A multi-objective collaborative configuration method based on regional power distribution network resources

Abstract

The present application relates to the technical field of power distribution network optimization operation, in particular to a multi-objective collaborative configuration method based on regional power distribution network resources, comprising: collecting real-time operation data of the power distribution network, and establishing a multi-time scale data coordination mechanism of minute, hour and day levels to form a hierarchical data set. Based on the set, a decision space is constructed with capacitor banks, on-load tap-changing transformers and distributed energy storage systems as dispatchable resources, and a multi-objective optimization problem is generated considering network loss, voltage deviation and power supply reliability. An adaptive weight adjustment strategy is adopted to dynamically adjust the weight coefficients of each objective according to the current operating state to solve the optimization problem, and the optimal scheme is issued for execution. Through monitoring the execution deviation and feedback adjustment strategy parameters, a closed-loop control is formed. The method improves the adaptability of the optimization strategy to different operating scenarios and the overall decision effect through multi-time scale data fusion and dynamic weight adaptation.

Description

Technical Field

[0001] This invention relates to the field of power distribution network optimization operation technology, and in particular to a method for multi-objective collaborative allocation of regional power distribution network resources. Background Technology

[0002] Existing methods for optimizing the allocation of regional distribution network resources typically rely on operational data at a single time scale, such as static optimization analysis using data collected at fixed intervals. These methods establish optimization models based on historical or short-term measured data, and use preset fixed weights to balance multiple optimization objectives during model solving. When dealing with the dynamic characteristics of distributed power generation, such as intermittent output and frequent load fluctuations in distribution networks, these technical solutions often struggle to simultaneously achieve both optimization speed and long-term operational economy.

[0003] Fixed-time-scale data aggregation methods cannot accurately reflect the operational characteristics of the distribution network across different time dimensions, potentially causing optimization decisions to overlook rapid fluctuations at the minute level or trend changes at the daily level. Fixed-weight multi-objective optimization strategies lack adaptability to changes in operating conditions and cannot automatically adjust the focus of objectives in specific scenarios such as voltage exceeding limits or sudden increases in network losses, potentially leading to optimization results that deviate from actual needs. The singularity of data processing and the rigidity of weight settings limit the real-time and accuracy of resource allocation, impacting the overall operational efficiency and safety of the distribution network.

[0004] This invention addresses how to leverage multi-time-granularity data to enhance the spatiotemporal adaptability of optimization models, and how to automatically adjust multi-objective weights based on real-time operating conditions to improve the flexibility of optimization strategies. This involves constructing hierarchical datasets at the minute, hour, and day levels to support dynamic optimization, and designing an adaptive solution mechanism where weight coefficients change in real-time with the operating conditions. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a multi-objective collaborative configuration method for regional distribution network resources.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for multi-objective collaborative allocation of regional distribution network resources, comprising:

[0007] Real-time operational data is collected by monitoring terminals deployed at various nodes of the distribution network. The real-time operational data includes line load, node voltage, distributed power output, and user electricity demand.

[0008] Establish a multi-timescale data coordination mechanism to classify and aggregate the collected real-time operational data according to minute-level, hour-level, and day-level time granularity to form a hierarchical data set;

[0009] A resource allocation decision space is constructed, which is defined by the types of schedulable resources in the distribution network and their operating parameter ranges. The schedulable resources include capacitor banks, on-load tap changers, and distributed energy storage systems.

[0010] A multi-objective optimization problem is generated based on a hierarchical dataset. The objective function of the multi-objective optimization problem simultaneously considers minimizing network loss, minimizing voltage deviation, and maximizing power supply reliability.

[0011] An adaptive weight adjustment strategy is adopted to solve the multi-objective optimization problem, and the weight coefficients of each objective function are dynamically adjusted according to the current operating status of the distribution network.

[0012] The optimal resource allocation scheme obtained from the solution is distributed to each resource control terminal for execution, and the deviation between the execution effect and the expected target is monitored in real time.

[0013] The parameters of the adaptive weight adjustment strategy are adjusted based on the feedback of the execution deviation to form a closed-loop optimization control.

[0014] As a further aspect of the present invention, the establishment of a multi-timescale data coordination mechanism includes:

[0015] Configure a clock synchronization module for each monitoring terminal to ensure that all real-time operating data has a unified time label;

[0016] Set data preprocessing rules to perform outlier detection and missing data imputation on raw sampling data at the minute-level granularity;

[0017] The processed minute-level data is aggregated by hourly average to generate an hourly data set, and the data fluctuation characteristic value for each hour is calculated.

[0018] Hourly data is processed by extracting key indicators on a daily basis to generate a daily data set. The key indicators include daily maximum load, daily minimum voltage, and daily average power supply reliability.

[0019] Establish a mapping relationship between the three granular data sets so that the upper-level data can be traced back to the detailed operational data of the lower level.

[0020] As a further aspect of the present invention, the step of aggregating the processed minute-level data by hourly average includes:

[0021] The real-time operating data collected every minute is stored in hourly segments, with the hour as the dividing line;

[0022] Calculate the arithmetic mean of each monitored parameter for each hour, and use it as the representative value for the corresponding hour;

[0023] Simultaneously, the standard deviation of each monitored parameter is calculated for each hour, serving as a quantitative indicator of the corresponding hourly data volatility.

