to remove noisy data. If the load data of a node exceeds a preset threshold range, it is marked as invalid data and discarded, resulting in a processed load dataset. Based on the processed load dataset, the k-means clustering algorithm is applied to group network nodes, generating multiple load clusters based on the similarity of load distribution, thus determining the load cluster set. For each load cluster set, the energy flow state within each cluster is analyzed. If the energy flow value of a node in a cluster deviates from the average value by more than a preset threshold, it is marked as an unbalanced cluster, resulting in an unbalanced cluster list.

[0022] ; This formula is used to determine whether the energy flow is balanced in a load cluster; Indicates the first Energy flow imbalance in individual clusters This represents the set of nodes within the k-th cluster. Indicates the total number of nodes in the cluster. Indicates the number within the cluster Energy flow value of each node The horizontal line above indicates the number. The average energy flow of all nodes within a cluster; when When the threshold is exceeded, the cluster is marked as an unbalanced cluster.

[0023] By analyzing the list of unbalanced clusters, load distribution and energy status data of nodes within each cluster are extracted to generate energy flow deviation records for each cluster, thus identifying the clusters with significant deviations. Based on these energy flow deviation records, nodes within the unbalanced clusters are categorized. If the load distribution of a node causes a sustained deviation in its energy status, it is recorded as a critical node, resulting in a set of critical nodes. For this set of critical nodes, a priority ranking scheme for load adjustment is generated. By analyzing the energy status and load distribution of the critical nodes, a list of adjustment sequences is determined.

[0024] In power distribution network management, based on the acquired dynamic load distribution information, information technology can be used to conduct in-depth analysis and processing of network nodes. First, for 120 network nodes distributed within a certain area, specially developed load analysis software automatically extracts the power data of each node over the past 24 hours. The data ranges from 500W to 8000W. These nodes are then grouped using a K-means clustering algorithm, with 6 clusters and 10 iterations, ultimately forming 6 load clusters. The calculated average power of one cluster is 3200W, with a standard deviation of 800W, indicating significant differences in internal load distribution. Next, the system automatically performs spatiotemporal characteristic analysis on the power data of each cluster, calculating the daily average power fluctuation rate for each cluster. For example, a cluster has an average power of 5500W during the morning peak hours and 1200W during the evening off-peak hours, with a fluctuation rate as high as 78.2%. By comparing this with a preset threshold of 50%, it is determined that this cluster exhibits an imbalance in energy flow. Subsequently, for the unbalanced cluster, the system further utilized historical data to analyze its geographical distribution characteristics. It was found that industrial load accounted for 60% of the area covered by this cluster. By comparing the load types with other clusters in the database, it was confirmed that concentrated industrial load was the main factor causing the imbalance. Finally, the system automatically generated identification records for the unbalanced cluster, marking it as a key monitoring target, and correlated it with the topology data of the distribution network. The calculated impact radius of this cluster on surrounding nodes was approximately 3.5 kilometers, providing data support for subsequent network planning. This entire process, through automated algorithms and system analysis, forms a complete logical chain, ensuring that every step from data grouping to imbalance judgment and impact assessment is driven by information technology and is interconnected.

[0025] S103. If the energy flow is uneven in the grouped load cluster, the loss value is calculated by simulating the energy transfer path to determine the location and severity of the critical loss nodes.

[0026] Real-time energy flow data is acquired from various load clusters using a pre-established power distribution network data acquisition system and stored in a data warehouse to obtain an initial energy flow dataset. For this initial dataset, a data cleaning method is used to remove outliers. If the energy flow data of a particular cluster exceeds a preset threshold, it is marked as invalid and removed, resulting in a cleaned energy flow dataset. Based on this cleaned dataset, a simulation model of the energy transfer paths is constructed to analyze the energy flow status along each path and determine the preliminary loss distribution of each path.

[0027] ; This formula is used to calculate the loss value of the energy transfer path; Representing a path Total loss value on, Indicates the number of node segments on the path. Indicates the first The resistance value of the segment, This represents the current value in segment j. Indicates the first The loss coefficient of the segment, Indicates the first The power value of the segment, Indicates the first The voltage value of the segment. This formula can be used to identify path segments with high loss values.

[0028] Based on the preliminary loss distribution, loss calculation tools are used to analyze the energy loss values ​​along each path, obtaining a detailed loss distribution record. For this detailed record, path segments with high loss values ​​are identified. If the loss value of a path segment exceeds a preset threshold, it is marked as a high-loss path, resulting in a set of high-loss paths. Based on this set, the corresponding key nodes within each path segment are located, and the energy flow deviations at each key node are analyzed to determine their distribution locations. For each key node, its severity is categorized. If the energy loss value of a node consistently exceeds the average, it is recorded as a high-severity node, resulting in a list of high-severity nodes.

