Large-scale energy storage aggregation method based on external network economic consistency principle

Abstract

The invention discloses a large-scale energy storage aggregation method based on an external network economic consistency principle. The method comprises the following steps: 1) constructing a power system economic dispatching model considering large-scale energy storage; 2) solving the power system economic dispatching model considering large-scale energy storage to obtain dispatching information before energy storage aggregation; 3) establishing a virtual energy storage double-layer optimization model considering multiple scenes; 4) converting the virtual energy storage double-layer optimization model considering multiple scenes into a single-layer mixed integer optimization model; and 5) solving the single-layer mixed integer optimization model to obtain large-scale energy storage aggregation scheduling information suitable for multiple scenes. According to the method, the feasible region after the large-scale energy storage is equivalent to the virtual energy storage can be obtained, then the virtual energy storage is connected to the grid, the economic dispatching behavior same as that of the large-scale energy storage is achieved, meanwhile, the virtual energy storage protects private information of large-scale energy storage users, user privacy leakage is prevented, and the energy storage users can be effectively promoted to participate in power grid dispatching.

Description

Technical Field

[0001] The present invention relates to the electric power field, and in particular to a large-scale energy storage aggregation method based on the principle of external network economic consistency. Background Art

[0002] With the continuous changes in the global energy landscape and the increasing attention to environmental sustainability, clean energy has gradually replaced traditional energy and become the key driving force for my country's energy transformation. However, new energy has strong volatility and randomness. In order to smooth out the random fluctuations of new energy and avoid energy waste, energy storage has become the key to solving the problem. In order to fully tap the economic value of large-scale energy storage after grid connection, effectively solve the widespread wind and solar power abandonment phenomenon of new energy, and reduce the overall operating cost of the power grid, the aggregation and scheduling of multiple energy storage resources and the provision of virtual energy storage for energy storage and use have become key issues in the new power system.

[0003] Energy storage aggregation can protect the information of energy storage users, while providing the grid with a feasible domain where all energy storage can be dispatched, which has become an industrial demand for energy storage aggregation. In the industry, large-scale energy storage is aggregated into virtual energy storage. The ideal aggregation goal is to obtain similar dispatch decision solutions and costs using the original energy storage model and the virtual energy storage model, thereby avoiding modeling errors caused by using the virtual energy storage model.

[0004] Existing methods mostly consider large-scale energy storage constraints and their network connection constraints, and obtain the feasible domain of energy storage aggregation by solving the optimization model. At present, research related to energy storage aggregation problems can be divided into physical model aggregation method and data-driven method. The core idea of ​​the physical model aggregation method is to optimize the approximate solution of the geometric polyhedron under the preset geometric polyhedron area, and then obtain the feasible domain of the energy storage system. Existing studies have proposed that the total feasible domain of virtual energy storage is the Euclidean space superposition of polyhedrons. By establishing an optimization problem, the maximum internal approximation and minimum external approximation of the original polyhedron problem are solved, or a spatial geometric model is constructed to solve its maximum inscribed cube or right-angled pyramid, and then the flexibility of "virtual energy storage" is obtained. Other scholars have proposed an approximate optimization aggregation method for virtual energy storage, which maps the constraint space of multiple energy sources to the constraint space of aggregated power, sacrificing the solution accuracy and obtaining a conservative solution for aggregated power. The literature proposes an outer approximation feasible domain aggregation method based on a virtual battery model. Through the linear programming aggregation feasible domain redundant constraint judgment method, the redundant constraints in the feasible domain are eliminated, and the performance of the aggregation model in the optimization solution is further improved, but the aggregation accuracy is sacrificed. This category of methods can effectively explain the aggregation process, but when facing the multi-period energy storage aggregation problem, the feasible domain considering the period coupling is too complex and difficult to quantify, and the aggregation dimension increases sharply with the aggregation period, and the computational burden is unbearable. In addition, there is an aggregation error between the feasible domain after energy storage aggregation and the feasible domain provided by the actual energy storage. This type of method is difficult to ensure the economic consistency of the scheduling results before and after aggregation. The core idea of ​​the data-driven method is to use data-driven methods to learn the mapping relationship between energy storage parameters and their feasible domains through historical operation data. Some scholars also predict the feasible domain of virtual energy storage through parallel decision tree algorithms and multi-time scale big data analysis. Due to the limitations of historical operation data, this type of method can only approximate some energy storage feasible domains that have appeared in historical operation data, and it lacks physical interpretability, and still needs further research to be applied to engineering practice. It can be seen that there is still a lack of theoretical methods that can effectively aggregate a large amount of energy storage into virtual energy storage while ensuring the economy of the power grid and that are clearly explainable.

[0005] In summary, the existing large-scale energy storage aggregation methods have problems such as low efficiency in solving the aggregation feasible domain, differences in the economic efficiency of grid-connected scheduling before and after energy storage aggregation, and insufficient model generalization ability and interpretability. Therefore, it is necessary to fully consider the economic efficiency and interpretability of the method and study a large-scale energy storage aggregation method based on the principle of external grid economic consistency. Summary of the invention

[0006] The purpose of the present invention is to provide a large-scale energy storage aggregation method based on the principle of external network economic consistency, comprising the following steps:

[0007] 1) Construct an economic dispatch model for power systems considering large-scale energy storage;

[0008] 2) Solve the economic dispatch model of the power system considering large-scale energy storage and obtain the dispatch information before energy storage aggregation;

[0009] 3) Based on the dispatch information before energy storage aggregation, a two-layer optimization model of virtual energy storage considering multiple scenarios is established based on the principle of external network economic consistency;

[0010] 4) Based on KKT condition derivation and big M method, the virtual energy storage two-layer optimization model considering multiple scenarios is converted into a single-layer mixed integer optimization model;

[0011] 5) Solve the single-layer mixed integer optimization model to obtain large-scale energy storage aggregation scheduling information suitable for multiple scenarios.

[0012] Furthermore, the objective function of the power system economic dispatch model considering large-scale energy storage is as follows:

[0013]

[0014] Where: C is the total cost of the original system operation, C gen,t is the power generation cost of the traditional unit in period t, C wind,t is the penalty fee for wind curtailment in period t, C light,t is the penalty fee for abandonment of light in period t, C storage,t N is the total cost of energy storage charging and discharging operation in period t; G 、N W 、N L and N B are the number of traditional units, wind power plants, photovoltaic plants and energy storage grid-connected units respectively; Δt is the time interval, T is the total number of time periods; a i The price of the traditional unit i is: W The penalty fee for unit wind abandonment; a L is the penalty fee for the unit abandoned light amount; c d,t is the unit energy storage discharge quotation in period t; c c,t P is the unit energy storage charging quotation in time period t; g,i,t is the output of the i-th traditional unit in period t; P w,k,t is the actual power of the kth wind power plant in period t; P l,r,t is the actual power of the rth photovoltaic plant in period t; P j,d,t is the discharge power of the jth energy storage period t, P j,c,t is the charging power of the jth energy storage period t; P wmax,k,t is the maximum power that can be generated by the k-th wind power plant in period t; P lmax,r,t is the maximum power that can be generated by the r-th photovoltaic plant in period t.

[0015] Furthermore, the constraints of the economic dispatch model of the power system considering large-scale energy storage include power balance constraints, upper and lower limit constraints of branch power, output constraints of traditional units, ramp constraints, wind power plant power constraints, photovoltaic power plant power constraints, and energy storage charging and discharging related constraints;

[0016] The power balance constraints are as follows:

[0017]

[0018] Where: N N is the total system load; P n,load,t is the load of the nth node in period t; t = 1, 2, ..., T; P g,i,t is the output of the i-th traditional unit in period t; P w,k,t is the actual power of the kth wind power plant in period t; P l,r,t is the actual power of the rth photovoltaic plant in period t; P j,d,t is the discharge power of the jth energy storage period t, P j,c,t is the charging power of the jth energy storage period t;

[0019] The upper and lower limit constraints of branch power are as follows:

[0020] P bpmin ≤P bp,t ≤P bpmax (7)

[0021] Where: P bpmax , P bpmin are the upper and lower limits of the power of the lth branch respectively; L is the total number of branches in the system; where t=1,2,…,T; l=1,2,…,L; P bp,t is the power of the lth branch in period t;

[0022] The output constraints of traditional units are as follows:

[0023] P gmin,i,t ≤P g,i,t ≤P gmax,i,t (8)

[0024] Where: P gmax,i,t , P gmin,i,t, are the upper and lower limits of the output of traditional units respectively; t=1,2,…,T; i=1,2,…,N G ;

[0025] The hill climbing constraints are as follows:

[0026] r gmin,i,t ≤P g,i,t -P g,i,t-1 ≤r gdwon,i,t (9)

[0027] Where: r gmin,i,t, 、r gdown,i,t are the upper and lower limits of the output ramp of the traditional unit; t=1,2,…,T; i=1,2,…,N G ;P g,i,t-1 is the output of the i-th conventional unit in period t-1;

[0028] The power constraints of the wind power plant and the photovoltaic power plant are as follows:

[0029] 0≤P w,k,t ≤P wmax,k,t (10)

[0030] 0≤P l,r,t ≤P lmax,r,t (11)

[0031] Where: t = 1, 2, ..., T; k = 1, 2, ..., N W ; r=1,2,…,N L ;P wmax,k,t is the maximum power that can be generated by the k-th wind power plant in period t; P lmax,r,t is the maximum power that can be generated by the r-th photovoltaic plant in period t;

[0032] The constraints related to energy storage charging and discharging are as follows:

[0033] 0≤P j,d,t ≤P j,Bmax (12)

[0034] 0≤P j,c,t ≤P j,Bmax (13)

[0035] E B,j,min ≤E B,j,t ≤E B,j,max (14)

[0036] E B,j,t -E B,j,t-1 -(η c P j,c,t -P j,d,t / η d )Δt=0(15)

[0037] P j,c,t P j,d,t =0 (16)

[0038] Where: P j,Bmax is the rated power value of charging and discharging of the jth energy storage; E B,j,max 、E B,j,min are the upper and lower limits of the energy that can be stored in the jth energy storage; E B,j,t 、EB,j,t-1 are the energy stored in the jth energy storage in period t and period t-1 respectively; η c , η d are the charging and discharging efficiency of energy storage respectively; when the energy storage is in the charging state, P j,d,t =0; when the energy storage is in the discharge state, P j,c,t =0; when the energy storage is in a static state, P j,d,t With P j,c,t All are 0. t=1,2,…,T;j=1,2,…,N B ; Δt is the time interval.

