A method and system for optimizing distributed resources in multiple distribution areas of a distribution network

Abstract

The present invention provides a method and system for optimizing the distributed resources of multiple distribution areas in a distribution network. The method includes the following steps: collecting historical load data, weather data, and network topology parameters of each distribution area in the distribution network for partitioning; constructing a multi-objective resource allocation model based on the partitioning results and network topology parameters; dynamically adjusting the output plan based on real-time load data and network status through a reinforcement learning algorithm, and feeding back to the distribution management system. By constructing a multi-objective resource allocation model with minimizing operating costs and voltage stability as the joint optimization objectives, the dual objectives of minimizing operating costs and enhancing voltage stability are taken into account, breaking through the limitations of traditional single-objective optimization; combined with the real-time feedback mechanism of the reinforcement learning algorithm, it can dynamically respond to load fluctuations and changes in network status, realize the coordinated optimization operation of multiple distribution areas in the distribution network, and improve the economic efficiency, safety and stability of the distribution network.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution networks, and in particular to a method and system for optimizing distributed resources in multiple distribution areas of a distribution network. Background Art

[0002] Traditional distribution network resource optimization methods often use static substation divisions and economic optimization as their target optimization strategies, but these strategies have significant drawbacks. First, substation clustering based on fixed-period load curves ignores the impact of weather factors on load dynamics, resulting in poor compatibility between the resulting zoning and actual operating scenarios. Second, using economic efficiency as the optimization objective makes it difficult to account for grid safety constraints such as voltage stability, and lacks the ability to adapt to real-time operating conditions. Furthermore, centralized optimization algorithms are prone to bottlenecks such as low computational efficiency and slow convergence when faced with large-scale, multi-substation coordinated scheduling, making them unable to meet dynamic control requirements, leading to regulation lags and energy efficiency losses. Summary of the Invention

[0003] In view of the shortcomings of the existing technology, the purpose of the present invention is to provide a method and system for optimizing the distributed resources of multiple distribution areas in a distribution network, aiming to solve the technical problem in the existing technology that the distribution network resource optimization cannot meet the dynamic regulation requirements, resulting in regulation lag and energy efficiency loss.

[0004] In order to achieve the above objectives, in a first aspect, the present invention provides: a method for optimizing distributed resources in multiple distribution areas of a distribution network, comprising the following steps:

[0005] Collect historical load data, weather data and network topology parameters of each distribution network area;

[0006] Calculating the similarity of load curves between substations based on the historical load data based on a dynamic time warping algorithm, and performing time series clustering in combination with a temperature difference factor to generate a partitioning result including dynamic substation similarity;

[0007] Based on the partitioning results and the network topology parameters, a multi-objective resource allocation model is constructed with the joint optimization goals of minimizing operating costs and voltage stability. A genetic algorithm is used to solve the output plan of distributed power generation and energy storage in each partition.

[0008] Based on real-time load data and network status, and on the basis of the output plan, the output of distributed resources is dynamically adjusted through a reinforcement learning algorithm and fed back to the distribution management system.

[0009] According to one aspect of the above technical solution, the calculation expression of the multi-objective resource allocation model is:

[0010] ;

[0011] ;

[0012] ;

[0013] Where, is the unit power generation cost of the distributed generation in the kth partition, K is the total number of partitions, is the unit power charging and discharging cost of energy storage in the kth partition, is the charging and discharging power of the energy storage in the kth partition, is the output power value of the distributed generation in the kth partition, is the unit network loss cost, is the total active power loss of distribution network lines and transformers, is the node voltage of the nth node, is the reference voltage, N is the number of nodes, is the voltage stability weight coefficient, is the output power value of the distributed power source, and are the upper and lower power limits of the distributed power supply, is the total output power of all distributed power sources, is the charging and discharging power of energy storage, is the total load demand power of the distribution network, is the total power loss of the distribution network.

