Three-phase unbalanced distribution network energy storage hierarchical configuration method and system considering multi-agent coordination

Abstract

This invention relates to the field of power system energy storage optimization configuration technology, and is a hierarchical configuration method and system for three-phase unbalanced distribution networks that considers multi-entity coordination. The method includes the following steps: establishing an accurate three-phase unbalanced distributed energy storage (DES) model; constructing a multi-entity differentiated objective function; wherein the multi-entity entities include the power grid company, distributed generation investors, and power users; designing a hierarchical coordination optimization solution framework based on the weight scheme of each entity's DES allocation and the multi-entity differentiated objective function, solving the accurate three-phase unbalanced DES model, and outputting an optimized configuration scheme; wherein the hierarchical coordination optimization solution framework includes an upper-level multi-entity game-theoretic coordination layer and a lower-level three-phase unbalanced physical optimization layer. A hierarchical architecture of "game-theoretic coordination-physical optimization" is proposed, where the upper level achieves interest coordination through multi-entity weight negotiation, and the lower level performs physical optimization based on the accurate three-phase unbalanced DES model, taking into account both economic efficiency and engineering feasibility.

Description

Technical Field

[0001] This invention relates to the field of power system energy storage optimization configuration technology, and in particular to a method and system for hierarchical configuration of energy storage in a three-phase unbalanced distribution network that takes into account the coordination of multiple entities. Background Technology

[0002] With the high proportion of renewable energy integration and the diversification of loads, distribution network operation exhibits significant three-phase imbalance characteristics. Distributed energy storage (DES), as a key flexible resource for improving power quality, promoting renewable energy consumption, and achieving peak shaving and valley filling, requires scientific allocation. Current research on DES configuration in distribution networks mainly faces the following problems: 1. Insufficient coordination of interests among stakeholders: Most existing configuration models are based on the economic interests of a single investment entity (such as power grid companies, power source owners or users), and have not established an effective mechanism for coordinating the interests of multiple parties, resulting in low acceptance of configuration schemes in actual promotion.

[0003] 2. Lack of three-phase imbalance modeling: Most studies are based on balanced network models, which ignore the asymmetry of three-phase parameters, load and distributed power output in actual distribution networks. The configuration results are difficult to effectively solve problems such as three-phase voltage exceeding limits and imbalance exceeding standards.

[0004] 3. Poor flexibility in allocation strategy: It generally adopts a "one-time" centralized allocation, without considering the timing and interactivity of investment decisions, making it difficult to adapt to the dynamic adjustment needs of different investment stages and different interests.

[0005] In the existing technology, some patent documents have involved methods for configuring energy storage in distribution networks. The "Multi-Objective Optimization Configuration Method and System for Distributed Energy Storage" (Patent Publication No. CN114254551A) considers multiple investment entities and constructs a mixed-integer nonlinear model for the power supply side, grid side, and user side. However, its model is based on the assumption of a balanced network and uses centralized optimization, failing to consider information privacy and independent decision-making rights among the entities. The "A Sequential Configuration Method for Distributed Energy Storage in Unbalanced Distribution Networks" (Patent Publication No. CN115048872A) proposes a sequential configuration strategy based on voltage sensitivity analysis, which can effectively avoid excessive local configuration. However, its core is still from the perspective of a single system operator, without involving the coordination of interests among multiple entities. Furthermore, its sensitivity analysis uses traditional active power-voltage sensitivity, failing to consider the impact of node voltage deviation on voltage regulation requirements. The patent publication number CN114614492A, titled "A Centralized-Distributed Energy Storage Location and Capacity Determination Method for Active Distribution Networks," uses the Analytic Hierarchy Process (AHP) and the Firefly Algorithm for partitioning configuration, but its partitioning is based solely on node distance and does not establish a clear cross-regional coordination mechanism.

[0006] In the application of game theory to power system optimization, the patent "Active Distribution Network Optimization Scheduling Method and System Considering Multi-Agent Cooperative Game" (Patent Publication No. CN120087714A) proposes a multi-agent cooperative game model based on Nash bargaining theory and introduces an allocation strategy based on distributed energy storage SOC equilibrium. However, its research object is optimization scheduling rather than energy storage planning, and it does not involve three-phase unbalanced network constraints. The patent "Optimization Method, System and Readable Storage Medium for Distribution Network Energy Storage System in Multiple Application Scenarios" (Patent Publication No. CN118281924A) proposes a multi-timescale framework of day-ahead distributed bar optimization, intraday rolling optimization and intraday real-time optimization, but its optimization objective focuses on a single operating entity and does not establish a multi-agent coordination mechanism.

[0007] Regarding sensitivity analysis, the patent publication CN115048872A, titled "A Method for Sequential Configuration of Distributed Energy Storage for Unbalanced Distribution Networks," employs a traditional active-voltage sensitivity matrix. However, this matrix only reflects the static sensitivity at a specific moment and cannot capture the temporal changes in the voltage regulation demand of the entire network. The patent publication CN110247393B, titled "A Method and System for Dynamic Partitioning and Optimized Configuration of Energy Storage Equipment in Active Distribution Systems," proposes defining node distance based on a voltage sensitivity factor and using a fast clustering algorithm based on peak density for dynamic partitioning. However, its sensitivity factor remains a static indicator and does not incorporate a time-series voltage offset coefficient.

[0008] In summary, existing technologies have not yet provided a method for optimizing distributed energy storage configuration that can simultaneously coordinate the interests of grid companies, distributed power generation investors, and power users under an accurate three-phase imbalance model, and that employs an improved sensitivity analysis and sequential configuration strategy based on time-series voltage offset coefficients.

[0009] In practical engineering, different stakeholders, such as power grid companies, photovoltaic / wind power investors, and industrial and commercial users, have different demands, making it difficult to implement simple technical optimization or centralized economic dispatch models. Therefore, there is an urgent need for a configuration method that can simultaneously meet the following requirements: 1) accurately characterize the operating characteristics of the three-phase unbalanced network at the physical level; 2) establish a model at the economic level that reflects the differentiated goals of multiple parties, including power grid companies (pursuing system safety and economy), power generators (pursuing smooth output and revenue), and users (pursuing electricity cost savings); 3) design a solution mechanism at the decision-making level that can achieve global coordination while respecting the independent decision-making rights and data privacy of each party.

