Double-layer planning method for ac-dc hybrid distribution network based on cooperation of vsc and ess

Abstract

The double-layer planning method of AC / DC hybrid distribution network based on VSC and ESS cooperation comprises the following steps: Latin hypercube sampling is used to obtain source and load initial scenes, and K-means clustering is used for reduction; an upper-layer planning model is established, a minimum system annual comprehensive cost is taken as an objective function, and installation positions and capacities of VSC and ESS are optimized; a lower-layer operation model is established, a minimum system comprehensive operation cost is taken as an objective under the configuration scheme given by the upper layer, the lower-layer optimal operation cost is fed back to the upper layer for individual fitness evaluation, and the VSC and ESS cooperative site selection and capacity determination scheme with the lowest system comprehensive cost is obtained through upper and lower layer cooperation iteration. The application provides a double-layer planning method of AC / DC hybrid distribution network based on VSC and ESS cooperation, which can provide a decision basis for cooperative configuration and operation scheduling of VSC and ESS in the AC / DC hybrid distribution network, and improve the economy of the planning scheme on the basis of meeting the system safe operation constraint.

Description

Technical Field

[0001] This invention relates to the field of power system planning and operation control technology, and in particular to a two-layer planning method for AC / DC hybrid distribution networks based on the collaboration of VSC and ESS. Background Technology

[0002] With the increasing integration of renewable energy, DC loads, and energy storage devices into distribution networks, traditional AC distribution networks face higher demands in terms of network losses, voltage control, and renewable energy consumption. DC distribution technology offers advantages such as lower losses, reduced AC / DC conversion stages, easier integration of DC power sources and storage devices, and greater flexibility in control. Therefore, building a hybrid AC / DC distribution network on the existing AC distribution network has become an important development direction for improving the overall operational performance of distribution networks.

[0003] However, most existing studies on distribution network site selection and capacity planning are limited to the capacity configuration of a single distributed power source or independent energy storage, and rarely involve the AC / DC flexible transformation of the original AC distribution network, especially lacking in-depth discussions on the coordinated site selection and capacity planning of VSC (voltage source converter) and ESS (energy storage system).

[0004] Meanwhile, the limited planning methods often still use deterministic power flow models or single static scenarios, failing to establish high-precision scenario reduction models, making it difficult to effectively adapt to the strong temporal fluctuations and uncertainties on the source and load sides of hybrid distribution networks. In addition, existing planning mostly adopts single-layer optimization models, which cannot simultaneously take into account the refined coordination of investment allocation and scheduling operation, making it difficult to maximize the overall comprehensive benefits of the system under complex AC and DC power flow constraints.

[0005] Therefore, there is an urgent need to propose a two-layer planning method for AC / DC hybrid distribution networks that takes into account the uncertainty of source-load timing and realizes the coordination of VSC and ESS, so as to significantly improve the economy and operational stability of the system. Summary of the Invention

[0006] The technical problem of this invention is that existing distribution network site selection and capacity planning mostly adopt a single static scenario, which is difficult to cope with strong temporal fluctuations and uncertainties on both the source and load sides; at the same time, there is a lack of grid structure planning for AC / DC flexible transformation of distribution networks and ESS system coordination, which fails to fully explore the active and reactive power support potential of equipment; in addition, existing planning mostly adopts a single-layer optimization model, which cannot simultaneously take into account the fine coordination of investment configuration and scheduling operation, and it is difficult to maximize the overall system benefits under complex AC / DC power flow constraints.

[0007] The purpose of this invention is to provide a two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration to address the above-mentioned problems. First, the source-load time-series uncertainty features are accurately extracted through scenario generation and reduction methods. Then, a two-layer planning model with upper-layer investment configuration and lower-layer operation scheduling collaborative interaction is constructed. A hybrid algorithm combining genetic algorithm with elite strategy and second-order cone programming is used to solve the problem, thereby obtaining the optimal addressing and capacity grading scheme that takes into account both economy and operational stability.

[0008] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A two-tier planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration includes the following steps: Step 1: Use the Latin hypercube sampling method to obtain the initial scene of the source load, and use the K-means clustering algorithm to reduce the scene; Step 2: Establish a higher-level planning model, with the goal of minimizing the system's annual overall cost, and optimize the installation location and capacity of VSC and ESS; Step 3: Establish a lower-level operation model. Under the VSC and ESS configuration scheme given by the upper-level planning model, optimize the operation status of the AC / DC hybrid distribution network with the goal of minimizing the overall system operating cost. Step 4: Use a genetic algorithm with an elite retention strategy to solve the upper-level planning model, and use a second-order cone programming method to solve the lower-level operating model. Feed the optimal operating cost calculated by the lower-level operating model back to the upper-level planning model for individual fitness evaluation. Through the coordinated iteration of the upper and lower levels, obtain the VSC and ESS collaborative location and occupancy scheme with the lowest overall system cost.

[0009] Step 1 specifically includes: Step 1.1: Obtain light intensity data and load data within the study area, divide the study period into four typical seasons: spring, summer, autumn, and winter, and determine the weight of the number of days corresponding to each typical season; Step 1.2: Normalize the light intensity data in Step 1.1 according to the standard test irradiance to obtain the photovoltaic power output per unit curve; normalize the load data in Step 1.1 according to the annual maximum load to obtain the load per unit curve; Step 1.3: For each time period of each typical season, establish a Beta distribution model based on the photovoltaic output sample of the corresponding time period; establish a normal distribution model based on the load sample of the corresponding time period to obtain the uncertainty distribution parameters of photovoltaic output and load in each season and time period; Step 1.4: Use the Latin hypercube sampling method to sample the Beta distribution model and the normal distribution model in Step 1.3 respectively, to generate the original photovoltaic output scenario set and the original load scenario set for each typical season; Step 1.5: Use the K-means clustering algorithm to reduce the original photovoltaic power output scenario set and the original load scenario set for each typical season in Step 1.4 to obtain the power output curves of four sets of seasonal source and load typical daily scenarios in spring, summer, autumn and winter.

[0010] Step 2 specifically includes: Step ①: The upper-level planning model aims to minimize the system's annual comprehensive cost. The objective function is as follows: (twenty two); In the formula: This represents the objective function value of the upper-level planning model, i.e., the system's annual comprehensive cost; This indicates the system access and construction cost; This represents the overall operating cost of the system; This indicates the annualized investment cost of VSC; This indicates the annualized investment cost of ESS; This indicates the operating loss cost of VSC; This indicates the operating cost of ESS charging and discharging; Indicates network loss cost; This indicates the cost of purchasing electricity from the upstream power grid; Indicates the cost of curtailment; Step 2: Construct the constraints for the upper-level planning model: 1) Constraints on the number of VSC and ESS candidate installation nodes.

[0011] (30); In the formula: For nodes i Number of VSC installations; For nodes j Number of ESS installations.

[0012] 2) VSC installation capacity constraints (31); In the formula: Represents a node The installation capacity of the VSC; and These represent the lower and upper limits of the installation capacity of a single VSC unit, respectively. Represents a node i The number of VSC installations.

[0013] 3) ESS installation capacity constraints (32); In the formula: and Representing nodes respectively The rated energy capacity and rated power capacity of the ESS; , These represent the lower and upper limits of the energy capacity of a single ESS unit, respectively. , These represent the lower and upper limits of the power capacity of a single ESS unit, respectively. Represents a node The number of ESS installations.

[0014] Step 3 specifically includes: Step 1: Construct the objective function of the lower-level operating model, with the lower level based on the overall system operating cost. Minimum target: (33); In the formula: This indicates that the lower-level operating model is based on the overall system operating cost. The minimum is the target.