[0014] According to one aspect of the above technical solution, the steps of using a genetic algorithm to solve the output plan of distributed power sources and energy storage in each zone specifically include:

[0015] Step S1: Based on the Pareto solution set output by the multi-objective resource allocation model, the optimal solutions of economy and voltage stability are mixed in a preset proportion to generate the core individuals of the initial population, and neighborhood disturbances are added;

[0016] Step S2, calculating the weighted sum of the economy and voltage stability of each group of solutions as a score to sort the groups;

[0017] Step S3: retain the top 20% of the solutions, eliminate the bottom 20% of the solutions, and select the middle 60% of the solutions as parents through roulette. The output plans of the two parents for the preset time periods are interchanged to generate offspring.

[0018] Step S4: Randomly modify the energy storage charging and discharging periods in some offspring or adjust the output value of the distributed power supply to perform random mutations, merge all offspring including the mutated offspring with the parent generation to form an extended population, and sort the extended population based on the scores, selecting several individuals with the highest scores as the next generation;

[0019] Step S5: loop through steps S2 to S4, optimizing the solution generation by generation until the termination condition is met, and obtaining the output plan of the distributed power supply and energy storage in each partition. After each round of iteration, the cost value and voltage stability of the optimal solution are fed back to the model to update the voltage stability weight coefficient for the next cycle.

[0020] According to one aspect of the above technical solution, the rule of random variation in step S4 is: adjusting the charging and discharging power of the energy storage within a first preset range, and adjusting the output power value of the distributed power source within a second preset range.

[0021] According to one aspect of the above technical solution, the rules for interchange in step S3 include: parents in the same partition exchange output plans during peak load periods, and only gene segments of devices of the same type are exchanged across partitions.

[0022] According to one aspect of the above technical solution, the calculation expression of the dynamic area similarity is as follows:

[0023] ;

[0024] ;

[0025] Where D is the dynamic area similarity, T is the total time period, and are the load powers of area i and area j in the current period t, is the difference weight coefficient of time period t, is the temperature difference weight, is the temperature difference factor, is the temperature difference, For humidity difference, and It is the temperature and humidity adaptation parameter based on regression fitting of historical load data.

[0026] According to one aspect of the above technical solution, the calculation expression of the difference weight coefficient is:

[0027] ;

[0028] Where, is the peak period of the distribution network load curve, is the Gaussian attenuation coefficient.

[0029] According to one aspect of the above technical solution, the calculation expression of the reward function in the reinforcement learning algorithm is:

[0030] ;

[0031] Where, is the total active power loss change of distribution network lines and transformers, is the voltage deviation change, is the voltage weight coefficient.

[0032] In another aspect, the present invention provides a distribution network multi-zone distributed resource optimization system, comprising:

[0033] The acquisition module is used to collect historical load data, weather data and network topology parameters of each distribution network area;

[0034] A partitioning module is used to calculate the similarity of load curves between substations based on the historical load data based on a dynamic time warping algorithm, and perform time series clustering in combination with a temperature difference factor to generate a partitioning result including dynamic substation similarity;

[0035] The output planning module is used to construct a multi-objective resource allocation model with the joint optimization goals of minimizing operating costs and voltage stability based on the partitioning results and the network topology parameters, and use a genetic algorithm to solve the output plan of distributed power sources and energy storage in each partition;

[0036] The feedback module is used to dynamically adjust the output of distributed resources based on real-time load data and network status and the output plan through a reinforcement learning algorithm, and feed back to the distribution management system.

[0037] According to one aspect of the above technical solution, the output planning module is specifically used to:

[0038] Step S1: Based on the Pareto solution set output by the multi-objective resource allocation model, the optimal solutions of economy and voltage stability are mixed in a preset proportion to generate the core individuals of the initial population, and neighborhood disturbances are added;

[0039] Step S2, calculating the weighted sum of the economy and voltage stability of each group of solutions as a score to sort the groups;

[0040] Step S3: retain the top 20% of the solutions, eliminate the bottom 20% of the solutions, and select the middle 60% of the solutions as parents through roulette. The output plans of the two parents for the preset time periods are interchanged to generate offspring.

