A power distribution network dynamic carrying capacity improvement method based on light storage coordination optimization

Abstract

A power distribution network dynamic carrying capacity improvement method based on light storage coordination optimization, comprising the following steps: step S1: based on historical photovoltaic output and load data, a photovoltaic-load typical time sequence scene set is constructed, which is hereinafter referred to as "typical time sequence scene set" in the following, and the scene data is subjected to abnormal elimination and standardization processing, providing input for subsequent light storage coordination optimization; step S2: on the basis of the typical time sequence scene set described in step S1, a light storage coordination multi-objective optimization model for dynamic carrying capacity improvement is constructed; step S3: based on the photovoltaic access scheme and energy storage configuration scheme obtained in step S2, a fitness function is constructed with light storage coordination adaptation degree as the core, and under the conditions of meeting the energy storage operation constraints and the safe operation constraints of the power distribution network, the energy storage time sequence charging and discharging strategy is optimized to smooth the photovoltaic output fluctuation and reduce the reverse power flow risk; S4: after time sequence power flow simulation verification, the optimal light storage coordination configuration scheme is output.

Description

Technical Field

[0001] This invention relates to the fields of distribution network operation and new energy consumption technology, specifically to a method for improving the dynamic carrying capacity of distribution networks based on photovoltaic-storage synergistic optimization. It is particularly applicable to high-penetration distributed photovoltaic access scenarios. By optimizing the spatial layout of photovoltaics and coordinating the timing of energy storage, the method can comprehensively improve the carrying capacity and operational economy of distribution networks. It can be widely applied to engineering practices such as distribution network planning and design, new energy grid connection scheduling, and energy storage system optimization configuration. Background Technology

[0002] In recent years, with the rapid growth of distributed photovoltaic (PV) installations, especially the high penetration rate in medium- and low-voltage distribution networks, the operation and planning of distribution networks face new uncertainties and challenges. Distributed PV is intermittent and fluctuating, and its output is significantly affected by weather conditions. Under light loads or localized concentrated access conditions, it can easily cause problems such as node voltage rise, changes in power flow direction, and reduced operating margins, thus limiting the carrying capacity of the distribution network. To improve the absorption capacity of distributed PV and enhance system operational flexibility, energy storage systems, with their peak-shaving, power regulation, and energy time-shifting capabilities, have gradually become one of the important means to improve the dynamic carrying capacity of the distribution network. Therefore, for high-penetration distributed PV access scenarios, how to achieve coordinated optimization of PV access layout, energy storage configuration, and operation strategies based on typical time-series characteristics has attracted much attention in the field of distribution network planning and operation.

[0003] Currently, research on improving carrying capacity has a certain foundation. For example, in terms of scene generation, to address the problem of excessive scene numbers caused by the temporal fluctuations of wind and solar power output and load, the paper "Optimal Configuration of Distributed Power Sources in Active Distribution Network Considering the Temporal Characteristics of Wind, Solar and Load" (document number 10.12204 / j.issn.1000-7229.2022.11.006) proposes to introduce the Davidson-Bolding Index (DBI) and combine it with the K-means algorithm to cluster and reduce the full-hour scenes. Based on this, a two-layer optimal configuration model for active distribution networks considering the temporal characteristics of wind, solar and load is established. Furthermore, in case studies of carrying capacity assessment and improvement, existing works generally adopt the approach of pre-setting access nodes or providing a candidate set of access nodes. For example, the paper "Carrying Capacity Assessment of Distributed Power Generation in Distribution Networks Based on APDE Algorithm" (document number 10.19912 / j.0254-0096.tynxb.2022-1075) uses the IEEE 33-node system as an example for case analysis. The paper specifies distributed photovoltaic grid-connected nodes as nodes 7, 11, 22, and 24, and conducts carrying capacity assessment and comparative analysis based on this setting. Meanwhile, some studies also incorporate energy storage into carrying capacity assessment and capacity configuration to improve safety and economy. Similarly, in the paper "Economic Carry-on Capacity Assessment of Distributed Power Generation in Distribution Networks" (document number 10.11930 / j.issn.1004-9649.202312019), a robust optimization method is used to optimize the configuration of distributed power generation and energy storage capacity in distribution networks, considering both safe operation and economic objectives. Multiple operating scenarios are generated based on historical wind and solar load data and distribution characteristics to construct an uncertainty set, thereby establishing a two-layer robust optimization model. In its IEEE 33-node example, the access nodes for wind power, photovoltaics, and energy storage are given (e.g., wind power accesses nodes 11, 19, and 30; photovoltaics accesses nodes 5, 12, and 24; and energy storage accesses nodes 6, 13, and 27), and the capacity configuration is optimized based on the set nodes.