[0010] Based on the shortcomings of existing technologies, this invention aims to propose a hierarchical coordinated optimization configuration method. Under an accurate three-phase imbalance model, by constructing a multi-agent game framework and a sequential decision-making process, the DES configuration scheme achieves equilibrium among multiple objectives such as overall economy, acceptance by all parties, and voltage quality improvement. Summary of the Invention

[0011] This invention provides a hierarchical configuration method and system for energy storage in a three-phase unbalanced distribution network that takes into account the coordination of multiple stakeholders. It overcomes the shortcomings of the prior art and can effectively solve the problem that the existing configuration model does not have a multi-party interest coordination mechanism, resulting in low acceptance of the configuration scheme in actual promotion.

[0012] To address the aforementioned problems, one of the technical solutions described in this invention is implemented through the following method: a hierarchical configuration method for energy storage in a three-phase unbalanced distribution network considering multi-entity coordination, comprising the following steps: Establish an accurate model of three-phase unbalanced DES; Construct a multi-stakeholder differentiated objective function; where the multiple stakeholders include power grid companies, distributed power generation investors, and electricity users; Based on the weight scheme of each subject's DES weight and the multi-subject differentiated objective function, a hierarchical coordination optimization solution framework is designed to solve the accurate model of three-phase unbalanced DES and output the optimized configuration scheme. The hierarchical coordination optimization solution framework includes an upper multi-subject game coordination layer and a lower three-phase unbalanced physical optimization layer.

[0013] The above-mentioned establishment of an accurate three-phase unbalanced DES model includes: A distribution network model is established, incorporating factors such as asymmetrical line parameters and uneven three-phase output between load and distributed generation. Three-phase power flow equations are used to describe the system's operating state. Specifically: Node power balance equations: , , in, The active power injection for phase m at node i at time t; Let m be the reactive power injection of phase m at node i at time t; m = a, b, c; Let m be the active power output of the DES at node i at time t; Let m be the active power of the load at node i at time t; Let the reactive power of phase m at node i be the load at time t. Let be the voltage amplitude of phase m at node i at time t; Let m be the phase angle difference between phase m of node i and phase p of node j at time t; Let m be the electrical conductance between phase m at node i and phase p at node j; The susceptance between phase m at node i and phase p at node j; Let be the amplitude of the p-phase voltage at node j at time t; Let be the cosine of the phase angle difference between phase m of node i and phase p of node j; Let be the sine value of the phase angle difference between phase m of node i and phase p of node j.

[0014] The above constructs a multi-stakeholder differentiated objective function; where the multiple stakeholders include power grid companies, distributed power generation investors, and electricity users, including: Operating costs of the power grid company and the objective function of voltage deviation penalty : , in: Let t be the electricity purchase price from the grid. Let t be the power purchased from the upstream power grid; The unit power generation cost of local generators; This is the voltage deviation penalty coefficient; Rated voltage; Let be the active power output of the generator at node i at time t; Let be the voltage amplitude of phase m at node i at time t; The profit objective function of distributed power investors : , in: For renewable energy feed-in tariffs; To provide power to the renewable energy source at node i at time t; This is the output fluctuation penalty coefficient; This represents the average output over the T time intervals preceding time t. T The length of the sliding window is defined to obtain the deviation of the output force from its sliding average value. This can measure the degree of deviation of the instantaneous output force from the defined recent average level, thereby reflecting the actual fluctuation characteristics. The calculation method is as follows. ; Electricity user's electricity cost objective function : , in, Let m be the active power of the load at node i at time t; Let m be the user-side DES discharge power of node i at time t; Let t be the electricity purchase price from the grid at time t.

[0015] The above-mentioned hierarchical coordination optimization solution framework, based on the weight scheme of each subject's DES allocation and the multi-subject differentiated objective function, solves the accurate model of the three-phase unbalanced DES and outputs the optimized configuration scheme. The hierarchical coordination optimization solution framework includes an upper-level multi-subject game coordination layer and a lower-level three-phase unbalanced physical optimization layer; including: Upper-level multi-agent game coordination layer: Each agent proposes a weighting scheme for the DES configuration based on its own interests, and the final weight combination is determined through cooperative game theory or Nash negotiation. as follows: , In the formula, Weighting of the power grid company; Weighting for distributed power investors; Weights for electricity users; where, weight Reflecting the bargaining power or decision-making priority of each party in the game, satisfying The objective function is constructed using a weighted social optimal cooperative game model. By adjusting the weights, different solutions on the Pareto front can be generated for the parties to negotiate and choose from. Based on the above weights, a comprehensive social cost function is established as follows: , In the formula, To integrate the social cost function, the interests of the three stakeholders are unified under a single minimization framework; the power grid company's operating costs and voltage deviation penalty objective function are considered. Electricity cost objective function for electricity users It is the metric that investors in distributed power generation aim to minimize; it is their profit objective function. The metric we want to maximize is reduced to a minimization problem by taking its negative value. Lower-level three-phase imbalance physical optimization layer: Based on the weights determined by the upper-level multi-agent game coordination layer, the improved time-series integrated voltage sensitivity analysis method is used to determine the DES location, and the DES capacity and operation strategy are optimized one by one by combining the sequential configuration strategy, and the optimized configuration scheme is output.

[0016] The above-mentioned method of determining DES location using improved timing-integrated voltage sensitivity analysis includes: Based on three-phase Newton-Raphson power flow calculations, the Jacobian submatrix N is extracted, and its inverse is used to obtain the active power-voltage sensitivity matrix. ; Introducing voltage offset coefficient As a weight, the overall voltage sensitivity of phase m at node t is calculated: , in, To improve overall voltage sensitivity, the comprehensive contribution potential of the m-phase of node i to the overall network voltage improvement at time t is described, taking into account both voltage regulation capability and voltage regulation requirements. For traditional active-voltage sensitivity matrix elements, the effect of injecting unit active power into phase m of node i at time t on the voltage of node j; Let be the actual voltage amplitude of phase m at node j at time t; The reference value for the node's rated voltage is taken as 1.0 pu; Based on improved overall voltage sensitivity Calculate the variance of the overall voltage sensitivity of each phase at each node throughout the 24 hours of the day. : , The node-phase combination with the largest variance in improved integrated voltage sensitivity is selected as the current access location for DES.