[0029] In power distribution network management, to address the energy flow imbalance problem discovered in grouped load clusters, the system employs a series of automated information technology methods for in-depth analysis and processing. First, for identified imbalanced clusters within a specific area, the system automatically retrieves energy transmission data from the past 48 hours, covering power transmission volumes ranging from 1000W to 10000W. Using specially developed energy path simulation software, based on the shortest path algorithm (Dijkstra's algorithm), it simulates energy transfer paths and calculates the energy loss value on each path. For example, the total loss value of a certain path is 2500W, accounting for 12.5% ​​of the total transmission volume. Next, the system quantitatively analyzes the simulation results, combining them with network topology data to automatically identify path segments with loss values ​​exceeding a preset threshold of 8%, and locates critical loss nodes. For instance, a node with a loss value reaching 1800W and a severity score of 85 (out of 100) is confirmed as a high-risk point by comparison with a historical loss database. Subsequently, the system further analyzes the surrounding connectivity characteristics of key nodes, calculating their impact coefficient on the loss of adjacent nodes. For example, an impact coefficient of 0.65 indicates that the loss of this node significantly reduces the transmission efficiency of surrounding nodes. To form a complete logical chain, the system also correlates the loss data of key nodes with the regional load type distribution, finding that the commercial load accounts for 70% of the area where the node is located, inferring that high-frequency load changes may be one of the reasons for the increased loss. Finally, the system automatically generates an analysis report of the loss-prone key node, including location coordinates, loss value, severity score, etc., and connects it with the network maintenance priority database, calculating that the optimization priority of this node is 92%, providing data basis for subsequent resource allocation. The above process, through automated algorithms and system analysis, ensures that the logic from path simulation to node location to impact assessment is rigorous and interconnected.

[0030] S104. For the identified critical loss nodes, obtain the transformer terminal operating status data, including temperature and load rate, and determine whether the transformer terminal is in an overload state.

[0031] For identified critical loss nodes, a pre-established monitoring system acquires operational status information from the transformer, including temperature data and load rate values, to obtain a preliminary operational status dataset. Based on this preliminary dataset, data cleaning tools are used to detect outliers in the temperature and load rate values. If a temperature or load rate value exceeds a preset threshold, it is marked as outlier and removed, resulting in a cleaned operational status dataset. For this cleaned dataset, an information analysis module comprehensively evaluates the transformer's temperature and load rate values. If the temperature consistently exceeds a preset threshold or the load rate exceeds a safe range, a potential overload condition is identified, yielding an overload risk assessment result. Based on this risk assessment, historical operational status information from the transformer is obtained and compared with the current data. If similar high-temperature or high-load-rate records exist in the historical data, the current state is confirmed as overload operation, and the distribution of overloaded nodes is determined. For nodes operating under overload conditions, a monitoring system is used to acquire real-time dynamic status information at the transformer end, analyze the changing trends of temperature data and load rate values, determine whether there is a risk of continuous deterioration, and obtain dynamic monitoring records. Based on the dynamic monitoring records, a support vector machine algorithm is used to classify and predict the overload operation risk at the transformer end. Combining the changing characteristics of temperature data and load rate values, the priority ranking of high-risk nodes is determined, resulting in a risk priority list.

[0032] ; This formula uses the support vector machine algorithm to classify and predict the overload operation risk at the transformer end; This represents the decision function value of the support vector machine, used for risk classification; Indicates the number of support vectors; Indicates the first The weight coefficients of each support vector; This represents the kernel function used to compute support vectors. The similarity between the vector and the current state vector x, where Includes variation characteristics of temperature data and load rate values; This represents the bias term; a larger decision function value indicates a higher risk of overload.

[0033] For the risk priority list, information analysis tools are used to correlate and map the operational status information of high-risk nodes with the distribution of loss-critical nodes to determine which nodes need to be monitored in a key manner, thus obtaining a set of key monitoring nodes.