[0004] However, while the aforementioned research has made some progress in scenario construction and energy storage synergy, it still relies mainly on empirical settings and local optimizations, making it difficult to simultaneously address the global decision-making needs of timing uncertainties and photovoltaic-energy storage synergy. The following technical shortcomings remain: First, existing typical scenario construction methods mainly rely on the statistical similarity of output or load for clustering or dimensionality reduction, which makes it difficult to reflect the differences in the sensitive characteristics of distribution network operation constraints. In particular, when considering feature dimensions that have a decisive impact on carrying capacity, such as reverse power flow risk and ramp risk, traditional "equal-weight distance" clustering is prone to causing key high-risk operating conditions to be mistakenly merged, thereby affecting the ability of subsequent planning and optimization to cover typical constraint boundaries. Secondly, existing photovoltaic-storage collaborative planning methods often treat the candidate locations for photovoltaic and energy storage access as pre-given conditions at the modeling level, and only perform local optimization on access capacity or operation strategy, resulting in a weakening of the coupling relationship between planning variables: on the one hand, the prior constraints on photovoltaic access locations mean that energy storage configuration and operation strategies can only "passively compensate" for the established access pattern, making it difficult to coordinate capacity, location and timing adjustment capabilities at the system level; on the other hand, the prior settings for energy storage locations also limit their targeted support for weak links in the power grid, which can easily lead to results that are "locally feasible but not optimal overall", making it difficult to obtain a globally optimal or near-global optimal photovoltaic-storage collaborative configuration scheme for improving dynamic carrying capacity.

[0005] Therefore, there is an urgent need for a method that can coordinate and optimize photovoltaic access layout, energy storage configuration and energy storage time-series operation strategy based on typical time-series scenario inputs, so as to improve the dynamic carrying capacity of the distribution network and take into account the economic efficiency and engineering feasibility of the solution. Summary of the Invention

[0006] The purpose of this invention is to solve the technical problems in existing methods for improving the dynamic carrying capacity of power distribution networks, such as insufficient consideration of the coordination of photovoltaic access layout, single multi-objective optimization dimension, and insufficient photovoltaic-storage synergy, which lead to limited carrying capacity improvement and imbalance of operational safety.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for improving the dynamic carrying capacity of distribution networks based on photovoltaic-storage synergistic optimization includes the following steps: Step S1: Based on historical photovoltaic power output and load data, construct a typical time series scenario set, and perform anomaly removal and standardization processing on the scenario data to provide input for subsequent photovoltaic-storage synergistic optimization; Step S2: Based on the typical time-series scenario set obtained in Step S1, construct a multi-objective optimization model for photovoltaic-storage synergy aimed at improving dynamic carrying capacity. With the objectives of maximizing the coordination degree of photovoltaic access layout, maximizing dynamic carrying capacity, and minimizing the total cost of photovoltaic-storage synergy scheme, jointly determine the set of photovoltaic access nodes and the access capacity of each node, and the set of energy storage access nodes. Step S3: Based on the photovoltaic access scheme and energy storage configuration scheme obtained in step S2, construct a fitness function with photovoltaic-energy storage synergy adaptability as the core. Under the conditions of meeting the energy storage operation constraints and the distribution network safety operation constraints, optimize the energy storage timing charging and discharging strategy to smooth photovoltaic power output fluctuations and reduce reverse power flow risk. Step S4: After time-series power flow simulation verification, the optimal optical-storage collaborative configuration scheme is output.

[0008] In step S1, based on historical photovoltaic power output and load data, a typical time-series scenario set is generated using improved weighted fuzzy k-means clustering, specifically including the following sub-steps: Step S1.1: Input the basic parameters of the distribution network, historical photovoltaic output and load data, and use... After removing outliers according to the criteria, the data units are standardized by min-max standardization. For photovoltaic power output data, samples with values ​​less than 0 or deviations from the mean exceeding 3 times the standard deviation are removed. For load data, samples with abnormal fluctuations exceeding 30% are removed and standardized. Step 1.2: Based on the standardized photovoltaic power output sequence and load sequence obtained in step S1.1, construct a joint sample set, i.e., a joint feature vector set; Step S1.3: Obtain the joint feature vector set in step S1.2. Based on this, fuzzy k-means is used to cluster the samples and obtain the weighted Euclidean distance; Step S1.4: Based on the distance obtained in step S1.3, the membership degree and cluster center are calculated using the standard iterative update form of fuzzy k-means.

[0009] In step S1.1, the formula used for standardization is: In the formula: This is the original data sample; and These are the minimum and maximum values ​​of the original data sequence, respectively. These are the standardized sample values; the standardized photovoltaic power output sequence is denoted as... The standardized load sequence is denoted as ,in s The historical sample day number, These represent typical daily discrete moments.

[0010] In step S1.2, when constructing the joint sample set, for the first... s A sequence of historical sample periods is represented as a feature vector. ,in For the first s One sample vector; T This represents the number of discrete time periods on a typical day. For the first s A historical sample in the time period The photovoltaic output value; For the first s A historical sample in the time period The photovoltaic output value; For the first s A historical sample in the time period The load value, For the first s A historical sample in the time period The load value; This is a characteristic quantity of reverse current risk. This represents the characteristic quantity of hill climbing risk; To enable scene clustering to prioritize high-risk conditions that are "prone to reverse currents and severe fluctuations," the risk characteristic quantity is calculated from the net injection sequence, and its expression is: ,in These are the serial numbers of typical daily discrete moments; For the sample s At any moment Net injection, For the sample s At any moment The photovoltaic output value, For the sample s At any moment The load value; Reverse current risk characteristic quantity is defined as ,in Used to extract inputs during periods where net injection is positive; The larger the value, the more likely the sample day will experience reverse current flow, and the greater the need for energy storage to absorb and reduce peak flows; The characteristic quantity of hill climb risk is defined as follows: , For the sample s At any moment The net injection, of which, The larger the value, the more drastic the change in net injection, and the more critical the role of subsequent energy storage timing charge and discharge strategies in smoothing out fluctuations.