[0015] Step 2: Construct the constraints for the lower-level runtime model: 1) DistFlow constraints in AC power grids (34); In the formula: Represented by node The set of the starting nodes of the branches of the terminal nodes; Represented by node It is the set of branch end nodes of the first node; Represents the set of AC branches; , They represent the first The first typical season Communication branch at each time period The active and reactive power transmitted upstream; , They represent AC branches respectively. Resistance and reactance; , Representing communication nodes Sum of squared voltages in AC branch Square of current; Indicates communication node The square of the voltage at the point; Indicates a branch The reactance.

[0016] (35); In the formula: , Representing communication nodes Net active and reactive loads. , Representing communication nodes Original active and reactive loads; , These represent the ESS charging power and discharging power, respectively. This indicates the actual power output of photovoltaic power generation. This indicates the parameters for the main network access node. When the node... j When it is a substation node, Otherwise, it is 0; , The first The first typical season During each time period, the upper-level power grid sends to the node j Injected active and reactive power.

[0017] 2) DC power flow constraints (36); (37); In the formula: Represents the set of DC branches; Indicates the first The first typical season DC branch during each time period Active power transmitted upstream; Indicates DC branch The resistance; , Representing DC nodes Sum of squared voltages in DC branch Square of current; Indicates DC node Net active load at the location; Indicates DC node Active load at point; , These represent the ESS charging power and discharging power, respectively. This indicates the actual photovoltaic power consumption. If the node... If no ESS or PV is connected at a node, the corresponding item is set to zero. Reactive power is not considered for DC nodes. Indicates DC node The square of the voltage at that point.

[0018] 3) VSC runtime constraints At the medium-voltage level, this paper adopts a unipolar symmetrical VSC structure for the transformation of AC and DC lines. During the transformation process, one VSC in the DC line must be in Vdc-Q control mode.

[0019] (38); In the formula: , They represent the first The VSC in the first The first typical season Active and reactive power on the AC side during each time period; Indicates the first The active power output of each VSC on the DC side; Indicates the first The reactive power control quantity of each VSC; , They represent the first The resistance and reactance of the VSC equivalent circuit; Indicates the flow through the first Current in a VSC equivalent circuit; , They represent the first Lower and upper limits of reactive power control for each VSC; Indicates the first The installation capacity of each VSC; Indicates the first Internal potential of each VSC; Indicates the first VSC DC-side voltage; This represents the voltage conversion factor under SVPWM modulation; Indicates DC voltage utilization rate; Indicates the first The modulation level of each VSC has a value range of [0,1]. 4) ESS running constraints When combined with time-of-use pricing, ESS can play a role in "peak shaving and valley filling" and can also improve the absorption capacity of PV.

[0020] (39); In the formula: Represents a node ESS at the The first typical season Net charging and discharging power for each time period; , These represent the ESS charging power and discharging power, respectively. express ESS rated power capacity; , These represent the charging and discharging state variables of the ESS, respectively, and are 0-1 variables; Indicates that ESS is in the The first typical season The amount of stored electricity in each time period; Represents a node ESS rated energy capacity; , These represent the ESS charging efficiency and discharging efficiency, respectively. Indicates that ESS is in the The first typical season The amount of stored electricity in each time period; Indicates that ESS is in the The first typical season The amount of stored electricity in each time period.

[0021] 5) PV operating constraints (40); In the formula: Represents the set of photovoltaic nodes; Indicates the first The photovoltaic node at the ... The first typical season The actual absorption capacity in each time period; Indicates the first The photovoltaic node at the ... The first typical season The predicted output for each time period.

[0022] 6) Power constraints for interaction with the main network The power interaction constraints between substation nodes and the upstream power grid are as follows: (41); In the formula: This represents the set of substation nodes connected to the upper-level power grid. , They represent the first The first typical season Substation nodes during each time period Active and reactive power interacting with the upstream power grid; , These represent the lower limit and upper limit of active power for main network interaction, respectively; , These represent the lower limit and upper limit of reactive power for main network interaction, respectively.

[0023] 7) Safety constraints Safety constraints include upper and lower limits for voltage and current, as well as transmission power constraints: (42); In the formula: , These represent the set of AC nodes and the set of DC nodes, respectively. , These represent the AC branch set and the DC branch set, respectively. , Representing nodes respectively Lower and upper limits of voltage amplitude; Indicates a branch The upper limit of the current amplitude; , They represent the first The first typical season Communication branch at each time period The active and reactive power transmitted upstream; Indicates the first The first typical season DC branch during each time period Active power transmitted upstream; , These represent the maximum transmission capacity of the AC branch and the DC branch, respectively. For the first s The first typical season t The flow through the branch road during each period ij The magnitude of the current.

[0024] Step 4 specifically includes the following: Step ①: Input initial values ​​for annual solar load dataset, network parameters, VSC, and ESS planning parameters; Step 2: Use a genetic algorithm with an elitist strategy to solve the upper-level planning model, setting the population size, maximum number of iterations, crossover probability, and mutation probability. Perform mixed encoding on the candidate installation nodes and capacities of VSC and ESS, and randomly generate the initial configuration population. Let the population size be The maximum number of iterations is The crossover probability is The mutation probability is The number of elite individuals retained is . No. The generation population is represented as: (43); In the formula: Indicates the first Generational population; Indicates the first The first generation of the population Individual; Indicates population size; , , They represent the first The first, second, and third offspring in the first generation population Individual; Indicates population size.

[0025] Each individual node uses a hybrid encoding of "candidate node installation status - capacity parameters", with the encoding format as follows: (44); In the formula: Indicates candidate nodes The decision variable for whether or not to install VSC; Indicates candidate nodes The installation capacity of the VSC; Indicates candidate nodes The decision variable for whether or not to install ESS; and Representing candidate nodes respectively The rated energy capacity and rated power capacity of the ESS. If Then the node VSC is not installed there; if Then the node ESS is not installed at this location.

[0026] Step 3: Pass the configuration scheme of each individual in the upper-level population as a known constraint to the lower-level operating model. With the goal of minimizing the overall system operating cost, derive the optimal overall operating cost and operating state under this configuration scheme, and return the cost to the upper level to form the main problem objective function: For the first The first generation of the population individual Decode the VSC and ESS configuration schemes: (45); In the formula: Represents the individual decoding function; Indicates the first VSC installation location and capacity plan for each individual; Indicates the first The installation location, rated energy capacity, and rated power capacity scheme for each individual ESS.

[0027] In the given and Under the given conditions, solve the lower-level operational model: (46); In the formula: Indicates the first The overall operating cost of the optimal lower-level system corresponding to each upper-level configuration scheme; This represents the corresponding set of optimal operating decision variables, including VSC operating power, ESS charging and discharging power, ESS power status, actual photovoltaic power absorption, branch power flow, node voltage, and main grid interaction power. This represents the feasible domain that satisfies the underlying operational constraints under a given VSC and ESS configuration scheme. This represents the set of operational decision variables in the lower-level operational model.

[0028] Based on the optimal operating cost feedback from the lower layer, calculate the first... The overall annual system cost corresponding to each upper-level configuration scheme: (47); In the formula: Indicates the first The system's annual comprehensive cost corresponding to each individual; Indicates the first Annualized investment costs for VSC and ESS for each individual; This represents the optimal overall operating cost of the system based on feedback from the lower layer.

[0029] Step 4: Calculate the fitness value of each individual in the population, rank the individuals in the population according to the calculated fitness value; use the elite retention strategy to enter the next generation with the best fitness of the current generation, and perform elite retention, crossover and mutation operations on the remaining population to generate a new offspring population. Since the upper-level planning model is a cost minimization problem, the first... The fitness of each individual is: (48); In the formula: Indicates the first The fitness of each individual; This represents the fitness correction parameter, used to avoid a denominator of 0.