[0034] In power distribution network management, for identified critical loss nodes, the system uses automated information technology to acquire transformer operating status data and performs comprehensive analysis to determine if an overload is present. First, the system automatically connects to the transformer monitoring database and extracts the operating parameters of the transformer associated with the critical node over the past 24 hours, including temperature and load rate. For example, if a transformer's current temperature is 75 degrees Celsius and its load rate is 88.3%, this data is compared to standard operating thresholds: 70 degrees Celsius for temperature and 85% for load rate. Next, based on a preset overload risk assessment algorithm, the system calculates a comprehensive overload index for temperature and load rate. For example, a weighted index of 0.78 (out of 1.0) is obtained, with temperature weighting at 0.4 and load rate weighting at 0.6, indicating a certain overload risk for the transformer. Subsequently, the system combines historical operating data and uses time series analysis algorithms to predict temperature and load rate trends over the next 6 hours. For example, it predicts the temperature may rise to 78 degrees Celsius and the load rate may reach 90.2%, and based on this, determines whether the overload risk will further escalate. To form a complete logical chain, the system also correlated and analyzed the transformer's operating status with load fluctuation data of key nodes. It was found that the peak load of this node reached 9500W in the past 12 hours, with a fluctuation frequency of 3 times per hour. This high load fluctuation was speculated to be a potential factor increasing the risk of transformer overload. Finally, the system automatically generated a transformer status assessment report, integrating information such as temperature, load rate, and overload index, and connecting it with the network operation stability database. The report calculated the transformer's operating risk level to be medium-high, with a priority score of 78 (out of 100), providing data support for subsequent network adjustments. This entire process, through automated systems and algorithmic analysis, ensures a logically rigorous and interconnected sequence from data acquisition to risk assessment.

[0035] S105. If the transformer is in an overload state, a deep reinforcement learning algorithm is used to optimize the network structure adjustment strategy to obtain an adjusted load distribution scheme that balances energy flow and transformer operating status.

[0036] To address the overload situation at the transformer end, a pre-established monitoring system acquires real-time operating status information and load distribution data from the transformer end, obtaining an initial operating status record. Based on this initial record, data processing tools are used to clean and organize the acquired operating status information, removing outlier data points to obtain a processed operating status dataset. For this processed dataset, an information analysis module extracts key features of load distribution and energy flow to determine if any imbalance exists, yielding a load distribution analysis result. If the load distribution in a certain area exceeds a preset threshold, a deep reinforcement learning algorithm is used to dynamically adjust the network structure, generating a preliminary load adjustment plan. For this preliminary plan, simulation analysis tools, combined with energy flow data, are used to conduct multi-scenario tests to determine its feasibility, resulting in an optimized load adjustment plan. Finally, based on the optimized plan, the operating monitoring system applies the plan to the actual operating environment at the transformer end, acquiring real-time operating status information after adjustment to determine if a balanced state has been achieved, providing the final operating status feedback.

[0037] In power distribution network management, when the system detects an overload at a transformer, the automation platform immediately initiates a deep reinforcement learning algorithm to optimize network structure adjustment strategies, aiming to achieve balanced load distribution and stable transformer operation. First, the system extracts the current distribution network structure data and load distribution from the network topology database. For example, in a region with five transformer nodes, the overloaded transformer has a load of 12000W, exceeding its rated capacity by 20%, while the adjacent nodes have a load of only 6000W, indicating a significant uneven distribution. Next, the system inputs this data into a deep reinforcement learning model. The model uses minimizing overload risk and energy transmission loss as its objective function, simulating adjustment strategies through multiple iterations. For instance, in the 50th iteration, the algorithm proposes transferring 15% (1800W) of the overloaded transformer's load to adjacent low-load nodes, while simultaneously adjusting the switching status of related lines. In each iteration, the model calculates the adjusted network stability index. For example, after adjustment, the overloaded transformer's load drops to 10200W, and the stability index increases from 0.65 to 0.82 (out of a maximum of 1.0). Subsequently, the system performed verification analysis based on historical load fluctuation data, discovering that the peak load fluctuation in the region over the past 48 hours was 25%. The algorithm dynamically adjusted the load transfer ratio accordingly to ensure that future peak periods would not trigger new overload risks. Finally, the system generated an optimized load distribution plan, such as redistributing the total load to maintain an average load rate of less than 75% for each node, and automatically pushed the plan to the network control center database, forming a complete closed-loop logic from overload detection to load adjustment, ensuring a balance between energy flow and equipment operating status.

[0038] S106. Based on the adjusted load distribution scheme obtained, the remote operation and maintenance system sends control commands to the intelligent interconnected devices to determine the reconfiguration of paths in the network structure and realize loss control.