[0011] In step S1.3, when performing clustering, let the number of clusters be... K , No. k The cluster center is ,sample s Cluster k The membership degree is The distance between a sample and the cluster center is defined as the weighted Euclidean distance. ,in, H The feature dimension; and The sample vector and the cluster center are respectively in the th... h Dimensional components; For the first h Dimensional weight coefficients; To unify the dispersion of different dimensions and highlight the importance of high-risk dimensions, the weighting coefficients are set to... ,in, For the first h The variance of the dimensional feature on the sample set, This represents the average of the variances in each dimension; Let be the risk amplification coefficient, where when the th hDimensional correspondence or Time to take The remaining dimensions are taken as follows: Through the above settings, the clustering has a higher resolution in the dimensions of "reverse current risk and ramp risk", thereby providing more constrained and representative typical scenario inputs for subsequent photovoltaic-storage synergistic optimization.

[0012] In step S1.4, when calculating membership and cluster centers using the standard iterative update form of fuzzy k-means, the specific steps are as follows: Membership degree updated to ,in, These are fuzzy control parameters used to adjust the "fuzziness level" of membership assignment; priority is given to selecting... ; The larger the value, the higher the value. r Samples in the next iteration s With the k The more similar the typical characteristics of the classes, For the first v Cluster center vector of the class; Cluster centers are updated using fuzzy means, expressed as follows: , ,in, For the first r The cluster center of the next iteration; For the first r Cluster center in -1 iteration; For the first s Feature vectors of historical samples; This is the convergence threshold; Finally, based on the converged cluster center vector Extract typical scenario curves, where the typical photovoltaic output curve and the typical load curve are respectively taken from the time-series components corresponding to the cluster center; and give the first... k The probability of occurrence of a typical time-series scene set is ,in, For the first k The probability of occurrence of a set of typical time-series "photovoltaic-load" scenarios satisfies ; This represents the number of historical sample days. The typical time-series scenario set and its probability will serve as input data for subsequent steps S2-S4 to optimize the synergy between photovoltaic and energy storage and improve dynamic carrying capacity, thereby achieving synergistic optimization of energy storage configuration and time-series strategies under various operational risk conditions.

[0013] In step S2, the objective function of the constructed photovoltaic-storage synergistic multi-objective optimization model is: ,in In the formula, For the coordination of photovoltaic grid connection layout; A set of photovoltaic access nodes; For nodes photovoltaic installed capacity, The total cost of the photovoltaic-storage synergy solution, , For the cost of photovoltaic grid connection, For the total life cycle cost of energy storage, This refers to the cost of upgrades related to the access of photovoltaic and energy storage to the distribution network.

[0014] The above-mentioned photovoltaic grid connection layout coordination Defined as ,in , , n For set The number of photovoltaic access nodes in the system The total number of photovoltaic access nodes allowed; For nodes i Electrical distance to the first node of the distribution network; These are the weighting coefficients; For capacity coordination items; For electrical position coordination items.

[0015] In step S3, with the photovoltaic-storage synergy adaptation degree as the core, the required fitness function is designed, and the energy storage timing charge and discharge strategy is optimized to smooth out photovoltaic power output fluctuations. The energy storage operation satisfies the following model constraints: ; ; ; ; In the formula, For nodes time Energy storage capacity, For nodes time Energy storage capacity; , They are nodes time The charging and discharging power of energy storage; For time step; , These represent the energy storage charging and discharging efficiencies, respectively, with values ​​ranging from 0.85 to 0.95. , They are nodes Rated power and rated capacity of energy storage; , The minimum and maximum threshold values ​​for the energy storage state of charge (SOC) are 0.2-0.3 and 0.8-0.9, respectively. Simultaneously meet the constraints of safe operation of the distribution network: ; ; ; In the formula, For nodes time voltage amplitude, , , Rated voltage; For the line time The current, For the line Maximum allowable flow rate For a moment Reverse load rate, Maximum permissible reverse load rate; The calculation function for the photonics-storage synergy fit is: ; In the formula, To ensure the coordination of photovoltaic grid connection layout, To maximize the coordination of photovoltaic grid connection layout; To mitigate fluctuations in photovoltaic power output, , To smooth out the standard deviation of previous photovoltaic output, To smooth out the standard deviation of the combined photovoltaic and energy storage output; Voltage deviation pass rate , The duration for which the voltage meets the constraint. Total duration; To improve the overall charging and discharging efficiency of energy storage; The maximum permissible life-cycle cost of energy storage; , , , , These are the weighting coefficients. .

[0016] In step S4, based on the energy storage time-series charging and discharging strategy and under the conditions of satisfying the energy storage operation constraints and the distribution network safety operation constraints, time-series power flow simulation verification is performed to obtain the node voltage amplitude, line current and reverse load rate at each time point, and to count the voltage over-limit rate, reverse power flow ratio and fluctuation smoothing rate, and output the power flow simulation verification results of the photovoltaic-storage collaborative scheme.