[0030] The first The most fittest in the generation population Individuals as a collection of elite individuals: (49); In the formula: Indicates the first A collection of elite individuals within a generational population; This means selecting the most fit population from the current population. Individual; This indicates the number of individuals retained by the elite.

[0031] Selection, crossover, and mutation operations are performed on non-elite individuals to generate offspring populations, which then form the next generation population together with elite individuals. (50); In the formula: Indicates the first Generational population; This represents the set of offspring individuals generated by selection, crossover, and mutation operations; Indicates the crossover probability; Indicates the proportion of variation; Indicates the probability of univariate mutation; This represents the variable asynchronous length coefficient.

[0032] Step 5: Repeat steps 2 to 4 to perform alternating optimization iterations between upper and lower layers until the algorithm reaches the set maximum number of iterations or meets the convergence condition; terminate the iteration and output the individual with the best global fitness, which is the optimal AC / DC hybrid distribution network planning scheme. Attached Figure Description

[0033] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a typical daily photovoltaic power output curve for an example of the present invention.

[0034] Figure 2 This is a typical daily load curve for an example of the present invention.

[0035] Figure 3 This is the equivalent circuit diagram of the VSC in an example of the present invention.

[0036] Figure 4 This is a logical relationship diagram of a two-level planning method in an example of the present invention.

[0037] Figure 5 This is a flowchart illustrating the solution process of the two-layer optimization model in an example of the present invention.

[0038] Figure 6 This is the topology diagram after planning for the IEEE 33-node example of this invention.

[0039] Figure 7 This is a comparison chart of voltage curves for an example of the present invention. Detailed Implementation

[0040] A two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration includes the following steps: Step 1: Obtain the initial scene of the source load using the Latin hypercube sampling method, and then use the K-means clustering algorithm to reduce the scene size. Step 1.1: Obtain light intensity data and load data within the study area, divide the study period into four typical seasons: spring, summer, autumn, and winter, and determine the weight of the number of days corresponding to each typical season; Acquire annual light intensity and load data for the study area, and divide the annual study period into four typical seasons: spring, summer, autumn, and winter, denoted as _____. Let the first... The number of days corresponding to each typical season is The weight of the number of days in this typical season is: (1); In the formula: Indicates the first s Weighting of the number of days in a typical season; Indicates the first The number of days included in a typical season; They represent the four typical seasons: spring, summer, autumn, and winter.

[0041] Step 1.2: Normalize the light intensity data according to the standard test irradiance to obtain the per-unit value curve of photovoltaic power output; normalize the load data according to the maximum annual load to obtain the per-unit value curve of load. Step ①: Place the first Heavenly Light intensity during the period Irradiance tested according to standards After normalization, the original per-unit photovoltaic output value is obtained: (2); In the formula: Indicates the first Heavenly The original per-unit value of photovoltaic output for the time period; Indicates the first Heavenly Light intensity during a given time period; This indicates the standard test irradiance.

[0042] Step 2: Limit and correct the original photovoltaic output per unit value: (3); In the formula: This indicates the upper limit of the per-unit value for photovoltaic power output.

[0043] Step 3: Place the first Heavenly Raw active load for the time period Based on the maximum active power load throughout the year Normalization is performed: (4); In the formula: Indicates the first Heavenly Per-unit value of active load for a given time period; Indicates the first Heavenly The original active load for the time period; This indicates the maximum active power load for the entire year.

[0044] Step 1.3: For each time period of each typical season, establish a Beta distribution model based on the photovoltaic output sample of the corresponding time period, and establish a normal distribution model based on the load sample of the corresponding time period to obtain the uncertainty distribution parameters of photovoltaic output and load in each season and time period; Step ①: For the first The first typical season A Beta distribution model is established for the photovoltaic power output samples during the time period, and its probability density function is: (5); In the formula: This indicates the per-unit value of photovoltaic power output; and They represent the first The first typical season Two shape parameters of the Beta distribution over a given time period; This represents the Gamma function.

[0045] Step 2: According to the first The first typical season Sample mean of photovoltaic power output during the time period and variance Calculate the intermediate variables for the Beta distribution parameters: (6); In the formula: Indicates the first The first typical season Intermediate variables in the calculation of Beta distribution parameters over a given time period; This represents the sample mean of photovoltaic power output; This represents the sample standard deviation of photovoltaic power output.

[0046] Step 3: When When the shape parameter of the Beta distribution is calculated, the following formula is used: (7); (8); when season: (9); In the formula: represents the lower limit value of the shape parameter of the Beta distribution; and These represent the two shape parameters of the Beta distribution.

[0047] Step 4: For the first The first typical season A normal distribution model is established for the active load samples during a time period, and its probability density function is: (10); In the formula: This represents the per-unit value of active power load; Indicates the first The first typical season The average active load sample for a given time period; Indicates the first The first typical season The standard deviation of active load samples for each time period.

[0048] Step 5: To avoid excessively small standard deviations in the loading samples leading to subsequent sampling anomalies, a lower limit correction is applied to the standard deviation of the loadings: (11); In the formula: Indicates the first The first typical season The original standard deviation of the active load sample for each time period; This represents the lower limit of the standard deviation of active power load.

[0049] Step 6: Convert the active load to reactive load based on the equivalent power factor angle: (12); In the formula: The power factor angle; This represents the ratio of reactive load to active load.

[0050] Step 1.4: Use the Latin hypercube sampling method to sample the photovoltaic output Beta distribution model and the load normal distribution model respectively, and generate the original photovoltaic output scenario set and the original load scenario set for each typical season; Step ①: Let the Latin hypercube sampling size be... The number of time periods within a day is Then the first ( ) the first time period ( The uniform sampling value corresponding to ) samples is: (13); In the formula: Indicates the first The first time period The uniformly sampled value of a sample; Indicates the size of the Latin hypercube sampling; Indicates the number of time periods within a day; Indicates the first The random sequence number corresponding to each time period; express Random numbers within the interval.

[0051] Step ②: Generate the first step based on the inverse cumulative distribution function of the photovoltaic output Beta distribution model. The first typical season The first time period Sample values ​​of photovoltaic power output: (14); In the formula: Indicates the first The first typical season The first time period A sample value of photovoltaic power output; The inverse cumulative distribution function of the Beta distribution; and They represent the first The first typical season The shape parameters of the Beta distribution over a given time period.

[0052] Step ③: Generate the inverse cumulative distribution function of the load normal distribution model. The first typical season The first time period Initial load sampling values: (15); In the formula: Indicates the first The first typical season The first time period Initial values ​​for load sampling; The inverse cumulative distribution function of the normal distribution; and They represent the first The first typical season The mean and standard deviation of active load samples for each time period.

[0053] Step 4: Correct the lower limit of the initial load sampling values ​​to obtain the original load scenario set: (16); In the formula: This represents the corrected load sample value; This indicates the lower limit of the load sampling value.

[0054] Step 1.5: Use the K-means clustering algorithm to reduce the original photovoltaic power output scenario set and the original load scenario set for each typical season to obtain four sets of seasonal source-load typical daily scenario power output curves for spring, summer, autumn and winter.

[0055] Step 1: The set of samples to be clustered Perform K-means clustering, where each sample For a containing The scene vectors for each time period. The K-means clustering objective function is: (17); In the formula: This represents the sum of squared clustering errors; Represents the number of clusters; Indicates the first A collection of scene types; Indicates the first Scene-centric; Indicates the first A sample of scenarios to be clustered.

[0056] Step 2: Determine the sample based on the principle of minimum distance. Cluster category: (18); In the formula: Indicates sample The cluster category to which it belongs; Represents the number of clusters; Indicates the first Scene center.

[0057] Step 3: The scene center is updated according to the following formula: (19); In the formula: Indicates the first The number of samples in the class scene set.