[0041] Step S4: Randomly modify the energy storage charging and discharging periods in some offspring or adjust the output value of the distributed power supply to perform random mutations, merge all offspring including the mutated offspring with the parent generation to form an extended population, and sort the extended population based on the scores, selecting several individuals with the highest scores as the next generation;

[0042] Step S5: loop through steps S2 to S4, optimizing the solution generation by generation until the termination condition is met, and obtaining the output plan of the distributed power supply and energy storage in each partition. After each round of iteration, the cost value and voltage stability of the optimal solution are fed back to the model to update the voltage stability weight coefficient for the next cycle.

[0043] Compared with the existing technology, the beneficial effects of the present invention are: dynamic clustering of substations is achieved by integrating temperature difference factors through a dynamic time warping algorithm, and by constructing a multi-objective resource allocation model with minimizing operating costs and voltage stability as the joint optimization goals, the dual goals of minimizing operating costs and enhancing voltage stability are taken into account, breaking through the limitations of traditional single-objective optimization; combined with the real-time feedback mechanism of the reinforcement learning algorithm, it can dynamically respond to load fluctuations and network status changes, significantly improve the robustness and adaptability of distributed resource regulation, realize the coordinated optimization operation of multiple substations in the distribution network, and improve the economy, safety and stability of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 Schematic diagram of the flow of the method for optimizing the distribution resources of multiple distribution areas in a distribution network according to the first embodiment of the present invention;

[0045] Figure 2 This is a structural block diagram of a distribution network multi-zone distributed resource optimization system according to a second embodiment of the present invention;

[0046] The following specific embodiments will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0047] To facilitate understanding of the present invention, the present invention will be described more fully below with reference to the accompanying drawings. The drawings illustrate several embodiments of the present invention. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the present invention.

[0048] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be an intermediate element. When an element is referred to as being "connected to" another element, it may be directly connected to the other element or there may be an intermediate element. The terms "vertical," "horizontal," "left," "right," and similar expressions used herein are for illustrative purposes only.

[0049] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one skilled in the art to which this invention pertains. The terms used in this specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0050] Example 1

[0051] See also Figure 1 , which is a flow chart of a method for optimizing the distribution resources of multiple distribution areas in a distribution network according to a first embodiment of the present invention. As shown in the figure, the method includes the following steps:

[0052] Step S100: collecting historical load data, weather data and network topology parameters of each substation in the distribution network.

[0053] In some application scenarios of this embodiment, historical load data includes active power curves (15-minute granularity), reactive power curves, and daily / weekly load peak and valley characteristics for each substation, collected in real time via smart meters, SCADA systems, and PMUs (synchronized phase measurement units). Weather data, including temperature and humidity, facilitates subsequent correction of similarity calculations using temperature and humidity difference factors, avoiding static partitioning biases that rely solely on historical loads. This, for example, occurs when air conditioning loads are strongly correlated with high temperatures in summer. This improves the adaptability of partitioning results to actual operating conditions and provides precise input for multi-objective optimization. Network topology parameters include the line impedance matrix (R / X value), transformer capacity and tap positions, node connectivity, distributed power generation parameters (such as maximum PV inverter output and energy storage charge and discharge efficiency curves), load node distribution, and capacity constraints. The line impedance matrix is used to calculate network losses and construct an economic objective function, while node connectivity and transformer capacity constrain voltage stability models.

[0054] In step S200 , the similarity of load curves between substations is calculated based on the historical load data using a dynamic time warping algorithm, and time series clustering is performed in combination with a temperature difference factor to generate a partitioning result including dynamic substation similarity.

[0055] Preferably, in this embodiment, the calculation expression of the dynamic area similarity is as follows:

[0056] ;

[0057] ;

[0058] Where D is the dynamic area similarity, T is the total time period, and are the load powers of area i and area j in the current period t, is the difference weight coefficient of time period t, is the temperature difference weight, is the temperature difference factor, is the temperature difference, For humidity difference, and is a temperature and humidity adaptation parameter derived from a regression fit of historical load data. It should be noted that in this embodiment, the temperature and humidity parameters and load power are normalized before calculation to eliminate dimensional differences. By calculating the similarity of the load curves between substations i and j, distortion caused by time offsets (e.g., differences in morning and evening peak times) is addressed.

