Dispatching method and system for relieving congestion of power distribution network through distributed energy storage aggregation

Abstract

The invention discloses a scheduling method and system for relieving congestion of a power distribution network through distributed energy storage aggregation, and belongs to the technical field of optimized operation of the power distribution network. The method comprises the following steps: establishing an aggregation model of various distributed energy storage based on a maximum internal approximation method, and calculating a reference load curve of each cluster; load flow calculation is carried out according to the branch load flow model, an objective function is minimization of the operation cost of the power distribution network, and constraint conditions comprise load flow constraint, new energy power generation constraint and energy storage system constraint; calculating and judging whether line overload or node voltage out-of-limit occurs in the power distribution network according to a load flow calculation result, if so, calculating a node marginal electricity price, and setting a compensation coefficient; and optimizing the objective function according to the compensation coefficient to obtain an updated distributed energy storage load curve, and performing load flow calculation again. And the iteration process is carried out until the conditions of line overload and node voltage out-of-limit do not occur in the power distribution network. The calculation amount is greatly reduced, and the problem of line congestion of the power distribution network is solved by a more efficient market mechanism.

Description

Technical Field

[0001] The present invention belongs to the field of optimized operation of distribution networks, and more specifically, relates to a scheduling method and system for alleviating distribution network congestion by aggregating distributed energy storage. Background Art

[0002] With the large-scale access of distributed energy storage (such as electric vehicles, variable-frequency air conditioners, and electrochemical energy storage), distribution networks face the problem of increased line congestion risks.

[0003] With the large-scale access of distributed power sources and flexible loads, the risk of congestion in distribution network lines is becoming increasingly prominent. Distributed energy storage (such as electric vehicles and air conditioning loads) is an adjustable resource with significant potential for alleviating congestion due to its flexible scheduling capabilities. However, the small capacity, large number, and discrete behavior of distributed energy storage units lead to the computational dimensionality disaster of traditional centralized optimization, making it difficult to meet efficient scheduling requirements. Existing congestion management methods have the following limitations:

[0004] 1. Aggregation Technology Bottlenecks: Current mainstream aggregation methods (such as the equivalent parameter method and the outer envelope box method) often oversimplify individual constraints to reduce computational complexity. The equivalent parameter method employs an "averaging" approach, ignoring differences in user behavior and resulting in a conservative aggregation model. While the outer envelope box method retains boundary constraints, it overexpands the feasible domain, making dispatch instructions prone to causing local voltage violations or line overloads. All of these methods struggle to accurately map the aggregate domain to the feasible domain of individual cells, limiting the accuracy of congestion control.

[0005] 2. Rigid market mechanisms: Traditional distribution network scheduling relies on fixed electricity prices or static safety constraints, lacking dynamic price signal guidance. Inadequate coordination between load aggregators and distribution network operators prevents real-time adjustments to energy storage charging and discharging strategies in response to line congestion, and is particularly slow to respond to reverse congestion caused by reverse power transmission from renewable energy sources.

[0006] 3. Lack of control dimension: Existing congestion management methods focus on forward flow congestion and lack effective means to deal with reverse overload problems caused by high penetration rates of photovoltaic and wind power. In addition, the scheduling results are difficult to decompose into individual devices for execution, which restricts the flexible use of distributed energy storage.

[0007] Existing dispatching methods have significant flaws when dealing with large-scale distributed energy storage: First, the large number of individual models leads to a computational dimensionality disaster for the optimization problem, making it difficult for traditional centralized optimization to solve efficiently; second, existing aggregation methods (such as the equivalent parameter method) oversimplify the model, sacrificing dispatching accuracy and easily causing line overloads or voltage limits. In addition, under the traditional market mechanism, there is a lack of dynamic coordination between load aggregators and distribution network operators, making it difficult to respond quickly to congestion problems. For example, the dispatching strategies that use fixed electricity prices or static constraints in existing technologies cannot guide load peak consumption through price signals, resulting in low dispatching efficiency. Summary of the Invention

[0008] In response to the shortcomings of related technologies, the purpose of the present invention is to provide a scheduling method and system for distributed energy storage aggregation to alleviate distribution network congestion, aiming to achieve rapid and economical scheduling of distribution networks, thereby accurately alleviating line congestion, and providing an innovative solution for distribution network scheduling with high-penetration distributed energy storage.

[0009] To achieve the above objectives, in a first aspect, the present invention provides a dispatching method for alleviating distribution network congestion by aggregating distributed energy storage, comprising:

[0010] Step S1: Establish an aggregation model of various distributed energy storage clusters based on the maximum inner approximation method (MIA), construct the aggregation feasible region of each cluster, calculate the baseline power of each cluster, and generate a benchmark load curve;

[0011] Step S2: Inputting the initial power value of each cluster set according to the baseline power into the branch power flow model for power flow calculation; the original objective function of the branch power flow model is to minimize the distribution network operating cost; the original constraints of the branch power flow model include power flow constraints, renewable energy generation constraints, and energy storage system constraints;

[0012] Step S3: Based on the power flow calculation results, the voltage sensitivity factor matrix and the power transfer factor matrix are used to calculate the load rate of each line and the voltage of each node in the distribution network, and determine whether a line overload or a node voltage exceeds a limit. If so, execute step S4; if not, execute step S6;

[0013] Step S4: Based on the original objective function, Lagrangian relaxation is performed on the power balance constraint, node voltage constraint, line capacity constraint, and power generation constraint in the power flow constraint. Combined with the current load curve corresponding to each cluster, the Lagrangian multiplier corresponding to each constraint is calculated, and the node marginal electricity price is calculated based on this. The compensation coefficient corresponding to each cluster is then set according to the node marginal electricity price and the compensation coefficient factor;

[0014] Step S5: Add the compensation cost for each cluster to the original objective function according to the compensation coefficient, and add the aggregate feasible region constraint of each cluster power to the original constraint condition to obtain an updated objective function and constraint condition; recalculate the power flow according to the updated objective function and constraint condition, and update the current load curve corresponding to each cluster, and return to step S3;

[0015] Step S6: Obtain a day-ahead dispatch plan based on the current load curve corresponding to each cluster.

[0016] Optionally, the various types of distributed energy storage clusters include electric vehicle clusters and air conditioning load clusters;

[0017] The method of establishing aggregation models of various distributed energy storage clusters based on the maximum inner approximation (MIA) method includes: establishing a single constraint model for electric vehicles and a single constraint model for air conditioning loads, and then establishing an aggregation model based on the maximum inner approximation (MIA) method according to the single constraint model;

[0018] The single-unit constraint model of the electric vehicle includes energy constraint, charging power constraint, and access-leave time constraint:

[0019] The dispatching operation model of a single electric vehicle i in the distribution network is as follows:

[0020]

[0021] Where, is the energy of the i-th EV in the cluster at time t; represents the variable charging power of the i-th EV at time t; represents the benchmark charging power of the i-th EV, which is equal to the average charging power obtained by dividing the total energy required for charging by the total connection time; They represent the maximum and minimum energy values ​​of the i-th EV in the cluster respectively; Respectively represent the arrival / departure time of the i-th EV charging station; denote the expected charging energy and initial arrival energy of the i-th EV respectively; represents the energy boundary of the i-th EV in the cluster at time t; represents the power boundary of the i-th EV in the cluster at time t;

[0022] The single constraint model of the air conditioning load includes virtual energy storage energy constraint, power constraint and indoor temperature constraint converted from the first-order thermal equivalent model:

[0023]

[0024] in, and is the maximum power and minimum power of the variable frequency air conditioning load, Δt is a single time step; t s and t e The time when the air conditioning load participates in scheduling and the time when scheduling ends.

