A virtual power plant optimization method for power grid deep peak load regulation auxiliary services

Abstract

The present invention relates to a virtual power plant optimization assembly method for deep peak-shaving auxiliary services of power grids, comprising the following steps: first, using a genetic algorithm to randomly generate an initial virtual power plant assembly plan population; second, simulating each virtual power plant to participate in the peak-shaving auxiliary service market based on a typical day scenario, using the linear programming solver ILOG CPLEX to solve the clearing result under a unified marginal clearing method, and calculating the peak-shaving revenue index; then, calculating the electrical distance between each node of the distribution network based on power and voltage sensitivity, and obtaining a modularity index that reflects the structural characteristics of the virtual power plant; finally, using the maximum value of the weighted sum of the peak-shaving revenue index and the modularity index as the objective function to establish a virtual power plant optimization assembly model and iteratively solve it using a genetic algorithm to formulate a virtual power plant optimization assembly plan. The method of the present invention solves the problem of aggregate assembly of flexible resources with dispersed demand and poor direct controllability, and meets the deep peak-shaving auxiliary service needs of the power grid.

Description

Technical Field

[0001] The present invention relates to the technical field of deep peak-shaving auxiliary services, and in particular to a method for optimizing the establishment of a virtual power plant for deep peak-shaving auxiliary services for power grids. Background Art

[0002] With the advent of the "dual carbon" era, new energy power generation such as wind and solar will develop rapidly and account for an increasing proportion, which will cause more serious power grid peak regulation problems. The development of traditional thermal power will gradually stagnate and decrease, and the deep peak regulation capacity on the source side will decrease instead of increase. However, demand-side flexible resources such as electric vehicles and energy storage are growing rapidly, and using them to participate in peak regulation can bring huge economic benefits.

[0003] As an effective way to aggregate demand-side flexibility resources, virtual power plants (VPPs) can achieve energy aggregation, storage, supply and use without changing their respective grid-connected methods and geographical locations, with the help of advanced communication, metering, control and other technologies. VPP aggregates various flexible resources with different characteristics and dispersion, and can further utilize their complementary characteristics to form more grid-friendly and controllable flexibility. It also plays a positive role in promoting the consumption of renewable energy and increasing the benefits of various flexible resources on the demand side. By optimizing and organizing massive decentralized demand-side units into multiple virtual power plants, deep peak-shaving auxiliary services can be better provided for the power grid.

[0004] Currently, most of the indicators for VPP formation are aimed at its structural and internal autonomy, and there are no relevant indicators for meeting the demand for grid ancillary services. Summary of the invention

[0005] The purpose of the present invention is to address the deficiencies in the prior art and provide a method for optimizing the establishment of a virtual power plant for deep peak-shaving auxiliary services in the power grid, thereby guiding dispatchers to formulate a virtual power plant establishment plan based on demand-side flexibility resources.

[0006] To achieve the above-mentioned purpose, the present invention adopts the following technical solution: a virtual power plant optimization construction method for power grid deep peak load auxiliary service, characterized in that it includes the following steps:

[0007] Step S1: Input the original data of the distribution network including load, line parameters, etc., as well as parameters such as renewable energy power generation under typical daily scenarios and the capacity of demand-side flexibility resources; use genetic algorithm to randomly generate an initial VPP formation plan population;

[0008] Step S2: simulate each virtual power plant participating in the peak load ancillary service market based on a typical day scenario, solve the clearing results using the linear programming solver ILOG CPLEX under a unified marginal clearing method, and calculate the peak load revenue index;

[0009] Step S3: Calculate the electrical distance between nodes of the distribution network based on power and voltage sensitivity to obtain a modularity index reflecting the structural characteristics of the virtual power plant;

[0010] Step S4: Calculate the fitness function of the virtual power plant optimization model and iterate and solve it according to the genetic algorithm.

[0011] Furthermore, the steps for calculating the peak load benefit index in step S2 are as follows:

[0012] S21. Calculate the deep peak-shaving power of the VPP in period t, which is the sum of the regulation capabilities of each internal flexibility resource:

[0013]

[0014] Among them, P VPP,t,dpr Indicates the load power that can be increased during the deep peak load period tVPP; P fl,m,t,dpr P represents the load power increased by the flexible load during a certain deep peak load period t; ess,n,t,dpr Indicates that the nth energy storage device increases charging or decreases discharging power during a certain deep peak load period t; M represents the total number of flexible loads in the VPP; N represents the total number of energy storage devices in the VPP; φ dpr represents the set of deep peak load periods;

[0015] S22. Calculate the declared electricity price of each VPP in period t, i.e. the marginal cost electricity price λ t Based on the marginal utility and the cost price c of each flexible resource in the period t t,i get:

[0016] λ t =max(c t,1 ,c t,2 ,c t,3 ,...,c t,n ,)

[0017] S23. Each VPP and thermal power unit simulates participating in the deep peak load regulation auxiliary service market based on a typical day scenario. After the electricity spot market based on a typical day scenario is cleared, if there is wind and solar power abandonment, the sequential deep peak load regulation market is started and the linear programming solver ILOG CPLEX is used to solve the clearing results.