[0024] The mean and standard deviation are stored together in an hourly dataset, and each data item is labeled with its corresponding collection time period;

[0025] When the missing data rate of a certain hour exceeds the preset threshold, the adjacent hour data compensation mechanism is activated to supplement the data using the weighted average of the preceding and following hours.

[0026] As a further aspect of the present invention, the construction of the resource allocation decision space includes:

[0027] Identify all resource types in the distribution network that can participate in scheduling, and establish a parameterized model for each resource;

[0028] For capacitor bank resources, establish a parametric model of its capacity, current switching status, and response speed;

[0029] For the tap changer of an on-load tap changer, establish a parametric model for its turns ratio range, adjustment step size, and adjustment delay time;

[0030] For distributed energy storage systems, establish parametric models for their rated capacity, current state of charge, maximum charge / discharge power, and cycle efficiency;

[0031] The parameter models of all resources are integrated into a unified resource allocation decision space, and the constraints for coordinated operation among the resources are defined.

[0032] As a further aspect of the present invention, the defined constraints for coordinated operation among resources include:

[0033] Analyze the mutual influence relationships of different types of resources during the adjustment process and establish a resource coupling constraint matrix;

[0034] Set resource adjustment priority rules to ensure that important resources have priority scheduling rights when adjusting conflicts;

[0035] Considering the impact of distribution network topology on resource coordination, establish a resource coordination feasibility verification rule based on power flow.

[0036] Define resource adjustment frequency limits to prevent frequent adjustments from adversely affecting equipment lifespan;

[0037] Establish global coordination constraints to ensure that the coordinated allocation of all resources meets the requirements for safe operation of the power distribution network.

[0038] As a further aspect of the present invention, the multi-objective optimization problem generated based on the hierarchical data set includes:

[0039] Extract the current operating status characteristics of the distribution network from minute-level data sets, including real-time load and voltage distribution at each node;

[0040] Extract load change trends and voltage fluctuation patterns from hourly datasets to serve as time-varying constraints for the optimization problem;

[0041] Historical performance indicators are extracted from daily datasets and used as benchmark reference values ​​for optimization targets.

[0042] Construct a multi-objective optimization model that includes network loss objective function, voltage deviation objective function, and power supply reliability objective function;

[0043] By using adjustable parameters in the resource allocation decision space as optimization variables, a mathematical relationship is established between the objective function and the optimization variables.

[0044] As a further aspect of the present invention, the construction of a multi-objective optimization model comprising a network loss objective function, a voltage deviation objective function, and a power supply reliability objective function includes:

[0045] The network loss objective function is based on the power flow calculation model of the distribution network, which expresses the total active power loss as a function of the power injected into each node;

[0046] The voltage deviation objective function calculates the sum of squares of the voltage deviations from the rated voltage at all nodes, and assigns differentiated weights to different nodes based on their importance.

[0047] The power supply reliability objective function is based on historical fault data and equipment reliability parameters, and a correlation model between resource allocation and power supply reliability indicators is established.

[0048] All three objective functions were normalized to eliminate the influence of dimensional differences on the optimization results;

[0049] The optimization model incorporates constraints on the safe operation of the distribution network, including line capacity constraints, voltage upper and lower limit constraints, and power output constraints.

[0050] As a further aspect of the present invention, the method of using an adaptive weight adjustment strategy to solve the multi-objective optimization problem includes:

[0051] Real-time monitoring of the distribution network operation status, identification of the most critical issues, and increase the weight of the voltage deviation target if the voltage deviation exceeds the standard.

[0052] Design a weight smoothing adjustment mechanism to avoid drastic fluctuations in resource allocation schemes caused by sudden weight changes;

[0053] Establish a non-linear mapping relationship between weights and deviations in operating indicators; use gradual adjustments for small deviations and rapid responses for large deviations.

[0054] During the weight adjustment process, the coupling relationship between the objective functions should be considered to avoid over-optimization of a single objective at the expense of other objectives;

[0055] Regularly evaluate the effectiveness of weight adjustments and optimize the parameter settings of the weight adjustment strategy based on the evaluation results.

[0056] As a further aspect of the present invention, the design weight smoothing adjustment mechanism includes:

[0057] Set a limit on the weight adjustment speed and specify the maximum change in the weights of each objective function per unit time.

[0058] A sliding window is used to record the historical weight sequence, calculate the weight change trend, and use it as a reference for current adjustments;

[0059] When a sudden change in the operating status is detected, the emergency weight adjustment mode is activated, and the adjustment speed limit is temporarily relaxed.

[0060] After the weights are adjusted, a stable observation period is set, during which the weights are kept unchanged to observe the effect of the adjustment.

[0061] Establish a weight rollback mechanism so that when the effect of the adjustment is not good, the weight settings before the adjustment can be quickly restored.

[0062] As a further aspect of the present invention, the parameter settings for adjusting the adaptive weight adjustment strategy based on execution deviation feedback include:

[0063] The actual performance of the resource allocation plan is compared with the expected goals in real time, and the deviation of each goal is calculated.

[0064] Analyze the causes of the deviation and distinguish whether it is due to unreasonable weight settings or changes in the external environment;

[0065] Establish a correlation model between deviation and weight adjustment parameters, and adjust the sensitivity parameters of the weight adjustment strategy according to the deviation characteristics;

[0066] Regularly conduct a comprehensive evaluation of the weight adjustment strategy and optimize the core parameters of the strategy based on historical data;

[0067] The optimized parameters are then updated into the adaptive weight adjustment strategy, completing one full cycle of closed-loop optimization.