[0025] Optionally, constructing the aggregate feasible domain of each cluster includes:

[0026] By solving the scaling coefficients and translation vectors corresponding to various distributed energy storage units through linear programming, the H polyhedron feasible domain of the unit is converted into the MIA feasible domain;

[0027] The polyhedron of the i-th distributed energy storage The feasible domain of MIA Expressed as:

[0028]

[0029] By calculating the MIA feasible region of each distributed energy storage in aggregator k The Minkowski sum (M-sum) of , we get the aggregation feasible region of aggregator k:

[0030]

[0031] in, and is the basic constraint matrix, are the optimal scaling factor and translation vector for the MIA problem.

[0032] Optionally, the expression of the original objective function is:

[0033] minC=C grid +C wind_loss +C pv_loss +C ES

[0034]

[0035] Where: C grid The cost of electricity purchased by the distribution network from the upper grid; C wind_loss and C pv_loss is the cost of curtailing wind and solar power; C ES is the operating cost of the energy storage system on the distribution network side; T is the time step set; N bus is the set of distribution network nodes; P t grid and is the active power and reactive power purchased by the root node from the upper power grid at time t; are the amount of wind and solar power curtailment at time t respectively; and They represent the discharge power and charging power of the energy storage system at node i at time t respectively; The price of active power and reactive power purchased by the distribution network from the upper power grid; c wind_loss 、c pv_loss are the penalty coefficients for wind and solar curtailment, respectively; c es is the unit operating cost of the energy storage system on the distribution network side; among them, each distributed energy storage consumes electricity according to the baseline power.

[0036] Optionally, the original constraints include: MIA aggregation feasible region constraints of each distributed energy storage cluster, power flow constraints, second-order cone constraints, line current constraints, new energy output constraints, and distribution network side energy storage system constraints;

[0037] The MIA aggregation feasible domain constraints of each distributed energy storage cluster are:

[0038]

[0039] in, are the variable active power vectors of the kth air conditioning load cluster and electric vehicle cluster respectively; are the translation vectors of the kth air conditioning load cluster and electric vehicle cluster respectively; are the scaling factors of the kth air conditioning load cluster and electric vehicle cluster, respectively;

[0040] The power flow constraint is:

[0041]

[0042] Among them, U i,t represents the voltage of node i at time t, u i,t represents the square of the voltage at node i at time t; I ij,t is the current of line ij at time t; w ij,t is the square of the current in line ij at time t; N line is the distribution network line set; φ par (i) and φ chi (i) represents the parent node set and child node set of node i, respectively, where the parent node refers to the node upstream of node i in the distribution network, and the child node refers to the node downstream of node i; h and g are the node numbers in the parent node set and child node set of node i, respectively; and They represent the active power and reactive power of line ig at time t respectively; r hi and x hi They represent the resistance and reactance of line hi respectively; and They represent the active power and reactive power of the conventional load at node i at time t respectively; and They represent the active power output of wind power generation and photovoltaic power generation at node i at time t. If there is no wind power station or photovoltaic power station connected to node i, the value is 0; They represent the variable active power and reference active power of the electric vehicle cluster at node i at time t respectively; They represent the variable active power and reference active power of the air conditioning load cluster at node i at time t respectively; They are respectively the active and reactive power purchased from the upper power grid;

[0043] The second-order cone constraint is:

[0044]

[0045] The line current constraint is:

[0046]

[0047] Among them, I ij,max Indicates the upper limit of the line ij current;

[0048] The new energy output constraints are:

[0049]

[0050] in, and are the minimum and maximum active power output of wind power generation at node i at time t, and are the minimum and maximum active power output of photovoltaic power generation at node i at time t;

[0051] The distribution grid side energy storage system constraints are:

[0052]

[0053] Where, is the energy stored on the distribution network side at time t; is the charging and discharging efficiency of the energy storage on the distribution network side; Δt is the optimization time interval; and are the charging and discharging power of the energy storage on the distribution network side at time t respectively; It is a binary variable. When the value is 1, it means that the ES is in the charging state. When the value is 0, it means that the ES is in the discharging state. The maximum charge and discharge power of the energy storage on the distribution network side; and They are the upper and lower limits of the energy storage capacity on the distribution network side; and are the energy stored on the distribution network side at the beginning and end of the period respectively.

[0054] Optionally, determining whether a line overload or a node voltage exceeds a limit includes:

[0055] Based on the linearized power flow model, according to the system topology and node power balance, the relationship between line flow and node injection amount is calculated as follows:

[0056] P f =F LSF ×P inj

[0057] Q f =F LSF ×Q inj

[0058] ΔU=RP f +XQ f

[0059] Among them, P inj and Q inj are the node active power injection matrix and the node reactive power injection matrix respectively; ΔU is the voltage change vector of each node in the distribution system relative to the root node; R and X are the matrices composed of the resistance and reactance between any two nodes in the distribution system respectively; F LSF is the load transfer factor, which represents the proportional coefficient of the line power flow change relative to the node injection load change;

[0060] If the load rate of each line in the distribution network does not meet the following judgment formula, it is determined that the line is overloaded;

[0061]

[0062] in, Indicates the maximum load of line ij;

[0063] If the voltage of each node does not satisfy the following judgment formula, it is determined that the node voltage exceeds the limit;

[0064]

[0065] Among them, ΔU i,max , ΔU i,min They represent the maximum and minimum voltage change limits of node i relative to the root node.

[0066] Optionally, the process of solving the node marginal electricity price includes:

[0067] Linearize the line capacity constraint using a 12-sided regular polygon to generate a linear constraint:

[0068]

[0069] Among them, α c,0 , α c,1 , α c,2 is a fixed parameter.

[0070] Optionally, the node marginal electricity price is obtained by calculating the node marginal electricity price, including:

[0071] Perform Lagrangian relaxation on the original objective function, and solve the dual multiplier based on the sensitivity factor and the transmission factor to calculate the node marginal price DLMP of active power:

[0072]

[0073] in, is the node marginal price DLMP; is the Lagrange multiplier for the power balance constraint; is the Lagrange multiplier of the node voltage constraint; is the Lagrange multiplier of line capacity constraint; S SFvp S is the sensitivity factor of node voltage change to node injected active power; SFlp The sensitivity factor of the line power flow change to the active power injected into the node; represents the transfer factor of active power at node i at time t.