[0017] Compared with the prior art, the present invention has the following technical effects: 1) In the process of constructing typical time-series scenarios and clustering scenarios, this invention adopts an improved fuzzy clustering method: Based on fuzzy k-means clustering, the distance metric between samples is improved by weighting and giving higher weights to feature dimensions that have a significant impact on carrying capacity, such as reverse power flow risk and ramp risk. This makes scenario merging more focused on preserving the differences of "high-risk and strong-constraint" conditions, reducing the erroneous merging of key constraint conditions. This improves the coverage of typical scenario sets for risks such as voltage overrun, line overload and reverse power flow, and provides more targeted input support for subsequent optical-storage collaborative optimization. 2) This invention introduces a photovoltaic grid connection layout coordination factor. As a core evaluation and optimization metric, capacity coordination items are used... Coordination with electrical position items The quantification enables coordinated optimization of photovoltaic access node selection and capacity configuration, making photovoltaic access schemes more suitable for the voltage sensitivity characteristics and electrical distance distribution of the distribution network, reducing the risks of voltage overruns, line overloads and reverse power flow caused by local centralized access, thereby improving the dynamic carrying capacity and operational margin of the distribution network for distributed photovoltaics. 3) This invention constructs a photovoltaic-storage synergy adaptability as a comprehensive quantitative criterion. Under the condition of meeting the energy storage operation boundary and the distribution network safety operation constraints, it guides the energy storage timing charge and discharge strategy to smooth out photovoltaic output fluctuations, and takes into account the voltage qualification level and the suppression of reverse power flow risk. This enables the photovoltaic access layout, energy storage configuration and operation strategy to achieve synergy matching, and improves the stability, feasibility and comprehensive benefits of the photovoltaic-storage synergy scheme. 4) This invention incorporates the overall cost of the photovoltaic-storage synergy solution into a unified optimization framework. While improving dynamic carrying capacity and operational safety, it simultaneously calculates and constrains the solution cost, avoiding the economic degradation caused by simply increasing equipment capacity or strengthening the network to improve carrying capacity. By comparing and constraining the photovoltaic access cost, energy storage-related costs, and distribution network access transformation costs under the same caliber, the output photovoltaic-storage synergy configuration solution can achieve a more reasonable trade-off between "carrying capacity improvement - operational safety - economy", thus enhancing the feasibility of engineering implementation. Attached Figure Description

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is an overall framework diagram of the present invention; Figure 2 The photovoltaic grid connection layout coordination degree in this invention And capacity convergence curve; Figure 3 This is the energy storage cost convergence curve in this invention; Figure 4 This is a graph showing the change in the number of Pareto solutions in this invention; Figure 5 This is a graph showing the range of maximum capacity and minimum cost in this invention; Figure 6 This is a diagram illustrating the Pareto front evolution of capacity and cost in this invention; Figure 7 The capacity and The evolutionary process of the Pareto frontier. Detailed Implementation

[0019] A method for improving the dynamic carrying capacity of distribution networks based on photovoltaic-storage synergistic optimization includes the following steps: Step S1: Based on historical photovoltaic power output and load data, construct a typical time series scenario set, and perform anomaly removal and standardization processing on the scenario data to provide input for subsequent photovoltaic-storage synergistic optimization; Step S2: Based on the typical time-series scenario set described in Step S1, construct a multi-objective optimization model for photovoltaic-storage synergy aimed at improving dynamic carrying capacity. With the objectives of maximizing the coordination degree of photovoltaic access layout, maximizing dynamic carrying capacity, and minimizing the total cost of photovoltaic-storage synergy scheme, jointly determine the photovoltaic access node set and the access capacity of each node, the energy storage access node set and the rated power and rated capacity of energy storage. Step S3: Based on the photovoltaic access scheme and energy storage configuration scheme obtained in step S2, construct a fitness function with photovoltaic-energy storage synergy adaptability as the core. Under the conditions of meeting the energy storage operation constraints and the distribution network safety operation constraints, optimize the energy storage timing charging and discharging strategy to smooth photovoltaic power output fluctuations and reduce reverse power flow risk. Step S4: After time-series power flow simulation verification, the optimal optical-storage collaborative configuration scheme is output.