[0068] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0069] The establishment of a multi-timescale data coordination mechanism categorizes and aggregates real-time operational data according to minute-level, hourly-level, and daily-level time granularities, forming a hierarchical data set. This technology, through parallel processing of data across different time dimensions, enables the optimization model to simultaneously respond to rapid instantaneous fluctuations and long-term trend changes, resulting in more comprehensive optimization decisions across time scales. Modeling based on hierarchical data effectively integrates short-term dynamic characteristics with long-term operational patterns, enhancing the model's ability to characterize the complex time-varying characteristics of the distribution network. Resource allocation schemes exhibit better adaptability and continuity in addressing operational needs across different time dimensions, avoiding decision-making lags or short-sightedness caused by a single data aggregation method.

[0070] The adaptive weight adjustment strategy dynamically adjusts the weight coefficients of each objective function based on the real-time operating status during the solution process. This technique automates and enables real-time optimization objective emphases by establishing a dynamic mapping relationship between operating state parameters and weights. The dynamic allocation of weight coefficients allows multi-objective optimization to autonomously adapt to different operating scenarios, automatically adjusting the priority of objectives such as network loss, voltage deviation, and reliability when the state changes. This real-time feedback adjustment mechanism eliminates the rigidity of fixed-weight strategies, tightly coupling the solution process with the actual system operating conditions. The optimization results can more accurately match the most pressing needs, enhancing the situational adaptability and overall decision-making quality of multi-objective optimization. Attached Figure Description

[0071] Figure 1 This is a flowchart of the multi-objective collaborative allocation method for regional distribution network resources described in this invention;

[0072] Figure 2 A flowchart for establishing a multi-timescale data coordination mechanism;

[0073] Figure 3 A flowchart for constructing the resource allocation decision space;

[0074] Figure 4 Dynamically adjust stacked bar charts for multi-objective optimization of distribution networks;

[0075] Figure 5 Line graph showing the deviation and adjustment effect of multi-objective optimization for power distribution networks. Detailed Implementation

[0076] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0077] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0078] See Figure 1 Real-time operational data is collected through monitoring terminals deployed at various nodes of the distribution network. This data includes line load, node voltage, distributed power generation output, and user electricity demand. A multi-timescale data coordination mechanism is established to classify and aggregate the collected real-time operational data according to minute, hourly, and daily time granularities, forming a hierarchical data set. A resource allocation decision space is constructed, defined by the types of schedulable resources in the distribution network and their operating parameter ranges. Schedulable resources include capacitor banks, on-load tap changers, and distributed energy storage systems. A multi-objective optimization problem is generated based on the hierarchical data set. The objective function of the multi-objective optimization problem simultaneously considers minimizing network losses, minimizing voltage deviation, and maximizing power supply reliability. An adaptive weight adjustment strategy is used to solve the multi-objective optimization problem, dynamically adjusting the weight coefficients of each objective function according to the current operating state of the distribution network. The optimal resource allocation scheme obtained is distributed to each resource control terminal for execution, and the deviation between the execution effect and the expected target is monitored in real time. The parameter settings of the adaptive weight adjustment strategy are adjusted based on the execution deviation feedback, forming a closed-loop optimization control.

[0079] See Figure 2In one embodiment of the present invention, monitoring terminals deployed at each node of the distribution network collect raw data on line load, node voltage, distributed power output, and user electricity demand at a frequency of once per minute. In specific implementation, each monitoring terminal is equipped with a clock synchronization module. This module receives a unified timing signal from the BeiDou Navigation Satellite System or Network Time Protocol, and adds a time stamp accurate to the millisecond level to each frame of real-time operating data collected, thereby ensuring that the data reported by all nodes are strictly aligned in the time dimension. In some embodiments, after the data is uploaded to the main station system, the system cleans the raw sampling data at the minute level according to preset data preprocessing rules. The preprocessing rules include outlier detection and missing data imputation. Outlier detection adopts a threshold judgment method based on the statistical three sigma principle. When a data point exceeds the range of plus or minus three standard deviations of the historical average for the same period, the data point is marked as outlier. For missing or outlier data points, the system uses linear interpolation to calculate and fill in the missing data points using the values ​​of the two valid sampling points before and after the missing data point.

[0080] In practical implementation, the core step in generating hourly datasets is to aggregate the processed minute-level data by hourly average. The system uses natural hours as boundaries, storing each minute's record in hourly segments. For each monitoring parameter, its arithmetic mean for each hour is calculated, and this average is stored as the representative value of that parameter for that hour in the hourly dataset. Simultaneously, the system calculates the standard deviation of the parameter for that hour, serving as a quantitative indicator reflecting the data volatility within that hour. The average and standard deviation together constitute a record in the hourly dataset, with each record clearly marked with its corresponding collection time period. It can be understood that when the missing data rate for a certain hour exceeds a preset threshold, such as 30%, the system will activate a data compensation mechanism for adjacent hours. Instead of directly calculating the average for that hour, it will use the weighted average of the data from the preceding and following hours to supplement it. The weighting coefficient is dynamically determined based on the time distance between the adjacent hour and the target hour, and the calculation formula is as follows:

[0081]

[0082] in: The compensation value represents the target hour. and These represent the average values ​​of the previous hour and the next hour, respectively. and It is a weighted coefficient and satisfies .