[0074] Optionally, adding a compensation cost for each cluster to the original objective function according to the compensation coefficient includes:

[0075] The compensation cost is:

[0076]

[0077] Among them, C EV is the compensation cost of electric vehicle cluster, C TCL is the compensation cost of the air conditioning load cluster, δ EV and δ TCL are the compensation coefficient factors of the electric vehicle cluster and the air conditioning load cluster respectively;

[0078] Update the current load curve corresponding to each cluster. The expression is:

[0079] minC=C grid +C wind_loss +C pv_loss +C ES +C EV +C TCL .

[0080] In a second aspect, the present invention further provides a dispatching system for alleviating distribution network congestion by aggregating distributed energy storage, comprising:

[0081] The aggregation model building module is used to establish the aggregation model of various distributed energy storage clusters based on the maximum inner approximation method (MIA), construct the aggregation feasible region of each cluster, calculate the baseline power of each cluster and generate the benchmark load curve;

[0082] A power flow calculation module is configured to input the initial power values ​​of each cluster set according to the baseline power into a branch power flow model for power flow calculation; the original objective function of the branch power flow model is to minimize the operating cost of the distribution network; the original constraints of the branch power flow model include power flow constraints, renewable energy generation constraints, and energy storage system constraints;

[0083] The judgment module is used to calculate the load rate of each line and the voltage of each node in the distribution network based on the power flow calculation results, using the voltage sensitivity factor matrix and the power transfer factor matrix, and determine whether the line is overloaded or the node voltage exceeds the limit. If so, it enters the node marginal electricity price compensation module to execute the corresponding steps; if not, it enters the scheduling plan generation module to execute the corresponding steps;

[0084] The node marginal electricity price compensation module is used to perform Lagrangian relaxation on the power balance constraint, node voltage constraint, line capacity constraint, and power generation constraint in the power flow constraint based on the original objective function, calculate the Lagrangian multiplier corresponding to each constraint in combination with the current load curve corresponding to each cluster, and calculate the node marginal electricity price based on this; and then set the compensation coefficient corresponding to each cluster according to the node marginal electricity price and the compensation coefficient factor;

[0085] a compensation update module, configured to add the compensation cost for each cluster to the original objective function according to the compensation coefficient, and to add the aggregate feasible region constraint of each cluster power to the original constraint condition, to obtain an updated objective function and constraint condition; recalculate the power flow according to the updated objective function and constraint condition, and update the current load curve corresponding to each cluster, and then return to the judgment module to execute the corresponding steps;

[0086] The scheduling plan generation module is used to obtain the day-ahead scheduling plan based on the current load curve corresponding to each cluster.

[0087] Compared with the prior art, the above technical solutions conceived by the present invention can achieve the following beneficial effects:

[0088] 1. An embodiment of the present invention provides a scheduling method for alleviating distribution network congestion by aggregating distributed energy storage. An aggregation model based on the maximum inner approximation (MIA) is established for various types of distributed energy storage (electric vehicles, air conditioning loads). The flow calculation is performed according to the branch flow model. Based on the flow calculation results, the voltage sensitivity factor matrix and the power transmission factor matrix are used to calculate the load rate of each line and the voltage of each node in the distribution network. If a line overload occurs or the node voltage exceeds the limit, the node marginal price is calculated, and the compensation coefficient of each distributed energy storage cluster is set according to the node marginal price, thereby adjusting its variable power. After compensation and adjustment, the flow calculation is performed again to obtain the load curve, line load, and node voltage. It is checked again whether the line is overloaded or the voltage exceeds the limit, and the iteration is performed until the above situation no longer occurs. This method can greatly reduce the amount of calculation, complete the iteration of the day-ahead scheduling plan between the distribution network operator and the market participants at a faster speed, and solve the distribution network line congestion problem with a more efficient market mechanism.

[0089] 2. Embodiments of the present invention provide a scheduling method for alleviating distribution network congestion by aggregating distributed energy storage. By integrating MIA aggregation technology with a node marginal price mechanism, this method alleviates distribution network line congestion, achieving the following beneficial effects: 1. Improved scheduling efficiency and accuracy: The aggregation model constructed based on MIA compresses the high-dimensional optimization problem to the aggregator dimension while preserving the operating constraints of individual units such as electric vehicles and air conditioning loads, reducing computational complexity. 2. Accurate decomposition of scheduling instructions: Using the scaling coefficients and translation vectors calculated during the MIA aggregation process, the optimized power at the aggregator level can be reversely decomposed into each distributed energy storage unit, generating a charging and discharging plan that meets individual constraints. Compared to traditional equivalent parameter methods (which only output cluster totals), this invention solves the "aggregation-decomposition" bidirectional mapping problem. 3. Coordinated management of bidirectional congestion: By combining the spatiotemporal signal characteristics of node marginal prices, it can simultaneously address both forward and reverse congestion. When the node marginal price is greater than 0, a negative compensation coefficient is used to suppress load congestion. When the node marginal price is less than 0 (such as in renewable energy reverse transmission scenarios), a positive compensation coefficient is used to incentivize energy storage charging to absorb excess power. This method expands the coverage of congestion management scenarios from the single forward mode of traditional methods to bidirectional full scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] Figure 1 A schematic flow chart of a method for alleviating distribution network congestion by aggregating distributed energy storage in the present invention;

[0091] Figure 2 This is a schematic diagram of the linearization principle of the quadratic capacity constraint in the present invention. DETAILED DESCRIPTION

[0092] In order to make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0093] The contents involved in the above embodiment are described below in conjunction with a preferred embodiment.

[0094] Example 1

[0095] This paper proposes a scheduling method for alleviating distribution network congestion by aggregating distributed energy storage. By constructing a precise aggregation feasible domain, the computational complexity is reduced while retaining the individual constraint characteristics. The locational marginal price (LMP) can reflect the supply and demand balance and operating costs of the power system from both spatial and temporal perspectives. Combined with its dynamic feedback mechanism, it can achieve rapid and economical scheduling of the distribution network, thereby accurately alleviating line congestion, providing an innovative solution for distribution network scheduling with high-penetration distributed energy storage.

[0096] A dispatching method for alleviating distribution network congestion by aggregating distributed energy storage, comprising:

[0097] Step S1: Establish an aggregation model of various distributed energy storage clusters based on the maximum inner approximation method (MIA), construct the aggregation feasible region of each cluster, calculate the baseline power of each cluster, and generate a benchmark load curve;

[0098] Step S2: Inputting the initial power value of each cluster set according to the baseline power into the branch power flow model for power flow calculation; the original objective function of the branch power flow model is to minimize the distribution network operating cost; the original constraints of the branch power flow model include power flow constraints, renewable energy generation constraints, and energy storage system constraints;

[0099] Step S3: Based on the power flow calculation results, the voltage sensitivity factor matrix and the power transfer factor matrix are used to calculate the load rate of each line and the voltage of each node in the distribution network, and determine whether a line overload or a node voltage exceeds a limit. If so, execute step S4; if not, execute step S6;

[0100] Step S4: Based on the original objective function, Lagrangian relaxation is performed on the power balance constraint, node voltage constraint, line capacity constraint, and power generation constraint in the power flow constraint. Combined with the current load curve corresponding to each cluster, the Lagrangian multiplier corresponding to each constraint is calculated, and the node marginal electricity price is calculated based on this. The compensation coefficient corresponding to each cluster is then set according to the node marginal electricity price and the compensation coefficient factor;

[0101] Step S5: Add the compensation cost for each cluster to the original objective function according to the compensation coefficient, and add the aggregate feasible region constraint of each cluster power to the original constraint condition to obtain an updated objective function and constraint condition; recalculate the power flow according to the updated objective function and constraint condition, and update the current load curve corresponding to each cluster, and return to step S3;

[0102] Step S6: Obtain a day-ahead dispatch plan based on the current load curve corresponding to each cluster.