[0020] In step S1, based on historical photovoltaic power output and load data, an improved weighted fuzzy k-means clustering method is used to generate a typical time-series scenario set of "photovoltaic-load", which specifically includes the following sub-steps: Step S1.1: Input the basic parameters of the distribution network, historical photovoltaic output and load data, and use... After removing outliers according to the criteria, the data units are standardized using min-max standardization. For photovoltaic output data, samples with values ​​less than 0 or deviations from the mean exceeding three standard deviations are removed. For load data, samples with abnormal fluctuations exceeding 30% are removed. The standardization formula is as follows: In the formula; This is the original data sample; and These are the minimum and maximum values ​​of the original data sequence, respectively. These are the standardized sample values. The standardized photovoltaic power output sequence is denoted as... The standardized load sequence is denoted as ,in s The historical sample day number, These are typical daily discrete moments; Step S1.2: Based on the standardized photovoltaic power output sequence and load sequence obtained in Step S1.1, construct a joint sample set. For the... s A sequence of historical sample periods is represented as a feature vector. ,in For the first s One sample vector; T This represents the number of discrete time periods on a typical day. For the first s A historical sample in the time period The photovoltaic output value; For the first s A historical sample in the time period The photovoltaic output value; For the first s A historical sample in the time period The load value, For the first s A historical sample in the time period The load value; This is a characteristic quantity of reverse current risk. This represents the characteristic quantity of hill climbing risk; To enable scene clustering to prioritize high-risk conditions that are "prone to reverse currents and severe fluctuations," the risk characteristic quantity is calculated from the net injection sequence, and its expression is: ,in For the sample s At any moment Net injection, For the sample s At any moment The photovoltaic output value, For the sample s At any moment The load value; Reverse current risk characteristic quantity is defined as ,in Used to extract inputs during periods where net injection is positive; The larger the value, the more likely the sample day will experience reverse current flow, and the greater the need for energy storage to absorb and reduce peak flows; The characteristic quantity of hill climb risk is defined as follows: , For the sample s At any moment The net injection, of which, The larger the value, the more drastic the change in net injection, and the more critical the role of subsequent energy storage timing charge and discharge strategies in smoothing out fluctuations. Step S1.3: Obtain the joint feature vector set in step S1.2. Based on this, fuzzy k-means is used to cluster the samples. Let the number of clusters be . K , No. k The cluster center is ,sample s Cluster k The membership degree is The distance between a sample and the cluster center is defined as the weighted Euclidean distance. ,in, H The feature dimension; and The sample vector and the cluster center are respectively in the th... h Dimensional components; For the first h Dimensional weight coefficients; To unify the dispersion of different dimensions and highlight the importance of high-risk dimensions, the weighting coefficients are set to... ,in, For the first h The variance of the dimensional feature on the sample set, This represents the average of the variances in each dimension; Let be the risk amplification coefficient, where when the th h Dimensional correspondence or Time to take The remaining dimensions are taken as follows: Through the above settings, clustering has a higher resolution in the dimensions of "reverse current risk" and "climbing risk", thus providing more constrained and representative typical scenario inputs for subsequent photovoltaic-storage synergistic optimization; Step S1.4: Based on the distance definition in step S1.3, the membership degree and cluster center are calculated using the standard iterative update form of fuzzy k-means.

[0021] Membership degree updated to ,in, These are fuzzy control parameters used to adjust the "fuzziness level" of membership assignment; priority is given to selecting... ; The larger the value, the higher the value. r Samples in the next iteration s With the k The more similar the typical characteristics of the classes, For the first v Cluster center vector of the class; Cluster centers are updated using fuzzy means, expressed as follows: , ,in, For the first r The cluster center of the next iteration; For the first r Cluster center in -1 iteration; For the first s Feature vectors of historical samples; This is the convergence threshold; Finally, based on the converged cluster center vector Extract typical scenario curves, where the typical photovoltaic output curve and the typical load curve are respectively taken from the time-series components corresponding to the cluster center; and give the first... k The probability of occurrence of a typical time-series scene set is ,in, For the first k The probability of occurrence of a set of typical time-series "photovoltaic-load" scenarios satisfies ; This represents the number of historical sample days. The typical time-series scenario set and its probability will serve as input data for subsequent steps S2-S4 to optimize the photovoltaic-storage synergy and improve dynamic carrying capacity, thereby achieving synergistic optimization of energy storage configuration and time-series strategies under various operational risk conditions.

[0022] In step S2, the objective function of the constructed photovoltaic-storage synergistic multi-objective optimization model is: ,in In the formula, For the coordination of photovoltaic grid connection layout; A set of photovoltaic access nodes; For nodes photovoltaic installed capacity, The total cost of the photovoltaic-storage synergy solution, , For the cost of photovoltaic grid connection, For the total life cycle cost of energy storage, This refers to the cost of upgrades related to the access of photovoltaic and energy storage to the distribution network.

[0023] The coordination degree of photovoltaic grid connection Defined as ,in , , n For set The number of photovoltaic access nodes in the system The total number of photovoltaic access nodes allowed; For nodes i Electrical distance to the first node of the distribution network; These are the weighting coefficients; For capacity coordination items; For electrical position coordination items.

[0024] With the synergistic adaptability of photovoltaic and energy storage as the core, the required fitness function is designed, and the energy storage timing charge and discharge strategy is optimized to smooth out photovoltaic power output fluctuations. The energy storage operation meets the following model constraints: ; ; ; ; In the formula, For nodes time Energy storage capacity, For nodes time Energy storage capacity; , They are nodes time The charging and discharging power of energy storage; For time step; , These represent the energy storage charging and discharging efficiencies, with values ​​ranging from 0.85 to 0.95. , They are nodes Energy storage rated power and rated capacity, values ​​range from 15 to 60 minutes; , These represent the minimum and maximum threshold values ​​for the energy storage state of charge (SOC), which are 0.2-0.3 and 0.8-0.9, respectively.

[0025] Simultaneously meet the constraints of safe operation of the distribution network: ; ; ; In the formula, For nodes time voltage amplitude, , , Rated voltage; For the line time The current, For the line Maximum allowable flow rate For a moment Reverse load rate, This represents the maximum permissible reverse load rate.