[0059] Furthermore, the calculation expression of the above difference weight coefficient is:

[0060] ;

[0061] Where, is the peak period of the distribution network load curve, is the Gaussian attenuation coefficient. Specifically, by setting the difference weight coefficient based on the peak period of the load curve, the weight reaches its maximum value near this period, strengthening the impact of load differences during peak periods on the similarity calculation. This can avoid noise interference from non-critical periods, and the results are more consistent with actual power consumption characteristics. A larger Gaussian attenuation coefficient results in a flatter weight distribution (covering a wider range of time periods), weakening differences during non-peak periods. A smaller Gaussian attenuation coefficient results in more concentrated weights near the peak period, focusing on differences during critical periods.

[0062] Step S300: Based on the partitioning results and the network topology parameters, a multi-objective resource allocation model is constructed with minimizing operating costs and voltage stability as the joint optimization goals, and a genetic algorithm is used to solve the output plan of distributed power sources and energy storage in each partition.

[0063] Specifically, in this embodiment, the calculation expression of the multi-objective resource allocation model is:

[0064] ;

[0065] ;

[0066] ;

[0067] Where, is the unit power generation cost of the distributed generation in the kth partition, K is the total number of partitions, is the unit power charging and discharging cost of energy storage in the kth partition, is the total active power loss of distribution network lines and transformers, is the charging and discharging power of the energy storage in the kth partition, is the output power value of the distributed generation in the kth partition, is the unit network loss cost, is the node voltage of the nth node, is the reference voltage, N is the number of nodes, is the voltage stability weight coefficient, is the output power value of the distributed power source, and are the upper and lower power limits of the distributed power supply, is the total output power of all distributed power sources, is the charging and discharging power of energy storage, is the total load demand power of the distribution network, is the total power loss of the distribution network.

[0068] It should be noted that For the economic part of the multi-objective resource allocation model, This is the voltage stability component of the multi-objective resource allocation model. The economic component aims to minimize the total operating cost of the distribution network. Distributed power sources include low-cost sources like photovoltaics and wind power, as well as high-cost sources like gas turbines. The unit power charging and discharging cost of energy storage reflects battery aging losses. The voltage stability component ensures grid security by minimizing node voltage deviations. The voltage stability weight coefficient is used to balance economic efficiency with voltage stability. High values strengthen voltage stability constraints, potentially limiting the output of distributed power sources (such as PV derating) and increasing operating costs. Low values prioritize economic efficiency and may lead to voltage over-limit risks.

[0069] Furthermore, in this embodiment, the steps of using a genetic algorithm to solve the output plan of the distributed power source and energy storage in each zone specifically include:

[0070] In step S1, based on the Pareto solution set output by the multi-objective resource allocation model, the optimal solutions for economy and voltage stability are mixed in a preset ratio to generate the initial population core individuals, and a neighborhood perturbation is added. Specifically, the preset ratio is 7:3, and the neighborhood perturbation is ±10% of the output power value of the distributed generation and ±20% of the charge and discharge power of the energy storage.

[0071] Step S2: Calculate the weighted sum of the economy and voltage stability of each solution group as a score to sort the groups. The scoring formula is:

[0072] ;

[0073] Where S is the score, h1 is the economic part of the above multi-objective resource allocation model, is the voltage stability part of the above multi-objective resource allocation model, where: is the dynamic voltage stability weight coefficient, and its initial value is the voltage stability weight coefficient output by the multi-objective resource allocation model and updated via S5.

[0074] In step S3, the top 20% of the scoring solutions are retained, the bottom 20% are eliminated, and the middle 60% of the scoring solutions are selected through a roulette wheel to serve as the parent generation. The output plans of the two parent generations for the preset time periods are then swapped and crossed to generate the child generation. Specifically, the rules for the swapping and crossing in step S3 include: parents within the same partition exchange output plans during peak load periods, and only genetic segments of devices of the same type are swapped across partitions. Because device capacities can vary significantly across partitions, only genetic segments of devices of the same type and similar capacity are swapped to avoid exceeding device capabilities in charge and discharge power. By swapping efficient output modes for similar devices, multi-partition coordinated scheduling can be explored (for example, discharging energy storage from multiple partitions during the same time period to support main grid voltage).

