A power distribution network energy storage double-layer scene random optimization configuration method and system for photovoltaic accommodation capacity improvement

Abstract

The application discloses a power distribution network energy storage double-layer scene random optimization configuration method and system for improving photovoltaic accommodation capacity, and belongs to the technical field of power distribution network planning and new energy consumption. The method firstly constructs four types of seasonal operation scenes, and performs annual weighting according to representative days in each season; secondly, candidate energy storage nodes are screened in combination with a load center, a photovoltaic access position, voltage weakness and network end topology characteristics; then, an upper energy storage planning model and a lower operation optimization model are established, power purchase savings, light abandonment reduction, network loss reduction, expansion delay income and life loss cost are counted, and annualized results are fed back to the upper layer; meanwhile, a strict feasibility discrimination mechanism based on voltage relaxation and line capacity relaxation is set to ensure that the planning scheme takes into account economy and network security. The application can improve photovoltaic consumption capacity, reduce power distribution network operation risk and improve the engineering implementability of energy storage planning results.

Description

Technical Field

[0001] This invention relates to the fields of distribution network energy storage planning and new energy consumption technology, specifically to a random optimization configuration method for a two-layer distribution network energy storage scenario aimed at improving photovoltaic acceptance capacity. Background Technology

[0002] As the scale of distributed photovoltaic (PV) grid integration continues to expand, the problem of source-load mismatch in time and space is becoming increasingly prominent. During the midday period, PV power output is concentrated, which can easily cause local voltage rises, reverse power flow to the main transformer, and branch overloads. During the evening peak period, PV power output drops rapidly, which can lead to a sharp increase in external power purchases, increasing the peak-valley difference in the distribution network and the pressure to expand capacity.

[0003] Most existing energy storage configuration methods are based on a single typical day or static load level, which fails to fully reflect the differences in peak load, photovoltaic output level and network operation boundary between seasons, resulting in biases in the annualized economic evaluation.

[0004] While some methods take economic considerations into account, they fail to adequately characterize the power flow, voltage, and line capacity constraints of the distribution network after energy storage is integrated. They often simplify the impact of network operation to a node power balance problem, making it difficult to guarantee the feasibility of the results in engineering.

[0005] Other methods use penalty functions instead of strict safety constraints, but do not separately check voltage relaxation and line over-limit relaxation. This means that although the resulting planning scheme may appear to have higher benefits, it may still have safety risks in actual operation.

[0006] Therefore, there is an urgent need to propose a two-layer optimization configuration method for energy storage that can simultaneously consider seasonal scenario differences, distribution network operation safety, full life cycle economics, and strict feasibility requirements, so as to improve photovoltaic acceptance capacity and enhance the reliability of planning results. Summary of the Invention

[0007] The technical problem this invention aims to solve is to address the shortcomings of existing distribution network energy storage configuration methods, which struggle to simultaneously consider photovoltaic (PV) absorption capacity, network security constraints, and life-cycle economic efficiency. This invention proposes a two-layer stochastic optimization configuration method for distribution network energy storage, geared towards enhancing PV absorption capacity. This method is based on seasonal scenario construction, with energy storage site selection and capacity determination as its core, and annualized feedback of operational optimization results as the link, forming a closed-loop solution framework to obtain the optimal energy storage configuration scheme.

[0008] To solve the above technical problems, the present invention adopts the following technical solution:

[0009] In a first aspect, this invention proposes a stochastic optimization configuration method for a two-layer distribution network energy storage scenario aimed at improving photovoltaic acceptance capacity, comprising the following steps:

[0010] First, four seasonal operating scenarios—spring, summer, autumn, and winter—are constructed, and load curves, reactive power load curves, and photovoltaic output curves are generated for each season. The operating results for representative days in each season are converted to an annual calculation based on the number of representative days. Under the current implementation guidelines, all four seasonal scenarios are counted as 90 days for the entire year. For any revenue or cost item, its annualized result can be expressed as:

[0011] ,

[0012] Where Ω_s represents the set of seasonal scenarios, D_s represents the number of representative days corresponding to the s-th seasonal scenario, and X_s represents the daily revenue or daily cost under that seasonal scenario.