[0058] Step 4: For the first... Clustering is performed on the original sets of photovoltaic output and load scenarios for each typical season, and the cluster centers are used as the typical daily photovoltaic output curve and typical daily load curve for that season: (20); (twenty one); In the formula: Indicates the first The first typical seasont Typical solar power output during a given period; Indicates the first The first typical season t Typical daily load capacity for each period; Indicates the first The photovoltaic output scenario cluster center in a typical season is in the first t The values ​​for each time period; Indicates the first The cluster centers of load scenarios under typical seasons are in the first... t The values ​​for each time period.

[0059] The typical daily photovoltaic power output curve and load power output curve obtained are as follows: Figure 1 , Figure 2 As shown.

[0060] Step 2: Establish a higher-level planning model, using the minimum annual overall system cost as the objective function, to optimize the installation location and capacity of VSC and ESS: Step ①: The upper-level planning model aims to minimize the system's annual comprehensive cost. The objective function is as follows: (twenty two); In the formula: This represents the objective function value of the upper-level planning model, i.e., the system's annual comprehensive cost; This indicates the system access and construction cost; This represents the overall operating cost of the system; This indicates the annualized investment cost of VSC; This indicates the annualized investment cost of ESS; This indicates the operating loss cost of VSC; This indicates the operating cost of ESS charging and discharging; Indicates network loss cost; This indicates the cost of purchasing electricity from the upstream power grid; Indicates the cost of curtailment; The methods for calculating each cost are as follows: 1) VSC Investment Costs

[0061] (twenty three); In the formula: The investment cost of VSC; r The discount rate; For the entire lifecycle of VSC; For the VSC candidate installation node set; The unit capacity investment cost of VSC; For nodes i The installation capacity of VSC.

[0062] 2) ESS Investment Costs

[0063] (twenty four); In the formula: The investment cost of ESS; r The discount rate; For the entire life cycle of ESS; For the ESS candidate installation node set; , These are the investment costs per unit energy capacity and per unit power capacity of the ESS, respectively. , They are nodes j The rated energy capacity and rated power capacity of the ESS.

[0064] 3) VSC power loss cost

[0065] (25); In the formula: Cost of VSC power loss; for t Real-time electricity pricing; T represents the number of typical seasons; D represents the number of typical daytime periods; s For the season s Total number of days; VSC active transmission loss coefficient; For nodes VSC at the The first typical season Active power during each time period; The unit time step is 1. h .

[0066] 4) ESS operating costs

[0067] (26); In the formula: For ESS operating costs; Cost per unit charge / discharge operation for ESS; , They are nodes ESS at the The first typical season The charging power and discharging power during each time period.

[0068] 5) Network loss cost

[0069] (27); In the formula: , These are respectively the AC branch set and the DC branch set; branch road ij The resistance value; For the first s The first typical season t The flow through the branch road during each period ij The magnitude of the current; 6) Cost of purchasing electricity from higher authorities

[0070] (28); In the formula: To reduce the cost of purchasing electricity from higher authorities; For the first The first typical season Power purchased from the upper-level power grid during each time period.

[0071] 7) Cost of abandoning light

[0072] (29); In the formula: Cost of wasted light; Represents the set of photovoltaic nodes; This represents the unit cost of light skipping penalty; , The first The photovoltaic node at the ... The first typical season The predicted output and actual absorption output for each time period.

[0073] Step 2: Construct the constraints for the upper-level planning optimization model: 1) Constraints on the number of VSC and ESS candidate installation nodes.

[0074] (30); In the formula: For nodes i Number of VSC installations; For nodes j Number of ESS installations.

[0075] 2) VSC installation capacity constraints (31); In the formula: Represents a node The installation capacity of the VSC; and These represent the lower and upper limits of the installation capacity of a single VSC unit, respectively.

[0076] 3) ESS installation capacity constraints (32); In the formula: and Representing nodes respectively The rated energy capacity and rated power capacity of the ESS; , These represent the lower and upper limits of the energy capacity of a single ESS unit, respectively. , These represent the lower and upper limits of the power capacity of a single ESS unit, respectively.

[0077] Step 3: Establish a lower-level operation model. Under the given VSC and ESS configuration schemes at the upper level, optimize the operation of the AC / DC hybrid distribution network with the goal of minimizing the overall system operating cost. Step 1: Construct the objective function of the lower-level operation optimization model, with the lower level using the overall system operating cost. Minimum target: (33); Step 2: Construct the constraints for the lower-level operational optimization model: 1) DistFlow constraints in AC power grids (34); (35); In the formula: Represented by node The set of the starting nodes of the branches of the terminal nodes; Represented by node It is the set of branch end nodes of the first node; Represents the set of AC branches; , They represent the first The first typical season Communication branch at each time period The active and reactive power transmitted upstream; , They represent AC branches respectively. Resistance and reactance; , Representing communication nodes Sum of squared voltages in AC branch Square of current; , Representing communication nodes Net active and reactive loads. , Representing communication nodes Original active and reactive loads; , These represent the ESS charging power and discharging power, respectively. This indicates the actual power output of photovoltaic power generation. This indicates the main network access node indicator parameter. When node j is a substation node, Otherwise, it is 0; , The first The first typical season During each time period, the upper-level power grid sends to the node Injected active and reactive power.

[0078] 2) DC power flow constraints (36); (37); In the formula: Represents the set of DC branches; Indicates the first The first typical season DC branch during each time period Active power transmitted upstream; Indicates DC branch The resistance; , Representing DC nodes Sum of squared voltages in DC branch Square of current; Indicates DC node Net active load at the location; Indicates DC node The active load is at the location; the meanings of the other symbols are consistent with the previous text. If the node... If no ESS or PV is connected at a node, the corresponding item is set to zero. Reactive power is not considered for DC nodes.

[0079] 3) VSC runtime constraints At the medium-voltage level, this paper employs a unipolar symmetrical VSC structure for the modification of AC / DC lines. During the modification process, one VSC in the DC line must be in Vdc-Q control mode. The equivalent circuit diagram of the VSC is shown below. Figure 3 As shown.

[0080] (38); In the formula: , They represent the first The VSC in the first The first typical season Active and reactive power on the AC side during each time period; Indicates the first The active power output of each VSC on the DC side; Indicates the first The reactive power control quantity of each VSC; , They represent the first The resistance and reactance of the VSC equivalent circuit; Indicates the flow through the first Current in a VSC equivalent circuit; , They represent the first Lower and upper limits of reactive power control for each VSC; Indicates the first The installation capacity of each VSC; Indicates the first Internal potential of each VSC; Indicates the first VSC DC-side voltage; This represents the voltage conversion factor under SVPWM modulation; Indicates DC voltage utilization rate; Indicates the first The modulation level of each VSC has a value range of [0,1]. 4) ESS running constraints When combined with time-of-use pricing, ESS can play a role in "peak shaving and valley filling" and can also improve the absorption capacity of PV.

[0081] (39); In the formula: Represents a node ESS at the The first typical season Net charging and discharging power for each time period; , These represent the ESS charging power and discharging power, respectively. express ESS rated power capacity; , These represent the charging and discharging state variables of the ESS, respectively, and are 0-1 variables; Indicates that ESS is in the The first typical season The amount of stored electricity in each time period; Represents a node ESS rated energy capacity; , These represent the ESS charging efficiency and discharging efficiency, respectively.

[0082] 5) PV operating constraints (40); In the formula: Represents the set of photovoltaic nodes; Indicates the first The photovoltaic node at the ... The first typical season The actual absorption capacity in each time period; Indicates the first The photovoltaic node at the ... The first typical season The predicted output for each time period.