[0103] In this invention, the loads connected to each node in the distribution network can be divided into two types: adjustable loads and conventional loads. Conventional loads are considered to have their power consumption plans not subject to change during both day-ahead and intraday scheduling. Adjustable loads are categorized into two typical load types: electric vehicle loads and variable-frequency air conditioning loads. Furthermore, the distribution network is equipped with energy storage power stations for dispatch by distribution network operators.

[0104] The maximum inner approximation-based aggregation process for each distributed energy storage cluster can be represented by the following general process:

[0105] First, the decision variables of each distributed energy storage i The feasible domain of can be expressed by H polyhedron as:

[0106]

[0107] in, and is the constraint matrix derived from the constraints of i.

[0108] First, a basic polyhedron is constructed by averaging the parameters of all distributed energy storages in the same aggregator k.

[0109]

[0110] in, and is the basic constraint matrix:

[0111]

[0112] Then, by scaling and translating the basic polyhedron Get the MIA feasible region of each distributed energy storage i And the corresponding scaling coefficient is obtained by solving the following linear programming and translation vectors This linear programming is based on Farkas' Lemma derivation.

[0113]

[0114] in, are the optimal scaling factor and translation vector for the MIA problem.

[0115] From this we can get the polyhedron of the i-th distributed energy storage The feasible domain of MIA Expressed as:

[0116]

[0117] Finally, the MIA feasible region of each distributed energy storage in aggregator k is calculated The Minkowski sum (M-sum) of , we can get the aggregation feasible domain of aggregator k:

[0118]

[0119] Among them, according to the scaling coefficient and translation vector of each monomer i, the relationship between the monomer state variable and the cluster state variable can also be derived:

[0120] Furthermore, the initial condition (1) of the monomer model in the aggregation model is derived from its own constraints. The constraints contained in various energy storage models are as follows: electric vehicle monomer model constraints: energy constraints, charging constraints, access-leave time constraints; variable frequency air conditioner monomer virtual energy storage model constraints: first-order thermal equivalent model heat-to-electricity conversion constraints, energy constraints, variable power constraints, indoor temperature constraints. In order to facilitate subsequent analysis and calculation, the first-order thermal equivalent model of the monomer model of the variable frequency air conditioner is converted into an electrical model without a thermodynamic model. The MIA aggregation derivation process of each distributed energy storage is as follows:

[0121] (1) Variable frequency air conditioning load model:

[0122] The first-order equivalent thermal parameter model is used to describe the dynamic response characteristics of a single variable-frequency air conditioner and is discretized. Its expression is:

[0123]

[0124] Where: T t is the indoor temperature of the air conditioning load at time t; T t out is the outdoor temperature at time t; R and C are the equivalent thermal resistance and heat capacity respectively; is the cooling capacity of the air conditioner at time t.

[0125] The electric power and cooling capacity of the air conditioning load can be approximately expressed as a linear function of the compressor frequency, and the expression is:

[0126] (8)

[0127] Among them, k1, k2, l1, and l2 are parameters related to the air conditioning load itself and are constant values. Users can set the most suitable temperature T* and fluctuation range ΔT according to their own comfort needs, thereby determining the maximum indoor temperature T max =T * +ΔT and minimum value T min =T * -ΔT. Energy state SOC value S of air conditioning load tIt represents the level of thermal energy stored, which is measured by the change of indoor temperature. Its expression is:

[0128]

[0129] is the virtual adjustable power value of the air conditioning load at time t, satisfying When the indoor temperature is continuously at the optimum temperature T*, the base power consumed by the air conditioning load is for:

[0130]

[0131] After the above mathematical derivation, the first-order thermal equivalent model of the i-th variable-frequency air conditioner can be converted into a virtual energy storage constraint model:

[0132]

[0133] in:

[0134] in, and is the maximum power and minimum power of the variable frequency air conditioning load, and Δt is a single time step. s and t e The time when the air conditioning load participates in the scheduling and the time when the scheduling ends, the SOC value at the beginning and end and Keep them equal and set them uniformly to 1, which means that the indoor temperature where the air conditioning load is located is the lowest point.

[0135] Based on this, MIA polymerization is performed to obtain the general formula (1) of the variable frequency air conditioner monomer model:

[0136]

[0137] Where T is the total time step, and the expression of the relevant parameter matrix is:

[0138]

[0139] Derived the load of variable frequency air conditioner unit After the matrix is ​​constructed, mathematical calculations of equations (2) to (4) are performed to obtain the feasible domain of the variable power unit i and the aggregator k of the air conditioning load:

[0140]

[0141] in, are the variable active power vectors of the kth air conditioning load cluster and electric vehicle cluster respectively; are the translation vectors of the kth air conditioning load cluster and electric vehicle cluster respectively; are the scaling factors of the kth air conditioning load cluster and electric vehicle cluster, respectively.

[0142] (2) Electric vehicle load model:

[0143] The plug-in and unplugging behavior of electric vehicles in the power grid can be regarded as an adjustable energy storage resource. In this invention, the electric vehicle cluster can only adjust its charging power without considering the discharge situation. The scheduling operation model of a single electric vehicle i in the distribution network is as follows:

[0144]

[0145] in,

[0146] Where, is the energy of the i-th EV in the cluster at time t; represents the variable charging power of the i-th EV at time t; represents the benchmark charging power of the i-th EV, which is equal to the average charging power obtained by dividing the total energy required for charging by the total connection time; They represent the maximum and minimum energy values ​​of the i-th EV in the cluster respectively; Respectively represent the arrival / departure time of the i-th EV charging station; denote the expected charging energy and initial arrival energy of the i-th EV respectively; represents the energy boundary of the i-th EV in the cluster at time t; represents the power boundary of the i-th EV in the cluster at time t; The physical meaning of the second term in the equation is to limit the maximum charging energy of the electric vehicle. The physical meaning of the second term in the calculation formula is to limit the energy of the electric vehicle to no less than the expected energy value when leaving the station.

[0147] Among them, the baseline power of the i-th air conditioning load unit at time t is: k1, k2, l1, l2 are the factory-provided correlation coefficients of variable frequency air conditioner load; T t out is the outdoor temperature at time t; T * is the user-set temperature; R is the equivalent thermal resistance.