[0017] The above-mentioned sequential configuration strategy optimizes the capacity and operation strategy of DES on a per-unit basis, and outputs an optimized configuration scheme, including: Step 1: Initialize the DES configuration quantity k=0, with a preset total number of configurations K; Step 2: Calculate the variance of the overall voltage sensitivity of each phase at all nodes. The node-phase combination with the largest comprehensive voltage sensitivity variance is selected as the access point of the (k+1)th DES. Step 3: Use an optimization algorithm to solve for the intraday operation plan of the DES node; the decision variable of the optimization algorithm is the 24-hour charging and discharging power sequence. The optimized 24-hour charge and discharge power sequence is the daily operation plan. Step 5: Introduce a hyperparameter alternation iteration mechanism, where the iteration update rules are as follows: , In the formula, This is the hyperparameter vector passed from the upper layer. ; This is the index vector fed back after the lower layer solution is obtained. , The gradient of the feedback indicator; To learn step size; when The iteration terminates at time; Step 6: Calculate the rated capacity and rated power of the DES; Based on the optimized power sequence, the absolute value of the cumulative energy during continuous charging and discharging periods is calculated, and the maximum value is taken as the rated capacity of the DES at that node. The calculation method is as follows: , In the formula, To optimize the obtained power sequence; For all continuous charging periods; This is the set of all continuous discharge periods; Among them, rated power The calculation method is as follows: ; Step 7: Update the system status, and input the configured k+1 DES units as fixed resources into the model. Let k=k+1. If k<K, return to step 2; otherwise, output the optimized configuration scheme. The optimized configuration scheme includes the installation location, rated capacity, rated power, and typical daily charge and discharge schedule of each DES unit.

[0018] The aforementioned lower-level three-phase imbalance physical optimization layer also includes: introducing a dynamic zoning mechanism, introducing the concept of energy storage marginal benefits and dispatchable domain to guide capacity optimization, introducing a dynamic current compensation control mechanism, and introducing a multi-mode operation strategy. In the introduction of the dynamic partitioning mechanism, the partition boundaries satisfy: , In the formula, For branch set; The partition threshold coefficient is set to 0.15-0.25. This is a regional evaluation index. ; Among them, the concepts of energy storage marginal benefits and dispatchable domain are introduced to guide capacity optimization. The energy storage marginal benefits are: , In the formula, The overall system benefit function after DES integration. To reduce network loss, This represents the reduction in voltage deviation. To increase the amount of new energy consumption, This refers to the corresponding revenue conversion factor; The schedulable domain is the power range that the system can flexibly schedule under the condition of satisfying safe operation constraints; Among them, a dynamic current compensation control mechanism is introduced, and the dynamic current compensation amount is... The calculation method is as follows: , In the formula, Let be the unbalanced current of phase m at time t. For reference balance current, These are PI control parameters; Among them, a multi-mode operation strategy is introduced, including constant reactive power mode, reactive power compensation mode, unbalanced current compensation mode, unbalanced and reactive power compensation mode, and reactive power and unbalanced compensation mode.

[0019] The second technical solution of this invention is achieved through the following method: a hierarchical configuration system for energy storage in a three-phase unbalanced distribution network that considers multi-entity coordination, comprising the following steps: Model building unit: Establish an accurate model of three-phase unbalanced DES; Establish function units and construct multi-entity differentiated objective functions; where the multiple entities include power grid companies, distributed power generation investors, and power users; The solution output unit designs a hierarchical coordination optimization solution framework based on the weight scheme of each subject's DES weight and the multi-subject differentiated objective function. It solves the accurate model of the three-phase unbalanced DES and outputs the optimized configuration scheme. The hierarchical coordination optimization solution framework includes an upper multi-subject game coordination layer and a lower three-phase unbalanced physical optimization layer.

[0020] Compared with the prior art, the present invention has the following advantages: 1. A hierarchical architecture of "game coordination-physical optimization" is proposed. The upper layer achieves interest coordination through multi-subject weight negotiation, while the lower layer performs physical optimization based on the three-phase unbalanced DES accurate model, taking into account both economic efficiency and engineering feasibility.

[0021] 2. A refined three-phase unbalanced DES accurate model was established, which takes into account the power flow constraints of three-phase asymmetry, voltage deviation and three-phase unbalance constraints, thus improving the applicability of the configuration scheme to the actual distribution network.

[0022] 3. A sequential configuration mechanism based on time-series integrated voltage sensitivity was designed to optimize DES unit by unit, avoid over-configuration in some areas, and improve equipment utilization and overall economy.

[0023] 4. Introduce a privacy protection mechanism. During the game coordination process, each party only needs to provide the objective function weights or objective function. The upper layer negotiates weights or distributes benefits based on these functions without disclosing its internal detailed operating data, thus protecting its commercial privacy. Attached Figure Description

[0024] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0025] Figure 1 This is a flowchart of the method in Embodiment 1 of the present invention.

[0026] Figure 2 This is a system structure block diagram of Embodiment 2 of the present invention. Detailed Implementation

[0027] The present invention is not limited to the following embodiments, and the specific implementation can be determined according to the technical solution of the present invention and the actual situation.

[0028] Example 1: As Figure 1As shown in the figure, this invention discloses a hierarchical optimization configuration method for distributed energy storage in a three-phase unbalanced distribution network that considers the coordination of multiple stakeholders, including the following steps: S101, Establish an accurate model of three-phase unbalanced DES; S102, construct a multi-entity differentiated objective function; where the multiple entities include power grid companies, distributed power generation investors, and electricity users; S103. Based on the weight scheme of each subject's DES weight and the multi-subject differentiated objective function, a hierarchical coordination optimization solution framework is designed to solve the three-phase unbalanced DES accurate model and output the optimized configuration scheme. The hierarchical coordination optimization solution framework includes an upper multi-subject game coordination layer and a lower three-phase unbalanced physical optimization layer.