[0075] The above preset time periods include the three consecutive hours with the highest load power (such as 18:00-21:00), the three consecutive hours with the lowest load power (such as 02:00-05:00), the period when the voltage deviation exceeds 5% in historical data (such as the voltage rise caused by the sudden increase in photovoltaic output from 12:00 to 14:00), and the period when the energy storage SOC is close to the upper and lower limits (such as the discharge period when SOC=20%).

[0076] In step S4, the energy storage charge and discharge periods in some offspring are randomly modified or the output value of the distributed power source is adjusted to perform random mutation. All offspring, including the mutated offspring, are merged with the parent generation to form an extended population. The extended population is sorted based on the scores, and several individuals with the highest scores are selected as the next generation. The rule for random mutation in step S4 is as follows: the charge and discharge power of the energy storage is adjusted within a first preset range, and the output power value of the distributed power source is adjusted within a second preset range. The first preset range is preferably ±20% of the current power value, and the second preset range is preferably ±10% of the current output power value. The number of individuals is the same as the number of core individuals in the initial population.

[0077] Step S5 loops through steps S2 to S4, optimizing the solution generation by generation until the termination criteria are met. This results in output plans for distributed generation and energy storage within each zone. After each iteration, the cost and voltage stability of the optimal solution are fed back into the model to update the voltage stability weight for the next iteration. Specifically, the termination criteria are preferably 100 iterations or convergence stability, meaning the optimal solution score improves by less than 1%.

[0078] Step S400 , based on the output plan and real-time load data and network status, dynamically adjusts the output of distributed resources using a reinforcement learning algorithm, and provides feedback to the distribution management system. Preferably, in this embodiment, the reinforcement learning algorithm employs a Q-Learning algorithm. The main steps include: capturing real-time grid operating status to define a state space, which includes load factor, voltage status, and the status of distributed generation (DG) and energy storage; defining actions as real-time output adjustments for distributed generation (DG) and energy storage (ESS); balancing economic efficiency and voltage stability through a reward function, and using an existing classic Q-value update rule.

[0079] Among them, the calculation expression of the reward function is:

[0080] ;

[0081] Where, is the total active power loss change of distribution network lines and transformers, is the voltage deviation change, is the voltage weight coefficient.

[0082] In summary, the method for optimizing distributed resources in multiple distribution areas of a distribution network in the above-mentioned embodiment of the present invention realizes dynamic clustering of distribution areas by integrating the temperature difference factor through a dynamic time warping algorithm. By constructing a multi-objective resource allocation model with the joint optimization objectives of minimizing operating costs and voltage stability, it takes into account the dual objectives of minimizing operating costs and enhancing voltage stability, breaking through the limitations of traditional single-objective optimization. Combined with the real-time feedback mechanism of the reinforcement learning algorithm, it can dynamically respond to load fluctuations and changes in network status, significantly improve the robustness and adaptability of distributed resource regulation, realize the coordinated optimization operation of multiple distribution areas in the distribution network, and improve the economy, safety and stability of the distribution network.

[0083] Example 2

[0084] The second embodiment of the present application also provides a distribution network multi-station distributed resource optimization system, which is used to implement the embodiments and preferred embodiments, and will not be repeated here. As used below, the terms "module", "unit", "sub-unit", etc. can implement a combination of software and / or hardware for a predetermined function. Although the systems described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and contemplated.

[0085] like Figure 2 As shown, the system includes: a collection module 100 , a partitioning module 200 , an output planning module 300 and a feedback module 400 .