[0083] 6) Power constraints for interaction with the main network The power interaction constraints between substation nodes and the upstream power grid are as follows: (41); In the formula: This represents the set of substation nodes connected to the upper-level power grid. , They represent the first The first typical season Substation nodes during each time period Active and reactive power interacting with the upstream power grid; , These represent the lower and upper limits of active power for main network interaction, respectively; , These represent the lower and upper limits of reactive power for main network interaction, respectively.

[0084] 7) Safety constraints Safety constraints include upper and lower limits for voltage and current, as well as transmission power constraints: (42); In the formula: , These represent the set of AC nodes and the set of DC nodes, respectively. , These represent the AC branch set and the DC branch set, respectively. , Representing nodes respectively Lower and upper limits of voltage amplitude; Indicates a branch The upper limit of the current amplitude; , They represent the first The first typical season Communication branch at each time period The active and reactive power transmitted upstream; Indicates the first The first typical season DC branch during each time period Active power transmitted upstream; , These represent the maximum transmission capacity of the AC branch and the DC branch, respectively.

[0085] Step 4: The upper-level planning model uses a genetic algorithm with an elite strategy to solve the problem, while the lower-level running model uses a second-order cone programming method to solve the optimal running schedule problem under the current configuration. The logical relationship diagram of the two-level planning is as follows: Figure 4 As shown; the optimal operating cost obtained from the lower layer is fed back to the upper layer and the individual fitness is calculated. Through collaborative optimization iteration between the upper and lower layers, the optimal AC / DC hybrid distribution network planning scheme with the lowest total system cost is finally obtained. The flowchart of the two-layer optimization model solution is shown below. Figure 5 As shown: Step ①: Input initial values ​​for annual solar load dataset, network parameters, VSC, and ESS planning parameters; Step 2: Use a genetic algorithm with an elitist strategy to solve the upper-level planning problem. Set the population size, maximum number of iterations, crossover probability, and mutation probability. Perform mixed encoding on the candidate installation nodes and capacities of VSC and ESS, and randomly generate the initial configuration population. Let the population size be The maximum number of iterations is The crossover probability is The mutation probability is The number of elite individuals retained is . No. The generation population is represented as: (43); In the formula: Indicates the first Generational population; Indicates the first The first generation of the population Individual; Indicates population size.

[0086] Each individual node uses a hybrid encoding of "candidate node installation status - capacity parameters", with the encoding format as follows: (44); In the formula: Indicates candidate nodes The decision variable for whether or not to install VSC; Indicates candidate nodes The installation capacity of the VSC; Indicates candidate nodes The decision variable for whether or not to install ESS; and Representing candidate nodes respectively The rated energy capacity and rated power capacity of the ESS. If Then the node VSC is not installed there; if Then the node ESS is not installed at this location.

[0087] Step 3: Pass the configuration scheme of each individual in the upper-level population as a known constraint to the lower-level operating model. With the goal of minimizing the overall system operating cost, derive the optimal overall operating cost and operating state under this configuration scheme, and return the cost to the upper level to form the main problem objective function: For the first The first generation of the population individual Decode the VSC and ESS configuration schemes: (45); In the formula: Represents the individual decoding function; Indicates the first VSC installation location and capacity plan for each individual; Indicates the first The installation location, rated energy capacity, and rated power capacity scheme for each individual ESS.

[0088] In the given and Under the given conditions, solve the lower-level operational optimization model: (46); In the formula: Indicates the first The overall operating cost of the optimal lower-level system corresponding to each upper-level configuration scheme; This represents the corresponding set of optimal operating decision variables, including VSC operating power, ESS charging and discharging power, ESS power status, actual photovoltaic power absorption, branch power flow, node voltage, and main grid interaction power. This represents the feasible domain that satisfies the underlying operational constraints under a given VSC and ESS configuration scheme.

[0089] Based on the optimal operating cost feedback from the lower layer, calculate the first... The system's annual comprehensive cost per individual: (47); In the formula: Indicates the first The system's annual comprehensive cost corresponding to each individual; Indicates the first Annualized investment costs for VSC and ESS for each individual; This represents the optimal overall operating cost of the system based on feedback from the lower layer.

[0090] Step 4: Calculate the fitness value of each individual in the population, rank the individuals in the population according to the calculated fitness value; use the elite retention strategy to enter the next generation with the best fitness of the current generation, and perform elite retention, crossover and mutation operations on the remaining population to generate a new offspring population. Since the upper-level planning model is a cost minimization problem, the first... The fitness of each individual is: (48); In the formula: Indicates the first The fitness of each individual; This represents the fitness correction parameter, used to avoid a denominator of 0.

[0091] The first The most fittest in the generation population Individuals as a collection of elite individuals: (49); In the formula: Indicates the first A collection of elite individuals within a generational population; This means selecting the most fit population from the current population. Individual; This indicates the number of individuals retained by the elite.

[0092] Selection, crossover, and mutation operations are performed on non-elite individuals to generate offspring populations, which then form the next generation population together with elite individuals. (50); In the formula: Indicates the first Generational population; This represents the set of offspring individuals generated by selection, crossover, and mutation operations; Indicates the crossover probability; Indicates the proportion of variation; Indicates the probability of univariate mutation; This represents the variable asynchronous length coefficient.

[0093] Step 5: Repeat steps 4.2 to 4.4 to perform alternating optimization iterations between upper and lower layers until the algorithm reaches the set maximum number of iterations or meets the convergence condition; terminate the iteration and output the individual with the best global fitness, which is the optimal AC / DC hybrid distribution network planning scheme.

[0094] Example 1: This invention uses a 10kV distribution network with IEEE 33 nodes as an example for analysis. The parameters of the VSC and ESS to be optimized are set. The genetic algorithm population size is 100, and the maximum number of iterations is 100. When the fitness remains unchanged, the iterations of the upper and lower layers end, and the planning results are output. Furthermore, to verify the superiority of the proposed model, three comparative scenarios are set up for analysis: Scenario 1 involves collaborative location and capacity determination of VSC and ESS, i.e., the model proposed in this example; Scenario 2 involves location and capacity determination of only VSC, without considering ESS collaboration; Scenario 3 involves location and capacity determination of only ESS, without considering VSC collaboration. Table 1 shows the planning results for VSC, Table 2 shows the running results for ESS planning, and Table 3 shows the running results for the planning schemes of the three models.

[0095]

[0096]

[0097]

[0098] The topology diagram of the IEEE 33-node example after planning using the method proposed in this example is as follows. Figure 6 As shown in Tables 1 to 3: The model proposed in this example has the best overall economic benefits. Although Model 3 has the lowest initial investment cost, it faces line overload during operation due to the lack of AC / DC flexible interconnection and power flow optimization support for VSC. Its power purchase cost, network loss cost, and curtailment cost all increase significantly.

[0099] Although Model 2 effectively improves the overall power flow distribution of the network through flexible interconnection, it lacks the role of energy storage resources in flexibly shaving peaks and filling valleys over time. To cope with extreme peak and valley conditions, it has to increase the rated transmission capacity of VSC compared to Model 1, resulting in a higher total investment cost and a higher overall cost than Model 1.

[0100] As can be seen, the planning model proposed in this example fully coordinates the spatial power flow regulation of VSC and the temporal peak shaving and valley filling of ESS, minimizing the network loss of the system and improving the renewable energy absorption capacity, thereby maximizing the overall benefits.

[0101] Figure 7 This is a comparison chart of voltage curves for an example of the present invention, provided by... Figure 7 It can be seen that: Scenario 3 has the lowest overall voltage level, especially with a significant voltage drop at the end of the AC feeder. This is due to the lack of AC / DC flexible interconnection and dynamic reactive power compensation of VSC. Relying solely on the active power charging and discharging of ESS is insufficient to fundamentally overcome the reactive power shortage and power flow transmission bottleneck of long-distance AC lines, resulting in weak voltage support capability at the end nodes.