[0039] Furthermore, the virtual energy storage two-layer optimization model considering multiple scenarios includes a virtual energy storage upper layer optimization model and a virtual energy storage lower layer optimization model;

[0040] The virtual energy storage upper layer optimization model is used to minimize the sum of the absolute values ​​of the differences between the operating costs of various types of resources in the original actual system at each time period and the costs of various types of resources in the virtual system at each time period;

[0041] The virtual energy storage lower layer optimization model is used to minimize the overall operating cost of the virtual system taking into account the virtual energy storage.

[0042] Furthermore, the objective function of the virtual energy storage upper optimization model is as follows:

[0043]

[0044] Where: D is the sum of the absolute values ​​of the errors between various resources in each period of the virtual system and various resources in each period of the original system; is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period; is the power generation cost of the traditional unit in period t, is the penalty fee for wind curtailment in period t, is the total cost of energy storage charging and discharging operation in period t; a represents the scenario;

[0045] The constraints of the virtual energy storage upper optimization model include virtual energy storage charging limit and virtual energy storage capacity limit;

[0046] The virtual energy storage charging restrictions are as follows:

[0047]

[0048] Where: It is the maximum value of the total charge and discharge during large-scale energy storage dispatch in the historical dispatch information; is the charging and discharging rated power value of the jth energy storage; It is the total amount of charging and discharging during large-scale energy storage dispatch;

[0049] The virtual energy storage capacity limits are as follows:

[0050]

[0051] In the formula, is the upper limit of the energy that can be stored in the jth energy storage; is the upper limit of virtual energy storage capacity.

[0052] Furthermore, the objective function minf of the virtual energy storage lower layer optimization model is as follows:

[0053]

[0054] Where: N VB is the number of virtual energy storage; is the output of the i-th traditional unit in the virtual system during period t; is the actual power of the kth wind power plant in the virtual system during period t; is the actual power of the rth photovoltaic plant in the virtual system during period t; is the discharge power of the Sjth virtual energy storage in period t, is the charging power of the Sjth virtual energy storage period t; is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period; Quote for the i-th unit of traditional unit; Penalty fee for unit wind abandonment; It is the penalty fee for the unit abandoned light amount; is the unit energy storage discharge quotation in time period t; The unit energy storage charging quotation in time period t; is the maximum power that can be generated by the k-th wind power plant in period t; is the maximum power that can be generated by the r-th photovoltaic plant in time period t; Δt is the time interval;

[0055] The constraints of the virtual energy storage lower-level optimization model include virtual system power balance constraints, branch power upper and lower limit constraints, and virtual energy storage related constraints;

[0056] The virtual system power balance constraints are as follows:

[0057]

[0058] In the formula, is the load of the nth node in period t;

[0059] The upper and lower limit constraints of branch power are as follows:

[0060]

[0061] Where: is the power of the lth branch of the virtual system;

[0062] The constraints related to virtual energy storage are as follows:

[0063]

[0064] In the formula, It is the maximum value of the total charge and discharge during large-scale energy storage dispatch in the historical dispatch information; is the upper limit of virtual energy storage capacity; is the virtual energy storage capacity during period t and period t-1; are the charging and discharging efficiencies of energy storage, respectively.

[0065] Furthermore, the steps of converting the virtual energy storage two-layer optimization model based on the economic consistency equivalence idea into a single-layer mixed integer optimization model are as follows:

[0066] s1) expressing the virtual energy storage upper layer optimization model and the virtual energy storage lower layer optimization model in a compact form;

[0067] The compact form of the virtual energy storage upper optimization model is as follows:

[0068]

[0069] In the formula, more is the number of scenes; is the optimal solution of the virtual energy storage lower optimization model under scenario a; min D is the objective function of the virtual energy storage upper optimization model; Quote for scenario a; is an equality constraint; is an inequality constraint;

[0070] The compact form of the virtual energy storage underlying optimization model is as follows:

[0071] min f a =KX a (34)

[0072]

[0073] Where K is the coefficient matrix; X a is a variable; min f a The objective function of the virtual energy storage lower layer optimization model; is an equality constraint; is an inequality constraint;

[0074] s2) Solve the optimal solution of the virtual energy storage lower layer optimization model and derive and construct the Lagrangian function L(X,λ j ,μ k ),Right now:

[0075]

[0076] Where: j is the vector of Lagrange multipliers, μ k is a vector composed of Lagrange multipliers. PTDF is the virtual system line transfer distribution factor matrix, Cg_N is the generator node association matrix, Cw_N is the wind power plant node association matrix, Cl_N is the photovoltaic plant node association matrix, λ j,t is the Lagrange multiplier of the lower model equality constraint, μ k,t is the Lagrange multiplier of the inequality constraint of the lower model; j = 1, 2; k = 1, 2, ..., 16; t = 1, 2, ..., T;

[0077] s3) The necessary and sufficient conditions for constructing the virtual energy storage lower layer optimization model to obtain the optimal solution, namely:

[0078]

[0079] In the formula, r gmin,i,t ,、r gdown,i,t They are the upper and lower limits of the output ramp of the traditional unit respectively;

[0080] s4) Use the Big M method to convert the nonlinear constraint μ in the necessary and sufficient conditions into k h k (X) = 0 is converted into a linear constraint, that is:

[0081] 0≤μ k ≤Mz k (39)

[0082] -M(1-z k )≤h k (X)≤0 (40)

[0083] Where: M is infinite, Z k It is a 0-1 integer variable.

[0084] s5) Build a single-layer mixed integer optimization model.

[0085] Further, the single-layer mixed integer optimization model is as follows:

[0086]

[0087] In the formula, It is a 0-1 integer variable.

[0088] Further, tools for solving single-level mixed integer optimization models include the commercial solver CPLEX.

[0089] A system based on the large-scale energy storage aggregation method includes:

[0090] Power system economic dispatch model construction module: Construct a power system economic dispatch model considering large-scale energy storage;

[0091] The objective function of the power system economic dispatch model considering large-scale energy storage is as follows:

[0092]

[0093] Where C is the total cost of the original system operation, C gen,t is the power generation cost of the traditional unit in period t, C wind,t is the penalty fee for wind curtailment in period t, C light,t is the penalty fee for abandonment of light in period t, C storage,t is the total cost of energy storage charging and discharging operation in period t;

[0094] Dispatch information calculation module: solves the economic dispatch model of the power system considering large-scale energy storage and obtains the dispatch information before energy storage aggregation;

[0095] Virtual energy storage two-layer optimization model construction module: Based on the dispatch information before energy storage aggregation, a virtual energy storage two-layer optimization model taking into account multiple scenarios is established based on the principle of external network economic consistency;

[0096] The virtual energy storage two-layer optimization model includes a virtual energy storage upper layer optimization model and a virtual energy storage lower layer optimization model;

[0097] The objective function of the virtual energy storage upper optimization model is as follows:

[0098]

[0099] Where: D is the sum of the absolute values ​​of the errors between various resources in each period of the virtual system and various resources in each period of the original system; is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period; is the power generation cost of the traditional unit in period t, is the penalty fee for wind curtailment in period t, is the total cost of energy storage charging and discharging operation in period t; a represents the scenario;

[0100] The objective function minf of the virtual energy storage lower layer optimization model is as follows:

[0101]

[0102] Where: is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period;

[0103] Model optimization module: Based on KKT condition derivation and big M method, the virtual energy storage double-layer optimization model considering multiple scenarios is converted into a single-layer mixed integer optimization model;

[0104] A single-layer mixed integer optimization model is shown below:

[0105]

[0106] In the formula, It is a 0-1 integer variable.

[0107] Large-scale energy storage aggregation scheduling information solution module: solves the single-layer mixed integer optimization model to obtain large-scale energy storage aggregation scheduling information suitable for multiple scenarios.

[0108] An electronic device comprises: a memory, a processor and a data bus, wherein the memory stores all machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the data bus, and the method described in the process of executing the machine-readable instructions when the processor is running.

[0109] A computer-readable storage medium stores a computer program, which is executed by a processor to execute the method.

[0110] The technical effect of the present invention is undoubted, and the beneficial effects of the present invention are as follows:

[0111] 1) The present invention can obtain the feasible domain after large-scale energy storage is equivalent to virtual energy storage, and then connect the virtual energy storage to the grid to achieve the same economic dispatch behavior as large-scale energy storage. At the same time, virtual energy storage protects the private information of large-scale energy storage users, prevents user privacy leakage, and can effectively promote energy storage users to participate in grid dispatch.