[0083] In practical implementation, based on the generated hourly dataset, the system further extracts key indicators on a daily basis to form a daily dataset. The system traverses all hourly data records throughout the day, extracting the daily maximum load value and its occurrence time from the line load data, the daily minimum voltage value and its occurrence time from the node voltage data, and calculating the daily average power supply reliability index from the power supply reliability logic judgment results. The daily average power supply reliability index is the ratio of the number of time points with normal power supply in all minute-level records to the total number of time points. These extracted key indicators are integrated into a single daily record and stored. Optionally, to support data traceability and analysis, the system establishes a mapping relationship between the minute-level, hourly-level, and daily-level datasets. This mapping relationship is achieved through foreign key links in the database and timestamp interval matching, allowing immediate drill-down when abnormal indicators are observed in the daily dataset to view the hourly fluctuation characteristics of the corresponding date and further locate the specific minute-level original operating data, thereby completing data traceability from macro indicators to micro details.

[0084] See Figure 3 In one embodiment of the present invention, the process of constructing the resource allocation decision space is achieved by identifying all dispatchable resource types in the distribution network and establishing a parameterized model for each resource as the first step. In a configuration task for a distribution network in a certain area, the system identifies three types of dispatchable resources: capacitor banks located at nodes 5, 8, and 11; on-load tap-changing transformer taps located at the main transformer outlet of a substation; and distributed energy storage systems installed at nodes 7 and 15. In specific implementation, for capacitor bank resources, the system establishes a parameter model for them that includes capacity, current switching status, and response speed. The response speed parameter model records the time from receiving the instruction to completing the switching action as 30 seconds. For on-load tap-changing transformer tap resources, the parameter model established by the system covers the turns ratio range, adjustment step size, and adjustment delay time. For example, the turns ratio range parameter model of the on-load tap-changing transformer tap is defined as 1.0 ± 8 × 1.25%, the adjustment step size parameter model is 1.25% per tap, and the adjustment delay time parameter model is 60 seconds for each tap adjustment. In some embodiments, for distributed energy storage system resources, the parameter model includes rated capacity, current state of charge, maximum charge / discharge power and cycle efficiency. The rated capacity parameter model of the distributed energy storage system at node 7 is 500 kWh, the current state of charge parameter model is 65% obtained in real time through the battery management system, the maximum charge / discharge power parameter model is 250 kW, and the cycle efficiency parameter model is recorded as 95% and 96% respectively during the charge and discharge process.

[0085] In practical implementation, the parameter models of all resources are integrated into a unified resource allocation decision space. This resource allocation decision space is a multi-dimensional vector, where each dimension represents a specific adjustable parameter of a particular resource. For example, it might be a set including the number of capacitor banks switched on / off, the tap positions of on-load tap-changing transformers, and the charging / discharging power of distributed energy storage systems. The definition of the resource allocation decision space also includes the feasible range of each parameter, such as the number of capacitor banks switched on / off being an integer from 0 to 2, the tap positions of on-load tap-changing transformers being an integer from -8 to +8, and the charging / discharging power of distributed energy storage systems being a continuous value from -250 kW to +250 kW. It is understood that the resource allocation decision space must match the physical constraints and operating rules of the distribution network; therefore, it is necessary to define the coordinated operation constraints between various resources. In practical implementation, when defining the coordinated operation constraints, the system analyzes the mutual influence relationships of different types of resources during the adjustment process and establishes a resource coupling constraint matrix. For example, the switching on / off of capacitor banks directly affects the node voltage, which in turn affects the adjustment demand of the on-load tap-changing transformer taps. This influence relationship is represented in the resource coupling constraint matrix by a correlation factor. Quantification is performed, among which This represents the unit impact of the action of resource i on the control objective of resource j.

[0086] In some embodiments, the system sets resource adjustment priority rules. When capacitor bank adjustment commands and on-load tap changer adjustment commands conflict in time, the rules determine that the on-load tap changer resource has priority scheduling rights because its adjustment impact is wider. Optionally, considering the impact of distribution network topology on resource coordination, the system establishes a power flow-based resource coordination feasibility verification rule. After generating any resource configuration scheme, the system calls a power flow calculation program to simulate and verify whether the scheme will cause any line overload or node voltage exceeding the limit. Only schemes that pass the verification are feasible solutions. Defining resource adjustment frequency limits is also a key constraint. For example, the rules limit the number of adjustments made by on-load tap changers to no more than 4 times in one consecutive hour to prevent excessive wear of mechanical parts. In specific implementation, the final established global coordination constraints require that all resource collaborative configuration schemes must simultaneously meet safe operation requirements such as maintaining node voltage between 0.95 and 1.05 per unit and feeder load rate not exceeding 90%, thereby ensuring that each point in the resource configuration decision space corresponds to a safe and feasible operating state.