[0026] The calculation function for the photonics-storage synergy fit is: ; In the formula, To ensure the coordination of photovoltaic grid connection layout, To maximize the coordination of photovoltaic grid connection layout; To mitigate fluctuations in photovoltaic power output, , To smooth out the standard deviation of previous photovoltaic output, To smooth out the standard deviation of the combined photovoltaic and energy storage output; Voltage deviation pass rate , The duration for which the voltage meets the constraint. Total duration; To improve the overall charging and discharging efficiency of energy storage; This represents the maximum total cost of a photovoltaic-storage synergistic solution. , , , , These are the weighting coefficients. .

[0027] In step S4, based on the energy storage timing charge and discharge strategy and under the conditions of satisfying the energy storage operation constraints and the distribution network safety operation constraints, a multi-objective evolutionary algorithm is used in one optional implementation to iteratively solve the problem, and a candidate photovoltaic-energy storage collaborative scheme is represented as a planning layer decision vector. ,in For the set of energy storage access nodes, the corresponding multi-objective function is: ,in, To facilitate a unified approach to maximizing / minimizing directions, they can be equivalently transposed into minimizing form. In the formula, To uniformly rewrite the original multi-objective optimization problem into a minimized form of the objective vector function; The target vectors are respectively The first, second, and third sub-objective functions; to uniformly handle the directions of maximization and minimization, define... , , .

[0028] The aforementioned energy storage operation constraints and distribution network safety operation constraints can be uniformly written as a set of inequality constraints. ,in For candidate solutions x In the k The constraint function values ​​under the inequality constraints; Candidate solutions for the configuration and operation strategy of photovoltaic-storage synergy; Indicates candidate solutions x Satisfy the firstk Inequality constraints; k The index is the number of the inequality constraint; This represents the total number of inequality constraints.

[0029] Define constraint violation function ,in This represents the positive part operator, which satisfies... , z Let be any real scalar; when The condition is 'when' indicates that the candidate solution satisfies all constraints. A penalty function is used for feasibility processing, expressed as follows: ,in The penalty coefficient vector is large enough to suppress infeasible solutions from entering the non-dominated frontier.

[0030] Non-dominated sorting and iterative update, let the i-th... o generation population , Population size; This represents the planning layer decision vector corresponding to the candidate photovoltaic-storage synergistic scheme. They represent the 1st, 2nd, and 3rd in the population, respectively. There are 1 candidate solutions; calculate for each individual And perform non-dominated sorting based on Pareto dominance. If Then it is called Dominate , recorded as ,in Indicates the candidate solution The j The results of feasibility processing of the objective function values ​​are obtained. This yields the hierarchical frontier. ,in For the current non-dominated solution set, To remove The non-dominated solution set of the remaining solutions is then calculated, and so on. To maintain the diversity of the solution set, the crowding distance is calculated for individuals within the same frontier. The selection is based on the principle of "rank priority, followed by crowd distance." Offspring generation and elite retention updates can be represented as follows: ; ; ; in Indicated based on parent population Selection, crossover, and mutation produce the first Offspring population; This represents the candidate population after merging the parent and offspring generations. Indicates the merging of populations Perform non-dominated sorting and select by level and crowding distance. Individuals form the next generation. When the number of iterations reaches... The process terminates when the update of the non-dominated solution set tends to stabilize, and the final non-dominated solution set is obtained for subsequent scheme screening and verification.

[0031] In this embodiment, based on the non-dominated solution set obtained from the above solution process, three representative schemes—optimal layout coordination, optimal access capacity, and optimal cost—are selected for comparison and explanation. The configuration results are shown in Table 1, and the key indicators are shown in Table 2.

[0032] Table 1 Comparison of Results of Photovoltaic-Energy Storage Collaborative Configuration Schemes

[0033] As shown in Table 1, among the three typical schemes, the sets of photovoltaic access nodes corresponding to "optimal layout coordination (Γ optimal)" and "optimal cost (minimum total cost)" are... Photovoltaic grid connection capacity and energy storage grid connection node set at each node Complete consistency indicates that, under the conditions of this example, improving the coordination of photovoltaic (PV) grid connection layout did not introduce additional site selection differences or capacity adjustments, and the results are highly consistent with the cost-optimal solution. In contrast, the "optimal grid connection capacity (maximum PV)" scheme concentrates PV capacity on another set of nodes (…). ), and simultaneously adjust the set of energy storage access nodes ( This demonstrates that when the objective leans towards "capacity maximization," the co-location of photovoltaic and energy storage will undergo a significant shift to support a higher scale of photovoltaic grid connection.

[0034] Table 2 Comparison of Key Performance Indicators

[0035] Table 2 further quantifies the performance differences among the three schemes. The "optimal grid connection capacity (maximum PV)" scheme has a total grid connection capacity of 7.57, which is about 28.3% higher than the "optimal Γ / minimum total cost" scheme's 5.90. However, its PV grid connection layout coordination degree Γ decreases to 0.040, indicating poorer coordination. Simultaneously, the energy storage cost increases to 5307.67 (about 28.4% higher than 4133.52), and a maximum voltage deviation of 0.046 pu is observed. Furthermore, the minimum reverse load factor (RPR) increases to 0.80, reflecting higher voltage deviation and reverse power flow risk. In contrast, the "optimal Γ" and "minimum total cost" schemes show consistent performance in key indicators: maximum voltage deviation is 0, RPR is 0.47, and energy storage cost remains at a low level. Meanwhile, Γ is 0.065 and 0.066 respectively, indicating that these schemes can achieve better economic efficiency and layout coordination while ensuring controllable voltage constraints and reverse power flow risk.