[0029] In step S101, an accurate model of the three-phase unbalanced DES is established, including: Based on a three-phase four-wire or three-phase three-wire system, a distribution network model is established that includes asymmetrical line parameters and uneven three-phase output between loads and distributed generation sources. Three-phase power flow equations are used to describe the system's operating state. Specifically: Node power balance equations: , , in, The active power injection for phase m at node i at time t; Let m be the reactive power injection of phase m at node i at time t; m = a, b, c; Let t be the active power output of the DES of node i phase m at time t (discharge is positive, charging is negative). Let m be the active power of the load at node i at time t; Let the reactive power of phase m at node i be the load at time t. Let be the voltage amplitude of phase m at node i at time t; Let m be the phase angle difference between phase m of node i and phase p of node j at time t; Let m be the electrical conductance between phase m at node i and phase p at node j; The susceptance between phase m at node i and phase p at node j; Let be the amplitude of the p-phase voltage at node j at time t; Let be the cosine of the phase angle difference between phase m of node i and phase p of node j; Let be the sine value of the phase angle difference between phase m of node i and phase p of node j.

[0030] In step S102, a multi-entity differentiated objective function is constructed; the multiple entities include the power grid company, distributed power generation investors, and electricity users, including: Operating costs of the power grid company and the objective function of voltage deviation penalty : , in: Let t be the electricity purchase price from the grid. Let t be the power purchased from the upstream power grid; The unit power generation cost of local generators; This is the voltage deviation penalty coefficient; Rated voltage; Let be the active power output of the generator at node i at time t; Let be the voltage amplitude of phase m at node i at time t; The profit objective function of distributed power investors : , in: For renewable energy feed-in tariffs; To provide power to the renewable energy source at node i at time t; This is the output fluctuation penalty coefficient; This represents the average output over the T time intervals preceding time t. T The length of the sliding window is defined to obtain the deviation of the output force from its sliding average value. This can measure the degree of deviation of the instantaneous output force from the defined recent average level, thereby reflecting the actual fluctuation characteristics. The calculation method is as follows. ; Electricity user's electricity cost objective function : , in, Let m be the active power of the load at node i at time t; Let m be the user-side DES discharge power of node i at time t; Let t be the electricity purchase price from the grid at time t.

[0031] In step S103, a hierarchical coordination optimization solution framework is designed based on the weight scheme of each subject's DES allocation and the multi-subject differentiated objective function. This framework solves the accurate three-phase unbalanced DES model and outputs the optimized configuration scheme. The hierarchical coordination optimization solution framework includes an upper-level multi-subject game coordination layer and a lower-level three-phase unbalanced physical optimization layer. Upper-level multi-stakeholder game coordination layer: The final weight combination is determined using the Nash negotiation model; where the payoffs of the power grid company, distributed power generation investors, and power users in a non-cooperative state are respectively... Then the Nash negotiation can be resolved as follows: , Determine the final weight combination as follows: , In the formula, Weighting of the power grid company; Weighting for distributed power investors; Weights for electricity users; where, weight Reflecting the bargaining power or decision-making priority of each party in the game, satisfying The objective function is constructed using a weighted social optimal cooperative game model. By adjusting the weights, different solutions on the Pareto front can be generated for the parties to negotiate and choose from. Based on the above weights, a comprehensive social cost function is established as follows: , In the formula, To integrate the social cost function, the interests of the three stakeholders are unified under a single minimization framework; the power grid company's operating costs and voltage deviation penalty objective function are considered. Electricity cost objective function for electricity users It is the metric that investors in distributed power generation aim to minimize; it is their profit objective function. The metric we want to maximize is reduced to a minimization problem by taking its negative value. Lower-level three-phase imbalance physical optimization layer: Based on the weights determined by the upper-level multi-agent game coordination layer, the improved time-series integrated voltage sensitivity analysis method is used to determine the DES location, and the DES capacity and operation strategy are optimized one by one by combining the sequential configuration strategy, and the optimized configuration scheme is output.

[0032] The above-mentioned method of determining DES location using improved timing-integrated voltage sensitivity analysis includes: Based on three-phase Newton-Raphson power flow calculations, the Jacobian submatrix N is extracted, and its inverse is used to obtain the active power-voltage sensitivity matrix. This element represents the effect of the unit active power injected into phase m of node i at time t on the voltage of node j. Introducing voltage offset coefficient As a weight, the overall voltage sensitivity of phase m at node t is calculated: , in, To improve overall voltage sensitivity, the comprehensive contribution potential of the m-phase of node i to the overall network voltage improvement at time t is described, taking into account both voltage regulation capability and voltage regulation requirements. For traditional active-voltage sensitivity matrix elements, the effect of injecting unit active power into phase m of node i at time t on the voltage of node j; Let be the actual voltage amplitude of phase m at node j at time t; The reference value for the node's rated voltage is taken as 1.0 pu; Based on improved overall voltage sensitivity Calculate the variance of the overall voltage sensitivity of each phase at each node throughout the 24 hours of the day. : , The node-phase combination with the largest variance in improved integrated voltage sensitivity is selected as the current access location for DES.

[0033] The above-mentioned sequential configuration strategy optimizes the capacity and operation strategy of DES on a per-unit basis, and outputs an optimized configuration scheme, including: Step 1: Initialize the DES configuration quantity k=0, with a preset total number of configurations K; Step 2: Calculate the variance of the overall voltage sensitivity of each phase at all nodes. The node-phase combination with the largest comprehensive voltage sensitivity variance is selected as the access point of the (k+1)th DES. Step 3: Use optimization algorithms (such as the improved Grey Wolf algorithm and second-order cone programming) to solve the intraday operation plan of the DES at this node; the decision variable of the optimization algorithm is the 24-hour charging and discharging power sequence. The optimized 24-hour charge and discharge power sequence is the daily operation plan. The decision variable is the 24-hour charge and discharge power sequence. The constraints include: Power balance constraints, node voltage approximation Three-phase unbalance constraint Energy storage energy balance constraints Energy storage power constraints Energy storage capacity constraints SOC constraints ; Step 5: Introduce a hyperparameter alternation iteration mechanism, where the iteration update rules are as follows: , In the formula, This is the hyperparameter vector passed from the upper layer. ; This is the index vector fed back after the lower layer solution is obtained. , The gradient of the feedback indicator; To learn step size; when The iteration terminates at time; Step 6: Calculate the rated capacity and rated power of the DES; Based on the optimized power sequence, the absolute value of the cumulative energy during continuous charging and discharging periods is calculated, and the maximum value is taken as the rated capacity of the DES at that node. The calculation method is as follows: , In the formula, To optimize the obtained power sequence; For all continuous charging periods; This is the set of all continuous discharge periods; Among them, rated power The calculation method is as follows: ; Step 7: Update the system status, and input the configured k+1 DES units as fixed resources into the model. Let k=k+1. If k<K, return to step 2; otherwise, output the optimized configuration scheme. The optimized configuration scheme includes the installation location, rated capacity, rated power, and typical daily charge and discharge schedule of each DES unit.