[0087] In one embodiment of the present invention, the process of generating a multi-objective optimization problem based on a hierarchical data set is implemented in the following way: the first step is to extract the current operating status characteristics of the distribution network from the minute-level data set. In a specific optimization cycle calculation, the system accesses the latest minute-level data set and extracts the real-time voltage amplitude of all 56 nodes. For example, the voltage of node 15 is 1.032 per unit and the voltage of node 42 is 0.957 per unit. At the same time, the real-time active load of all 78 branches is extracted. For example, the load of branch 5-6 is 1.85 MW and the load of branch 20-21 is 0.73 MW. These data together constitute the current network snapshot on which the optimization calculation depends. In some embodiments, load change trends and voltage fluctuation patterns are extracted from hourly datasets as time-varying constraints for the optimization problem. The system analyzes the hourly datasets of the past 6 hours to identify a pattern where load shows a continuous upward trend in the afternoon and voltage fluctuations intensify in the evening. These trends and patterns are transformed into constraints on the prediction range for the next hour. For example, the allowable fluctuation range of node voltage in the next hour is set to ±0.02 per unit of the current value, and the upper limit of the branch active load growth is set to 10% of the current value. Historical operating indicators are extracted from daily datasets as benchmark reference values ​​for the optimization target. The system retrieves daily datasets of the past 7 days and calculates an average daily network loss rate of 2.3%, an average voltage deviation index of 0.015, and an average power supply reliability rate of 99.92%. These historical averages serve as the baseline for the expected improvement in this optimization.

[0088] In practical implementation, the core task is to construct a multi-objective optimization model that includes network loss objective function, voltage deviation objective function, and power supply reliability objective function. The network loss objective function, based on the distribution network power flow calculation model, expresses the total active power loss as a function of the injected power at each node. Its purpose is to find a resource allocation that minimizes the total active power loss of the system after power flow calculation. The voltage deviation objective function calculates the sum of squares of the voltage deviations of all nodes from their rated voltages and assigns differentiated weights based on the importance of different nodes; for example, higher weight coefficients are assigned to the voltage deviations of critical load nodes and network end nodes. The power supply reliability objective function, based on historical fault data and equipment reliability parameters, establishes a correlation model between resource allocation and power supply reliability indicators. This model is constructed by analyzing the relationship between the number of operations of equipment such as capacitor banks and on-load tap changers, the charging and discharging depth of energy storage systems, equipment failure rates, and the resulting probability of power outages for users. All three objective functions are normalized to eliminate the influence of dimensional differences on the optimization results. Each objective function value is divided by its corresponding historical benchmark value to transform it into a dimensionless relative index.

[0089] In practical implementation, a specific mathematical expression for the voltage deviation objective function is as follows:

[0090]

[0091] in: This represents the target function value of the voltage deviation. It is the total number of distribution network nodes. It is the node's index number. It is to assign nodes Voltage deviation weighting factor, It is a node voltage amplitude, This is the system's rated voltage reference value. Optional, weighting factor. The settings are determined based on the importance of the nodes. For nodes connecting critical loads such as hospitals and data centers, their... The value is set to 2.0; for ordinary residential load nodes, its The value is set to 1.0. Introducing distribution network safety operation constraints into the optimization model is an indispensable step. Line capacity constraints require that the active and reactive power flowing through each branch must not exceed its thermal stability limit; voltage upper and lower limit constraints require that the voltage amplitude of all nodes must be maintained between 0.95 and 1.05 per unit; and power output constraints require that the actual output of distributed photovoltaic, wind power, and other power sources must not exceed their predicted maximum power output. In some embodiments, adjustable parameters in the resource allocation decision space are used as optimization variables, including integer variables of the switching states of each capacitor bank, integer variables of the tap positions of on-load tap-changing transformers, and continuous variables of the charging and discharging power of each distributed energy storage system. Mathematical relationships are established between these optimization variables and the aforementioned three objective functions and constraints, thus fully defining the multi-objective optimization problem.

[0092] In one embodiment of the present invention, the process of solving a multi-objective optimization problem using an adaptive weight adjustment strategy is implemented as follows: real-time monitoring of the distribution network operation status and identification of the most pressing contradiction is the trigger condition for initiating weight adjustment. During one operation, the system detects that the voltage amplitude of eight nodes is below the lower limit of 0.95 per unit value, with the lowest voltage at node 33 being 0.928 per unit value. Simultaneously, the total active power loss rate is 2.5%, and the power supply reliability rate remains at 99.91%. At this point, the system identifies excessive voltage deviation as the most pressing contradiction. In specific implementation, the adaptive weight adjustment strategy responds immediately, increasing the weight coefficient of the voltage deviation objective function in the multi-objective optimization. The original weights of the network loss objective function, voltage deviation objective function, and power supply reliability objective function are set to 0.4, 0.3, and 0.3, respectively. After adjustment, the weight of the voltage deviation objective function increases to 0.6, while the weights of the network loss objective function and the power supply reliability objective function decrease to 0.25 and 0.15, respectively. This makes the optimization algorithm more inclined to generate resource allocation schemes that can quickly improve voltage levels. In some embodiments, a weight smoothing adjustment mechanism is designed to avoid drastic fluctuations in resource allocation schemes caused by sudden weight changes. This mechanism stipulates that the maximum change in the weight of any objective function per unit time shall not exceed 0.2. The system uses a sliding window of length 5 to record the historical weight sequence and calculates the average change trend of the last five weight adjustments. If the currently calculated weight adjustment requirement is opposite to the historical trend, the adjustment magnitude is reduced.