[0039] The adjusted load distribution scheme data is obtained through a remote operation and maintenance system. Combined with a pre-established network structure model, the initial requirements for path planning are analyzed, and the priority sequence for path adjustments is determined. Based on the priority sequence, control commands are issued to the intelligent interconnected devices via the remote operation and maintenance system to dynamically reconstruct the network structure, obtaining the reconstructed network topology information. For the reconstructed network topology information, the operation monitoring module collects real-time operating data from the intelligent interconnected devices to determine whether device management meets preset threshold ranges and obtains feedback results on device status. If the device status feedback results show that the operating data exceeds the preset threshold range, new control commands are generated through the remote operation and maintenance system to adjust the operating parameters of the intelligent interconnected devices, determining the adjusted device operating status. Based on the adjusted device operating status, combined with the loss control module, the energy transmission loss in the network structure is analyzed to determine whether the loss is within an acceptable range, obtaining loss assessment data. Based on the loss assessment data, the assessment results are uploaded to the remote operation and maintenance system through a status feedback mechanism. Combined with real-time monitoring of load distribution, the final path planning optimization scheme is determined.

[0040] In remote operation and maintenance management of power distribution networks, the system automatically initiates the process of issuing control commands and reconfiguring paths based on the generated adjusted load distribution scheme to effectively manage network losses. First, the remote operation and maintenance system generates specific control commands based on the optimization scheme. For example, for three transformer nodes with a total load of 8000W in a certain area, the scheme requires adjusting the load of one node from 3500W to 2800W, and transferring the remaining load to the other two nodes. The system calculates the line paths that need to be adjusted and generates corresponding switch control signals. Next, the system sends the commands in encrypted data packets through the communication interface of the intelligent interconnected devices. Each data packet contains, for example, the target device ID, the switch status change command (e.g., from "closed" to "open"), and a timestamp, ensuring the accuracy and real-time nature of command transmission, with transmission delay controlled within 0.5 seconds. Subsequently, upon receiving the instruction, the intelligent interconnected device automatically performs path reconfiguration. For example, by adjusting the switching status of a line, the load transfer path is switched from line A to line B. The system monitors the current changes during the switching process in real time, finding that the loss rate of line B decreased from 3.2% to 2.8% after the switch, and feeds the data back to the operation and maintenance system for recording. Simultaneously, the system calls a loss analysis algorithm to predict the loss trend over the next 12 hours based on historical data (such as an average loss rate of 3.5% over the past 24 hours) and the current network topology. It calculates that without path adjustment, the loss might increase to 4.1%, thus verifying the necessity of reconfiguration. Finally, the system automatically archives all operation logs and loss data into the database, forming a complete closed loop from instruction issuance to path adjustment to loss monitoring, ensuring efficient network operation and providing data support for subsequent optimization.

[0041] S107. Based on the reconfigured network structure, monitor the changes in the power-side energy efficiency indicators, obtain the optimized energy balance results, and determine whether the preset governance threshold has been reached.

[0042] By using a pre-established network structure model, the reconfigured network structure information is obtained, and combined with relevant power-side energy efficiency data, the initial changes in indicators are determined. For these initial changes, a real-time monitoring module continuously collects dynamic data on power-side energy efficiency to obtain a preliminary energy balance state. Based on this preliminary energy balance state, a support vector machine algorithm is used to classify the dynamic data and determine whether it is close to a preset governance threshold range. If the classification results show that the energy balance state has not reached the preset governance threshold range, a data processing module performs stratified analysis on the real-time monitored data to obtain specific deviation information. For this specific deviation information, an automated adjustment tool is used to generate targeted network structure optimization instructions to determine the trend of the adjusted power-side energy efficiency indicators.

[0043] ; This formula is used to generate an automated adjustment strategy for network structure optimization instructions; where This represents the adjusted network structure parameter vector. This represents the current network structure parameter vector. This indicates adjusting the step size coefficient. This represents the gradient of the objective function with respect to the parameters. Let T represent the objective function for the deviation, and let T represent the total number of monitoring periods. This represents the preset target energy balance value. Indicates at time Based on parameters The actual energy balance value. This formula determines the trend of the adjusted power-side energy efficiency index using the gradient descent method.

[0044] Based on the changing trend of the adjusted power-side energy efficiency indicators, the optimized data is uploaded to the monitoring system through the state feedback mechanism to obtain the final balance state assessment result.