[0013] Secondly, a candidate node set for energy storage is constructed based on load center nodes, photovoltaic access nodes, voltage-weak nodes, and network end nodes, thereby narrowing the optimization search space and improving solution efficiency. After the candidate nodes are determined, an upper-level energy storage planning model is established. The upper-level model uses energy storage construction variables, rated power variables, and rated capacity variables as decision variables, and takes maximizing the net present value over the entire life cycle as the optimization objective. Its objective function can be expressed as:

[0014] ,

[0015] in, For initial investment costs, For the total return in year y, The maintenance cost for year y is... Let be the depreciation cost in year y, r be the discount rate, and Y be the project lifespan. Residual value A penalty imposed on planning schemes that do not meet strict feasibility requirements.

[0016] Furthermore, the rated capacity, rated power, and configuration duration of energy storage satisfy the following relationship:

[0017] ,

[0018] in, Let i be the rated capacity of the energy storage device at node i. Let i be the rated power of the energy storage device at node i. Configure the duration for energy storage.

[0019] Given the upper-level planning scheme, a lower-level operation optimization model is established. The lower-level model uses energy storage charging and discharging power, state of charge (SOC), curtailed solar power, branch power flow, and node voltage as operating variables to optimize the operation of the distribution network under various seasonal scenarios. The SOC of energy storage satisfies the following recursive relationship:

[0020] ,

[0021] in, Let i be the state of charge of the energy stored at node i during time period t. and These are charging power and discharging power, respectively. and These are charging efficiency and discharging efficiency, respectively. This represents the length of the time period.

[0022] The lower-level model uses DistFlow-based power flow relationships and second-order cone constraints to describe the operating boundaries of distribution network branches, where the branch capacity constraints can be expressed as:

[0023] ,

[0024] in, and Let be the active power and reactive power of branch ij in time period t, respectively. Let be the square of the voltage at node i during time period t. Let be the square of the current in branch ij during time period t. This constraint couples the energy storage access location, charging and discharging behavior with node voltage and line capacity boundaries.

[0025] In terms of economic evaluation, this invention incorporates the benefits of electricity purchase savings, reduced curtailment of solar power, reduced grid losses, delayed capacity expansion, and depreciation costs into the annualized benefit calculation. The comprehensive annualized benefit can be expressed as:

[0026] ,

[0027] in, This indicates the savings from electricity purchases. This indicates that abandoning light reduces revenue. This indicates that network losses reduce revenue. This indicates a delay in the benefits of expansion. This indicates the cost of energy storage lifespan reduction.

[0028] After the lower-level operational model is solved, a strict feasibility judgment mechanism is further set up to check the maximum voltage relaxation and maximum line capacity relaxation in each scenario. The current energy storage planning scheme is deemed to meet the strict feasibility requirements if and only if both do not exceed the preset tolerance. Its judgment form can be expressed as:

[0029] strictFeasible = 1, when and Otherwise, it is 0.

[0030] Secondly, the present invention provides an electronic system including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0031] In addition, the present invention also proposes a computer program product, including a computer program / instruction, wherein the computer program / instruction is executed by a processor using the method proposed in the present invention.

[0032] This invention achieves a shift from "solvable" to "feasible" energy storage configuration schemes through the coordinated efforts of upper-level planning variable search, lower-level operational optimization solution, annualized return feedback, and rigorous feasibility assessment. This method integrates seasonal scenario construction, candidate node selection, upper-level planning, lower-level operational optimization, annualized return calculation, and rigorous feasibility assessment, and can be used for energy storage site selection and capacity determination in high-penetration photovoltaic distribution networks.

[0033] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0034] (1) By constructing four types of seasonal scenarios and weighting them throughout the year, the source load change characteristics and energy storage value differences under different seasons can be more accurately reflected.

[0035] (2) By establishing a two-layer optimization model that couples upper-layer site selection and capacity determination with lower-layer operation safety, the energy storage configuration results can take into account both economic efficiency and network security.