[0086] The collection module 100 is used to collect historical load data, weather data and network topology parameters of each substation in the distribution network;

[0087] The partitioning module 200 is used to calculate the similarity of load curves between substations based on the historical load data based on a dynamic time warping algorithm, and perform time series clustering in combination with a temperature difference factor to generate a partitioning result including dynamic substation similarity;

[0088] The output planning module 300 is used to construct a multi-objective resource allocation model with the joint optimization goals of minimizing operating costs and voltage stability based on the partitioning results and the network topology parameters, and use a genetic algorithm to solve the output plan of distributed power sources and energy storage in each partition;

[0089] The feedback module 400 is used to dynamically adjust the output of distributed resources based on the real-time load data and network status and the output plan through a reinforcement learning algorithm, and feed back to the distribution management system.

[0090] Preferably, in this embodiment, the output planning module 300 is specifically used to:

[0091] Step S1: Based on the Pareto solution set output by the multi-objective resource allocation model, the optimal solutions of economy and voltage stability are mixed in a preset proportion to generate the core individuals of the initial population, and neighborhood disturbances are added;

[0092] Step S2, calculating the weighted sum of the economy and voltage stability of each group of solutions as a score to sort the groups;

[0093] Step S3: retain the top 20% of the solutions, eliminate the bottom 20% of the solutions, and select the middle 60% of the solutions as parents through roulette. The output plans of the two parents for the preset time periods are interchanged to generate offspring.

[0094] Step S4: Randomly modify the energy storage charging and discharging periods in some offspring or adjust the output value of the distributed power supply to perform random mutations, merge all offspring including the mutated offspring with the parent generation to form an extended population, and sort the extended population based on the scores, selecting several individuals with the highest scores as the next generation;

[0095] Step S5: loop through steps S2 to S4, optimizing the solution generation by generation until the termination condition is met, and obtaining the output plan of the distributed power supply and energy storage in each partition. After each round of iteration, the cost value and voltage stability of the optimal solution are fed back to the model to update the voltage stability weight coefficient for the next cycle.

[0096] The technical features of the above-described embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0097] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for optimizing distributed resources in multiple distribution areas of a distribution network, characterized in that: The following steps are involved: Collect historical load data, weather data and network topology parameters of each distribution network area; Calculating the similarity of load curves between substations based on the historical load data based on a dynamic time warping algorithm, and performing time series clustering in combination with a temperature difference factor to generate a partitioning result including dynamic substation similarity; Based on the partitioning results and the network topology parameters, a multi-objective resource allocation model is constructed with the joint optimization goals of minimizing operating costs and voltage stability. A genetic algorithm is used to solve the output plan of distributed power generation and energy storage in each partition. Based on the real-time load data and network status, and on the basis of the output plan, the distributed resource output is dynamically adjusted through a reinforcement learning algorithm and fed back to the distribution management system; The calculation expression of dynamic area similarity is as follows: ； ； Where D is the dynamic area similarity, T is the total time period, and are the load powers of area i and area j in the current period t, is the difference weight coefficient of time period t, is the temperature difference weight, is the temperature difference factor, is the temperature difference, For humidity difference, and It is the temperature and humidity adaptation parameter based on regression fitting of historical load data; The calculation expression of the difference weight coefficient is: ； Where, is the peak period of the distribution network load curve, is the Gaussian attenuation coefficient.

2. The method for optimizing distributed resources in multiple distribution areas of a distribution network according to claim 1, characterized in that: The calculation expression of the multi-objective resource allocation model is: ； ； ； Where, is the unit power generation cost of the distributed generation in the kth partition, K is the total number of partitions, is the unit power charging and discharging cost of energy storage in the kth partition, is the total active power loss of distribution network lines and transformers, is the charging and discharging power of the energy storage in the kth partition, is the output power value of the distributed generation in the kth partition, is the unit network loss cost, is the node voltage of the nth node, is the reference voltage, N is the number of nodes, is the voltage stability weight coefficient, is the output power value of the distributed power source, and are the upper and lower power limits of the distributed power supply, is the total output power of all distributed power sources, is the charging and discharging power of energy storage, is the total load demand power of the distribution network, is the total power loss of the distribution network.