[0093] In practical implementation, a nonlinear mapping relationship is established between the weights and the deviations of the operating indicators. For the voltage deviation indicator, a threshold is set when the absolute value of the maximum voltage deviation from the limit is reached. When the value is less than 0.01 per unit, the weight adjustment factor A gradual adjustment mode is adopted, resulting in a low rate of change; when When the per-unit value is greater than or equal to 0.02, the weighting adjustment factor is... It employs a fast response mode with a high rate of change. This relationship is described by a piecewise function:

[0094]

[0095] in: It is the weighting adjustment coefficient of the voltage deviation objective function. This represents the maximum absolute value of the deviation of all node voltages from the rated value (1.0 per unit) or the safety limit (0.95 / 1.05 per unit) under the current operating conditions. It is understandable that the coupling relationship between the objective functions must be considered during weight adjustment. For example, to avoid excessive switching of capacitors to increase voltage, which could lead to a surge in network losses, the system sets a protective lower limit weight of 0.2 for the network loss objective function within the same calculation cycle when increasing the weight of the voltage deviation objective function, preventing its weight from dropping to zero. See Table 1 for a simplified dynamic weight adjustment example.

[0096] Table 1: Dynamic Adjustment Table of Weights for Multi-Objective Functions

[0097] Running status identification Network loss target weight Voltage deviation target weight Power supply reliability target weight Adjustment Basis Explanation Initial equilibrium state 0.40 0.30 0.30 There are no major contradictions among the objectives. Voltage deviation exceeds standard 0.25 0.60 0.15 The voltage at 8 nodes is below 0.95 pu. Network loss increased significantly 0.50 0.35 0.15 Total active power loss rose to 3.2%. Restored to a stable state 0.40 0.35 0.25 All indicators have returned to normal range.

[0098] In practical implementation, when a sudden change in operating status is detected, an emergency weight adjustment mode is activated, temporarily relaxing the adjustment speed limit. For example, when a large-scale distributed photovoltaic grid disconnection causes a voltage surge, the maximum weight change limit per unit time is temporarily relaxed from 0.2 to 0.4 to allow for rapid weight response. A stable observation period is set after weight adjustment, for example, keeping the new weights unchanged for three consecutive optimization cycles after the adjustment to observe and evaluate the actual performance of the optimization scheme under this weight configuration. A weight backtracking mechanism is established. After the stable observation period ends, if the evaluation finds that voltage quality has not improved while grid loss has worsened beyond expectations, the system can trigger the backtracking mechanism to quickly restore the weights of the three objectives to the settings before the adjustment. Optionally, the weight adjustment effect is evaluated periodically, and the parameter settings of the weight adjustment strategy are optimized based on the evaluation results. Every 24 hours, the system performs correlation analysis on the weight adjustment records of the past day and the corresponding changes in operating indicators. If the analysis finds that the weights are too sensitive to voltage deviations, causing frequent oscillations, the coefficient of the fast response mode in the nonlinear mapping function is automatically reduced.

[0099] See Figure 4 This is a stacked bar chart showing the dynamic adjustment of weights in multi-objective optimization of a distribution network. It clearly illustrates the changes in weight allocation for the three objective functions—network loss, voltage deviation, and power supply reliability—under four different operating states, perfectly matching the scenario of your previously provided patent implementation. This chart intuitively reflects the core logic of the adaptive weight adjustment strategy; the system dynamically allocates optimization priorities based on the most prominent operational contradiction. The total weights across all states are always 1, meeting the professional requirement of weight normalization in multi-objective optimization. It clearly demonstrates how the weights dynamically change with the operating state of the distribution network, allowing technicians to directly verify whether the strategy's response logic matches the design expectations, providing a visual basis for algorithm iteration. The weight allocation for each state corresponds to specific operational indicator anomalies, enabling subsequent tracing of the correlation between weight adjustments and operational effects, supporting continuous improvement of closed-loop optimization control.

[0100] In one embodiment of the present invention, adjusting the parameter settings of the adaptive weight adjustment strategy based on the execution deviation feedback is a key step in forming closed-loop optimization control. Real-time comparison of the actual execution effect of the resource allocation scheme with the expected target is the starting point for deviation calculation. In one optimization cycle, the resource allocation scheme generated by the adaptive weight adjustment strategy is expected to increase the minimum node voltage to 0.98 per unit and reduce the network loss rate to 2.1%. Five minutes after the scheme is issued and executed, the system collects actual operating data and calculates that the actual minimum node voltage is 0.965 per unit, the actual network loss rate is 2.4%, and the voltage deviation target execution deviation is... The per-unit value represents the deviation from the network loss target. In some embodiments, analyzing the causes of deviations and distinguishing between unreasonable weight settings and changes in the external environment forms the basis for subsequent adjustments. For example, a system check revealed that while the voltage increase was not as expected, the actual discharge power of a distributed energy storage system was 15% lower than the commanded value. Further investigation revealed that the actual usable capacity of this distributed energy storage system was lower than the nominal value in the parameter model. This situation was determined to be a deviation caused by changes in the external environment, such as the equipment status, rather than an underestimation of the weight of the voltage deviation target.

[0101] In practical implementation, establishing a correlation model between deviation and weight adjustment parameters and adjusting the sensitivity parameters of the weight adjustment strategy according to deviation characteristics are the core operations. The adaptive weight adjustment strategy includes a sensitivity parameter that controls the response speed, such as a coefficient in the voltage deviation target weight adjustment formula. The system defines the adjustment amount of the weight adjustment sensitivity parameter. Deviation from Execution and deviation reason identification The correlation model can be represented as follows:

[0102]

[0103] in: This represents the adjustment amount of the weight adjustment sensitivity parameter. It is a basic adjustment coefficient. It is the absolute value of the target execution deviation after normalization. It is the deviation cause identifier. It is an influence factor function based on cause determination. This can be understood as, when the cause of the deviation... When it is determined to be a "change in the external environment", A value of 0.5 indicates that the sensitivity parameter will not be significantly adjusted for the time being; when the deviation is due to... When judged as having "unreasonable weight settings", A value of 1.5 indicates a need to significantly enhance the sensitivity of weight adjustments to correct biases more quickly. The system uses this calculation... The value is used to update the sensitivity parameters.