[0102] In contrast, Scenario 2 introduces a VSC through AC / DC conversion, fully leveraging its reactive power support and active power transmission capabilities, resulting in a significant increase in the overall grid voltage level. However, due to the lack of ESS (Emerging Power Supply) for peak shaving and valley filling over time, relying solely on VSC regulation can easily lead to excessively high voltages at some feeder nodes when facing extreme source-load interaction conditions such as localized large-scale photovoltaic power generation, increasing the risk of localized overvoltages. Scenario 1, as presented in this invention, demonstrates the optimal voltage stabilization effect and steady-state control capability.

Claims

1. A two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration, characterized in that, Includes the following steps: Step 1: Use the Latin hypercube sampling method to obtain the initial scene of the source load, and use the K-means clustering algorithm to reduce the scene; Step 2: Establish a higher-level planning model, with the goal of minimizing the system's annual overall cost, and optimize the installation location and capacity of VSC and ESS; Step 3: Establish a lower-level operation model. Under the VSC and ESS configuration scheme given by the upper-level planning model, optimize the operation status of the AC / DC hybrid distribution network with the goal of minimizing the overall system operating cost. Step 4: Solve the upper-level planning model and use the second-order cone programming method to solve the lower-level operation model. Feed the optimal operating cost calculated by the lower-level operation model back to the upper-level planning model for individual fitness evaluation. Through the coordinated iteration of the upper and lower levels, obtain the VSC and ESS collaborative location and occupancy scheme with the lowest overall system cost.

2. The two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration as described in claim 1, characterized in that: Step 1 specifically includes: Step 1.1: Obtain light intensity data and load data within the study area, divide the study period into four typical seasons: spring, summer, autumn, and winter, and determine the weight of the number of days corresponding to each typical season; Step 1.2: Normalize the light intensity data in Step 1.1 according to the standard test irradiance to obtain the photovoltaic power output per unit curve; normalize the load data in Step 1.1 according to the annual maximum load to obtain the load per unit curve; Step 1.3: For each time period of each typical season, establish a Beta distribution model based on the photovoltaic output sample of the corresponding time period; establish a normal distribution model based on the load sample of the corresponding time period to obtain the uncertainty distribution parameters of photovoltaic output and load in each season and time period; Step 1.4: Use the Latin hypercube sampling method to sample the Beta distribution model and the normal distribution model in Step 1.3 respectively, to generate the original photovoltaic output scenario set and the original load scenario set for each typical season; Step 1.5: Use the K-means clustering algorithm to reduce the original photovoltaic power output scenario set and the original load scenario set for each typical season in Step 1.4 to obtain the power output curves of four sets of seasonal source and load typical daily scenarios in spring, summer, autumn and winter.

3. The two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration as described in claim 2, characterized in that: Step 1.1 specifically includes: Acquire annual light intensity and load data for the study area, and divide the annual study period into four typical seasons: spring, summer, autumn, and winter, denoted as _____. Let the first The number of days corresponding to each typical season is The weight of the number of days in this typical season is: （1）； In the formula: Indicates the first s Weighting of the number of days in a typical season; Indicates the first The number of days included in a typical season; They represent the four typical seasons: spring, summer, autumn, and winter.

4. The two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration as described in claim 2, characterized in that: Step 1.2 specifically includes: Step ①: Place the first Heavenly Light intensity during the period Irradiance tested according to standards After normalization, the original per-unit photovoltaic output value is obtained: （2）； In the formula: Indicates the first Heavenly The original per-unit value of photovoltaic output for the time period; Indicates the first Heavenly Light intensity during a given time period; Indicates standard test irradiance; Step 2: Limit and correct the original photovoltaic output per unit value: （3）； In the formula: This indicates the upper limit of the per-unit value for photovoltaic power output; Step 3: Place the first Heavenly Raw active load for the time period Based on the maximum active power load throughout the year Normalization is performed: （4）； In the formula: Indicates the first Heavenly Per-unit value of active load for a given time period; Indicates the first Heavenly The original active load for the time period; This indicates the maximum active power load for the entire year.

5. The two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration as described in claim 2, characterized in that: Step 1.3 specifically includes: Step ①: For the first The first typical season A Beta distribution model is established for the photovoltaic power output samples during the time period, and its probability density function is: （5）； In the formula: This indicates the per-unit value of photovoltaic power output; and They represent the first The first typical season Two shape parameters of the Beta distribution over a given time period; Represents the Gamma function. Indicates the per-unit value of photovoltaic power output The power function term has an exponent of . , used to characterize the distribution pattern of the probability density function of the Beta distribution near the low output range; The power function term representing the difference between the per-unit value of photovoltaic power output and its upper limit is used to characterize the distribution pattern near the high-output range; Indicates the first The first typical season Per unit value of photovoltaic output during the time period The probability density function corresponding to a Beta distribution; Step 2: According to the first The first typical season Sample mean of photovoltaic power output during the time period and variance Calculate the intermediate variables for the Beta distribution parameters: （6）； In the formula: Indicates the first The first typical season Intermediate variables in the calculation of Beta distribution parameters over a given time period; This represents the sample mean of photovoltaic power output; This represents the sample standard deviation of photovoltaic power output; Step 3: When When the shape parameter of the Beta distribution is calculated, the following formula is used: （7）； （8）； when season: （9）； In the formula: This represents the lower limit of the shape parameter of the Beta distribution; and These represent the two shape parameters of the Beta distribution; Step 4: For the first The first typical season A normal distribution model is established for the active load samples during a time period, and its probability density function is: （10）； In the formula: Indicates the first The first typical season Per-unit value of active load during the time period The probability density function corresponding to a normal distribution; This represents the per-unit value of active power load; Indicates the first The first typical season The average active load sample for a given time period; Indicates the first The first typical season Standard deviation of active load sample for a given time period; Step 5: To avoid excessively small standard deviations in the loading samples leading to subsequent sampling anomalies, a lower limit correction is applied to the standard deviation of the loadings: （11）； In the formula Indicates the first The first typical season The original standard deviation of the active load sample for each time period; This represents the lower limit of the standard deviation of active power load; Step 6: Convert the active load to reactive load based on the equivalent power factor angle: （12）； In the formula: Indicates the first The first typical season Reactive load during a given period; Indicates the first The first typical season Active load during a given time period; The power factor angle; This represents the ratio of reactive load to active load.