[0036] In summary, this study introduces a photovoltaic (PV) grid connection location coordination index into the PV-storage synergistic optimization framework, unifying and collaboratively solving the factors of "capacity enhancement, spatial layout rationality, and economic efficiency." Comparative results show that the proposed method can improve the grid's distributed PV capacity while optimizing the grid's grid connection location and capacity configuration. This makes PV grid connection more aligned with grid operating characteristics, effectively suppressing voltage deviation and reverse power flow risks, and avoiding a significant increase in storage and retrofit costs due to pursuing capacity growth. Therefore, this study achieves a synergistic optimization effect that balances grid capacity enhancement and overall cost reduction while ensuring safe operation, providing quantifiable data for planning and operational decisions in scenarios of large-scale distributed power generation integration.

Claims

1. A method for improving the dynamic carrying capacity of a distribution network based on photovoltaic-storage synergistic optimization, characterized in that, Includes the following steps: Step S1: Based on historical photovoltaic power output and load data, construct a typical time series scenario set, and perform anomaly removal and standardization processing on the scenario data to provide input for subsequent photovoltaic-storage synergistic optimization; Step S2: Based on the typical time-series scenario set obtained in Step S1, construct a multi-objective optimization model for photovoltaic-storage synergy aimed at improving dynamic carrying capacity. With the objectives of maximizing the coordination degree of photovoltaic access layout, maximizing dynamic carrying capacity, and minimizing the total cost of photovoltaic-storage synergy scheme, jointly determine the set of photovoltaic access nodes and the access capacity of each node, and the set of energy storage access nodes. Step S3: Based on the photovoltaic access scheme and energy storage configuration scheme obtained in step S2, construct a fitness function with photovoltaic-energy storage synergy adaptability as the core. Under the conditions of meeting the energy storage operation constraints and the distribution network safety operation constraints, optimize the energy storage timing charging and discharging strategy to smooth photovoltaic power output fluctuations and reduce reverse power flow risk. Step S4: After time-series power flow simulation verification, the optimal optical-storage collaborative configuration scheme is output.

2. The method according to claim 1, characterized in that, In step S1, based on historical photovoltaic power output and load data, a typical time-series scenario set is generated using improved weighted fuzzy k-means clustering, specifically including the following sub-steps: Step S1.1: Input the basic parameters of the distribution network, historical photovoltaic output and load data, and use... After removing outliers according to the criteria, the data units are standardized by min-max standardization. For photovoltaic power output data, samples with values ​​less than 0 or deviations from the mean exceeding N times the standard deviation are removed. For load data, samples with abnormal fluctuations exceeding a certain percentage are removed and standardized. Step 1.2: Based on the standardized photovoltaic power output sequence and load sequence obtained in step S1.1, construct a joint sample set, i.e., a joint feature vector set; Step S1.3: Obtain the joint feature vector set in step S1.

2. Based on this, fuzzy k-means is used to cluster the samples and obtain the weighted Euclidean distance; Step S1.4: Based on the distance obtained in step S1.3, the membership degree and cluster center are calculated using the standard iterative update form of fuzzy k-means.

3. The method according to claim 2, characterized in that, In step S1.1, the formula used for standardization is: In the formula: This is the original data sample; and These are the minimum and maximum values ​​of the original data sequence, respectively. These are the standardized sample values; the standardized photovoltaic power output sequence is denoted as... The standardized load sequence is denoted as ,in s The historical sample day number, These represent typical daily discrete moments.

4. The method according to claim 2, characterized in that, In step S1.2, when constructing the joint sample set, for the first... s A sequence of historical sample periods is represented as a feature vector. ,in For the first s One sample vector; T This represents the number of discrete time periods on a typical day. For the first s A historical sample in the time period The photovoltaic output value; For the first s A historical sample in the time period The photovoltaic output value; For the first s A historical sample in the time period The load value, For the first s A historical sample in the time period The load value; This is a characteristic quantity of reverse current risk. This represents the characteristic quantity of hill climbing risk; To enable scene clustering to prioritize high-risk conditions that are "prone to reverse currents and severe fluctuations," the risk characteristic quantity is calculated from the net injection sequence, and its expression is: ,in These are the serial numbers of typical daily discrete moments; For the sample s At any moment Net injection, For the sample s At any moment The photovoltaic output value, For the sample s At any moment The load value; Reverse current risk characteristic quantity is defined as ,in Used to extract inputs during periods where net injection is positive; The larger the value, the more likely the sample day will experience reverse current flow, and the greater the need for energy storage to absorb and reduce peak flows; The characteristic quantity of hill climb risk is defined as follows: , For the sample s At any moment The net injection, of which, The larger the value, the more drastic the change in net injection, and the more critical the role of subsequent energy storage timing charge and discharge strategies in smoothing out fluctuations.