[0036] (3) By introducing multi-dimensional benefit and cost items such as electricity purchase savings, reduced curtailment of solar power, reduced grid loss, delayed capacity expansion and lifespan reduction, the completeness of planning evaluation has been improved.

[0037] (4) By setting up a strict feasibility judgment mechanism, pseudo-optimal solutions that have high returns but do not meet the engineering operation boundary can be effectively avoided, thereby improving the reliability of the results.

[0038] (5) This invention can be widely applied to energy storage optimization configuration scenarios of park distribution networks, regional distribution networks and typical test systems with a high proportion of distributed photovoltaic access. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a flowchart of the overall process for the random optimization configuration method for a two-layer energy storage scenario as described in this invention.

[0041] Figure 2This is a data interaction structure diagram of the upper-level planning, lower-level operation, and result feedback of this invention.

[0042] Figure 3 This is a flowchart of the annualized return feedback and rigorous feasibility assessment process for this invention.

[0043] Figure 4 This is a graph showing the system load and photovoltaic output under four seasons.

[0044] Figure 5 This is a diagram showing the combined changes in energy storage power, state of charge, and node voltage. Detailed Implementation

[0045] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but the scope of protection of the present invention is not limited to the following embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0046] Example 1: This example proposes a stochastic optimization configuration method for a two-layer distribution network energy storage scenario aimed at improving photovoltaic (PV) integration capacity. (Refer to...) Figure 1 As shown, it includes the following steps:

[0047] S1. Construct four types of seasonal operation scenarios, obtain the load curve, reactive load curve and photovoltaic output curve under each scenario, and perform annualized weighting according to the number of representative days of each season.

[0048] S2. Select candidate energy storage nodes based on the load level of distribution network nodes, photovoltaic access location, voltage weakness, and network end topology characteristics;

[0049] S3. Establish an upper-level planning model, encode whether energy storage sites should be built, their rated power and rated capacity, and use an improved particle swarm optimization algorithm to search for candidate solutions.

[0050] S4. Establish a lower-level operation model and, under the given upper-level planning scheme, perform mixed integer second-order cone optimization to solve for energy storage charging and discharging power, state of charge, curtailed power, branch power flow, and node voltage.

[0051] S5. The operating results obtained from each seasonal scenario are converted into annual benefits to obtain benefits such as electricity purchase savings, reduced curtailment of solar power, reduced grid losses, delayed capacity expansion, and lifespan depreciation costs.

[0052] S6. Perform a rigorous feasibility assessment based on the voltage relaxation and line capacity relaxation, and then feed the annualized revenue results and feasibility results back to the upper-level model.

[0053] S7. Output the optimal energy storage configuration scheme based on maximizing the net present value over the entire life cycle.

[0054] like Figure 2 As shown, the implementation scheme provided by this invention includes an upper-level planning-lower-level operation-feedback interaction structure comprising an upper-level planning module, a lower-level operation optimization module, an annualized benefit calculation module, and a rigorous feasibility judgment module. The upper-level planning module outputs the energy storage access node, rated power, and rated capacity. The lower-level operation optimization module calculates the energy storage charging and discharging, curtailment, power flow, and voltage results for each seasonal scenario based on the planning scheme, and feeds back the annualized benefit, lifetime depreciation cost, and feasibility judgment results to the upper level.

[0055] like Figure 3 As shown, the annualized return feedback and strict feasibility assessment process includes statistical analysis of seasonal scenario operation results, weighted average of representative days, summary of revenue and costs, verification of maximum voltage relaxation, and verification of maximum line capacity relaxation. Only when both voltage relaxation and line capacity relaxation do the corresponding planning schemes are marked as strictly feasible and participate in the upper-level optimal scheme update.

[0056] like Figure 4 As shown, the system load and photovoltaic output curves for the four seasons are used to characterize the differences in source load changes in spring, summer, autumn, and winter. The load curve reflects the temporal changes in electricity demand in each season, while the photovoltaic output curve reflects the differences in the power output of distributed photovoltaic systems under different seasonal illumination conditions. These curves together serve as the scenario input for the lower-level operation optimization model.