[0112] 2) The present invention further considers the operation conditions of multiple scenarios, and proposes a large-scale energy storage aggregation system-distribution network collaborative economic dispatch method considering multiple scenarios. It fully explores the application of large-scale energy storage in different scenarios, and realizes the multi-scenario equivalence of virtual energy storage parameters to ensure that the aggregated virtual energy storage can effectively respond to the system economic dispatch in different scenarios. At the same time, the WGAN method is used to generate new scenarios for theoretical verification. BRIEF DESCRIPTION OF THE DRAWINGS

[0113] Figure 1 It is a virtual energy storage parameter decision framework based on the principle of external grid economic consistency;

[0114] Figure 2 To clarify the virtual energy storage parameter decision process for multiple scenarios;

[0115] Figure 3 (a)- Figure 3 (d) The economic scheduling of various resources of M0 and M2 methods in four scenarios of IEEE-5 nodes;

[0116] Figure 4 (a)- Figure 4 (d) Energy storage behavior of M0 and M2 methods under four scenarios of IEEE-5 nodes.

[0117] Figure 5 (a)- Figure 5 (d) Energy storage behavior of M0 and M1 methods under four scenarios of IEEE-5 nodes

[0118] Figure 6 (a)- Figure 6 (d) Energy storage behavior of M0 and M1 methods in some scenarios of IEEE-118 nodes

[0119] Figure 7 The relative error diagram of the cost of IEEE-118 nodes in each of the 100 scenarios;

[0120] Figure 8 The Euclidean distance between the generated scenario and the historical scenario and the relative error of the overall operating costs of the two are plotted. DETAILED DESCRIPTION

[0121] The present invention is further described below in conjunction with the embodiments, but it should not be understood that the above subject matter of the present invention is limited to the following embodiments. Without departing from the above technical ideas of the present invention, various substitutions and changes are made according to the common technical knowledge and customary means in the art, which should all be included in the protection scope of the present invention.

[0122] Embodiment 1:

[0123] See also Figures 1 to 8 , a large-scale energy storage aggregation method based on the principle of external grid economic consistency, comprising the following steps:

[0124] 1) Construct an economic dispatch model for power systems considering large-scale energy storage;

[0125] 2) Solve the economic dispatch model of the power system considering large-scale energy storage and obtain the dispatch information before energy storage aggregation;

[0126] 3) Based on the dispatch information before energy storage aggregation, a two-layer optimization model of virtual energy storage considering multiple scenarios is established based on the principle of external network economic consistency;

[0127] 4) Based on KKT condition derivation and big M method, the virtual energy storage two-layer optimization model considering multiple scenarios is converted into a single-layer mixed integer optimization model;

[0128] 5) Solve the single-layer mixed integer optimization model to obtain large-scale energy storage aggregation scheduling information suitable for multiple scenarios.

[0129] Embodiment 2:

[0130] A large-scale energy storage aggregation method based on the principle of external network economic consistency, the technical content is the same as that of Example 1, and further, the objective function of the power system economic dispatch model considering large-scale energy storage is as follows:

[0131]

[0132] Where: C is the total cost of the original system operation, C gen,t is the power generation cost of the traditional unit in period t, C wind,t is the penalty fee for wind curtailment in period t, C light,t is the penalty fee for abandonment of light in period t, C storage,t N is the total cost of energy storage charging and discharging operation in period t; G 、N W 、N L and N B are the number of traditional units, wind power plants, photovoltaic plants and energy storage grid-connected units respectively; Δt is the time interval, T is the total number of time periods; a i The price of the i-th traditional unit; a W The penalty fee for unit wind abandonment; aL is the penalty fee for the unit abandoned light amount; c d,t is the unit energy storage discharge quotation in period t; c c,t P is the unit energy storage charging quotation in time period t; g,i,t is the output of the i-th traditional unit in period t; P w,k,t is the actual power of the kth wind power plant in period t; P l,r,t is the actual power of the rth photovoltaic plant in period t; P j,d,t is the discharge power of the jth energy storage period t, P j,c,t is the charging power of the jth energy storage period t; P wmax,k,t is the maximum power that can be generated by the k-th wind power plant in period t; P lmax,r,t is the maximum power that can be generated by the r-th photovoltaic plant in period t.

[0133] Embodiment 3:

[0134] A large-scale energy storage aggregation method based on the principle of external network economic consistency, the technical content is the same as any one of embodiments 1-2, further, the constraints of the power system economic dispatch model considering large-scale energy storage include power balance constraints, branch power upper and lower limit constraints, traditional unit output constraints, climbing constraints, wind power plant power constraints, photovoltaic power plant power constraints, and energy storage charging and discharging related constraints;

[0135] The power balance constraints are as follows:

[0136]

[0137] Where: N N is the total system load; P n,load,t is the load of the nth node in period t; t = 1, 2, ..., T; P g,i,t is the output of the i-th traditional unit in period t; P w,k,t is the actual power of the kth wind power plant in period t; P l,r,t is the actual power of the rth photovoltaic plant in period t; P j,d,t is the discharge power of the jth energy storage period t, P j,c,t is the charging power of the jth energy storage period t;

[0138] The upper and lower limit constraints of branch power are as follows:

[0139] P bpmin ≤P bp,t ≤P bpmax (7)

[0140] Where: P bpmax , P bpminare the upper and lower limits of the power of the lth branch respectively; L is the total number of branches in the system; where t=1,2,…,T; l=1,2,…,L; P bp,t is the power of the lth branch in period t;

[0141] The output constraints of traditional units are as follows:

[0142] P gmin,i,t ≤P g,i,t ≤P gmax,i,t (8)

[0143] Where: P gmax,i,t , P gmin,i,t, are the upper and lower limits of the output of traditional units respectively; t=1,2,…,T; i=1,2,…,N G ;

[0144] The hill climbing constraints are as follows:

[0145] r gmin,i,t ≤P g,i,t -P g,i,t-1 ≤r gdwon,i,t (9)

[0146] Where: r gmin,i,t ,、r gdown,i,t are the upper and lower limits of the output ramp of the traditional unit; t=1,2,…,T; i=1,2,…,N G ;P g,i,t-1 is the output of the i-th conventional unit in period t-1;

[0147] The power constraints of the wind power plant and the photovoltaic power plant are as follows:

[0148] 0≤P w,k,t ≤P wmax,k,t (10)

[0149] 0≤P l,r,t ≤P lmax,r,t (11)

[0150] Where: t = 1, 2, ..., T; k = 1, 2, ..., N W ; r=1,2,…,N L ;P wmax,k,t is the maximum power that can be generated by the k-th wind power plant in period t; P lmax,r,t is the maximum power that can be generated by the r-th photovoltaic plant in period t;

[0151] The constraints related to energy storage charging and discharging are as follows:

[0152] 0≤P j,d,t ≤P j,Bmax (12)

[0153] 0≤P j,c,t ≤P j,Bmax (13)

[0154] E B,j,min ≤E B,j,t ≤E B,j,max (14)

[0155] E B,j,t -E B,j,t-1 -(η c P j,c,t -P j,d,t / η d )Δt=0(15)

[0156] P j,c,t P j,d,t =0 (16)

[0157] Where: P j,Bmax is the rated power value of charging and discharging of the jth energy storage; E B,j,max 、E B,j,min are the upper and lower limits of the energy that can be stored in the jth energy storage; E B,j,t 、E B,j,t-1 are the energy stored in the jth energy storage in period t and period t-1 respectively; η c , η d are the charging and discharging efficiency of energy storage respectively; when the energy storage is in the charging state, P j,d,t =0; when the energy storage is in the discharge state, P j,c,t =0; when the energy storage is in a static state, P j,d,t With P j,c,t All are 0. t=1,2,…,T;j=1,2,…,N B ; Δt is the time interval.

[0158] Embodiment 4:

[0159] A large-scale energy storage aggregation method based on the principle of external network economic consistency, the technical content is the same as any one of embodiments 1-3, further, the virtual energy storage double-layer optimization model considering multiple scenarios includes a virtual energy storage upper layer optimization model and a virtual energy storage lower layer optimization model;

[0160] The virtual energy storage upper optimization model is used to minimize the absolute value of the difference between the operating cost of various resources in each period of the original actual system and the cost of various resources in each period of the virtual system;

[0161] The virtual energy storage lower layer optimization model is used to minimize the overall operating cost of the virtual system taking into account the virtual energy storage.

[0162] Embodiment 5:

[0163] A large-scale energy storage aggregation method based on the principle of external network economic consistency, the technical content is the same as any one of embodiments 1-4, and further, the objective function of the virtual energy storage upper layer optimization model is as follows:

[0164]

[0165] Where: D is the sum of the absolute values ​​of the errors between various resources in each period of the virtual system and various resources in each period of the original system; is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period; is the power generation cost of the traditional unit in period t, is the penalty fee for wind curtailment in period t, is the total cost of energy storage charging and discharging operation in period t; a represents the scenario;

[0166] The constraints of the virtual energy storage upper optimization model include virtual energy storage charging limit and virtual energy storage capacity limit;

[0167] The virtual energy storage charging restrictions are as follows:

[0168]

[0169] Where: It is the maximum value of the total charge and discharge during large-scale energy storage dispatch in the historical dispatch information; is the charging and discharging rated power value of the jth energy storage; It is the total amount of charging and discharging during large-scale energy storage dispatch;

[0170] The virtual energy storage capacity limits are as follows:

[0171]

[0172] In the formula, is the upper limit of the energy that can be stored in the jth energy storage; is the upper limit of virtual energy storage capacity.