[0034] The above-mentioned DES installation locations are node-phase combinations determined in each iteration through improved timing-integrated voltage sensitivity analysis; the typical daily charge-discharge schedule is the daily operation plan obtained through optimization, which is organized into a table format according to a 24-hour time series, that is, the charge and discharge power values ​​of energy storage at each moment.

[0035] In the lower-level three-phase imbalance physical optimization layer, a dynamic partitioning mechanism is introduced. Based on the comprehensive voltage sensitivity variance, a threshold triggering method is used for dynamic partitioning. A partitioning evaluation index is defined. When the difference in the evaluation index of adjacent nodes exceeds the threshold At that time, the partition boundaries are defined. The partition boundaries satisfy: , in For branch road collection, The partition threshold coefficient is set to 0.15-0.25.

[0036] Furthermore, the concepts of energy storage marginal benefits and dispatchable domain are introduced to guide capacity optimization. The energy storage marginal benefit is defined as: , in The overall system benefit function after DES integration. To reduce network loss, This represents the reduction in voltage deviation. To increase the amount of new energy consumption, This is the corresponding revenue conversion factor.

[0037] The schedulable domain is defined as the range of power that the system can flexibly schedule under the condition of satisfying safe operation constraints. , Among them, safety constraints include node voltage constraints. Three-phase unbalance constraint Line current carrying constraints And the operational constraints of energy storage itself (power, capacity, SOC, etc.). For distributed renewable energy output vectors, This represents the energy storage charge / discharge power vector. The schedulable domain volume. Through the It is obtained by sampling or convex hull approximation calculation.

[0038] Add a schedulable domain maximization term to the capacity optimization objective: , in For the volume metric of the schedulable domain, This is a flexibility weighting coefficient.

[0039] A dynamic current compensation control mechanism is introduced. The current compensation control decision is determined based on the proportional relationship between the three-phase negative-sequence and positive-sequence currents. The compensation control quantity is calculated as follows: , in Let be the unbalanced current of phase m at time t. For reference balance current, This is a PI control parameter. The compensation amount is achieved through the charging and discharging power of the DES. , A multi-mode operation strategy is introduced, including constant reactive power mode, reactive power compensation mode, unbalanced current compensation mode, unbalanced and reactive power compensation mode, and reactive power and unbalanced compensation mode. The DES operation commands are calculated under different modes as follows: Constant reactive power mode: DES outputs constant reactive power, while active power is executed according to dispatch instructions. , in, This is a preset reactive power setting, in kVar. The active power command obtained from the optimized scheduling is expressed in kW. Reactive power compensation mode: DES performs dynamic reactive power compensation based on local reactive power demand. , in, This is the reactive power compensation coefficient, which is dimensionless and ranges from 0 to 1. The reactive power of phase m at node i at time t is expressed in kVar. For reference reactive power, it is usually taken as 0, with the unit being kVar; Unbalanced current compensation mode: DES prioritizes compensating for three-phase unbalanced current, and the active power dispatch command can be adjusted appropriately. , in, This is the unbalanced current compensation coefficient, in kV / A, used to convert current deviation into active power; Let be the unbalanced current of phase m at time t, in A. Imbalance and Reactive Power Compensation Mode: DES simultaneously provides active power dispatch, current compensation, and reactive power compensation. The operating command is as follows: , in, For dynamic current compensation power, by Calculation, unit: kW; This refers to reactive power compensation, measured in kVar. Reactive power and unbalanced compensation modes: DES prioritizes reactive power compensation, with remaining capacity used for unbalanced current compensation. , , in, The maximum reactive power capacity of DES is determined by the inverter capacity and must satisfy... , The rated apparent power of DES, in kVA; For the remaining capacity, Calculation, unit: kVA; Let be the voltage amplitude of phase m at node i at time t, in kV.

[0040] The above content describes the specific technical means of the lower physical optimization layer of the present invention, which supports the implementation of the upper game coordination results at the engineering level. Among them, dynamic partitioning affects which two locations the DES is installed in, marginal benefits and schedulable domain jointly determine the capacity of each DES, and current compensation control and multi-mode operation determine the operation mode of the DES.

[0041] In summary, compared with the prior art, the present invention has the following advantages: 1. A hierarchical architecture of "game coordination-physical optimization" is proposed. The upper layer achieves interest coordination through multi-subject weight negotiation, while the lower layer performs physical optimization based on the three-phase unbalanced DES accurate model, taking into account both economic efficiency and engineering feasibility.

[0042] 2. A refined three-phase unbalanced DES accurate model was established, which takes into account the power flow constraints of three-phase asymmetry, voltage deviation and three-phase unbalance constraints, thus improving the applicability of the configuration scheme to the actual distribution network.

[0043] 3. A sequential configuration mechanism based on time-series integrated voltage sensitivity was designed to optimize DES unit by unit, avoid over-configuration in some areas, and improve equipment utilization and overall economy.

[0044] 4. Introduce a privacy protection mechanism. During the game coordination process, each party only needs to provide the objective function weights or objective function. The upper layer negotiates weights or distributes benefits based on these functions without disclosing its internal detailed operating data, thus protecting its commercial privacy.