[0045] In the energy efficiency management of power distribution networks, the system automatically initiates monitoring and analysis of power-side energy efficiency indicators based on the reconfigured network structure to assess the energy balance status and determine whether preset governance thresholds are met. First, the system collects real-time power-side data using smart sensors deployed at key nodes. For example, in a certain area, the total input power of five distribution nodes is 10,000 watts, the output power is 9,200 watts, and the initial energy efficiency ratio is 92%. This data is then uploaded to a cloud-based analysis platform. Next, the system calls an energy efficiency assessment algorithm, combined with the current network topology, to calculate the energy loss distribution of each node. It finds that one node has a loss ratio as high as 5.3%, exceeding the average by 2.1%, and generates a loss heatmap to aid subsequent analysis. Then, the system compares and analyzes the collected real-time data with historical data (such as an average energy efficiency ratio of 90.5% over the past 48 hours) to calculate the optimized energy balance result. The current energy efficiency ratio is improved to 93.2%, and it predicts that if the current configuration is maintained, the energy efficiency ratio may stabilize between 93% and 94% in the next 24 hours. To ensure a closed-loop logic, the system further correlates energy efficiency data with environmental variables (such as temperature and humidity), discovering that for every 5 degrees Celsius increase in temperature, the energy efficiency ratio decreases by approximately 0.3%. This automatically adjusts the data model weights, improving prediction accuracy. Finally, the system compares the optimized energy efficiency ratio of 93.2% with the preset governance threshold of 95%, finding that it still falls short. It then automatically triggers the energy efficiency optimization strategy generation module, calculating recommendations for the next round of network adjustments based on current data. This forms a complete process from monitoring and analysis to decision-making, ensuring continuous improvement.

[0046] This invention provides a power-side energy efficiency management and loss control system under dynamic load characteristics, mainly comprising: The data acquisition module is used to collect voltage, current and power data of each node in the power distribution network in real time through the sensor network, obtain dynamic load distribution information, and determine the degree of influence of load connection location on energy flow. The node grouping module is used to group network nodes using the k-means clustering algorithm based on the acquired dynamic load distribution information, obtain the grouped load clusters, and determine which clusters have uneven energy flow. The loss calculation module is used to calculate the loss value by simulating the energy transfer path if the energy flow in the grouped load cluster is uneven, and to determine the location and severity of the loss critical nodes. The transformer status judgment module is used to acquire transformer terminal operating status data, including temperature and load rate, for identified key loss nodes, and to determine whether the transformer terminal is in an overload state. The network optimization module is used to optimize the network structure adjustment strategy using a deep reinforcement learning algorithm if the transformer is in an overload state, so as to obtain an adjusted load distribution scheme and balance energy flow and transformer operating status. The path configuration module is used to send control commands from the remote operation and maintenance system to the intelligent interconnected devices based on the obtained adjusted load distribution scheme, determine the reconfiguration of paths in the network structure, and realize loss control. The energy efficiency monitoring module is used to monitor changes in power-side energy efficiency indicators based on the reconfigured network structure, obtain optimized energy balance results, and determine whether the preset governance threshold has been reached.

[0047] The above description is only a preferred embodiment of the present invention. It should be noted that those skilled in the art can make several improvements and additions without departing from the principle of the present invention, and these improvements and additions should also be considered within the scope of protection of the present invention.

Claims

1. A method for energy efficiency management and loss control on the power side under dynamic load characteristics, characterized in that, The method includes: By collecting voltage, current and power data of each node in the power distribution network in real time through sensor networks, dynamic load distribution information can be obtained to determine the degree of influence of load connection location on energy flow. Based on the obtained dynamic load distribution information, the k-means clustering algorithm is used to group the network nodes to obtain the grouped load clusters, and it is determined which clusters have uneven energy flow. If the energy flow is uneven in the grouped load cluster, the loss value is calculated by simulating the energy transfer path to determine the location and severity of the critical loss nodes; For the identified critical loss nodes, obtain the transformer terminal operating status data, including temperature and load rate, to determine whether the transformer terminal is in an overload state. If the transformer is in an overload state, a deep reinforcement learning algorithm is used to optimize the network structure adjustment strategy to obtain an adjusted load distribution scheme that balances energy flow and transformer operating status. Based on the adjusted load distribution scheme, the remote operation and maintenance system sends control commands to the intelligent interconnected devices to determine the reconfiguration of paths in the network structure and achieve loss control. Based on the reconfigured network structure, monitor changes in power-side energy efficiency indicators to obtain optimized energy balance results and determine whether the preset governance threshold has been reached.