[0173] Embodiment 6:

[0174] A large-scale energy storage aggregation method based on the principle of external network economic consistency, the technical content is the same as any one of embodiments 1-5, and further, the objective function minf of the virtual energy storage lower layer optimization model is as follows:

[0175]

[0176] Where: NVB is the number of virtual energy storage; is the output of the i-th traditional unit in the virtual system during period t; is the actual power of the kth wind power plant in the virtual system during period t; is the actual power of the rth photovoltaic plant in the virtual system during period t; is the discharge power of the Sjth virtual energy storage in period t, is the charging power of the Sjth virtual energy storage period t; is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period; Quote for the i-th unit of traditional unit; Penalty fee for unit wind abandonment; It is the penalty fee for the unit abandoned light amount; is the unit energy storage discharge quotation in period t; The unit energy storage charging quotation in time period t; is the maximum power that can be generated by the k-th wind power plant in period t; is the maximum power that can be generated by the r-th photovoltaic plant in time period t; Δt is the time interval;

[0177] The constraints of the virtual energy storage lower-level optimization model include virtual system power balance constraints, branch power upper and lower limit constraints, and virtual energy storage related constraints;

[0178] The virtual system power balance constraints are as follows:

[0179]

[0180] In the formula, is the load of the nth node in period t;

[0181] The upper and lower limit constraints of branch power are as follows:

[0182]

[0183] Where: is the power of the lth branch of the virtual system;

[0184] The constraints related to virtual energy storage are as follows:

[0185]

[0186] In the formula, It is the maximum value of the total charge and discharge during large-scale energy storage dispatch in the historical dispatch information; is the upper limit of virtual energy storage capacity; is the virtual energy storage capacity during period t and period t-1; are the charging and discharging efficiencies of energy storage, respectively.

[0187] Embodiment 7:

[0188] A large-scale energy storage aggregation method based on the principle of external network economic consistency, the technical content is the same as any one of Examples 1-6, and further, the steps of converting the virtual energy storage double-layer optimization model based on the economic consistency equivalence idea into a single-layer mixed integer optimization model are as follows:

[0189] s1) expressing the virtual energy storage upper layer optimization model and the virtual energy storage lower layer optimization model in a compact form;

[0190] The compact form of the virtual energy storage upper optimization model is as follows:

[0191]

[0192] In the formula, more is the number of scenes; is the optimal solution of the virtual energy storage lower optimization model under scenario a; min D is the objective function of the virtual energy storage upper optimization model; Quote for scenario a; is an equality constraint; is an inequality constraint;

[0193] The compact form of the virtual energy storage underlying optimization model is as follows:

[0194] min f a =KX a (34)

[0195]

[0196] Where K is the coefficient matrix; X a is a variable; min f a The objective function of the virtual energy storage lower layer optimization model; is an equality constraint; is an inequality constraint;

[0197] s2) Solve the optimal solution of the virtual energy storage lower layer optimization model and derive and construct the Lagrangian function L(X,λ j ,μ k ),Right now:

[0198]

[0199]

[0200] Where:j is the vector of Lagrange multipliers, μ k is a vector composed of Lagrange multipliers. PTDF is the virtual system line transfer distribution factor matrix, Cg_N is the generator node association matrix, Cw_N is the wind power plant node association matrix, Cl_N is the photovoltaic plant node association matrix, λ j,t is the Lagrange multiplier of the lower model equality constraint, μ k,t is the Lagrange multiplier of the inequality constraint of the lower model; j = 1, 2; k = 1, 2, ..., 16; t = 1, 2, ..., T;

[0201] s3) The necessary and sufficient conditions for constructing the virtual energy storage lower layer optimization model to obtain the optimal solution, namely:

[0202]

[0203] In the formula, r gmin,i,t, 、r gdown,i,t They are the upper and lower limits of the output ramp of the traditional unit respectively;

[0204] s4) Use the Big M method to convert the nonlinear constraint μ in the necessary and sufficient conditions into k h k (X) = 0 is converted into a linear constraint, that is:

[0205] 0≤μ k ≤Mz k (39)

[0206] -M(1-z k )≤h k (X)≤0 (40)

[0207] Where: M is infinite, Z k It is a 0-1 integer variable.

[0208] s5) Build a single-layer mixed integer optimization model.

[0209] Embodiment 8:

[0210] A large-scale energy storage aggregation method based on the principle of external network economic consistency, the technical content is the same as any one of Examples 1-7, and further, the single-layer mixed integer optimization model is as follows:

[0211]

[0212] In the formula, It is a 0-1 integer variable.

[0213] Embodiment 9:

[0214] A large-scale energy storage aggregation method based on the principle of external grid economic consistency, the technical content is the same as any one of Examples 1-2, and further, the tool for solving the single-layer mixed integer optimization model includes the commercial solver CPLEX.

[0215] Embodiment 10:

[0216] A system based on the large-scale energy storage aggregation method according to any one of embodiments 1-9 comprises:

[0217] Power system economic dispatch model construction module: Construct a power system economic dispatch model considering large-scale energy storage;

[0218] The objective function of the power system economic dispatch model considering large-scale energy storage is as follows:

[0219]

[0220] Where C is the total cost of the original system operation, C gen,t is the power generation cost of the traditional unit in period t, C wind,t is the penalty fee for wind curtailment in period t, C light,t is the penalty fee for abandonment of light in period t, C storage,t is the total cost of energy storage charging and discharging operation in period t;

[0221] Dispatch information calculation module: solves the economic dispatch model of the power system considering large-scale energy storage and obtains the dispatch information before energy storage aggregation;

[0222] Virtual energy storage two-layer optimization model construction module: Based on the dispatch information before energy storage aggregation, a virtual energy storage two-layer optimization model taking into account multiple scenarios is established based on the principle of external network economic consistency;

[0223] The virtual energy storage two-layer optimization model includes a virtual energy storage upper layer optimization model and a virtual energy storage lower layer optimization model;

[0224] The objective function of the virtual energy storage upper optimization model is as follows:

[0225]

[0226] Where: D is the sum of the absolute values ​​of the errors between various resources in each period of the virtual system and various resources in each period of the original system; is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period; is the power generation cost of the traditional unit in period t, is the penalty fee for wind curtailment in period t, is the total cost of energy storage charging and discharging operation in period t; a represents the scenario;

[0227] The objective function minf of the virtual energy storage lower layer optimization model is as follows:

[0228]

[0229] Where: is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period;

[0230] Model optimization module: Based on KKT condition derivation and big M method, the virtual energy storage double-layer optimization model considering multiple scenarios is converted into a single-layer mixed integer optimization model;

[0231] A single-layer mixed integer optimization model is shown below:

[0232]

[0233] In the formula, It is a 0-1 integer variable.

[0234] Large-scale energy storage aggregation scheduling information solution module: solves the single-layer mixed integer optimization model to obtain large-scale energy storage aggregation scheduling information suitable for multiple scenarios.

[0235] Embodiment 11:

[0236] An electronic device comprises: a memory, a processor and a data bus, wherein the memory stores all machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the data bus, and the machine-readable instructions are executed by the processor to execute the method described in any one of Examples 1-9.

[0237] Embodiment 12:

[0238] A computer-readable storage medium stores a computer program, which executes the method described in any one of embodiments 1-9 when executed by a processor.

[0239] Embodiment 13:

[0240] A large-scale energy storage aggregation method based on the principle of external grid economic consistency is as follows:

[0241] Considering the characteristics of large-scale grid-connected energy storage, the operating costs of various resources in each period of the optimal dispatch of the power system are calculated through historical data. Based on the idea of ​​economic equivalence, a large-scale energy storage aggregation method based on the principle of external grid economic consistency is proposed. First, considering the high proportion of wind and solar new energy grid connection, the optimal operating cost of the system with large-scale energy storage participating in grid dispatch is calculated. Based on this, large-scale energy storage is aggregated into virtual energy storage, and a two-layer optimization model of virtual energy storage is proposed according to the principle of external grid economic consistency. Secondly, considering the problem of model solution efficiency, based on the derivation of KKT conditions, the lower-level optimization model is converted into constraint conditions and embedded in the upper-level optimization model, realizing the conversion of the two-layer optimization model into a single-layer optimization model, and the large M method is used to eliminate the nonlinear constraints in the model to speed up the solution efficiency.

[0242] The specific contents are as follows:

[0243] 1 Economic dispatch model of power system considering large-scale energy storage

[0244] When a high proportion of wind and solar energy is connected to the grid, large-scale energy storage participating in power system dispatch can effectively reduce the overall operating cost of the system, and effectively play a role in peak load reduction and valley filling, ensuring the safe and economical operation of the system. Considering the actual status of the industry, this paper uses the DC power flow model to model the economic dispatch of the power system. Therefore, this paper establishes a power system economic dispatch model considering large-scale energy storage connected to the grid.