[0045] Example 2: As Figure 2 As shown, this invention discloses a hierarchical optimization configuration system for distributed energy storage in a three-phase unbalanced distribution network that considers the coordination of multiple stakeholders, including the following steps: Model building unit: Establish an accurate model of three-phase unbalanced DES; Establish function units and construct multi-entity differentiated objective functions; where the multiple entities include power grid companies, distributed power generation investors, and power users; The solution output unit designs a hierarchical coordination optimization solution framework based on the weight scheme of each subject's DES weight and the multi-subject differentiated objective function. It solves the accurate model of the three-phase unbalanced DES and outputs the optimized configuration scheme. The hierarchical coordination optimization solution framework includes an upper multi-subject game coordination layer and a lower three-phase unbalanced physical optimization layer.

[0046] Example 3: DES hierarchical coordination configuration based on an improved IEEE 33-node system This invention uses a typical improved IEEE 33-node three-phase unbalanced distribution network as an example to demonstrate the complete implementation process of the method of this invention. This embodiment follows the principles of describability and derivability, illustrating its feasibility through the logical explanation of the method steps and parameter settings.

[0047] 1. System basic data and parameter settings Assume a regional power distribution network topology is a standard IEEE 33-bus system, with three-phase imbalance modifications. The system base voltage is 12.66kV. Three-phase photovoltaic (PV) power sources with capacities of 1000kW and 500kW are connected to nodes 11 and 28, respectively. Load, PV output, and grid electricity price are all considered based on typical daily curves. Key parameters are shown in Table 1.

[0048] Table 1 Key Parameters of Implementation Case 2. The method steps in this embodiment are as follows: Step 1: Initialization and Data Preparation Input the three-phase unbalanced line impedance matrix of the 33-node system, the 24-hour three-phase load curve, the 24-hour photovoltaic output prediction curve, and the time-of-use electricity price curve. Calculate the initial power flow to obtain the system state without DES. 10,000 yuan / year 10,000 yuan / year Ten thousand yuan per year.

[0049] Step 2: Upper-level game coordination to determine weights Solve the Nash negotiation model: , satisfy Solve this problem. The overall objective function is: , Introducing bargaining factors The improved objective function is: , Step 3: Configuring the first DES unit Sub-step 1: Calculate the improved integrated voltage sensitivity of each phase at each node under DES-free conditions. Based on the three-phase power flow results, extract the Jacobian submatrix at each time step and inverse it to obtain... Calculate the voltage offset coefficient. The overall voltage sensitivity is obtained. .

[0050] Sub-step 2: Calculate the variance of the overall voltage sensitivity. Variance of phase B at node 18. The highest across the entire network. It employs a dynamic partitioning mechanism to calculate the partition evaluation index. Adjacent node 17 ,difference Nodes 17-18 are divided into different zones, with the zone containing node 18 being the weakest area at the end of the voltage range.

[0051] Sub-step 3: Select phase B of node 18 as the DES#1 access point. The optimization variable is the 24-hour charging and discharging power. Considering all constraints, the optimal power sequence is obtained by solving the problem.

[0052] Sub-step 4: Based on the optimization results, the continuous discharge period is 17:00-21:00 (4 hours), and the continuous charging period is 22:00-5:00 (7 hours). The calculated maximum continuous discharge capacity is 195kWh, and the maximum continuous charging capacity is 182kWh. The maximum value of 195kWh is taken as the rated capacity. The maximum charge / discharge power is 300kW, reaching the power limit.

[0053] Step 4: Sequence configuration of the second DES unit Sub-step 1: Substitute DES#1 (node ​​18-B phase, 195kWh / 300kW) and its operation plan into the system model and update the power flow status.

[0054] Sub-step 2: Re-perform the comprehensive voltage sensitivity analysis. At this point, the C-phase variance of node 33... It became the highest in the entire network. Dynamic partitioning identified the branch where node 33 was located as a new weak voltage area.

[0055] Sub-step 3: Select phase C of node 33 as the DES#2 access point. Optimization solution yields a rated capacity of 197kWh and a rated power of 300kW.

[0056] Sub-step 4: Calculate the marginal benefit of energy storage. For DES#2, the overall system benefit... ,in , , ,Pick The marginal benefit is calculated. The price is approximately RMB 1700 / kWh, which is close to the unit capacity cost of RMB 1700 / kWh, thus reaching the optimal level.

[0057] Sub-step 5: Calculate the schedulable domain. After DES#2 access, the schedulable domain volume... It increased by 32.5% compared to when there was no DES.

[0058] Step 5: Iterate alternately with hyperparameters Upper-level hyperparameters Lower-level feedback indicators After iterative updates, middle Adjusted to 0.012. The value was set to 0.19. The algorithm converged after 3 iterations.

[0059] Step Six: Dynamic Current Compensation Control Based on the three-phase unbalanced current monitoring data, the compensation control quantity is calculated. Taking node 18 as an example, at t=18:00, the unbalanced current of phase B is... Reference balance current ,Pick Compensation current is required Compensation power is provided via DES#1. This is superimposed on the peak discharge power.

[0060] Step 7: Multi-mode operation strategy Using unbalanced and reactive power compensation mode, the DES operation command is as follows: , in To optimize scheduling power, For dynamic power compensation, This is the reactive power compensation power.

[0061] Step 8: Output Configuration Scheme Table 2 Output of DES Final Configuration Scheme Compared with two traditional technical solutions, the results of this invention are as follows: 1. Traditional technical solution A (without DES) During the midday peak of solar power generation, the voltage at some nodes may exceed the upper limit (>1.05 pu); during the evening peak load, the voltage at the end nodes may exceed the lower limit (<0.95 pu); and the solar curtailment rate is relatively high, resulting in high total system operating costs.

[0062] 2. Traditional technical solution B (centralized configuration, considering only a single economic objective) It's possible to concentrate both DES units near the nodes with the highest electricity price arbitrage profits. While the overall economic cost may be lower, it cannot effectively solve the voltage quality problem across the entire grid, especially since the voltage at the end of branches without DES units may still exceed the limit, and the smoothing effect of photovoltaic fluctuations is limited.