[0148] Based on this, MIA polymerization was performed to obtain the general formula (1) of the electric vehicle monomer model:

[0149]

[0150] Where T represents the total time step,

[0151] Derived electric vehicle single load After the matrix is ​​constructed, mathematical calculations of equations (2) to (4) are performed to obtain the feasible domain of the variable power monomer i and aggregator k of the electric vehicle load:

[0152]

[0153] Furthermore, after the aggregation model is built, during the day-ahead phase, the distribution network operator optimizes the system with the objective function of minimizing operating costs. The costs include the cost of purchasing electricity, the cost of curtailing wind and solar power, and the cost of operating the energy storage system. Each distributed energy storage system uses electricity according to the benchmark power:

[0154]

[0155] Where: C grid The cost of electricity purchased by the distribution network from the upper grid; C wind_loss and C pv_loss is the cost of curtailing wind and solar power; C ES is the operating cost of the energy storage system on the distribution network side; T is the time step set; N bus is the set of distribution network nodes; P t grid and is the active power and reactive power purchased by the root node from the upper power grid at time t; are the amount of wind and solar power curtailment at time t, respectively. If there is no wind power station or photovoltaic power station connected to node i, the values ​​are both 0; and They represent the discharge power and charging power of the energy storage system at node i at time t. If there is no energy storage system connected to node i, the values ​​are both 0; The price of active power and reactive power purchased by the distribution network from the upper power grid; c wind_loss 、c pv_loss are the penalty coefficients for wind and solar curtailment, respectively; c es is the unit operating cost of the energy storage system on the distribution network side.

[0156] By integrating MIA aggregation technology with a node marginal price mechanism to alleviate distribution network line congestion, the following beneficial effects can be achieved: 1. Improved dispatch efficiency and accuracy: The aggregation model constructed based on MIA compresses the high-dimensional optimization problem to the aggregator dimension while retaining the operating constraints of individual units such as electric vehicles and air conditioning loads, reducing computational complexity. 2. Achieve precise decomposition of dispatch instructions: The scaling coefficients and translation vectors calculated through the MIA aggregation process can reversely decompose the optimized power at the aggregator level to each distributed energy storage unit, generating a charging and discharging plan that meets individual constraints. Compared with the traditional equivalent parameter method (which can only output the total cluster amount), this invention solves the "aggregation-decomposition" bidirectional mapping problem. 3. Coordinated management of bidirectional congestion: Combining the spatiotemporal signal characteristics of node marginal prices, it can simultaneously respond to forward and reverse congestion: When the node marginal price is greater than 0, a negative compensation coefficient is used to suppress the load from exacerbating congestion; when the node marginal price is less than 0 (such as in the scenario of reverse power transmission from new energy sources), a positive compensation coefficient is used to incentivize energy storage charging to absorb excess power. This method expands the coverage of congestion management scenarios from the single forward mode of traditional methods to bidirectional full scenarios.

[0157] Furthermore, the original constraints include the MIA aggregation feasible domain constraints of each distributed energy storage cluster, power flow constraints, line current constraints, new energy output constraints, and distribution network side energy storage system constraints.

[0158] (1) Constraints on the MIA aggregation feasible region of each adjustable load cluster:

[0159]

[0160] in, are the variable active power vectors of the kth air conditioning load cluster and electric vehicle cluster respectively; are the translation vectors of the kth air conditioning load cluster and electric vehicle cluster respectively; are the scaling factors of the kth air conditioning load cluster and electric vehicle cluster, respectively.

[0161] (2) Power flow constraints:

[0162] The branch flow model has high accuracy when used for radial power line network calculations. Therefore, the present invention selects the branch flow model for power flow calculations. The branch flow model has the following constraints:

[0163]

[0164] Among them: U i,t represents the voltage of node i at time t, u i,t represents the square of the voltage at node i at time t; I ij,t is the current of line ij at time t; w ij,tis the square of the current in line ij at time t; N line is the distribution network line set; φ par (i) and φ chi (i) represents the parent node set and child node set of node i, respectively, where the parent node refers to the node upstream of node i in the distribution network, and the child node refers to the node downstream of node i; h and g are the node numbers in the parent node set and child node set of node i, respectively; and They represent the active power and reactive power of line ig at time t respectively; r hi and x hi They represent the resistance and reactance of line hi respectively; and They represent the active power and reactive power of the conventional load at node i at time t respectively; and They represent the active power output of wind power generation and photovoltaic power generation at node i at time t. If there is no wind power station or photovoltaic power station connected to node i, the value is 0; They represent the variable active power and reference active power of the electric vehicle cluster at node i at time t. If there is no electric vehicle cluster connected to node i, the values ​​are both 0; They represent the variable active power and reference active power of the air-conditioning load cluster at node i at time t. If there is no air-conditioning load cluster connected to node i, the values ​​are both 0; They are respectively the active and reactive power purchased from the upper power grid, and are not 0 only at the root node.

[0165] The above power flow constraints contain quadratic terms and are non-convex. Conventional algorithms are not effective for solving optimization problems with non-convex constraints. However, when the objective function is convex, commercial solvers provide exact optimal solutions to second-order cone programming problems. Therefore, non-convex power flow constraints can be converted to second-order cone constraints through phase relaxation and second-order cone relaxation:

[0166] Relaxing Equation (28) to a rotational second-order cone constraint yields:

[0167]

[0168] The above formula can be written in standard second-order cone form, namely:

[0169]

[0170] (3) Line current constraints:

[0171]

[0172] Among them, I ij,max Indicates the upper limit of the line ij current.

[0173] (4) Constraints on new energy output:

[0174]

[0175] in, and are the minimum and maximum active power output of wind power generation at node i at time t, and are the minimum and maximum active power output of photovoltaic power generation at node i at time t, respectively.

[0176] (5) Constraints on the energy storage system on the distribution network side:

[0177]

[0178] Where, is the energy stored on the distribution network side at time t; is the charging and discharging efficiency of the energy storage on the distribution network side; Δt is the optimization time interval; and are the charging and discharging power of the energy storage on the distribution network side at time t respectively; It is a binary variable. When the value is 1, it means that the ES is in the charging state. When the value is 0, it means that the ES is in the discharging state. The maximum charge and discharge power of the energy storage on the distribution network side; and They are the upper and lower limits of the energy storage capacity on the distribution network side; and The energy stored on the distribution network side at the beginning and end of the period is required to return to the initial value after a scheduling cycle to ensure the sustainability of scheduling.

[0179] Furthermore, based on the linearized power flow model, we can clearly calculate the relationship between line flow and node injection according to the system topology and node power balance:

[0180] P f =F LSF ×P inj (35)

[0181] Q f =F LSF ×Q inj (36)

[0182] ΔU=RP f +XQ f (37)

[0183] Among them, P inj and Q injare the node active power injection matrix and the node reactive power injection matrix respectively; ΔU is the voltage change vector of each node in the distribution system relative to the root node; R and X are the matrices composed of the resistance and reactance between any two nodes in the distribution system respectively; F LSF is the load transfer factor, which represents the proportional coefficient of line power flow change relative to the node injection load change.