2. The method for power-side energy efficiency management and loss control under dynamic load characteristics according to claim 1, characterized in that, The process of acquiring real-time voltage, current, and power data of each node in the power distribution network through a sensor network to obtain dynamic load distribution information and determine the degree of influence of load connection location on energy flow includes: The sensor network is used to collect data from each node in the power distribution network in real time, and the raw data streams of node voltage, node current and node power are obtained and stored in a pre-established data warehouse to obtain the initial data collection dataset. For the initial collected dataset, a data cleaning method is used to remove outliers and missing values. If the voltage or current of a node exceeds the preset threshold range, it is marked as abnormal data and removed, resulting in a cleaned node parameter dataset. Based on the cleaned node parameter dataset, the dynamic load value of each node is calculated, and the distribution of load over time is extracted using time series analysis to determine the load distribution feature set of each node. ； This formula is used to calculate the dynamic load value of each node; where Represents a node At any moment The dynamic load value, Represents a node At any moment Power data, Represents a node At any moment voltage value, Represents a node At any moment The current value, Indicates the length of the time series. Indicates the time weighting coefficient. Represents a node At any moment The formula extracts the load distribution over time using time series analysis to determine the load distribution characteristics of each node. For the load distribution feature set, the correlation between the load access location and each node is analyzed. If the load value change at a certain access location causes the power fluctuation of adjacent nodes to exceed a preset threshold, then the location is recorded as a key access point, and a set of key access points is obtained. By using a set of key access points and combining the support vector machine algorithm, the energy flow path is modeled, the effect of each key access point on the direction and intensity of energy flow is analyzed, and the degree of influence of the load access location on energy flow is determined. ； This formula is used to calculate the degree to which energy flow is affected by the location of load access; where Indicates access location The degree of influence on energy flow. This indicates the total number of critical access points. Indicates the first The weight coefficients of key access points Indicates the first Energy flow intensity at key access points Indicates access location The load capacity, The penalty coefficient represents the energy deviation. This represents the reference energy flow value at the j-th critical access point. This formula, combined with a support vector machine algorithm, models the energy flow path and analyzes the influence of each critical access point on the direction and intensity of energy flow. Based on the impact value of energy flow, a priority ranking of each key access point is generated. If the impact value of a certain access point is higher than the preset threshold, it is marked as a high priority point, thus obtaining the basis for optimizing and adjusting the load access points in the distribution network. Based on the optimization and adjustment criteria, a load balance distribution scheme for each node in the power distribution network is generated. By simulating changes in energy flow paths, the final load access location adjustment strategy is determined.

3. The method for power-side energy efficiency management and loss control under dynamic load characteristics according to claim 1, characterized in that, Based on the acquired dynamic load distribution information, the network nodes are grouped using the k-means clustering algorithm to obtain the grouped load clusters. The process then involves determining which clusters exhibit uneven energy flow, including: By using a pre-established data acquisition system, dynamic load and distribution information of each node in the power distribution network is obtained and stored in a data warehouse to obtain an initial load dataset. For the initial load dataset, a data filtering method is used to remove noisy data. If the load data of a certain node exceeds the preset threshold range, it is marked as invalid data and removed to obtain the processed load dataset. Based on the processed load dataset, the k-means clustering algorithm is applied to group the network nodes, and multiple load clusters are generated based on the similarity of load distribution to determine the load cluster set. For a set of load clusters, the energy flow status within each cluster is analyzed. If the energy flow value of a node in a cluster deviates from the average value by more than a preset threshold, it is marked as an unbalanced cluster, and a list of unbalanced clusters is obtained. ； This formula is used to determine whether the energy flow is balanced in a load cluster; Indicates the first Energy flow imbalance in individual clusters This represents the set of nodes within the k-th cluster. Indicates the total number of nodes in the cluster. Indicates the number within the cluster Energy flow value of each node The horizontal line above indicates the number. The average energy flow of all nodes within a cluster; when When the threshold is exceeded, the cluster is marked as an unbalanced cluster; By using the unbalanced cluster list, load distribution and energy status data of nodes within each cluster are extracted, energy flow deviation records of each cluster are generated, and the range of clusters with large deviations is determined. Based on the energy flow deviation records, nodes in the unbalanced cluster are classified and processed. If the load distribution of a certain node causes the energy state to deviate continuously, it is recorded as a critical node, and a set of critical nodes is obtained. For the set of critical nodes, a priority ranking scheme for load adjustment is generated. By analyzing the energy status and load distribution of the critical nodes, the adjustment order list is determined.