[0063] The technical effects of this invention are as follows: Through the weight combination determined by the upper-level Nash negotiation model, the comprehensive objective function takes into account the interests of all three parties. In the lower-level physical optimization layer, through two site selections based on comprehensive voltage sensitivity, DES#1 and DES#2 are precisely deployed on the two most critical weak phases, achieving accurate identification of voltage weak points. Simulation results show that after configuration, the voltage of each phase of all nodes in the entire network is maintained between [0.96, 1.04] pu throughout the day, and the three-phase imbalance is reduced to below 2.6%. The dynamic partitioning mechanism divides the system into multiple voltage characteristic partitions, avoiding excessive concentration of energy storage resources in the same area. Capacity optimization introduces the marginal benefits of energy storage, achieving economic optimization. The increased dispatchable domain volume improves the system's flexibility in responding to photovoltaic output fluctuations. Dynamic current compensation control achieves real-time compensation of three-phase unbalanced current, and the multi-mode operation strategy enables DES to simultaneously provide active power dispatch, current compensation, and reactive power compensation, improving energy storage utilization.

[0064] The economic and multi-stakeholder benefits of this invention are as follows: 1) Power grid companies: System network losses are reduced by approximately 30%, voltage qualification rate is increased to 100%, and safety benefits are significant. 2) Photovoltaic investors: Through the smoothing effect of DES, the volatility of photovoltaic output is reduced, grid-connected electricity is more stable, and returns are guaranteed. 3) Electricity users: Through peak-valley arbitrage of DES, the overall electricity cost in their region decreases. Under the weights determined by the Nash negotiations, the overall objective is... It is superior to both Scheme A and Scheme B, achieving a Pareto improvement that benefits all parties, with the benefits of cooperation from all parties exceeding those of non-cooperation.

[0065] In summary, the embodiments of this invention detail the complete logical process of configuring two distributed energy storage systems in a three-phase unbalanced distribution network using the method of this invention. Through the complete technical chain of "upper-level Nash negotiation to determine weights - dynamic partitioning - improved sensitivity analysis - marginal benefit balancing - sequence optimization - alternating hyperparameter iteration - dynamic current compensation - multi-mode operation," the invention verifies that the method can effectively coordinate the interests of multiple parties and output an energy storage configuration scheme that simultaneously improves system voltage quality, increases the renewable energy absorption rate, enhances energy storage utilization, and considers the economics of all stakeholders, thus possessing certain engineering implementation guidance value.

Claims

1. A method for hierarchical configuration of energy storage in a three-phase unbalanced distribution network considering multi-entity coordination, characterized in that, Includes the following steps: Establish an accurate model of three-phase unbalanced DES; Construct a multi-stakeholder differentiated objective function; where the multiple stakeholders include power grid companies, distributed power generation investors, and electricity users; Based on the weight scheme of each subject's DES weight and the multi-subject differentiated objective function, a hierarchical coordination optimization solution framework is designed to solve the accurate model of three-phase unbalanced DES and output the optimized configuration scheme. The hierarchical coordination optimization solution framework includes an upper multi-subject game coordination layer and a lower three-phase unbalanced physical optimization layer.

2. The method for hierarchical configuration of energy storage in a three-phase unbalanced distribution network considering multi-entity coordination as described in claim 1, characterized in that, The establishment of an accurate three-phase unbalanced DES model includes: A distribution network model is established, incorporating factors such as asymmetrical line parameters and uneven three-phase output between load and distributed generation. Three-phase power flow equations are used to describe the system's operating state. Specifically: Node power balance equations: ， ， in, The active power injection for phase m at node i at time t; Let m be the reactive power injection of phase m at node i at time t; m = a, b, c; Let m be the active power output of the DES at node i at time t; Let m be the active power of the load at node i at time t; Let the reactive power of phase m at node i be the load at time t. Let be the voltage amplitude of phase m at node i at time t; Let m be the phase angle difference between phase m of node i and phase p of node j at time t; Let m be the electrical conductance between phase m at node i and phase p at node j; The susceptance between phase m at node i and phase p at node j; Let be the amplitude of the p-phase voltage at node j at time t; Let be the cosine of the phase angle difference between phase m of node i and phase p of node j; Let be the sine value of the phase angle difference between phase m of node i and phase p of node j.

3. The method for hierarchical configuration of energy storage in a three-phase unbalanced distribution network considering multi-entity coordination as described in claim 1, characterized in that, The construction of a multi-entity differentiated objective function; wherein the multiple entities include power grid companies, distributed power generation investors, and electricity users, including: Operating costs of the power grid company and the objective function of voltage deviation penalty : ， in: Let t be the electricity purchase price from the grid. Let t be the power purchased from the upstream power grid; The unit power generation cost of local generators; This is the voltage deviation penalty coefficient; Rated voltage; Let be the active power output of the generator at node i at time t; Let be the voltage amplitude of phase m at node i at time t; The profit objective function of distributed power investors : ， in: For renewable energy feed-in tariffs; To provide power to the renewable energy source at node i at time t; This is the output fluctuation penalty coefficient; This represents the average output over the T time intervals preceding time t. T The length of the sliding window is defined to obtain the deviation of the output force from its sliding average value. This can measure the degree of deviation of the instantaneous output force from the defined recent average level, thereby reflecting the actual fluctuation characteristics. The calculation method is as follows. ； Electricity user's electricity cost objective function : ， in, Let m be the active power of the load at node i at time t; Let m be the user-side DES discharge power of node i at time t; Let t be the electricity purchase price from the grid at time t.