[0184] If any line in the power flow calculation results does not meet the following requirements:

[0185]

[0186] in, Indicates the maximum load of line ij.

[0187] Or any node does not satisfy:

[0188]

[0189] Among them, ΔU i,max , ΔU i,min They represent the maximum and minimum voltage change limits of node i relative to the root node.

[0190] At this point, it can be considered that a line overload or node voltage exceeding the limit has occurred in the distribution network, and the calculation of the node marginal electricity price begins.

[0191] Furthermore, the calculation process of the node marginal price is as follows: Lagrangian relaxation is performed on the original objective function, and the node marginal price is calculated by the dual multiplier obtained by optimization. The following is the model for calculating the node marginal price:

[0192] (1) Linearized capacity constraints

[0193] The feasible region defined by constraint (38) is the interior of a circle whose radius is equal to like Figure 2 As shown. Using a regular polygon with 12 sides to linearize the quadratic capacity constraint, we get the linear constraint:

[0194]

[0195] Since the vertices of the regular polygon are all located on the circumference, the coordinates of the regular polygon and the coefficients of the constraint condition (40) can be obtained. The specific parameters are shown in Table 1.

[0196] Table 1

[0197] c <![CDATA[α c,0 ]]> <![CDATA[α c,1 ]]> <![CDATA[α c,2 ]]> 1 1 0.2679 -1 2 1 1 -1.366 3 0.2679 1 -1 4 -0.2679 1 -1 5 -1 1 -1.366 6 -1 0.2679 -1 7 -1 -0.2679 -1 8 -1 -1 -1.366 9 -0.2679 -1 -1 10 0.2679 -1 -1 11 1 -1 -1.366 12 1 -0.2679 -1

[0198] (2) Calculation of power consumption and output

[0199] According to equations (25) and (26), the active power consumption and active power output of node i at time t are defined as:

[0200]

[0201] The reactive power consumption and reactive power output of node i at time t are:

[0202]

[0203] (3) Sensitivity factor

[0204] According to formulas (35), (36), and (37), the following sensitivity factors can be derived:

[0205]

[0206] Where S SFvp and S SFvq S are the sensitivity factors of node voltage change to node injected active power and node injected reactive power respectively; SFlp and S SFlq are the sensitivity factors of line power flow changes to the node injected active power and node injected reactive power, respectively.

[0207] (4) Transfer factor:

[0208] The transfer factor represents the ratio of the power transmitted in the network to the power injected into the node, taking into account network losses. The transfer factor of active power and reactive power at node i can be expressed as:

[0209]

[0210] Where, and They represent the active power and reactive power of line l calculated in the distribution network optimization scheduling model, and are constants in the process of calculating the node marginal electricity price.

[0211] (5) Power loss and virtual node load requirements:

[0212]

[0213] Where N(i) represents the set of nodes on the same line as node i, P loss,t and Q loss,t They represent the active and reactive network losses calculated in the distribution network optimization scheduling model, and are constants in the process of calculating the node marginal electricity price.

[0214] The concept of virtual node load demand is to evenly distribute the power loss of a line to the nodes on both sides of the line, and use virtual node injection power to represent the line network loss. The virtual node load demand of active power and reactive power can be expressed as:

[0215]

[0216] The constraints for calculating the node marginal electricity price model are as follows. The variables in brackets after the constraints are dual variables.

[0217] (1) Power balance constraints:

[0218]

[0219]

[0220] (2) Node voltage constraints:

[0221]

[0222] Where U 1,t Represents the voltage of the root node of the distribution line at time t.

[0223] (3) Line capacity constraints:

[0224]

[0225] (4) Maximum and minimum power generation constraints:

[0226]

[0227] According to the above optimization model, the final relaxed Lagrangian function is:

[0228]

[0229] Among them, C(x,y * ) is the initial objective function, where x is the variable and y * is a constant. The variables include the Lagrange multipliers and Constants include P loss,t 、 Q loss,t 、 and other known parameters.

[0230] DLMP is the Lagrangian function of active power consumption The first-order partial derivative of can be calculated as follows:

[0231]

[0232] From the constraints corresponding to each Lagrange multiplier, the above formula can be decomposed into energy price Congestion Price Voltage support price and loss price

[0233]

[0234] When π i,t ≥0 reflects one or more of the following situations: at time t, node i is located in a load-concentrated area and must purchase electricity from the upstream grid, resulting in a high wholesale market LMP; the capacity of the line downstream of node i is limited, and increasing generation or reducing demand can alleviate congestion (a positive congestion price); node i is far from the power source, resulting in higher line loss costs for power supply (a positive loss price). In these situations, price signals are needed to suppress demand or incentivize supply, so the variable active power of the distributed energy storage cluster is optimized towards a reduced value.

[0235] When π i,t ≤0 reflects one or more of the following situations: at time t, the wind and photovoltaic output at node i exceeds the local load demand, resulting in reverse power flow and line overload; the system encourages energy storage charging through negative electricity prices to absorb excess power; node i is close to the power source, and injected power reduces network losses (the loss price is negative). In these cases, price signals are needed to suppress demand or incentivize supply, so the variable active power of the distributed energy storage cluster will be optimized towards increasing it.

[0236] According to the DLMP obtained by the above calculation, the compensation coefficients of the electric vehicle cluster and the air conditioning load cluster are set, and the compensation cost for the distributed energy storage cluster is added to the original objective function (20):

[0237] minC=C grid +C wind_loss +C pv_loss +C ES +C EV +C TCL (67)

[0238]

[0239] Among them, δ EV and δ TCL are the compensation coefficient factors of the electric vehicle cluster and the air conditioning load cluster respectively.

[0240] Combined with the constraints of equations (21) to (34), the power flow calculation is re-performed to obtain the updated load curves of each distributed energy storage cluster. The load of each line and the voltage offset of each node are calculated by equations (35) to (39) and the over-limit situation is checked. The above iterative process is continued until there is no line overload and node voltage over-limit in the distribution network.

[0241] An embodiment of the present invention provides a scheduling method for alleviating distribution network congestion by aggregating distributed energy storage. This method establishes a maximum inner approximation (MIA)-based aggregation model for various types of distributed energy storage (electric vehicles, air conditioning loads), performs power flow calculations based on branch power flow models, and uses a voltage sensitivity factor matrix and a power transfer factor matrix to calculate the load factor of each line and the voltage at each node in the distribution network based on the power flow calculation results. If a line is overloaded or a node voltage exceeds a limit, the node marginal price is calculated and used to set the compensation coefficient for each distributed energy storage cluster, thereby adjusting its variable power. After compensation adjustment, power flow calculations are performed again to obtain load curves, line loads, and node voltages. Line overloads and voltage limits are then checked again, and iterations are repeated until these conditions no longer occur. This method significantly reduces computational complexity, enabling faster iteration of day-ahead scheduling plans between distribution network operators and market participants, and resolving distribution network line congestion issues with a more efficient market mechanism. This method accurately alleviates line congestion and provides an innovative solution for distribution network scheduling with high penetration of distributed energy storage.