6. The two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration as described in claim 2, characterized in that: Step 1.4 specifically includes: Step ①: Let the Latin hypercube sampling size be... The number of time periods within a day is Then the first ( ) the first time period ( The uniform sampling value corresponding to ) samples is: （13）； In the formula: Indicates the first The first time period The uniformly sampled value of a sample; Indicates the size of the Latin hypercube sampling; Indicates the number of time periods within a day; Indicates the first The random sequence number corresponding to each time period; express Random numbers within the interval; Step ②: Generate the first step based on the inverse cumulative distribution function of the Beta distribution model. The first typical season The first time period Sampling values ​​of photovoltaic output for each sample: （14）； In the formula: Indicates the first The first typical season The first time period The sampled photovoltaic output value of each sample; The inverse cumulative distribution function of the Beta distribution; Indicates the first The first time period The uniformly sampled value of a sample; and They represent the first The first typical season Shape parameters of the Beta distribution over a given time period; The first The photovoltaic output sample values ​​of the same sample in each time period under a typical season are combined in chronological order to obtain the first... Original photovoltaic output curves for each sample: ； In the formula: Indicates the first Under a typical season Original photovoltaic output curves for each sample; , , They represent the first Under a typical season The samples were in the 1st, 2nd, and 3rd... T Sample values ​​of photovoltaic power output for each time period; Furthermore, when At that time, the first A collection of original photovoltaic power output scenarios under a typical season: ； In the formula: Indicates the first A collection of original photovoltaic power output scenarios under a typical season; , , They represent the first Articles 1, 2, and 3 under a typical season N A graph showing the original scene of photovoltaic power output; Step ③: Generate the inverse cumulative distribution function of the load normal distribution model. The first typical season The first time period Initial sampling load values ​​for each sample: （15）； In the formula: Indicates the first The first typical season The first time period Initial sampling load values ​​for each sample; The inverse cumulative distribution function of the normal distribution; Indicates the first The first time period The uniformly sampled value of a sample; and They represent the first The first typical season The mean and standard deviation of active load samples over a given time period; Step 4: Correct the initial load sampling values ​​to obtain the corrected active load sample values: （16）； In the formula: This represents the corrected load sample value; Indicates the first The first typical season The first time period Initial active load sampling values ​​for each sample without lower bound correction; Indicates the lower limit of the load sampling value; The first The same sample number under a typical season The active load sample values ​​for each time period are combined in chronological order to obtain the first... Original load scenario curves for each sample: ； In the formula: Indicates the first Under a typical season The original scenario load curve for each sample; Furthermore, when At that time, the first A collection of original load scenarios under typical seasons: ； In the formula: Indicates the first A collection of original load scenarios under a typical season; , , They represent the first Articles 1, 2, and 3 under a typical season N The original load scenario curve.

7. The two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration as described in claim 2, characterized in that: Step 1.5 specifically includes: Step 1: The set of samples to be clustered Perform K-means clustering, where each sample For a containing Scene vectors for each time period; Among them, the sample set to be clustered The original photovoltaic power output scenario set obtained from step 1.4 Or load the original scene set Composition; when clustering the original set of photovoltaic power output scenarios, Indicates the first Under a typical season A raw photovoltaic output curve; when clustering the raw load scenario set... Indicates the first Under a typical season Original load scenario curve; This represents the first original scene curve in the set of samples to be clustered; This represents the second original scene curve in the set of samples to be clustered; Describes the first sample in the set to be clustered. Original scene curves; each All include Daily scene vectors for each time period; The objective function for K-means clustering is: （17）； In the formula: Represented by the sum of squared clustering errors Minimum is the objective; This represents the sum of squared clustering errors; Represents the number of clusters; Indicates the first A collection of scene types; Indicates the first Scene-centric; Indicates the first One clustering scenario sample; Step 2: Determine the sample based on the principle of minimum distance. Cluster category: （18）； In the formula: Indicates sample The cluster category to which it belongs; Represents the number of clusters; Indicates the first Scene-centric; Step 3: The scene center is updated according to the following formula: （19）； In the formula: Indicates the first The number of samples in the class scene set; Step 4: For the first... Clustering is performed on the original sets of photovoltaic output and load scenarios for each typical season, and the cluster centers are used as the typical daily photovoltaic output curve and typical daily load curve for that season: （20）； （21）； In the formula: Indicates the first The first typical season t Typical solar power output during a given period; Indicates the first The first typical season t Typical daily load capacity for each period; Indicates the first The photovoltaic output scenario cluster center in a typical season is in the first t The values ​​for each time period; Indicates the first The cluster centers of load scenarios under typical seasons are in the first... t The values ​​for each time period.

8. The two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration as described in claim 1, characterized in that: Step 2 specifically includes: Step ①: The upper-level planning model aims to minimize the system's annual comprehensive cost. The objective function is as follows: （22）； In the formula: This represents the objective function value of the upper-level planning model, i.e., the system's annual comprehensive cost; This indicates the system access and construction cost; This represents the overall operating cost of the system; This indicates the annualized investment cost of VSC; This indicates the annualized investment cost of ESS; This indicates the operating loss cost of VSC; This indicates the operating cost of the ESS during charging and discharging; Indicates network loss cost; This indicates the cost of purchasing electricity from the upstream power grid; Indicates the cost of curtailment; The methods for calculating each cost are as follows: 1) VSC Investment Costs （23）； In the formula: The investment cost of VSC; r The discount rate; For the entire lifecycle of VSC; For the VSC candidate installation node set; The unit capacity investment cost of VSC; For nodes i The installation capacity of the VSC; 2) ESS Investment Costs （24）； In the formula: The investment cost of ESS; r The discount rate; For the entire life cycle of ESS; For the ESS candidate installation node set; , These are the investment costs per unit energy capacity and per unit power capacity of the ESS, respectively. , They are nodes j The rated energy capacity and rated power capacity of the ESS; 3) VSC power loss cost （25）； In the formula: Cost of VSC power loss; for t Real-time electricity pricing; T represents the number of typical seasonal periods; D represents the number of typical daily time periods; s For the season s Total number of days; VSC active transmission loss coefficient; For nodes VSC at the The first typical season Active power during each time period; The unit time step is 1. h ; 4) ESS operating costs （26）； In the formula: For ESS operating costs; Cost per unit charge / discharge operation for ESS; , They are nodes ESS at the The first typical season Charging power and discharging power for each time period; 5) Network loss cost （27）； In the formula: , These are respectively the AC branch set and the DC branch set; branch road ij The resistance value; For the first s The first typical season t The flow through the branch road during each period ij The magnitude of the current; 6) Cost of purchasing electricity from higher authorities （28）； In the formula: To reduce the cost of purchasing electricity from higher authorities; For the first The first typical season Power purchased from the upper-level power grid during each time period; 7) Cost of abandoning light （29）； In the formula: Cost of abandoning light; Represents the set of photovoltaic nodes; This represents the unit cost of light skipping penalty; , The first The photovoltaic node at the ... The first typical season Forecasted output and actual absorption output for each time period; Step 2: Construct the constraints for the upper-level planning model: 1) Constraints on the number of VSC and ESS candidate installation nodes; （30）； In the formula: For nodes i Number of VSC installations; For nodes j Number of ESS installations; 2) VSC installation capacity constraints （31）； In the formula: Represents a node The installation capacity of the VSC; and These represent the lower and upper limits of the installation capacity of a single VSC unit, respectively. Represents a node i The number of VSC installations; 3) ESS installation capacity constraints （32）； In the formula: and Representing nodes respectively The rated energy capacity and rated power capacity of the ESS; , These represent the lower and upper limits of the energy capacity of a single ESS unit, respectively. , These represent the lower and upper limits of the power capacity of a single ESS unit, respectively. Represents a node The number of ESS installations.