[0057] refer to Figure 5 As shown, a graph illustrating the combined changes in energy storage power, state of charge (SBC), and node voltage is used to explain the impact of energy storage operation on the distribution network. When energy storage charges during periods of surplus or low photovoltaic power, its SBC increases and absorbs local surplus power; when energy storage discharges during periods of peak load or low voltage, its SBC decreases and supports node voltage, thus achieving a synergistic effect of photovoltaic power absorption, peak shaving and valley filling, and voltage improvement.

[0058] This embodiment uses an IEEE 33-node distribution network as an example to construct a verification system. Based on node and branch parameters, the system conducts research on energy storage optimization configuration by combining the load and photovoltaic output curves of representative days for four seasonal scenarios. Each representative day is discretized into 48 time periods, with a time scale of 0.5 hours. The four seasonal scenarios (spring, summer, autumn, and winter) are all included in the annual weighted average as 90 days.

[0059] In this embodiment, the basic network data of the distribution network is first read to establish the parent-child relationship of nodes, branch impedance parameters, reference capacity and voltage boundary; then, the active load, reactive load and photovoltaic output matrices for four seasonal scenarios of spring, summer, autumn and winter are generated as inputs to the lower-level operation model.

[0060] During the candidate node selection process, identification was conducted from four dimensions: node base load, known photovoltaic access location, voltage weakness under baseline conditions, and deep topology at the end. Load center nodes are used to demonstrate the value of energy storage in peak shaving and valley filling; key photovoltaic nodes are used to demonstrate the value of energy storage in local consumption; and voltage-weak nodes and end nodes are used to enhance voltage support and alleviate branch overload.

[0061] The upper-level planning model is solved using an improved particle swarm optimization (PSO) algorithm. This improved PSO algorithm includes a decreasing inertia weight mechanism, a mutation perturbation mechanism, and a stagnation-early termination mechanism. After the search, it performs expansion repair, weak node replacement, and rigorous feasibility refinement on the current optimal planning scheme. For each particle, the construction length is 3... Encoding vector: front The component represents the website's score, with the middle component being... Each component represents the original value of the rated power corresponding to the candidate node, and then... Each component represents the original configuration duration value corresponding to the candidate node. After threshold mapping and boundary repair, the site construction decision, rated power, and rated capacity are obtained.

[0062] In this embodiment, the candidate node set is obtained by taking the union of load center nodes, photovoltaic key nodes, voltage-weak nodes, and deep end nodes. Taking the IEEE 33-node example, candidate nodes may include nodes 18, 33, 17, 16, 15, 14, 13, 32, 25, 30, 8, 7, and 24. During particle encoding, the preceding... Each component serves as a site construction score for each candidate node; a higher score indicates a greater likelihood that the node will be selected as an energy storage site. Each component serves as a candidate value for the rated power of each candidate node; subsequently... Each component serves as a candidate value for the configuration duration of each candidate node. This represents the number of candidate nodes.

[0063] During particle decoding, the station construction variables are first determined based on the station construction score threshold or sorting rules, and then the rated capacity is calculated based on the rated power and configuration duration. If the decoding result does not meet the boundaries of the number of stations, total power, total capacity, or configuration duration, the scheme repair is performed. For the optimal particle in this embodiment, the energy storage stations obtained after decoding are nodes 16, 17, 30, and 32, with a total rated power of 1800.00kW and a total rated capacity of 6137.52kWh, which are further submitted to the lower-level mixed integer second-order cone model for seasonal scenario operation verification.

[0064] ,

[0065] ,

[0066] When the number of stations, total power, or total capacity of a particle is lower than a preset lower bound, a fast screening penalty is applied to the particle; when the number of stations of a particle exceeds the upper limit, only nodes with higher scores are retained, and their power and capacity boundaries are simultaneously corrected, thereby reducing the probability of invalid solutions entering the lower-level optimization model.