3. The method for optimizing the distribution resources of multiple distribution areas in a power distribution network according to claim 2, characterized in that: The steps of using genetic algorithm to solve the output plan of distributed power generation and energy storage in each zone include: Step S1: Based on the Pareto solution set output by the multi-objective resource allocation model, the optimal solutions of economy and voltage stability are mixed in a preset proportion to generate the core individuals of the initial population, and neighborhood disturbances are added; Step S2, calculating the weighted sum of the economy and voltage stability of each group of solutions as a score to sort the groups; Step S3: retain the top 20% of the solutions, eliminate the bottom 20% of the solutions, and select the middle 60% of the solutions as parents through roulette. The output plans of the two parents for the preset time periods are interchanged to generate offspring. Step S4: Randomly modify the energy storage charging and discharging periods in some offspring or adjust the output value of the distributed power supply to perform random mutations, merge all offspring including the mutated offspring with the parent generation to form an extended population, and sort the extended population based on the scores, selecting several individuals with the highest scores as the next generation; Step S5: loop through steps S2 to S4, optimizing the solution generation by generation until the termination condition is met, and obtaining the output plan of the distributed power supply and energy storage in each partition. After each round of iteration, the cost value and voltage stability of the optimal solution are fed back to the model to update the voltage stability weight coefficient for the next cycle.

4. The method for optimizing the distribution resources of multiple distribution areas in a power distribution network according to claim 3, characterized in that: The rule of random variation in step S4 is: adjusting the charging and discharging power of the energy storage within a first preset range, and adjusting the output power value of the distributed power source within a second preset range.

5. The method for optimizing the distribution resources of multiple distribution areas in a power distribution network according to claim 3, characterized in that: The rules for the exchange and crossover in step S3 include: the parents of the same partition exchange the output plans during the load peak period, and only exchange the gene segments of the same type of devices across partitions.

6. The method for optimizing distributed resources in multiple distribution areas of a power distribution network according to claim 3, characterized in that: The calculation expression of the reward function in the reinforcement learning algorithm is: ； Where, is the total active power loss change of distribution network lines and transformers, is the voltage deviation change, is the voltage weight coefficient.

7. A distribution network multi-zone distributed resource optimization system for implementing the method according to any one of claims 1 to 6, characterized in that: include: The acquisition module is used to collect historical load data, weather data and network topology parameters of each distribution network area; A partitioning module is used to calculate the similarity of load curves between substations based on the historical load data based on a dynamic time warping algorithm, and perform time series clustering in combination with a temperature difference factor to generate a partitioning result including dynamic substation similarity; The output planning module is used to construct a multi-objective resource allocation model with the joint optimization goals of minimizing operating costs and voltage stability based on the partitioning results and the network topology parameters, and use a genetic algorithm to solve the output plan of distributed power sources and energy storage in each partition; The feedback module is used to dynamically adjust the output of distributed resources based on real-time load data and network status and the output plan through a reinforcement learning algorithm, and feed back to the distribution management system.

8. The distribution network multi-zone distributed resource optimization system according to claim 7, characterized in that: The output planning module is specifically used for: Step S1: Based on the Pareto solution set output by the multi-objective resource allocation model, the optimal solutions of economy and voltage stability are mixed in a preset proportion to generate the core individuals of the initial population, and neighborhood disturbances are added; Step S2, calculating the weighted sum of the economy and voltage stability of each group of solutions as a score to sort the groups; Step S3: retain the top 20% of the solutions, eliminate the bottom 20% of the solutions, and select the middle 60% of the solutions as parents through roulette. The output plans of the two parents for the preset time periods are interchanged to generate offspring. Step S4: Randomly modify the energy storage charging and discharging periods in some offspring or adjust the output value of the distributed power supply to perform random mutations, merge all offspring including the mutated offspring with the parent generation to form an extended population, and sort the extended population based on the scores, selecting several individuals with the highest scores as the next generation; Step S5: loop through steps S2 to S4, optimizing the solution generation by generation until the termination condition is met, and obtaining the output plan of the distributed power supply and energy storage in each partition. After each round of iteration, the cost value and voltage stability of the optimal solution are fed back to the model to update the voltage stability weight coefficient for the next cycle.

Images