[0104] Regularly evaluating the weight adjustment strategy and optimizing its core parameters based on historical data is a mechanism to ensure the strategy's continued effectiveness. The system sets an evaluation cycle of 24 hours. In practice, at the end of each evaluation cycle, the system retrieves all execution deviation data, weight adjustment records, and corresponding external environment logs from the past 24 hours for correlation analysis. In some embodiments, the analysis found that during periods when distributed power output fluctuations exceed 30%, deviations caused by "external environment changes" account for as much as 80%, while during periods of stable output, deviations caused by "unreasonable weight settings" account for an even higher proportion. Based on this historical data pattern, the system optimized the threshold condition for "external environment changes" in the deviation cause determination logic. Optionally, the system also optimized the basic adjustment coefficient in the correlation model. This allows the system to adaptively fine-tune based on historical average deviation levels over different time periods. In practice, updating the optimized parameters into the adaptive weight adjustment strategy completes one full cycle of closed-loop optimization. After evaluation and optimization, the new deviation cause determination logic and updated sensitivity parameters are written into the configuration library of the adaptive weight adjustment strategy. The strategy will apply these new parameters at the start of the next optimization cycle, thus achieving continuous iteration and improvement of the strategy itself based on actual performance.

[0105] See Figure 5 This is a line graph showing the relationship between multi-objective optimization deviation and adjustment effect in a distribution network. It illustrates the changes in three types of operational deviations—voltage, network loss, and reliability—over multiple feedback cycles, as well as the optimization effect corresponding to the weight adjustment strategy. The orange curve, representing the "adjustment effect," shows a significant positive correlation with fluctuations in network loss deviation. As network loss deviation increases, the adjustment effect also improves, indicating that the adaptive weight strategy can quickly respond and enhance optimization when network loss is abnormal. During periods of low network loss deviation, the adjustment effect also declines, reflecting the dynamic adaptability of the strategy—optimization efforts are only increased when the core contradiction is prominent. The graph clearly demonstrates the dynamic response capability of the adaptive weight adjustment strategy, proving its ability to accurately optimize major deviations. Large fluctuations in network loss deviation can serve as early warning signals for abnormal distribution network operation, providing maintenance personnel with intuitive monitoring data. By analyzing the correlation between deviation and adjustment effect, the sensitivity of weight adjustment can be further optimized. For example, for rapid fluctuations in network loss deviation, the lower limit of the weight for the network loss target can be appropriately increased to avoid over-optimization.

[0106] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications 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 protection scope of the present invention.

Claims

1. A method for multi-objective collaborative allocation of regional distribution network resources, characterized in that, include: Real-time operational data is collected by monitoring terminals deployed at various nodes of the distribution network. The real-time operational data includes line load, node voltage, distributed power output, and user electricity demand. Establish a multi-timescale data coordination mechanism to classify and aggregate the collected real-time operational data according to minute-level, hour-level, and day-level time granularity to form a hierarchical data set; A resource allocation decision space is constructed, which is defined by the types of schedulable resources in the distribution network and their operating parameter ranges. The schedulable resources include capacitor banks, on-load tap changers, and distributed energy storage systems. A multi-objective optimization problem is generated based on a hierarchical dataset. The objective function of the multi-objective optimization problem simultaneously considers minimizing network loss, minimizing voltage deviation, and maximizing power supply reliability. An adaptive weight adjustment strategy is adopted to solve the multi-objective optimization problem, and the weight coefficients of each objective function are dynamically adjusted according to the current operating status of the distribution network. The optimal resource allocation scheme obtained from the solution is distributed to each resource control terminal for execution, and the deviation between the execution effect and the expected target is monitored in real time. The parameters of the adaptive weight adjustment strategy are adjusted based on the feedback of the execution deviation to form a closed-loop optimization control.

2. The method for multi-objective collaborative allocation of regional distribution network resources according to claim 1, characterized in that, The establishment of a multi-timescale data coordination mechanism includes: Configure a clock synchronization module for each monitoring terminal to ensure that all real-time operating data has a unified time label; Set data preprocessing rules to perform outlier detection and missing data imputation on raw sampling data at the minute-level granularity; The processed minute-level data is aggregated by hourly average to generate an hourly data set, and the data fluctuation characteristic value for each hour is calculated. Hourly data is processed by extracting key indicators on a daily basis to generate a daily data set. The key indicators include daily maximum load, daily minimum voltage, and daily average power supply reliability. Establish a mapping relationship between the three granular data sets so that the upper-level data can be traced back to the detailed operational data of the lower level.

3. The method for multi-objective collaborative allocation of regional distribution network resources according to claim 2, characterized in that, The step of aggregating the processed minute-level data by hourly average includes: The real-time operating data collected every minute is stored in hourly segments, with the hour as the dividing line; Calculate the arithmetic mean of each monitored parameter for each hour, and use it as the representative value for the corresponding hour; Simultaneously, the standard deviation of each monitored parameter is calculated for each hour, serving as a quantitative indicator of the corresponding hourly data volatility. The mean and standard deviation are stored together in an hourly dataset, and each data item is labeled with its corresponding collection time period; When the missing data rate of a certain hour exceeds the preset threshold, the adjacent hour data compensation mechanism is activated to supplement the data using the weighted average of the preceding and following hours.