[0067] Given the planning results from the upper layer, the lower-level operation model establishes energy storage operation and distribution network safety constraints for four seasonal scenarios. The energy storage operation component mainly includes charging power. Discharge power Charging status Discharge state and state of charge The details are as follows:

[0068] ,

[0069] ,

[0070] ,

[0071] ;

[0072] In the formula, i is the energy storage access node number; t is the discrete scheduling period number within a typical day; T is the last scheduling period of a typical day; Let be the rated power of the energy storage device at node i; The length of a single scheduling period; and These are the energy storage charging efficiency and discharging efficiency, respectively. The state of charge of the energy stored at node i in the previous time period; and These are the minimum and maximum allowable state of charge ratios for energy storage at node i, respectively; Let be the rated capacity of the energy storage device at node i; and Let be the state of charge of the energy stored at node i during the initial and final periods of a typical day, respectively. In the formula, ... Constrain the energy storage to maintain a consistent state of charge at the beginning and end of the day to ensure that typical daily operating results can represent a repeatable steady-state cycle process.

[0073] The main components of the power distribution network operation include curtailed solar power. Branch active power Branch reactive power Line current square and the square of the node voltage Specifically, it includes:

[0074] ,

[0075] ,

[0076] ,

[0077] ;

[0078] Where i, j, and k are the distribution network node numbers, branch ij represents the distribution line from node i to node j, and branch jk represents the distribution line from node j to its downstream node k; t is the discrete scheduling period number. This represents the set of downstream adjacent nodes starting from node j, which is the set of child nodes of node j. and These represent the active power and reactive power of branch jk within time period t, respectively. and These are the net active power demand and net reactive power demand of node j in time period t, respectively, which comprehensively reflect the equivalent power demand after the combined effect of the node's load, photovoltaic output, curtailed power and energy storage charging and discharging power. and Let be the resistance and reactance of branch ij, respectively; Let be the square of the voltage amplitude of node j during time period t.

[0079] In terms of economic evaluation, the lower level first calculates the daily electricity purchase cost, curtailment penalty cost, risk cost, and lifespan loss cost for each seasonal scenario, and then converts them into annual costs according to the representative number of days in each season. At the same time, by comparing the benchmark operating conditions with the energy storage configuration operating conditions, the annual electricity purchase savings, curtailment reduction benefits, grid loss reduction benefits, and delayed capacity expansion benefits can be obtained.

[0080] ,

[0081] ,

[0082] ,

[0083] ,

[0084] in, This indicates the savings from electricity purchases. This indicates that abandoning light reduces revenue. This indicates that network losses reduce revenue. This indicates a delay in the benefits of expansion. This indicates the cost of energy storage lifespan reduction. Delaying the expansion value per unit capacity; This refers to the initial investment cost; The maintenance cost for year y; The cost of life loss in year y; This is the residual value at the end of the useful life. A penalty imposed on planning schemes that fail to meet strict feasibility conditions.

[0085] The strict feasibility determination process is as follows: After the lower-level model is solved, the maximum voltage relaxation (maxVoltSlack) and the maximum line capacity relaxation (maxLineSlack) are calculated for each seasonal scenario throughout the year. If and only if maxVoltSlack is not greater than the preset voltage tolerance and maxLineSlack is not greater than the preset line tolerance, then strictFeasible is set to 1; otherwise, strictFeasible is set to 0. This mechanism effectively distinguishes between "nearly feasible solutions that can be tolerated by penalty terms" and "strictly feasible solutions that satisfy engineering safety boundaries."

[0086] strictFeasible={1, and ; 0, others}.

[0087] To improve the solution efficiency and the probability of obtaining a strictly feasible solution, this embodiment introduces the following mechanisms into the improved particle swarm optimization algorithm: First, an inertial weight that decreases with the number of iterations to balance global exploration in the early stage of the search with local convergence in the later stage; second, a small-probability mutation perturbation is applied after particle updates to avoid getting trapped in local optima; third, the search is terminated early if no improvement is found for several consecutive generations; and fourth, after the search is completed, the current optimal solution is expanded and repaired, weak nodes are filled, and strict feasibility refinement is performed.