[0245] Objective function:

[0246]

[0247] Where: C is the total cost of the original system operation, C gen,t is the power generation cost of the traditional unit in period t, C wind,t is the penalty fee for wind curtailment in period t, C light,t is the penalty fee for abandonment of light in period t, C storage,t N is the total cost of energy storage charging and discharging operation in period t; G 、N W 、N L and N B are the number of traditional units, wind power plants, photovoltaic plants and energy storage grid-connected units respectively; Δt is the time interval, the interval is 1 hour, T is the total number of time periods, a total of 24 hours; i The price of the traditional unit i is: W is the penalty fee for unit wind abandonment; a L is the penalty fee for the unit abandoned light amount; c d,t is the unit energy storage discharge quotation and c in period t c,t P is the unit energy storage charging quotation in time period t; g,i,t is the output of the i-th traditional unit in period t; P w,k,tis the actual power of the kth wind power plant in period t; P l,r,t is the actual power of the rth photovoltaic plant in period t; P j,d,t is the discharge power of the jth energy storage period t, P j,c,t is the charging power of the jth energy storage period t; P wmax,k,t is the maximum power that can be generated by the k-th wind power plant in period t; P pmax,r,t is the maximum power that can be generated by the r-th photovoltaic plant in period t.

[0248] Original system constraints:

[0249] Power balance constraints

[0250]

[0251] Where: N N is the total system load; P n,load,t is the load of the nth node in period t; where t = 1, 2, …, T.

[0252] Branch power upper and lower limit constraints

[0253] P bpmin ≤P bp,t ≤P bpmax (7)

[0254] Where: P bpmax , P bpmin are the upper and lower limits of the power of the lth branch respectively; L is the total number of all branches contained in the system; where t=1,2,…,T; l=1,2,…,L.

[0255] Output constraints of traditional units

[0256] P gmin,i,t ≤P g,i,t ≤P gmax,i,t (8)

[0257] Where: P gmax,i,t , P gmin,i,t, They are the upper and lower limits of the output of the traditional unit respectively; t=1,2,…,T;

[0258] i=1,2,…,N G ;

[0259] Hill climbing constraint

[0260] r gmin,i,t ≤P g,i,t -P g,i,t-1 ≤r gdwon,i,t (9)

[0261] Where: r gmin,i,t, 、r gdown,i,tThey are the upper and lower limits of the output ramp of the traditional unit. t=1,2,…,T;

[0262] i=1,2,…,N G ;

[0263] Wind power plant power constraints and photovoltaic power plant power constraints

[0264] The actual power used by wind and solar power cannot be higher than the maximum power generated by the wind and solar power plants at the current moment, that is:

[0265] 0≤P w,k,t ≤P wmax,k,t (10)

[0266] 0≤P l,r,t ≤P lmax,r,t (11)

[0267] Where: t = 1, 2, ..., T; k = 1, 2, ..., N W ; r=1,2,…,N L .

[0268] Energy storage charging and discharging related constraints

[0269] During the actual charging and discharging process of energy storage, the charging and discharging power cannot be higher than the rated power value of the energy storage; the energy storage device stores energy continuously, that is, the energy stored in the current period is equal to the capacity of the previous period and is determined by the charging and discharging efficiency and actual charging and discharging power of the current period; the energy storage has upper and lower limits; and, the energy storage can only exist in one state in the same period, namely charging, discharging or static.

[0270] 0≤P j,d,t ≤P j,Bmax (12)

[0271] 0≤P j,c,t ≤P j,Bmax (13)

[0272] E B,j,min ≤E B,j,t ≤E B,j,max (14)

[0273] E B,j,t -E B,j,t-1 -(η c P j,c,t -P j,d,t / η d )Δt=0(15)

[0274] P j,c,t P j,d,t =0 (16)

[0275] Where: P j,Bmaxis the rated power value of charging and discharging of the jth energy storage; E B,j,max 、E B,j,min are the upper and lower limits of the energy that can be stored in the jth energy storage. B,j,t 、E B,j,t-1 are the energy stored in the jth energy storage in period t and period t-1 respectively; η c , η d are the charging and discharging efficiencies of the energy storage, respectively. In addition, for the nonlinear equation (16), when the energy storage is in the charging state, P j,d,t =0; when the energy storage is in the discharge state, P j,c,t =0; when the energy storage is in a static state, P j,d,t With P j,c,t All are 0. t=1,2,…,T;j=1,2,…,N B ;

[0276] The above equations (1)-(16) are the economic dispatch model of the power system considering large-scale energy storage grid connection. Since the virtual energy storage system in the following text also adopts the above model, it is necessary to derive the KKT conditions in the future. However, the original model has nonlinear constraints. By removing the nonlinear constraints, the optimal solution of the optimization problem (1)-(15) is equivalent. That is, the economic dispatch model of the power system considering large-scale energy storage grid connection is equation (1)-(15).

[0277] 2 Virtual energy storage parameter decision method based on external network economic consistency principle

[0278] This section makes decisions on the parameters of large-scale energy storage equivalent to virtual energy storage based on the principle of external grid economic consistency. According to the power system economic dispatch model in Section 2.1, the variables and constraints related to large-scale energy storage are replaced with virtual energy storage variables and constraints, and a two-layer optimization model for virtual energy storage based on the idea of ​​economic consistency equivalence is established. The upper model minimizes the sum of the absolute values ​​of the difference between the operating costs of various resources in each period of the original actual system and the costs of various resources in each period of the virtual system, and the lower model minimizes the overall operating cost of the virtual system considering virtual energy storage. The specific model is as follows:

[0279] Upper objective function:

[0280]

[0281] Where: D is the sum of the absolute values ​​of the errors between the various resources in the virtual system at each time period and the various resources in the original system at each time period; f gen,t is the power generation cost of the traditional unit in the virtual system at the tth period, f wind,t is the penalty fee for wind curtailment in the virtual system at the tth period, f light,t is the penalty fee for abandonment of light in the virtual system at time t, f storage,t is the total cost of charging and discharging operation of the virtual energy storage in the virtual system during the tth period.

[0282] The upper-level model minimizes the absolute value and cost error of various types of resources in each time period between the actual system and the virtual system, aiming to minimize the difference in scheduling results before and after energy storage aggregation from a cost perspective.

[0283] Upper-level optimization model constraints:

[0284] Virtual energy storage charging restrictions:

[0285] P Sj,d,t ≤P Sj,max (18)

[0286]

[0287] Where: P Sj,max It is the maximum total charge and discharge value during large-scale energy storage scheduling in the historical scheduling information, that is, the maximum value of virtual energy storage charge and discharge.

[0288] Virtual energy storage capacity limit:

[0289]

[0290] The lower objective function is:

[0291]

[0292] Where: N VB is the number of virtual energy storage; P fg,i,t is the output of the i-th traditional unit in the virtual system during period t; P fw,k,t is the actual power of the kth wind farm in the virtual system during period t; P fl,r,t is the actual power of the rth photovoltaic plant in the virtual system during period t; P Sj,d,t is the discharge power of the Sjth virtual energy storage period t, P Sj,c,t is the charging power of the Sjth virtual energy storage in period t; in the virtual system, P fg,i,t , P fw,k,t , P fl,r,t The constraints of the three types of resources are the same as those of the original scheduling system.

[0293] The lower-level model minimizes the total operating cost of various resources in the virtual system at each time period, aiming to achieve the equivalent of large-scale energy storage as virtual energy storage while enabling the virtual system to achieve optimal economic scheduling.

[0294] The constraints of the lower optimization model are:

[0295] Virtual system power balance constraints

[0296]

[0297] Branch power upper and lower limit constraints

[0298] P fbpmin ≤P fbp,t ≤P fbpmax (27)

[0299] Where: P fbpmax , P fbpmin are the upper and lower limits of the power of the lth branch of the virtual system respectively; L is the total number of all branches contained in the system; where t=1,2,…,T; l=1,2,…,L.

[0300] Virtual energy storage related constraints:

[0301] 0≤P Sj,d,t ≤P Sj,max (28)

[0302] 0≤P Sj,c,t ≤P Sj,max (29)

[0303] E Sj,min ≤E Sj,t ≤E Sj,max (30)

[0304] E Sj,t -E Sj,t-1 -(η c P Sj,c,t -P Sj,d,t / η d )Δt=0(31)

[0305] The objective function of the lower model is equations (21)-(25), and the constraints are composed of equations (8)-(11) and (26)-(31).

[0306] In summary, equations (8)-(11) and (17)-(31) constitute a large-scale energy storage aggregation optimization model based on the principle of external grid economic consistency.

[0307] 3. Double-layer model transformation method for virtual energy storage parameter decision based on KKT condition and big M method

[0308] Because the large-scale energy storage aggregation model based on the principle of external grid economic consistency belongs to a two-layer optimization problem, when the scale of the problem is large, the iterative method often cannot solve the problem, so other methods are needed to solve it. Analyzing the essence of the above model, the two objective functions of the proposed model are both linear functions, and the constraints are also linear constraints. Therefore, the two-layer model is a convex problem. Therefore, the Karush-Kuhn-Tucker (KKT) condition is derived for the lower optimization model, and the optimal solution that satisfies the KKT condition of the lower model is equivalent to the optimal solution of the original lower problem. Then, the KKT condition of the lower layer is embedded in the constraint condition of the upper optimization model to realize the conversion of the two-layer optimization model to a single-layer optimization model. The large M method is further used to eliminate the nonlinear constraints in the KKT condition, and finally the two-layer optimization problem is converted into a single-layer mixed integer optimization problem, realizing the simplification of the large-scale energy storage aggregation problem.