4. The method for power-side energy efficiency management and loss control under dynamic load characteristics according to claim 1, characterized in that, If the energy flow is uneven in the grouped load cluster, the loss value is calculated by simulating the energy transfer path to determine the location and severity of the critical loss nodes, including: By using a pre-established power distribution network data acquisition system, real-time energy flow data is obtained from each load cluster and stored in a data warehouse to obtain an initial energy flow dataset. For the initial energy flow dataset, a data cleaning method is used to remove outliers. If the energy flow data of a certain cluster exceeds the preset threshold range, it is marked as invalid data and removed, resulting in a cleaned energy flow dataset. Based on the cleaned energy flow dataset, a simulation model of the energy transfer path is constructed to analyze the energy flow state on each path and determine the preliminary loss distribution of each path. ； This formula is used to calculate the loss value of the energy transfer path; Representing a path Total loss value on, Indicates the number of node segments on the path. Indicates the first The resistance value of the segment, This represents the current value in segment j. Indicates the first The loss coefficient of the segment, Indicates the first The power value of the segment, Indicates the first The voltage value of the segment. This formula can be used to identify path segments with high loss values; Based on the preliminary loss distribution, loss calculation tools are used to analyze the energy loss values ​​on each path one by one to obtain a detailed loss distribution record. Based on detailed loss distribution records, path segments with high loss values ​​are identified. If the loss value of a certain path segment exceeds a preset threshold, it is marked as a high-loss path, thus obtaining a set of high-loss paths. Based on the set of high-loss paths, locate the corresponding key nodes within the path segments, analyze the energy flow deviation of each key node, and determine the distribution location of the loss key nodes. Based on the distribution location of critical loss nodes, the severity of each node is classified and processed. If the energy loss value of a certain node is consistently higher than the average value, it is recorded as a high-severity node, and a list of high-severity nodes is obtained.

5. The method for power-side energy efficiency management and loss control under dynamic load characteristics according to claim 1, characterized in that, For the identified critical loss nodes, the process involves acquiring transformer terminal operating status data, including temperature and load rate, to determine whether the transformer terminal is in an overload state, including: For the identified critical loss nodes, the operating status information, including temperature data and load rate values, is obtained from the transformer through a pre-established monitoring system to obtain a preliminary operating status dataset. Based on the preliminary operational status dataset, a data cleaning tool is used to detect outliers in the temperature data and load rate values. If the temperature or load rate value of a certain data exceeds the preset threshold range, it is marked as outlier data and removed, resulting in a cleaned operational status dataset. Based on the cleaned operating status dataset, the information analysis module comprehensively evaluates the temperature data and load rate value at the transformer end. If the temperature data continues to be higher than the preset threshold or the load rate value exceeds the safe range, it is determined to be a potential overload operating state, and the overload risk assessment result is obtained. Based on the overload risk assessment results, obtain the historical operating status information of the transformer end, and compare and analyze it with the current data. If there are similar high temperature or high load rate records in the historical data, the current status is confirmed to be overload operation, and the distribution of nodes in overload operation is determined. For nodes operating under overload, the dynamic status information of the transformer end is obtained in real time through the operation monitoring system. The changing trends of temperature data and load rate values ​​are analyzed to determine whether there is a risk of continuous deterioration and to obtain dynamic monitoring records. Based on dynamic monitoring records, the overload operation risk at the transformer end is classified and predicted using the support vector machine algorithm. Combining the temperature data and load rate value change characteristics, the priority ranking of high-risk nodes is determined, resulting in a risk priority list. ； This formula uses the support vector machine algorithm to classify and predict the overload operation risk at the transformer end; This represents the decision function value of the support vector machine, used for risk classification; Indicates the number of support vectors; Indicates the first The weight coefficients of each support vector; This represents the kernel function used to compute support vectors. The similarity between the vector and the current state vector x, where Includes variation characteristics of temperature data and load rate values; This represents the bias term; a larger decision function value indicates a higher risk of overload. For the risk priority list, information analysis tools are used to correlate and map the operational status information of high-risk nodes with the distribution of loss-critical nodes to determine which nodes need to be monitored in a key manner, thus obtaining a set of key monitoring nodes.

6. The method for power-side energy efficiency management and loss control under dynamic load characteristics according to claim 1, characterized in that, If the transformer is in an overload state, a deep reinforcement learning algorithm is used to optimize the network structure adjustment strategy to obtain an adjusted load distribution scheme that balances energy flow and transformer operating status, including: In the case of an overload at the transformer end, a pre-established monitoring system is used to obtain real-time operating status information and load distribution data from the transformer end to obtain an initial operating status record. Based on the initial running status records, data processing tools are used to clean and organize the acquired running status information, remove abnormal data points, and obtain the processed running status dataset. For the processed operational status dataset, the information analysis module extracts key features of load distribution and energy flow, determines whether there is an imbalance, and obtains load distribution analysis results. Based on the load distribution analysis results, if the load distribution in a certain area is found to exceed the preset threshold range, the network structure will be dynamically adjusted through a deep reinforcement learning algorithm to generate a preliminary load adjustment plan. For the initial load adjustment plan, simulation analysis tools were used in conjunction with energy flow data to conduct multi-scenario tests on the plan, determine its feasibility, and obtain an optimized load adjustment plan. Based on the optimized load adjustment scheme, the adjustment scheme is applied to the actual operating environment of the transformer through the operation monitoring system. The adjusted operating status information is obtained in real time to determine whether a balanced state has been reached and to obtain the final operating status feedback.