[0088] One solution result of this embodiment shows that the optimal solution, which configures energy storage at 4 nodes, yields an annualized electricity purchase savings of 4,505,093.78 yuan, reduced curtailment of solar power of 268,557.93 yuan, reduced grid losses of 64,741.47 yuan, delayed capacity expansion of 612,812.85 yuan, and lifetime depreciation costs of 34,215.20 yuan. The net present value over the entire life cycle reaches 29,579,096.67 yuan, and the strictFeasible criterion result is 1, indicating that the solution simultaneously meets the requirements of economy and safety.

[0089] Table 1 Optimization Results

[0090] index result Optimal address node 16、17、30、32 Total rated power 1800.00 kW Total fixed capacity 6137.52 kWh Electricity purchase savings 4,505,093.78 yuan / year Wasting light reduces revenue 268,557.93 yuan / year Network loss reduces revenue 64741.47 yuan / year Delaying expansion benefits 612,812.85 yuan / year Lifespan depreciation cost 34215.20 yuan / year Net Present Value (NPV) over the entire life cycle 29,579,096.67 yuan Strict feasibility strictFeasible=1

[0091] Furthermore, in the example results, the total rated power of energy storage is approximately 1800.00 kW, and the total rated capacity of energy storage is approximately 6137.52 kWh. The selected sites are located at nodes 16, 17, 30, and 32. This demonstrates that the present invention can improve photovoltaic integration capacity while reducing peak power purchase pressure and operational risks in the distribution network through the synergistic optimization of energy storage space layout and capacity parameters.

[0092] Example 2: This example provides an electronic system including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the method proposed in the above example.

[0093] It should be noted that the electronic system can use terminal devices such as desktop computers, laptops, or cloud servers. Furthermore, terminal devices include, but are not limited to, processors and memory. For example, terminal devices can also include input / output devices, network access devices, and buses.

[0094] Furthermore, the processor can be a central processing unit (CPU). Of course, depending on the actual use, other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. can also be used. The general-purpose processor can be a microprocessor or any conventional processor, etc., and this application does not limit it in this regard.

[0095] Furthermore, the memory can be an internal storage unit of the terminal device, such as the hard disk or RAM of the terminal device, or an external storage device of the terminal device, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD), or flash memory card (FC) equipped on the terminal device. In addition, the memory can also be a combination of the internal storage unit and the external storage device of the terminal device. The memory is used to store computer programs and other programs and data required by the terminal device. The memory can also be used to temporarily store data that has been output or will be output. This application does not limit this.

[0096] Furthermore, through this electronic system, any one of the methods in the above embodiments can be stored in the memory of the electronic system and loaded and executed on the processor of the terminal device for convenient use.

[0097] Example 3: This example also discloses a computer program product, including a computer program / instructions, wherein when the computer program / instructions are executed by a processor, the steps of any of the methods in the above examples are adopted.

[0098] The computer program can be stored in a computer-readable medium. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or certain middleware. The computer-readable medium includes any entity or device capable of carrying computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the computer-readable medium includes, but is not limited to, the above-mentioned components.

[0099] This invention is not limited to the embodiments described above. For different regional distribution networks, scenario data and parameters can be replaced according to actual load, photovoltaic output, branch capacity, voltage boundary, energy storage cost, and expansion cost; without departing from the core idea of ​​this invention, equivalent changes to candidate node generation rules, revenue parameters, and solver settings should all fall within the protection scope of this invention.

Claims

1. A stochastic optimization configuration method for a two-layer energy storage system in a distribution network to enhance photovoltaic (PV) integration capabilities, characterized in that... Includes the following steps: S1. Construct four types of seasonal operation scenarios, obtain the load curve, reactive load curve and photovoltaic output curve under each scenario, and perform annualized weighting according to the number of representative days of each season. S2. Select candidate energy storage nodes based on the load level of distribution network nodes, photovoltaic access location, voltage weakness, and network end topology characteristics; S3. Establish an upper-level planning model, encode whether energy storage sites are to be built, their rated power and rated capacity, and use an improved particle swarm optimization algorithm to search for candidate solutions. The improved particle swarm optimization algorithm includes an inertia weight reduction mechanism, a mutation perturbation mechanism and a stagnation early termination mechanism, and performs capacity expansion repair, weak node supplementation and strict feasibility refinement on the current optimal planning solution after the search is completed. S4. Establish a lower-level operation model and, under the given upper-level planning scheme, perform mixed integer second-order cone optimization to solve for energy storage charging and discharging power, state of charge, curtailed power, branch power flow, and node voltage. S5. The operating results obtained from each seasonal scenario are converted into annual benefits to obtain benefits such as electricity purchase savings, reduced curtailment of solar power, reduced grid losses, delayed capacity expansion, and lifespan depreciation costs. S6. Perform a rigorous feasibility assessment based on the voltage relaxation and line capacity relaxation, and then feed the annualized revenue results and feasibility results back to the upper-level model. S7. Output the optimal energy storage configuration scheme based on maximizing the net present value over the entire life cycle.