[0309] To facilitate the derivation in this section, the above optimization problem is expressed in the following compact form:

[0310]

[0311] min f=KX (34)

[0312]

[0313] In order to solve the optimal solution of the lower-level model that minimizes the overall operating cost of the virtual system, the Lagrangian function L is derived and constructed for the lower-level model:

[0314]

[0315] Where: j is the vector of Lagrange multipliers, μ k is the vector of Lagrange multipliers.

[0316] The specific expression for the derivation of the Lagrangian function of the underlying optimization problem is:

[0317]

[0318] Where: PTDF is the virtual system line transfer distribution factor matrix, Cg_N is the generator node association matrix, Cw_N is the wind power plant node association matrix, Cl_N is the photovoltaic plant node association matrix, λ j,t is the Lagrange multiplier of the lower model equality constraint, μ k,t is the Lagrange multiplier of the inequality constraint of the lower model, where j = 1, 2; k = 1, 2, ..., 16; t = 1, 2, ..., T.

[0319] Since the objective function and constraints of the lower-level model constitute a convex optimization problem, the necessary and sufficient conditions for the lower-level optimization model to obtain the optimal solution are:

[0320]

[0321] Since the constraint condition (38) contains the nonlinear constraint μ k h k (X) = 0, which cannot be solved directly using commercial solvers, so the Big M method is used to convert nonlinear constraints into linear constraints:

[0322] 0≤μ k ≤Mz k (39)

[0323] -M(1-z k )≤h k (X)≤0 (40)

[0324] Where: M is infinite, Z k It is a 0-1 integer variable.

[0325] In summary, the large-scale energy storage aggregation two-layer model based on the principle of external grid economic consistency is transformed into the following single-layer mixed integer optimization model, which is written in the corresponding compact form as follows:

[0326]

[0327] Based on the idea of ​​economic consistency, this chapter proposes a large-scale energy storage aggregation method based on the principle of external network economic consistency, which can achieve the equivalence of the feasible domain of virtual energy storage in a single scenario and provide a theoretical basis for virtual energy storage applications in multiple scenarios.

[0328] 4 Virtual energy storage parameter decision model considering multiple scenarios

[0329] The previous article proposed a large-scale energy storage aggregation method based on the principle of external network economic consistency. This method simplifies the energy storage modeling process and protects the privacy of energy storage users by aggregating scattered energy storage units into a virtual energy storage. However, this method only achieves equivalence in a single scenario for the aggregation of energy storage, and is difficult to apply to a wide range of multi-situation dispatches. In order to overcome this limitation, based on the method proposed in Chapter 2, the operating characteristics of energy storage in different scenarios are comprehensively considered, which can be combined with the dispatch strategy of the distribution network to achieve the coordinated optimization of energy storage resources and grid operation. This chapter proposes a virtual energy storage parameter decision method that takes into account multiple scenarios.

[0330] In view of the above problems, this section considers the universality of this method, and further considers multiple scenarios. Large-scale energy storage is equivalent to a virtual energy storage parameter decision method for virtual energy storage. The upper model is the minimum sum of the absolute values ​​of the difference between the virtual system cost and the actual system cost under multiple scenarios. The lower model is the optimal economic dispatch model under multiple scenarios. Based on the compact model expansion in the first section, the specific model is as follows:

[0331]

[0332] min f a =KX a (44)

[0333]

[0334] Where: superscript a represents the ath scene, and more represents the total number of scenes.

[0335] The above model is further transformed based on the theory proposed in Section 2.3 to obtain a single-layer optimization model of virtual energy storage parameters taking into account multiple scenarios:

[0336]

[0337] In the formula: Similarly, the superscript a represents the ath scene.

[0338] According to the single-layer optimization model of virtual energy storage parameters taking into account multiple scenarios, the virtual energy storage parameters suitable for multiple scenarios can be solved, and the virtual energy storage parameters are used to replace the actual large-scale energy storage. The economic dispatch of various resources in the case of virtual energy storage grid connection is compared with the economic dispatch of various resources in the actual large-scale energy storage grid connection.

[0339] The specific steps of the virtual energy storage parameter decision method considering multiple scenarios are as follows:

[0340] Step 1: Solve the dispatch plan of the actual system before energy storage aggregation. Based on the large-scale energy storage economic dispatch model, solve the dispatch information before energy storage aggregation

[0341] Step 2: Construct a virtual energy storage parameter decision model. Based on the energy storage pre-aggregation dispatch information obtained in step 1 and the idea of ​​economic consistency, a two-layer optimization model for virtual energy storage parameter decision taking into account multiple scenarios is established, namely, equations (42)-(45).

[0342] Step 3: Two-layer model conversion method. Based on the KKT condition derivation and the big M method, the two-layer optimization model established in step 2 is converted into a single-layer optimization model to eliminate the nonlinear constraints in the model, that is, formula (46).

[0343] Step 4: Solving virtual energy storage parameters. The commercial solver CPLEX is used to solve the converted single-layer optimization model of virtual energy storage parameters taking into account multiple scenarios, and obtain virtual energy storage parameters that are suitable for multiple scenarios while ensuring consistent economic performance.

[0344] Embodiment 14:

[0345] A validation of a large-scale energy storage aggregation method based on the principle of external grid economic consistency is as follows:

[0346] The simulation of this example is carried out in MATLAB 2018b software. The computer used for the simulation is tested under the hardware environment of Intel(R) Core(TM) i7-12700H CPU@2.30GHz and 16GB RAM. The CPLEX commercial solver is used to solve the converted linear programming model in the article. W 300 yuan, abandonment penalty a L The equipment parameters of the energy storage system in each node are shown in Table 1.

[0347] First, the 4-node distribution network system is connected through the IEEE-5 node system, and the energy storage parameter decision model based on the external network economic consistency principle mentioned in Chapter 2 is verified by comparing the following economic dispatch methods (M0-M2). The accuracy and effectiveness of the virtual energy storage parameter decision method considering multiple scenarios mentioned in any one of Examples 1-10 are verified by connecting the 4-node distribution network system through the IEEE-118 node:

[0348] Table 1 Distribution network system energy storage parameters

[0349]

[0350] 1. Validation of the virtual energy storage parameter decision model based on the external grid economic consistency principle

[0351] For the verification of the method proposed in Chapter 2, in the IEEE-5-node system, a wind power plant is connected to node 2, a photovoltaic plant is connected to node 3, a 4-node distribution network system is connected to node 5, 2 energy storages are installed on node 3 in the distribution network system, and 2 energy storages are installed on node 4.

[0352] This section first simulates and analyzes the energy storage parameter decision model based on the principle of external grid economic consistency mentioned in Section 2, and demonstrates the accuracy of the virtual energy storage parameter equivalence, as well as the consistency between the feasible domain provided by the virtual energy storage to the power system after grid connection and the feasible domain provided by the actual large-scale energy storage to the power system. The M0-M2 example is used for comparison and verification:

[0353] M0: Economic dispatch method for power system with actual large-scale energy storage grid connection, taking into account the system power limitation of actual energy storage grid connection.

[0354] M1: This paper proposes a large-scale energy storage aggregation method based on the principle of external grid economic consistency.

[0355] M2: Directly add the individual feasible regions provided by large-scale energy storage and adopt the same power system economic dispatch method as M1.

[0356] Table 2 shows the optimization results of M0-M2. As can be seen from Table 2, the total operating cost of the power grid in multiple scenarios using the M1 method is equal to the total operating cost of the power grid in the corresponding scenario of M0, which shows that the M1 method can effectively guarantee the economy of actual large-scale energy storage aggregation and grid connection, achieve cost equivalence, and avoid cost waste. However, the operating cost of multiple scenarios of the M2 method is not equal to the operating cost of the corresponding scenario of the M0 method. The overall operating cost has been reduced, and economic consistency has not been effectively guaranteed. Figure 3 The comparison of the operating costs of M1 and M2 for each resource at each moment in four scenarios is given. Figure 3 It can be seen that the cost of M2 method in scenarios 1-4 all deviated at the 24th moment, and the cost of scenario 4 deviated between moments 5-10, etc. Therefore, further comparison is needed to verify the actual power operation characteristics of the system.

[0357] Table 2 Comparison of total operating costs of the M0-M2 method system under IEEE-5 nodes

[0358]

[0359] To further verify the power characteristics of large-scale energy storage aggregation, Figure 4 The power operation characteristics of the above four scenarios M0 at each moment and M2 at each moment are given. Figure 4 It can be seen that the M2 method will have a large difference between the power dispatch and the actual dispatch method at the 1-10 moment in scenario 2 and the 5-10 moment in scenario 4, and fails to effectively ensure the safety of the power grid.

[0360] Figure 5 The power operation characteristics of the above four scenarios M0 and M1 at each moment are given. As can be seen from the figure, the M1 method can effectively realize the aggregation of the feasible domain of large-scale energy storage in each scenario at each moment, and the power results of the system dispatching energy storage are completely consistent with the dispatching energy storage power results of the M0 method. While ensuring the economic consistency of the power grid, the effective aggregation of the feasible domain of large-scale energy storage is achieved.

[0361] In summary, the economic dispatch results and energy storage power dispatch results of the M0 method and the M1 method at each moment are consistent. This verifies that the large-scale energy storage aggregation method based on the principle of external network economic consistency proposed in this paper can effectively convert large-scale energy storage into virtual energy storage.

[0362] 2 Analysis of virtual energy storage parameter decision-making method considering multiple scenarios

[0363] This section conducts a simulation analysis on the proposed virtual energy storage parameter decision method considering multiple scenarios.