[0242] Example 2

[0243] A dispatching system for alleviating distribution network congestion by aggregating distributed energy storage, comprising:

[0244] The aggregation model building module is used to establish the aggregation model of various distributed energy storage clusters based on the maximum inner approximation method (MIA), construct the aggregation feasible region of each cluster, calculate the baseline power of each cluster and generate the benchmark load curve;

[0245] A power flow calculation module is configured to input the initial power values ​​of each cluster set according to the baseline power into a branch power flow model for power flow calculation; the original objective function of the branch power flow model is to minimize the operating cost of the distribution network; the original constraints of the branch power flow model include power flow constraints, renewable energy generation constraints, and energy storage system constraints;

[0246] The judgment module is used to calculate the load rate of each line and the voltage of each node in the distribution network based on the power flow calculation results, using the voltage sensitivity factor matrix and the power transfer factor matrix, and determine whether the line is overloaded or the node voltage exceeds the limit. If so, it enters the node marginal electricity price compensation module to execute the corresponding steps; if not, it enters the scheduling plan generation module to execute the corresponding steps;

[0247] The node marginal electricity price compensation module is used to perform Lagrangian relaxation on the power balance constraint, node voltage constraint, line capacity constraint, and power generation constraint in the power flow constraint based on the original objective function, calculate the Lagrangian multiplier corresponding to each constraint in combination with the current load curve corresponding to each cluster, and calculate the node marginal electricity price based on this; and then set the compensation coefficient corresponding to each cluster according to the node marginal electricity price and the compensation coefficient factor;

[0248] a compensation update module, configured to add the compensation cost for each cluster to the original objective function according to the compensation coefficient, and to add the aggregate feasible region constraint of each cluster power to the original constraint condition, to obtain an updated objective function and constraint condition; recalculate the power flow according to the updated objective function and constraint condition, and update the current load curve corresponding to each cluster, and then return to the judgment module to execute the corresponding steps;

[0249] The scheduling plan generation module is used to obtain the day-ahead scheduling plan based on the current load curve corresponding to each cluster.

[0250] A scheduling system for alleviating distribution network congestion through distributed energy storage aggregation provided by an embodiment of the present invention is used to execute a scheduling method for alleviating distribution network congestion through distributed energy storage aggregation, and has similar beneficial effects.

[0251] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for alleviating distribution network congestion by using distributed energy storage aggregation, characterized in that: include: Step S1: Establish an aggregation model of various distributed energy storage clusters based on the maximum inner approximation method (MIA), construct the aggregation feasible region of each cluster, calculate the baseline power of each cluster, and generate a benchmark load curve; Step S2: Inputting the initial power value of each cluster set according to the baseline power into the branch power flow model for power flow calculation; the original objective function of the branch power flow model is to minimize the distribution network operating cost; the original constraints of the branch power flow model include power flow constraints, renewable energy generation constraints, and energy storage system constraints; Step S3: Based on the power flow calculation results, the voltage sensitivity factor matrix and the power transfer factor matrix are used to calculate the load rate of each line and the voltage of each node in the distribution network, and determine whether a line overload or a node voltage exceeds a limit. If so, execute step S4; if not, execute step S6; Step S4: Based on the original objective function, Lagrangian relaxation is performed on the power balance constraint, node voltage constraint, line capacity constraint, and power generation constraint in the power flow constraint. Combined with the current load curve corresponding to each cluster, the Lagrangian multiplier corresponding to each constraint is calculated, and the node marginal electricity price is calculated based on this. The compensation coefficient corresponding to each cluster is then set according to the node marginal electricity price and the compensation coefficient factor; Step S5: Add the compensation cost for each cluster to the original objective function according to the compensation coefficient, and add the aggregate feasible region constraint of each cluster power to the original constraint condition to obtain an updated objective function and constraint condition; recalculate the power flow according to the updated objective function and constraint condition, and update the current load curve corresponding to each cluster, and return to step S3; Step S6: Obtain a day-ahead dispatch plan based on the current load curve corresponding to each cluster.

2. The scheduling method according to claim 1, wherein: The various types of distributed energy storage clusters include electric vehicle clusters and air conditioning load clusters; The method of establishing aggregation models of various distributed energy storage clusters based on the maximum inner approximation (MIA) method includes: establishing a single constraint model for electric vehicles and a single constraint model for air conditioning loads, and then establishing an aggregation model based on the maximum inner approximation (MIA) method according to the single constraint model; The single-unit constraint model of the electric vehicle includes energy constraint, charging power constraint, and access-leave time constraint: The dispatching operation model of a single electric vehicle i in the distribution network is as follows: Where, is the energy of the i-th EV in the cluster at time t; represents the variable charging power of the i-th EV at time t; represents the benchmark charging power of the i-th EV, which is equal to the average charging power obtained by dividing the total energy required for charging by the total connection time; They represent the maximum and minimum energy values ​​of the i-th EV in the cluster respectively; Respectively represent the arrival / departure time of the i-th EV charging station; denote the expected charging energy and initial arrival energy of the i-th EV respectively; represents the energy boundary of the i-th EV in the cluster at time t; represents the power boundary of the i-th EV in the cluster at time t; The single constraint model of the air conditioning load includes virtual energy storage energy constraint, power constraint and indoor temperature constraint converted from the first-order thermal equivalent model: in, and is the maximum power and minimum power of the variable frequency air conditioning load, Δt is a single time step; t s and t e The time when the air conditioning load participates in scheduling and the time when scheduling ends.

3. The scheduling method according to claim 1, wherein: The aggregation feasible domain of each cluster is constructed as follows: By solving the scaling coefficients and translation vectors corresponding to various distributed energy storage units through linear programming, the H polyhedron feasible domain of the unit is converted into the MIA feasible domain; The polyhedron of the i-th distributed energy storage The feasible domain of MIA Expressed as: By calculating the MIA feasible region of each distributed energy storage in aggregator k The Minkowski sum (M-sum) of , we get the aggregation feasible region of aggregator k: in, and is the basic constraint matrix, are the optimal scaling factor and translation vector for the MIA problem.

4. The scheduling method according to claim 1, wherein: The expression of the original objective function is: minC=C grid +C wind_loss +C pv_loss +C ES Where: C grid The cost of electricity purchased by the distribution network from the upper grid; C wind_loss and C pv_loss is the cost of curtailing wind and solar power; C ES is the operating cost of the energy storage system on the distribution network side; T is the time step set; N bus is the set of distribution network nodes; P t grid and is the active power and reactive power purchased by the root node from the upper power grid at time t; are the amount of wind and solar power curtailment at time t respectively; and They represent the discharge power and charging power of the energy storage system at node i at time t respectively; The price of active power and reactive power purchased by the distribution network from the upper power grid; c wind_loss 、c pv_loss are the penalty coefficients for wind and solar curtailment, respectively; c es is the unit operating cost of the energy storage system on the distribution network side; among them, each distributed energy storage consumes electricity according to the baseline power.