2. The method as described in claim 1, characterized in that, In step S1, the four seasonal operating scenarios correspond to representative days for spring, summer, autumn and winter, respectively. The annual conversion value is obtained by multiplying the operating results of the representative days by the corresponding number of representative days.

3. The method as described in claim 1, characterized in that, In step S2, candidate nodes are obtained through multi-source feature fusion, and include at least: Select load center nodes from largest to smallest based on active power load; Select key photovoltaic nodes based on actual photovoltaic grid connection locations or high-penetration node sets; Under baseline operating conditions, weak nodes are selected based on the degree of low node voltage. Filter terminal deep nodes based on network hierarchy depth, path impedance, and leaf node labels; The union of the above node sets is taken as the set of candidate energy storage nodes.

4. The method as described in claim 1, characterized in that, In step S3, the upper-level planning model uses a length of 3 The particle vectors are encoded, where The number of candidate nodes, the first Wei is the website building score variable, and the middle Wei is a candidate value for rated power, and then The dimension is a candidate value for the configuration duration; Whether a website needs to be built is obtained through threshold mapping. and by rated power and configuration duration H i Calculate rated capacity ; Simultaneously, the solution is repaired by combining the boundaries of the number of stations, total power, and total capacity.

5. The method as described in claim 1, characterized in that, The lower-level operation model in step S4 includes the following constraints: upper limit constraint of energy storage charging and discharging power, mutual exclusion constraint of charging and discharging, recursive constraint of state of charge, upper and lower bound constraints of state of charge, and consistency constraint of state of charge in the first and last periods; active power balance constraint, reactive power balance constraint, voltage recursive constraint, and second-order cone constraint of branch capacity based on the DistFlow relationship of the distribution network; upper and lower limit constraints of node voltage, constraint of non-reverse power transmission from balancing nodes, and non-negative constraint of curtailed power.

6. The method as described in claim 1, characterized in that, The annual benefits in step S5 shall include at least: electricity purchase savings. Wasted light reduces revenue Network loss reduces revenue and delaying expansion benefits The annual cost should include at least the cost of depreciation over life. Operation and maintenance costs and initial investment costs; Among them, the benefits of delaying expansion This is obtained by multiplying the maximum peak electricity purchase reduction by the value per unit capacity.

7. The method as described in claim 1, characterized in that, In step S6, the strict feasibility determination adopts a component tolerance verification mechanism, which calculates the maximum voltage relaxation amount maxVoltSlack and the maximum line capacity relaxation amount maxLineSlack for each scenario throughout the year. The current planning scheme is determined to be a strictly feasible scheme only if maxVoltSlack is not greater than the preset voltage tolerance and maxLineSlack is not greater than the preset line tolerance.

8. The method as described in claim 1, characterized in that, In step S7, the total life-cycle net present value (NPV) is calculated using the following formula: ， In the formula, For initial investment costs, For the total return in year y, The maintenance cost for year y is... Let be the depreciation cost in year y, r be the discount rate, and Y be the project lifespan. Residual value A penalty imposed on planning schemes that do not meet strict feasibility requirements.

9. A computer program product comprising a computer program / instructions, characterized in that, When a computer program / instruction is executed by a processor, the method described in any one of claims 1 to 8 is employed.

10. An electronic system comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the method according to any one of claims 1 to 9.

Images