4. The method for hierarchical configuration of energy storage in a three-phase unbalanced distribution network considering multi-entity coordination as described in claim 1, characterized in that, The hierarchical coordination optimization solution framework, designed based on the weight scheme of each subject's DES allocation and the multi-subject differentiated objective function, solves the accurate model of the three-phase unbalanced DES and outputs an optimized configuration scheme. The hierarchical coordination optimization solution framework includes an upper-level multi-subject game coordination layer and a lower-level three-phase unbalanced physical optimization layer. Upper-level multi-agent game coordination layer: Each agent proposes a weighting scheme for the DES configuration based on its own interests, and the final weight combination is determined through cooperative game theory or Nash negotiation. as follows: ， In the formula, Weighting of the power grid company; Weighting for distributed power investors; Weights for electricity users; where, weight Reflecting the bargaining power or decision-making priority of each party in the game, satisfying The objective function is constructed using a weighted social optimal cooperative game model. By adjusting the weights, different solutions on the Pareto front can be generated for the parties to negotiate and choose from. Based on the above weights, a comprehensive social cost function is established as follows: ， In the formula, To integrate the social cost function, the interests of the three stakeholders are unified under a single minimization framework; the power grid company's operating costs and voltage deviation penalty objective function are considered. Electricity cost objective function for electricity users It is the metric that investors in distributed power generation aim to minimize; it is their profit objective function. The metric we want to maximize is reduced to a minimization problem by taking its negative value. Lower-level three-phase imbalance physical optimization layer: Based on the weights determined by the upper-level multi-agent game coordination layer, the improved time-series integrated voltage sensitivity analysis method is used to determine the DES location, and the DES capacity and operation strategy are optimized one by one by combining the sequential configuration strategy, and the optimized configuration scheme is output.

5. The method for hierarchical configuration of energy storage in a three-phase unbalanced distribution network considering multi-entity coordination as described in claim 4, characterized in that, The DES location is determined using an improved timing-integrated voltage sensitivity analysis method, including: Based on three-phase Newton-Raphson power flow calculations, the Jacobian submatrix N is extracted, and its inverse is used to obtain the active power-voltage sensitivity matrix. ; Introducing voltage offset coefficient As a weight, the overall voltage sensitivity of phase m at node t is calculated: ， in, To improve overall voltage sensitivity, the comprehensive contribution potential of the m-phase of node i to the overall network voltage improvement at time t is described, taking into account both voltage regulation capability and voltage regulation requirements. For traditional active-voltage sensitivity matrix elements, the effect of injecting unit active power into phase m of node i at time t on the voltage of node j; Let be the actual voltage amplitude of phase m at node j at time t; The reference value for the node's rated voltage is taken as 1.0 pu; Based on improved overall voltage sensitivity Calculate the variance of the overall voltage sensitivity of each phase at each node throughout the 24 hours of the day. : ， The node-phase combination with the largest variance in improved integrated voltage sensitivity is selected as the current access location for DES.

6. The method for hierarchical configuration of energy storage in a three-phase unbalanced distribution network considering multi-entity coordination as described in claim 5, characterized in that, By combining sequential configuration strategies to optimize DES capacity and operation strategies on a per-unit basis, an optimized configuration scheme is output, including: Step 1: Initialize the DES configuration quantity k=0, with a preset total number of configurations K; Step 2: Calculate the variance of the overall voltage sensitivity of each phase at all nodes. The node-phase combination with the largest comprehensive voltage sensitivity variance is selected as the access point of the (k+1)th DES. Step 3: Use an optimization algorithm to solve for the intraday operation plan of the DES node; the decision variable of the optimization algorithm is the 24-hour charging and discharging power sequence. The optimized 24-hour charge and discharge power sequence is the daily operation plan. Step 5: Introduce a hyperparameter alternation iteration mechanism, where the iteration update rules are as follows: ， In the formula, This is the hyperparameter vector passed from the upper layer. ; This is the index vector fed back after the lower layer solution is obtained. , The gradient of the feedback indicator; To learn step size; when The iteration terminates at time; Step 6: Calculate the rated capacity and rated power of the DES; Based on the optimized power sequence, the absolute value of the cumulative energy during continuous charging and discharging periods is calculated, and the maximum value is taken as the rated capacity of the DES at that node. The calculation method is as follows: ， In the formula, To optimize the obtained power sequence; For all continuous charging periods; This is the set of all continuous discharge periods; Among them, rated power The calculation method is as follows: ； Step 7: Update the system status, and input the configured k+1 DES units as fixed resources into the model. Let k=k+1. If k<K, return to step 2; otherwise, output the optimized configuration scheme. The optimized configuration scheme includes the installation location, rated capacity, rated power, and typical daily charge and discharge schedule of each DES unit.

7. The method for hierarchical configuration of energy storage in a three-phase unbalanced distribution network considering multi-entity coordination as described in claim 6, characterized in that, The lower three-phase imbalance physical optimization layer also includes: introducing a dynamic zoning mechanism, introducing the concept of energy storage marginal benefits and dispatchable domain to guide capacity optimization, introducing a dynamic current compensation control mechanism, and introducing a multi-mode operation strategy. In the introduction of the dynamic partitioning mechanism, the partition boundaries satisfy: ， In the formula, For branch set; The partition threshold coefficient is set to 0.15-0.

25. This is a regional evaluation index. ; Among them, the concepts of energy storage marginal benefits and dispatchable domain are introduced to guide capacity optimization. The energy storage marginal benefits are: ， In the formula, The overall system benefit function after DES integration. To reduce network loss, This represents the reduction in voltage deviation. To increase the amount of new energy consumption, This refers to the corresponding revenue conversion factor; The schedulable domain is the power range that the system can flexibly schedule under the condition of satisfying safe operation constraints; Among them, a dynamic current compensation control mechanism is introduced, and the dynamic current compensation amount is... The calculation method is as follows: ， In the formula, Let be the unbalanced current of phase m at time t. For reference balance current, These are PI control parameters; Among them, a multi-mode operation strategy is introduced, including constant reactive power mode, reactive power compensation mode, unbalanced current compensation mode, unbalanced and reactive power compensation mode, and reactive power and unbalanced compensation mode.

8. A hierarchical energy storage configuration system for a three-phase unbalanced distribution network considering multi-entity coordination, characterized in that, Includes the following steps: Model building unit: Establish an accurate model of three-phase unbalanced DES; Establish function units and construct multi-entity differentiated objective functions; where the multiple entities include power grid companies, distributed power generation investors, and power users; The solution output unit designs a hierarchical coordination optimization solution framework based on the weight scheme of each subject's DES weight and the multi-subject differentiated objective function. It solves the accurate model of the three-phase unbalanced DES and outputs the optimized configuration scheme. The hierarchical coordination optimization solution framework includes an upper multi-subject game coordination layer and a lower three-phase unbalanced physical optimization layer.

Images