[0364] For multi-scenario load data, wind farm data, and photovoltaic plant data, this example generates IEEE-118 node data based on the actual data ratio of a power company in a certain province in China. In the simulation example IEEE-118 node, a wind farm is incorporated into node 2, a photovoltaic plant is incorporated into node 3, and a 4-node distribution network system is incorporated into node 9. Five energy storages are installed on node 3 in the distribution network system, and five energy storages are installed on node 4.

[0365] Considering multiple scenarios, we first calculate the economic dispatch of large-scale energy storage grid connection under multiple scenarios based on the model in Section 2.1, where the maximum value of the charging and discharging period under multiple scenarios is 320MV. Then, according to the method proposed in Chapter 2, the rated power parameter of the aggregated equivalent virtual energy storage is obtained as 320MV, which verifies the method proposed in this article. In multiple scenarios, the feasible domain after large-scale energy storage is aggregated into virtual energy storage can also be accurately obtained. In the same scenario, the economic dispatch model of the virtual system after the virtual energy storage is incorporated into the power system is also solved. Four scenarios are selected to draw graphs to compare the dispatch results of energy storage using the M0 and M1 methods, as shown below: Figure 6 shown.

[0366] In multiple scenarios, by comparing the large-scale energy storage dispatch of the M0 method and the virtual energy storage dispatch of the M1 method proposed in this paper in the same scenario, it can be intuitively seen that in the same scenario, the dispatch behavior of the virtual energy storage grid-connected power system is consistent with the dispatch behavior of the actual large-scale energy storage, which verifies the effective equivalence of virtual energy storage grid-connected dispatch in multiple scenarios. In this scenario, the overall operating cost of various resources is shown in Table 3. Tables 3 and 4 give the economic dispatch results of the power system with large-scale energy storage grid-connected on some days and the economic dispatch results of the power system with virtual energy storage grid-connected under multiple scenarios.

[0367] Table 3 Results of economic scheduling of various resources in actual systems under some scenarios

[0368]

[0369] Tab.4 Economic Dispatch Results of Various Resources in Virtual Systems under Partial Scenarios

[0370]

[0371] It can be seen from Tables 3 and 4 that the proposed large-scale energy storage aggregation method based on the principle of external grid economic consistency can achieve the equivalence of virtual system operation cost and actual system operation cost in multiple scenarios. Including multiple scenarios, the upper optimization result is 8.8×10 -9This shows that while energy storage aggregation is achieved, the economic dispatch behavior of the large-scale energy storage grid-connected power system is not changed, which illustrates the effectiveness of the large-scale energy storage aggregation system-distribution network coordinated economic dispatch method under multiple scenarios proposed in this paper.

[0372] To further illustrate the adaptability of this method to multiple scenarios, 100 scenarios were tested and the cost errors between the original system and the virtual system were calculated under 100 different scenarios. The results are shown in the figure below. Figure 7 At the same time, different numbers of scenes were tested, and the optimization results and calculation time were shown in Table 5.

[0373] Table 5 Error optimization results and calculation time of M1 method in multiple scenarios

[0374]

[0375] Depend on Figure 7 It can be seen that for 100 scenarios, the relative error of the total operating cost between the original system and the virtual system in each scenario is less than 6.0×10 -16 ,Under this method, the error of each scenario is approximately 0, which shows that this method can effectively perform virtual energy storage equivalence in each scenario and achieve the same scheduling results.

[0376] It can be seen from Table 5 that the errors obtained by solving different numbers of scenes are much less than 1.0×10 -5 , each optimization result is approximately 0. The proposed method can effectively aggregate virtual energy storage parameters under different numbers of scenarios, and because the upper objective function takes into account the scheduling information of each historical scenario, the optimization results of different scenarios under this parameter can show scheduling behaviors similar to their actual systems, indicating that this method has good results for the calculation of different numbers of scenarios. In addition, the calculation time of the proposed method increases with the increase of the number of scenarios. The average calculation time is 3.32s, 5.24s, 7.98s, 8.68s, and 10.44s, respectively. It is easy to find that the fewer the number of scenarios, the shorter the calculation time.

[0377] Since the above case analysis all uses historical data, there are bound to be differences between the new scenarios that appear in actual scheduling and the above historical scenarios, but there is still a certain correlation. Considering factors such as the characteristics of historical load and wind and solar power scenarios and their respective time correlation relationships, the WGAN method is further used to generate new load and wind and solar power scenarios to verify the adaptability of the proposed method in unseen scenarios.

[0378] Figure 8 The relationship between the Euclidean distance of the newly generated 100 scenarios and the historical scenarios and the relative error of the overall operating costs of the two scenarios is given using the method proposed in this paper. Figure 8It can be seen that the relative errors of the total operating costs of the 100 newly generated scenarios obtained by the method proposed in this paper are all less than 0.3, and the relative errors of the total operating costs of each newly generated scenario are concentrated in 0.01-0.15, which is within an acceptable range. Moreover, as the Euclidean distance between the newly generated scenario and the historical scenario increases, the relative error of the cost also shows a trend of gradually increasing. This shows that the more similar the new scenario is to the historical scenario, the smaller the Euclidean distance is, and the smaller the corresponding relative error of the cost is. Therefore, the virtual energy storage parameters obtained by the method proposed in this paper are still valid in the new scenario and have universal applicability.

[0379] in conclusion:

[0380] Large-scale energy storage grid connection can effectively reduce peak loads and fill valleys, reducing the overall operating cost of the system. There is an information barrier between large-scale energy storage and the power system, while virtual energy storage can effectively aggregate large-scale energy storage. Based on the idea of ​​cost equivalence, this paper proposes a large-scale energy storage aggregation method based on the principle of external network economic consistency. Considering multiple scenarios, a large-scale energy storage aggregation system-distribution network collaborative economic dispatch method is proposed. Finally, the effectiveness of the method proposed in this paper is verified in the IEEE-5 and IEEE-118 node systems.

Claims

1. A large-scale energy storage aggregation method based on the principle of external network economic consistency, characterized in that: The following steps are involved: 1) Construct an economic dispatch model for power systems considering large-scale energy storage; 2) Solve the economic dispatch model of the power system considering large-scale energy storage and obtain the dispatch information before energy storage aggregation; 3) Based on the dispatch information before energy storage aggregation, a two-layer optimization model of virtual energy storage considering multiple scenarios is established based on the principle of external network economic consistency; 4) Based on KKT condition derivation and big M method, the virtual energy storage two-layer optimization model considering multiple scenarios is converted into a single-layer mixed integer optimization model; 5) Solve the single-layer mixed integer optimization model to obtain large-scale energy storage aggregation scheduling information suitable for multiple scenarios.

2. A large-scale energy storage aggregation method based on the external network economic consistency principle according to claim 1, characterized in that: The objective function of the power system economic dispatch model considering large-scale energy storage is as follows: Where: C is the total cost of the original system operation, C gen,t is the power generation cost of the traditional unit in period t, C wind,t is the penalty fee for wind curtailment in period t, C light,t is the penalty fee for abandonment of light in period t, C storage,t N is the total cost of energy storage charging and discharging operation in period t; G 、N W 、N L and N B They are the number of traditional units, wind power plants, photovoltaic plants and energy storage grid-connected; Δt is the time interval, T is the total number of time periods; a i The price of the i-th traditional unit; a W is the penalty fee for unit wind abandonment; a L is the penalty fee for the unit abandoned light amount; c d,t is the unit energy storage discharge quotation in time period t; c c,t The unit energy storage charging quotation in time period t; P g,i,t is the output of the i-th traditional unit in period t; P w,k,t is the actual power of the kth wind power plant in period t; P l,r,t is the actual power of the rth photovoltaic plant in period t; P j,d,t is the discharge power of the jth energy storage period t, P j,c,t is the charging power of the jth energy storage period t; P wmax,k,t is the maximum power that can be generated by the k-th wind power plant in period t; P lmax,r,t is the maximum power that can be generated by the r-th photovoltaic plant in period t.