9. The two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration as described in claim 1, characterized in that: Step 3 specifically includes: Step 1: Construct the objective function of the lower-level operating model, with the lower level based on the overall system operating cost. Minimum target: （33）； In the formula: This indicates that the lower-level operating model is based on the overall system operating cost. The minimum is the target; Step 2: Construct the constraints for the lower-level runtime model: 1) DistFlow constraints in AC power grids （34）； In the formula: Represented by node The set of the starting nodes of the branches of the terminal nodes; Represented by node It is the set of branch end nodes of the first node; Represents the set of AC branches; , They represent the first The first typical season Communication branch at each time period The active and reactive power transmitted upstream; , They represent AC branches respectively. Resistance and reactance; , Representing communication nodes Sum of squared voltages in AC branches Square of current; Indicates communication node The square of the voltage at the point; Indicates a branch The reactance; （35）； In the formula: , Representing communication nodes Net active and reactive loads; , Representing communication nodes Original active and reactive loads; , These represent the ESS charging power and discharging power, respectively. This indicates the actual power output of photovoltaic power generation. This indicates the parameters for the main network access node. When the node... j When it is a substation node, Otherwise, it is 0; , The first The first typical season During each time period, the upper-level power grid sends to the node j Injected active and reactive power; 2) DC power flow constraints （36）； （37）； In the formula: Represents the set of DC branches; Indicates the first The first typical season DC branch during each time period Active power transmitted upstream; Indicates DC branch The resistance; , Representing DC nodes Sum of squared voltages in DC branch Square of current; Indicates DC node Net active load at the location; Indicates DC node Active load at point; , These represent the ESS charging power and discharging power, respectively. This indicates the actual photovoltaic power consumption; if the node If no ESS or PV is connected at a point, the corresponding item is set to zero; reactive power is not considered for DC nodes. Indicates DC node The square of the voltage at the point; 3) VSC runtime constraints At the medium voltage level, this paper adopts a single-pole symmetrical VSC structure to transform AC and DC lines; during the transformation process, one VSC in the DC line must be in Vdc-Q control mode. （38）； In the formula: , They represent the first The VSC in the first The first typical season Active and reactive power on the AC side during each time period; Indicates the first The active power output of each VSC on the DC side; Indicates the first The reactive power control quantity of each VSC; , They represent the first The resistance and reactance of the VSC equivalent circuit; Indicates the flow through the first Current of the VSC equivalent circuit; , They represent the first Lower and upper limits of reactive power control for each VSC; Indicates the first The installation capacity of each VSC; Indicates the first Internal potential of each VSC; Indicates the first VSC DC-side voltage; This represents the voltage conversion factor under SVPWM modulation; Indicates DC voltage utilization rate; Indicates the first The modulation level of each VSC has a value range of [0,1]. 4) ESS running constraints When combined with time-of-use pricing, ESS can play a role in "peak shaving and valley filling" and can also improve the absorption capacity of PV. （39）； In the formula: Represents a node ESS at the The first typical season Net charge / discharge power for each time period; , These represent the ESS charging power and discharging power, respectively. express ESS rated power capacity; , These represent the charging and discharging state variables of the ESS, respectively, and are 0-1 variables; Indicates that ESS is in the The first typical season The amount of stored electricity in each time period; Represents a node ESS rated energy capacity; , These represent the ESS charging efficiency and discharging efficiency, respectively. Indicates that ESS is in the The first typical season The amount of stored electricity in each time period; Indicates that ESS is in the The first typical season The amount of stored electricity in each time period; 5) PV operating constraints （40）； In the formula: Represents the set of photovoltaic nodes; Indicates the first The photovoltaic node at the ... The first typical season The actual absorption capacity in each time period; Indicates the first The photovoltaic node at the ... The first typical season Forecast output for each time period; 6) Power constraints for interaction with the main network The power interaction constraints between substation nodes and the upstream power grid are as follows: （41）； In the formula: This represents the set of substation nodes connected to the upper-level power grid. , They represent the first The first typical season Substation nodes during each time period Active and reactive power interacting with the upstream power grid; , These represent the lower limit and upper limit of active power for main network interaction, respectively; , These represent the lower limit and upper limit of reactive power for main grid interaction, respectively; 7) Safety constraints Safety constraints include upper and lower limits for voltage and current, as well as transmission power constraints: （42）； In the formula: , These represent the set of AC nodes and the set of DC nodes, respectively. , These represent the AC branch set and the DC branch set, respectively. , Representing nodes respectively Lower and upper limits of voltage amplitude; Indicates a branch The upper limit of the current amplitude; , They represent the first The first typical season Communication branch at each time period The active and reactive power transmitted upstream; Indicates the first The first typical season DC branch during each time period Active power transmitted upstream; , These represent the maximum transmission capacity of the AC branch and the DC branch, respectively. For the first s The first typical season t The flow through the branch road during each period ij The magnitude of the current.

10. The two-layer planning method for AC / DC hybrid distribution networks based on VSC and ESS collaboration according to claim 1, characterized in that: Step 4 specifically includes the following steps: Step ①: Input the annual solar load dataset, network parameters, VSC, and initial values ​​of ESS planning parameters; Step 2: Use a genetic algorithm with an elitist strategy to solve the upper-level planning model, setting the population size, maximum number of iterations, crossover probability, and mutation probability. Perform mixed encoding on the candidate installation nodes and capacities of VSC and ESS, and randomly generate the initial configuration population. Let the population size be The maximum number of iterations is The crossover probability is The mutation probability is The number of elite individuals retained is ;No. The generation population is represented as: （43）； In the formula: Indicates the first Generational population; Indicates population size; , , They represent the first The first, second, and third offspring in the first generation population Individual; Indicates population size; Each individual node uses a hybrid encoding of "candidate node installation status - capacity parameters", with the encoding format as follows: （44）； In the formula: Indicates candidate nodes The decision variable for whether or not to install VSC; Indicates candidate nodes The installation capacity of the VSC; Indicates candidate nodes The decision variable for whether or not to install ESS; and Representing candidate nodes respectively The rated energy capacity and rated power capacity of the ESS; if Then the node VSC is not installed there; if Then the node ESS is not installed at this location; Step 3: Pass the configuration scheme of each individual in the upper-level population as a known constraint to the lower-level operating model. With the goal of minimizing the overall system operating cost, derive the optimal overall operating cost and operating state under this configuration scheme, and return the cost to the upper level to form the main problem objective function: For the first The first generation of the population individual Decode the VSC and ESS configuration schemes: （45）； In the formula: Represents the individual decoding function; Indicates the first VSC installation location and capacity plan for each individual; Indicates the first The installation location, rated energy capacity, and rated power capacity scheme for each individual ESS; In the given and Under the given conditions, solve the lower-level operational model: （46）； In the formula: Indicates the first The overall operating cost of the optimal lower-level system corresponding to each upper-level configuration scheme; This represents the corresponding set of optimal operating decision variables, including VSC operating power, ESS charging and discharging power, ESS power status, actual photovoltaic power absorption, branch power flow, node voltage, and main grid interaction power. This represents the feasible domain that satisfies the underlying operational constraints under a given VSC and ESS configuration scheme. This represents the set of operational decision variables in the lower-level operational model; Based on the optimal operating cost feedback from the lower layer, calculate the first... The overall annual system cost corresponding to each upper-level configuration scheme: （47）； In the formula: Indicates the first The system's annual comprehensive cost corresponding to each individual; Indicates the first Annualized investment costs for VSC and ESS for each individual; This represents the optimal overall operating cost of the system based on feedback from the lower layer. Step 4: Calculate the fitness value of each individual in the population, rank the individuals in the population according to the calculated fitness value; use the elite retention strategy to enter the next generation with the best fitness of the current generation, and perform elite retention, crossover and mutation operations on the remaining population to generate a new offspring population. Since the upper-level planning model is a cost minimization problem, the first... The fitness of each individual is: （48）； In the formula: Indicates the first The fitness of each individual; This represents the fitness correction parameter, used to avoid a denominator of 0; The first The most fittest in the generation population Individuals as a collection of elite individuals: （49）； In the formula: Indicates the first A collection of elite individuals within a generational population; This means selecting the most fit population from the current population. Individual; This indicates the number of individuals retained by the elite. Selection, crossover, and mutation operations are performed on non-elite individuals to generate offspring populations, which then form the next generation population together with elite individuals. （50）； In the formula: Indicates the first Generational population; This represents the set of offspring individuals generated by selection, crossover, and mutation operations; Indicates the crossover probability; Indicates the proportion of variation; Indicates the probability of univariate mutation; Indicates the variable-asynchronous length coefficient; Step 5: Repeat steps 2 to 4 to perform alternating optimization iterations between upper and lower layers until the algorithm reaches the set maximum number of iterations or meets the convergence condition; terminate the iteration and output the individual with the best global fitness, which is the optimal AC / DC hybrid distribution network planning scheme.

Images