7. The method for power-side energy efficiency management and loss control under dynamic load characteristics according to claim 1, characterized in that, The adjusted load distribution scheme is used to send control commands from the remote operation and maintenance system to the intelligent interconnected devices, determine the reconfiguration of paths in the network structure, and achieve loss control, including: By using a remote operation and maintenance system, we can obtain data on the adjusted load distribution scheme, combine it with a pre-established network structure model, analyze the preliminary requirements for path planning, and determine the priority sequence of path adjustments. Based on the priority sequence of path adjustment, a remote operation and maintenance system is used to issue control commands to intelligent interconnected devices to complete the dynamic reconstruction of the network structure and obtain the reconstructed network topology information. Based on the reconstructed network topology information, the monitoring module is run to collect real-time operating data of intelligent interconnected devices, determine whether device management meets the preset threshold range, and obtain feedback results on device status. If the feedback result of the device status shows that the operating data exceeds the preset threshold range, a new control command is generated through the remote operation and maintenance system to adjust the operating parameters of the smart interconnected device and determine the adjusted device operating status. Based on the adjusted equipment operating status, and in conjunction with the loss control module, the energy transmission loss in the network structure is analyzed to determine whether the loss is within an acceptable range, and loss assessment data is obtained. Based on the loss assessment data, the assessment results are uploaded to the remote operation and maintenance system through a status feedback mechanism. Combined with real-time monitoring of load distribution, the final path planning optimization scheme is determined.

8. The method for power-side energy efficiency management and loss control under dynamic load characteristics according to claim 1, characterized in that, The step of monitoring changes in power-side energy efficiency indicators based on the reconfigured network structure, obtaining optimized energy balance results, and determining whether a preset governance threshold has been reached includes: By using a pre-established network structure model, the reconfigured network structure information is obtained, and combined with relevant data on power-side energy efficiency, the initial changes in indicators are determined. In response to the initial changes in indicators, a real-time monitoring module is used to continuously collect dynamic data on the power-side energy efficiency to obtain the preliminary state of energy balance. Based on the initial state of energy balance, the dynamic data is classified using the support vector machine algorithm to determine whether it is close to the preset governance threshold range. If the classification results show that the energy balance status has not reached the preset governance threshold range, the data processing module will perform hierarchical analysis on the real-time monitored data to obtain specific deviation information. For specific deviation information, an automated adjustment tool is used to generate targeted network structure optimization instructions and determine the trend of changes in the adjusted power-side energy efficiency indicators. ； This formula is used to generate an automated adjustment strategy for network structure optimization instructions; where This represents the adjusted network structure parameter vector. This represents the current network structure parameter vector. This indicates adjusting the step size coefficient. This represents the gradient of the objective function with respect to the parameters. Let T represent the objective function for the deviation, and let T represent the total number of monitoring periods. This represents the preset target energy balance value. Indicates at time Based on parameters The actual energy balance value. This formula determines the trend of the adjusted power-side energy efficiency index using the gradient descent method; Based on the changing trend of the adjusted power-side energy efficiency indicators, the optimized data is uploaded to the monitoring system through the state feedback mechanism to obtain the final balance state assessment result.

9. A power-side energy efficiency management and loss control system under dynamic load characteristics, characterized in that, The system includes: The data acquisition module is used to collect voltage, current and power data of each node in the power distribution network in real time through the sensor network, obtain dynamic load distribution information, and determine the degree of influence of load connection location on energy flow. The node grouping module is used to group network nodes using the k-means clustering algorithm based on the acquired dynamic load distribution information, obtain the grouped load clusters, and determine which clusters have uneven energy flow. The loss calculation module is used to calculate the loss value by simulating the energy transfer path if the energy flow in the grouped load cluster is uneven, and to determine the location and severity of the loss critical nodes. The transformer status judgment module is used to acquire transformer terminal operating status data, including temperature and load rate, for identified key loss nodes, and to determine whether the transformer terminal is in an overload state. The network optimization module is used to optimize the network structure adjustment strategy using a deep reinforcement learning algorithm if the transformer is in an overload state, so as to obtain an adjusted load distribution scheme and balance energy flow and transformer operating status. The path configuration module is used to send control commands from the remote operation and maintenance system to the intelligent interconnected devices based on the obtained adjusted load distribution scheme, determine the reconfiguration of paths in the network structure, and realize loss control. The energy efficiency monitoring module is used to monitor changes in power-side energy efficiency indicators based on the reconfigured network structure, obtain optimized energy balance results, and determine whether the preset governance threshold has been reached.

Images