5. The method according to claim 2, characterized in that, In step S1.3, when performing clustering, let the number of clusters be... K , No. k The cluster center is ,sample s Cluster k The membership degree is The distance between a sample and the cluster center is defined as the weighted Euclidean distance. ,in, H The feature dimension; and The sample vector and the cluster center are respectively in the th... h Dimensional components; For the first h Dimensional weight coefficients; To unify the dispersion of different dimensions and highlight the importance of high-risk dimensions, the weighting coefficients are set to... ,in, For the first h The variance of the dimensional feature on the sample set, This represents the average of the variances in each dimension; Let be the risk amplification coefficient, where when the th h Dimensional correspondence or Time to take The remaining dimensions are taken as follows: Through the above settings, clustering has a higher resolution in the dimensions of "reverse current risk and ramp risk", thereby providing more constrained and representative typical scenario inputs for subsequent photovoltaic-storage synergistic optimization.

6. The method according to claim 2, characterized in that, In step S1.4, when calculating membership and cluster centers using the standard iterative update form of fuzzy k-means, the specific steps are as follows: Membership degree updated to ,in, These are fuzzy control parameters used to adjust the "fuzziness level" of membership assignment; priority is given to selecting... ; The larger the value, the higher the value. r Samples in the next iteration s With the k The more similar the typical characteristics of the classes, For the first v Cluster center vector of the class; Cluster centers are updated using fuzzy means, expressed as follows: , ,in, For the first r The cluster center of the next iteration; For the first r Cluster center in -1 iteration; For the first s Feature vectors of historical samples; This is the convergence threshold; Finally, based on the converged cluster center vector Extract typical scenario curves, where the typical photovoltaic output curve and the typical load curve are respectively taken from the time-series components corresponding to the cluster center; and give the first... k The probability of occurrence of a typical time-series scene set is ,in, For the first k The probability of occurrence of a set of typical time-series "photovoltaic-load" scenarios satisfies ; This represents the number of historical sample days. The typical time-series scenario set and its probability will serve as input data for subsequent steps S2-S4 to optimize the synergy between photovoltaic and energy storage and improve dynamic carrying capacity, thereby achieving synergistic optimization of energy storage configuration and time-series strategies under various operational risk conditions.

7. The method according to claim 1, characterized in that, In step S2, the objective function of the constructed photovoltaic-storage synergistic multi-objective optimization model is: ,in In the formula, For the coordination of photovoltaic grid connection layout; A set of photovoltaic access nodes; For nodes photovoltaic installed capacity, The total cost of the photovoltaic-storage synergy solution, , For the cost of photovoltaic grid connection, For the total life cycle cost of energy storage, This refers to the cost of upgrades related to the access of photovoltaic and energy storage to the distribution network.

8. The method according to claim 7, characterized in that, The coordination degree of photovoltaic grid connection Defined as ,in , , n For set The number of photovoltaic access nodes in the system The total number of photovoltaic access nodes allowed; For nodes i Electrical distance to the first node of the distribution network; These are the weighting coefficients; For capacity coordination items; For electrical position coordination items.

9. The method according to any one of claims 1 to 8, characterized in that, In step S3, with the photovoltaic-storage synergy adaptation degree as the core, the required fitness function is designed, and the energy storage timing charge and discharge strategy is optimized to smooth out photovoltaic power output fluctuations. The energy storage operation satisfies the following model constraints: ； ； ； ； In the formula, For nodes time Energy storage capacity, For nodes time Energy storage capacity; , They are nodes time The charging and discharging power of energy storage; For time step; , These refer to the energy storage charging and discharging efficiencies, respectively. , They are nodes Rated power and rated capacity of energy storage; , The minimum and maximum threshold values ​​for the State of Charge (SOC) of energy storage are 0.2-0.3 and 0.8-0.9, respectively. Simultaneously meet the constraints of safe operation of the distribution network: ； ； ； In the formula, For nodes time voltage amplitude, , , Rated voltage; For the line time The current, For the line Maximum allowable flow rate For a moment Reverse load rate, Maximum permissible reverse load rate; The calculation function for the photonics-storage synergy fit is: ; In the formula, To ensure the coordination of photovoltaic grid connection layout, To maximize the coordination of photovoltaic grid connection layout; To mitigate fluctuations in photovoltaic power output, , To smooth out the standard deviation of previous photovoltaic output, To smooth out the standard deviation of the combined photovoltaic and energy storage output; Voltage deviation pass rate , The duration for which the voltage meets the constraint. Total duration; To improve the overall charging and discharging efficiency of energy storage; The maximum permissible life-cycle cost of energy storage; , , , , These are the weighting coefficients. .

10. The method according to claim 5, characterized in that, In step S4, based on the energy storage time-series charging and discharging strategy and under the conditions of satisfying the energy storage operation constraints and the distribution network safety operation constraints, time-series power flow simulation verification is performed to obtain the node voltage amplitude, line current and reverse load rate at each time point, and to count the voltage over-limit rate, reverse power flow ratio and fluctuation smoothing rate, and output the power flow simulation verification results of the photovoltaic-storage collaborative scheme.

Images