3. A large-scale energy storage aggregation method based on the external network economic consistency principle according to claim 1, characterized in that: The constraints of the power system economic dispatch model considering large-scale energy storage include power balance constraints, upper and lower limit constraints of branch power, output constraints of traditional units, ramp constraints, wind power plant power constraints, photovoltaic power plant power constraints, and energy storage charging and discharging related constraints; The power balance constraints are as follows: Where: N N is the total system load; P n,load,t is the load of the nth node in period t; t = 1, 2, ..., T; P g,i,t is the output of the i-th traditional unit in period t; P w,k,t is the actual power of the kth wind power plant in period t; P l,r,t is the actual power of the rth photovoltaic plant in period t; P j,d,t is the discharge power of the jth energy storage period t, P j,c,t is the charging power of the jth energy storage period t; The upper and lower limit constraints of branch power are as follows: P bpmin ≤P bp,t ≤P bpmax (7) Where: P bpmax , P bpmin are the upper and lower limits of the power of the lth branch respectively; L is the total number of branches in the system; where t=1,2,…,T; l=1,2,…,L; P bp,t is the power of the lth branch in period t; The output constraints of traditional units are as follows: P gmin,i,t ≤P g,i,t ≤P gmax,i,t (8) Where: P gmax,i,t , P gmin,i,t, They are the upper and lower limits of the output of the traditional unit respectively; t=1,2,…,T; i=1,2,…,N G ; The hill climbing constraints are as follows: r gmin,i,t ≤P g,i,t -P g,i,t-1 ≤r gdwon,i,t (9) Where: r gmin,i,t, 、r gdown,i,t are the upper and lower limits of the output ramp of the traditional unit; t=1,2,…,T; i=1,2,…,N G ;P g,i,t-1 is the output of the i-th conventional unit in period t-1; The power constraints of the wind power plant and the photovoltaic power plant are as follows: 0≤P w,k,t ≤P wmax,k,t (10) 0≤P l,r,t ≤P lmax,r,t (11) Where: t = 1, 2, ..., T; k = 1, 2, ..., N W ; r=1,2,…,N L ;P wmax,k,t is the maximum power that can be generated by the k-th wind power plant in period t; P lmax,r,t is the maximum power that can be generated by the r-th photovoltaic plant in period t; The constraints related to energy storage charging and discharging are as follows: 0≤P j,d,t ≤P j,Bmax (12) 0≤P j,c,t ≤P j,Bmax (13) AND B,j,min ≤E B,j,t ≤E B,j,max (14) E B,j,t -E B,j,t-1 -(or c P j,c,t -P j,d,t / or d )Δt=0 (15) P j,c,t P j,d,t =0 (16) Where: P j,Bmax is the rated power value of charging and discharging of the jth energy storage; E B,j,max 、E B,j,min are the upper and lower limits of the energy that can be stored in the jth energy storage; E B,j,t 、E B,j,t-1 are the energy stored in the jth energy storage in period t and period t-1 respectively; η c , η d are the charging and discharging efficiency of energy storage respectively; when the energy storage is in the charging state, P j,d,t =0; when the energy storage is in the discharge state, P j,c,t =0; when the energy storage is in a static state, P j,d,t With P j,c,t All are 0. t=1,2,…,T;j=1,2,…,N B ; Δt is the time interval.

4. A large-scale energy storage aggregation method based on the external network economic consistency principle according to claim 1, characterized in that: The virtual energy storage two-layer optimization model considering multiple scenarios includes a virtual energy storage upper layer optimization model and a virtual energy storage lower layer optimization model; The virtual energy storage upper optimization model is used to minimize the absolute value of the difference between the operating cost of various resources in each period of the original actual system and the cost of various resources in each period of the virtual system; The virtual energy storage lower layer optimization model is used to minimize the overall operating cost of the virtual system taking into account the virtual energy storage.

5. A large-scale energy storage aggregation method based on the external network economic consistency principle according to claim 4, characterized in that: The objective function of the virtual energy storage upper optimization model is as follows: Where: D is the sum of the absolute values ​​of the errors between various resources in each period of the virtual system and various resources in each period of the original system; is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period; is the power generation cost of the traditional unit in period t, is the penalty fee for wind curtailment in period t, is the total cost of energy storage charging and discharging operation in period t; a represents the scenario; The constraints of the virtual energy storage upper optimization model include virtual energy storage charging limit and virtual energy storage capacity limit; The virtual energy storage charging restrictions are as follows: Where: It is the maximum value of the total charge and discharge during large-scale energy storage dispatch in the historical dispatch information; is the charging and discharging rated power value of the jth energy storage; It is the total amount of charging and discharging during large-scale energy storage dispatch; The virtual energy storage capacity limits are as follows: In the formula, is the upper limit of the energy that can be stored in the jth energy storage; is the upper limit of virtual energy storage capacity.

6. A large-scale energy storage aggregation method based on the external network economic consistency principle according to claim 4, characterized in that: The objective function minf of the virtual energy storage lower layer optimization model is as follows: Where: N VB is the number of virtual energy storage; is the output of the i-th traditional unit in the virtual system during period t; is the actual power of the kth wind power plant in the virtual system during period t; is the actual power of the rth photovoltaic plant in the virtual system during period t; is the discharge power of the Sjth virtual energy storage in period t, is the charging power of the Sjth virtual energy storage period t; is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period; Quote for the i-th unit of traditional unit; Penalty fee for unit wind abandonment; It is the penalty fee for the unit abandoned light amount; is the unit energy storage discharge quotation in period t; The unit energy storage charging quotation in time period t; is the maximum power that can be generated by the k-th wind power plant in period t; is the maximum power that can be generated by the r-th photovoltaic plant in time period t; Δt is the time interval; The constraints of the virtual energy storage lower-level optimization model include virtual system power balance constraints, branch power upper and lower limit constraints, and virtual energy storage related constraints; The virtual system power balance constraints are as follows: In the formula, is the load of the nth node in period t; The upper and lower limit constraints of branch power are as follows: Where: l=1,2,…,L; is the power of the lth branch of the virtual system; The constraints related to virtual energy storage are as follows: In the formula, It is the maximum value of the total charge and discharge during large-scale energy storage dispatch in the historical dispatch information; is the upper limit of virtual energy storage capacity; is the virtual energy storage capacity during period t and period t-1; are the charging and discharging efficiencies of energy storage, respectively.

7. A large-scale energy storage aggregation method based on the external network economic consistency principle according to claim 4, characterized in that: The steps to convert the virtual energy storage two-layer optimization model based on the economic consistency equivalence idea into a single-layer mixed integer optimization model are as follows: 1) Expressing the virtual energy storage upper optimization model and the virtual energy storage lower optimization model in a compact form; The compact form of the virtual energy storage upper optimization model is as follows: In the formula, more is the number of scenes; f i a is the optimal solution of the virtual energy storage lower optimization model under scenario a; min D is the objective function of the virtual energy storage upper optimization model; Quote for scenario a; is an equality constraint; is an inequality constraint; The compact form of the virtual energy storage underlying optimization model is as follows: min f a =KX a (34) Where K is the coefficient matrix; X a is a variable; min f a The objective function of the virtual energy storage lower layer optimization model; is an equality constraint; is an inequality constraint; 2) Solve the optimal solution of the virtual energy storage lower layer optimization model and derive and construct the Lagrangian function L(X,λ j ,μ k ),Right now: Where: j is the vector of Lagrange multipliers, μ k is a vector composed of Lagrange multipliers. PTDF is the virtual system line transfer distribution factor matrix, Cg_N is the generator node association matrix, Cw_N is the wind power plant node association matrix, Cl_N is the photovoltaic plant node association matrix, λ j,t is the Lagrange multiplier of the lower model equality constraint, μ k,t is the Lagrange multiplier of the inequality constraint of the lower model; j = 1, 2; k = 1, 2, ..., 16; t = 1, 2, ..., T; 3) Construct the necessary and sufficient conditions for the virtual energy storage lower layer optimization model to obtain the optimal solution, namely: In the formula, r gmin,i,t, 、r gdown,i,t They are the upper and lower limits of the output ramp of the traditional unit respectively; 4) Use the Big M method to convert the nonlinear constraint μ in the necessary and sufficient conditions into k h k (X) = 0 is converted into a linear constraint, that is: 0≤μ k ≤Mz k (39) -M(1-z k )≤h k (X)≤0 (40) Where: M is infinite, Z k It is a 0-1 integer variable. 5) Construct a single-layer mixed integer optimization model, namely: In the formula, It is a 0-1 integer variable.

8. A system based on the large-scale energy storage aggregation method according to any one of claims 1 to 7, characterized in that: include: Power system economic dispatch model construction module: Construct a power system economic dispatch model considering large-scale energy storage; The objective function of the power system economic dispatch model considering large-scale energy storage is as follows: Where C is the total cost of the original system operation, C gen,t is the power generation cost of the traditional unit in period t, C wind,t is the penalty fee for wind curtailment in period t, C light,t is the penalty fee for abandonment of light in period t, C storage,t is the total cost of energy storage charging and discharging operation in period t; Dispatch information calculation module: solves the economic dispatch model of the power system considering large-scale energy storage and obtains the dispatch information before energy storage aggregation; Virtual energy storage two-layer optimization model construction module: Based on the dispatch information before energy storage aggregation, a virtual energy storage two-layer optimization model taking into account multiple scenarios is established based on the principle of external network economic consistency; The virtual energy storage two-layer optimization model includes a virtual energy storage upper layer optimization model and a virtual energy storage lower layer optimization model; The objective function of the virtual energy storage upper optimization model is as follows: Where: D is the sum of the absolute values ​​of the errors between various resources in each period of the virtual system and various resources in each period of the original system; is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period; is the power generation cost of the traditional unit in period t, is the penalty fee for wind curtailment in period t, is the total cost of energy storage charging and discharging operation in period t; a represents the scenario; The objective function minf of the virtual energy storage lower layer optimization model is as follows: Where: is the power generation cost of the traditional unit in the virtual system at the tth period, is the penalty fee for wind curtailment in the virtual system at period t, is the penalty fee for abandonment of light in the virtual system at time period t, is the total cost of charging and discharging virtual energy storage in the virtual system during the t period; Model optimization module: Based on KKT condition derivation and big M method, the virtual energy storage double-layer optimization model considering multiple scenarios is converted into a single-layer mixed integer optimization model; A single-layer mixed integer optimization model is shown below: In the formula, It is a 0-1 integer variable. Large-scale energy storage aggregation scheduling information solution module: solves the single-layer mixed integer optimization model to obtain large-scale energy storage aggregation scheduling information suitable for multiple scenarios.

9. An electronic device, characterized in that: include: A memory, a processor and a data bus, wherein the memory stores all machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the data bus, and the machine-readable instructions are executed by the processor to execute the method described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is executed.

Images