5. The scheduling method according to claim 1, wherein: The original constraints include: MIA aggregation feasible region constraints of each distributed energy storage cluster, power flow constraints, second-order cone constraints, line current constraints, new energy output constraints, and distribution network side energy storage system constraints; The MIA aggregation feasible domain constraints of each distributed energy storage cluster are: in, are the variable active power vectors of the kth air conditioning load cluster and electric vehicle cluster respectively; are the translation vectors of the kth air conditioning load cluster and electric vehicle cluster respectively; are the scaling factors of the kth air conditioning load cluster and electric vehicle cluster, respectively; The power flow constraint is: Among them, U i,t represents the voltage of node i at time t, u i,t represents the square of the voltage at node i at time t; I ij,t is the current of line ij at time t; w ij,t is the square of the current of line ij at time t; N line is the distribution network line set; φ par (i) and φ chi (i) represents the parent node set and child node set of node i, respectively, where the parent node refers to the node upstream of node i in the distribution network, and the child node refers to the node downstream of node i; h and g are the node numbers in the parent node set and child node set of node i, respectively; and They represent the active power and reactive power of line ig at time t respectively; r hi and x hi They represent the resistance and reactance of line hi respectively; and They represent the active power and reactive power of the conventional load at node i at time t respectively; and They represent the active power output of wind power generation and photovoltaic power generation at node i at time t. If there is no wind power station or photovoltaic power station connected to node i, the value is 0; They represent the variable active power and reference active power of the electric vehicle cluster at node i at time t respectively; They represent the variable active power and reference active power of the air conditioning load cluster at node i at time t respectively; They are respectively the active and reactive power purchased from the upper power grid; The second-order cone constraint is: The line current constraint is: Among them, I ij,max Indicates the upper limit of the line ij current; The new energy output constraints are: in, and are the minimum and maximum active power output of wind power generation at node i at time t, and are the minimum and maximum active power output of photovoltaic power generation at node i at time t; The distribution grid side energy storage system constraints are: Where, is the energy stored on the distribution network side at time t; is the charging and discharging efficiency of the energy storage on the distribution network side; Δt is the optimization time interval; P t ES,ch and P t ES,dis are the charging and discharging power of the energy storage on the distribution network side at time t respectively; It is a binary variable. When the value is 1, it means that the ES is in the charging state, and when the value is 0, it means that the ES is in the discharging state. The maximum charging and discharging power of the energy storage on the distribution network side; and They are the upper and lower limits of the energy storage capacity on the distribution network side; and are the energy stored on the distribution network side at the beginning and end of the period respectively.

6. The scheduling method according to claim 1, wherein: Determine whether the line is overloaded or the node voltage exceeds the limit, including: Based on the linearized power flow model, according to the system topology and node power balance, the relationship between line flow and node injection amount is calculated as follows: P f =F LSF ×P inj Q f =F LSF ×Q inj ΔU=RP f +XQ f Among them, P inj and Q inj are the node active power injection matrix and the node reactive power injection matrix respectively; ΔU is the voltage change vector of each node in the distribution system relative to the root node; R and X are the matrices composed of the resistance and reactance between any two nodes in the distribution system respectively; F LSF is the load transfer factor, which represents the proportional coefficient of the line power flow change relative to the node injection load change; If the load rate of each line in the distribution network does not meet the following judgment formula, it is determined that the line is overloaded; in, Indicates the maximum load of line ij; If the voltage of each node does not satisfy the following judgment formula, it is determined that the node voltage exceeds the limit; Among them, ΔU i,max , ΔU i,min They represent the maximum and minimum voltage change limits of node i relative to the root node.

7. The scheduling method according to claim 1, wherein: The process of solving the node marginal electricity price includes: Linearize the line capacity constraint using a 12-sided regular polygon to generate a linear constraint: Among them, α c,0 , α c,1 , α c,2 is a fixed parameter.

8. The scheduling method according to claim 7, wherein: The node marginal electricity price is obtained by calculating: Perform Lagrangian relaxation on the original objective function, and solve the dual multiplier based on the sensitivity factor and the transmission factor to calculate the node marginal price DLMP of active power: in, is the node marginal price DLMP; is the Lagrange multiplier for the power balance constraint; is the Lagrange multiplier of the node voltage constraint; is the Lagrange multiplier of line capacity constraint; S SFvp S is the sensitivity factor of node voltage change to node injected active power; SFlp The sensitivity factor of the line power flow change to the active power injected into the node; represents the transfer factor of active power at node i at time t.

9. The scheduling method according to claim 1, wherein: Adding the compensation cost for each cluster to the original objective function according to the compensation coefficient includes: The compensation cost is: Among them, C EV is the compensation cost of electric vehicle cluster, C TCL is the compensation cost of the air conditioning load cluster, δ EV and δ TCL are the compensation coefficient factors of the electric vehicle cluster and the air conditioning load cluster respectively; Update the current load curve corresponding to each cluster. The expression is: minC=C grid +C wind_loss +C pv_loss +C ES +C EV +C TCL 。 10. A dispatching system for alleviating distribution network congestion by using distributed energy storage aggregation, characterized in that: include: The aggregation model building module is used to establish the aggregation model of various distributed energy storage clusters based on the maximum inner approximation method (MIA), construct the aggregation feasible region of each cluster, calculate the baseline power of each cluster and generate the benchmark load curve; A power flow calculation module is configured to input the initial power values ​​of each cluster set according to the baseline power into a branch power flow model for power flow calculation; the original objective function of the branch power flow model is to minimize the operating cost of the distribution network; the original constraints of the branch power flow model include power flow constraints, renewable energy generation constraints, and energy storage system constraints; The judgment module is used to calculate the load rate of each line and the voltage of each node in the distribution network based on the power flow calculation results, using the voltage sensitivity factor matrix and the power transfer factor matrix, and determine whether the line is overloaded or the node voltage exceeds the limit. If so, it enters the node marginal electricity price compensation module to execute the corresponding steps; if not, it enters the scheduling plan generation module to execute the corresponding steps; The node marginal electricity price compensation module is used to perform Lagrangian relaxation on the power balance constraint, node voltage constraint, line capacity constraint, and power generation constraint in the power flow constraint based on the original objective function, calculate the Lagrangian multiplier corresponding to each constraint in combination with the current load curve corresponding to each cluster, and calculate the node marginal electricity price based on this; and then set the compensation coefficient corresponding to each cluster according to the node marginal electricity price and the compensation coefficient factor; a compensation update module, configured to add the compensation cost for each cluster to the original objective function according to the compensation coefficient, and to add the aggregate feasible region constraint of each cluster power to the original constraint condition, to obtain an updated objective function and constraint condition; recalculate the power flow according to the updated objective function and constraint condition, and update the current load curve corresponding to each cluster, and then return to the judgment module to execute the corresponding steps; The scheduling plan generation module is used to obtain the day-ahead scheduling plan based on the current load curve corresponding to each cluster.

Images