4. The method for multi-objective collaborative allocation of regional distribution network resources according to claim 1, characterized in that, The constructed resource allocation decision space includes: Identify all resource types in the distribution network that can participate in scheduling, and establish a parameterized model for each resource; For capacitor bank resources, establish a parametric model of its capacity, current switching status, and response speed; For the tap changer of an on-load tap changer, establish a parametric model for its turns ratio range, adjustment step size, and adjustment delay time; For distributed energy storage systems, establish parametric models for their rated capacity, current state of charge, maximum charge / discharge power, and cycle efficiency; The parameter models of all resources are integrated into a unified resource allocation decision space, and the constraints for coordinated operation among the resources are defined.

5. A method for multi-objective collaborative allocation of regional distribution network resources according to claim 4, characterized in that, The defined constraints for coordinated operation among various resources include: Analyze the mutual influence relationships of different types of resources during the adjustment process and establish a resource coupling constraint matrix; Set resource adjustment priority rules to ensure that important resources have priority scheduling rights when adjusting conflicts; Considering the impact of distribution network topology on resource coordination, establish a resource coordination feasibility verification rule based on power flow. Define resource adjustment frequency limits to prevent frequent adjustments from adversely affecting equipment lifespan; Establish global coordination constraints to ensure that the coordinated allocation of all resources meets the requirements for safe operation of the power distribution network.

6. The method for multi-objective collaborative allocation of regional distribution network resources according to claim 1, characterized in that, The multi-objective optimization problem generated based on hierarchical datasets includes: Extract the current operating status characteristics of the distribution network from minute-level data sets, including real-time load and voltage distribution at each node; Extract load change trends and voltage fluctuation patterns from hourly datasets to serve as time-varying constraints for the optimization problem; Historical performance indicators are extracted from daily datasets and used as benchmark reference values ​​for optimization targets. Construct a multi-objective optimization model that includes network loss objective function, voltage deviation objective function, and power supply reliability objective function; By using adjustable parameters in the resource allocation decision space as optimization variables, a mathematical relationship is established between the objective function and the optimization variables.

7. A method for multi-objective collaborative allocation of regional distribution network resources according to claim 6, characterized in that, The construction of the multi-objective optimization model, which includes network loss objective function, voltage deviation objective function, and power supply reliability objective function, includes: The network loss objective function is based on the power flow calculation model of the distribution network, which expresses the total active power loss as a function of the power injected into each node; The voltage deviation objective function calculates the sum of squares of the voltage deviations from the rated voltage at all nodes, and assigns differentiated weights to different nodes based on their importance. The power supply reliability objective function is based on historical fault data and equipment reliability parameters, and a correlation model between resource allocation and power supply reliability indicators is established. All three objective functions were normalized to eliminate the influence of dimensional differences on the optimization results; The optimization model incorporates constraints on the safe operation of the distribution network, including line capacity constraints, voltage upper and lower limit constraints, and power output constraints.

8. A method for multi-objective collaborative allocation of regional distribution network resources according to claim 1, characterized in that, The method of using an adaptive weight adjustment strategy to solve the multi-objective optimization problem includes: Real-time monitoring of the distribution network operation status, identification of the most critical issues, and increase the weight of the voltage deviation target if the voltage deviation exceeds the standard. Design a weight smoothing adjustment mechanism to avoid drastic fluctuations in resource allocation schemes caused by sudden weight changes; Establish a non-linear mapping relationship between weights and deviations in operating indicators; use gradual adjustments for small deviations and rapid responses for large deviations. During the weight adjustment process, the coupling relationship between the objective functions should be considered to avoid over-optimization of a single objective at the expense of other objectives; Regularly evaluate the effectiveness of weight adjustments and optimize the parameter settings of the weight adjustment strategy based on the evaluation results.

9. A method for multi-objective collaborative allocation of regional distribution network resources according to claim 8, characterized in that, The design weight smoothing adjustment mechanism includes: Set a limit on the weight adjustment speed and specify the maximum change in the weights of each objective function per unit time. A sliding window is used to record the historical weight sequence, calculate the weight change trend, and use it as a reference for current adjustments; When a sudden change in the operating status is detected, the emergency weight adjustment mode is activated, and the adjustment speed limit is temporarily relaxed. After the weights are adjusted, a stable observation period is set, during which the weights are kept unchanged to observe the effect of the adjustment. Establish a weight rollback mechanism so that when the effect of the adjustment is not good, the weight settings before the adjustment can be quickly restored.

10. A method for multi-objective collaborative allocation of regional distribution network resources according to claim 1, characterized in that, The parameter settings for adjusting the adaptive weight adjustment strategy based on execution deviation feedback include: The actual performance of the resource allocation plan is compared with the expected goals in real time, and the deviation of each goal is calculated. Analyze the causes of the deviation and distinguish whether it is due to unreasonable weight settings or changes in the external environment; Establish a correlation model between deviation and weight adjustment parameters, and adjust the sensitivity parameters of the weight adjustment strategy according to the deviation characteristics; Regularly conduct a comprehensive evaluation of the weight adjustment strategy and optimize the core parameters of the strategy based on historical data; The optimized parameters are then updated into the adaptive weight adjustment strategy, completing one full cycle of closed-loop optimization.

Images