[0018] S24. Analyze the market clearing results to obtain the benefits of each VPP. VPP benefits refer to the net benefits obtained by each VPP from selling its own flexible resource load increment based on the deep peak-shaving auxiliary service market clearing electricity price and its own declared electricity price, minus the benefits obtained after reducing the output cost of purchasing thermal power during non-deep peak-shaving periods due to the transfer characteristics. Then the benefits of VPPj are:

[0019]

[0020] Among them, T dpr Indicates the number of deep peak load regulation periods; dpr,t represents the clearing price of the deep peak load ancillary service market in period t; j,t P represents the declared electricity price of VPPj in period t; VPP,j,t,zdpr It represents the deep peak load regulation power of VPPj in period t;

[0021] S25. Calculate the deep peak load benefit index:

[0022]

[0023] Among them, N VPP Indicates the number of VPP groups; I max =max{I j} represents the maximum peak-shaving benefit among all VPPs. Peak-shaving benefit index

[0024] Furthermore, the steps of clearing the sequential deep peak load ancillary service market in step S23 are as follows:

[0025] S231. The dispatch center releases the deep peak-shaving demand. If only thermal power units apply for deep peak-shaving capacity, the dispatch center will clear it. If all wind and solar power can be absorbed, the next step of deep peak-shaving market will not be started. If there is still abandonment of wind and solar power, the next step of deep peak-shaving market will be started.

[0026] S232. Each VPP declares that the deep peak-shaving capacity that cannot be met even after thermal power participates in deep peak-shaving during the deep peak-shaving period will be cleared by the dispatching center.

[0027] (1) The objective function of this clearing model is to minimize cost:

[0028]

[0029] Among them, λ j,t,ex and P VPP,j,t,ex are the thermal power reduction price and purchase quantity of VPPj during the non-deep peak load period t; φ ndpr Represents a collection of non-deep peak-shaving periods.

[0030] (2) The clearing model must also meet the following constraints:

[0031] ① Two types of flexibility resources and VPP constraints after establishment:

[0032]

[0033] Among them, T ndprIndicates the number of non-deep peak-shaving periods; P fl,m,t,ndpr P represents the load power reduced by the flexible load for power balance during a non-deep peak load period t; fl ' ,m,t represents the actual power of the mth flexible load in period t; P fl,m,t,base represents the baseline power of the mth flexible load in period t; M j represents the flexible load set of the jth VPP; P fl,m,max , P fl,m,min They represent the upper and lower limits of the mth flexible load power respectively; φ represents the set of 24 time periods in a day; ΔP fl,m,U , ΔP fl,m,D They represent the maximum upward and downward climbing rates of the mth flexible load respectively; P e ' ss,n,t represents the actual power of the nth energy storage device in time period t; P ess,n,t,base represents the baseline power of the nth energy storage device in period t; P ess,n,t,ndpr Indicates that in a non-deep peak load period t, the nth energy storage device increases the discharge power or reduces the charging power for power balance; N j represents the energy storage device set of the jth VPP; P ess,n,t,out , P ess,n,t,in They represent the discharge and charging power of the nth energy storage device in time period t respectively; E ess,n,t represents the capacity state of the nth energy storage device at the initial time of period t; η n,out , η n,in Respectively represent the discharge and charging efficiency of the nth energy storage device; T out,t , T in,t They represent the total number of time periods in which the energy storage device is in the discharging and charging states before time period t; T out , T in They represent the total number of time periods during which the energy storage device is in the discharging and charging states in one day; E ess,n,max represents the maximum capacity of the nth energy storage device; P ess,n,out,max , P ess,n,in,max Respectively represent the maximum discharge and charge power of the nth energy storage device; ΔP ess,n,U , ΔP ess,n,D Respectively represent the maximum upward and downward climbing rates of the nth energy storage device; P VPP,min Indicates the minimum deep peak-shaving capacity of the VPP.

[0034] ②Balance of deep peak load regulation demand and network security constraints:

[0035]

[0036] Among them, R tIt indicates the deep peak load regulation demand that cannot be met even after the thermal power units participate in the deep peak load regulation during period t; G l-n G represents the power transfer distribution factor of the node where the energy storage device n is located to the line l; l-m G represents the power transfer distribution factor of the node where the flexible load m is located to the line l; l-d N represents the power transfer distribution factor of node d to line l; DP Indicates the number of nodes on the distribution network side; D d,t represents the rigid load of node d in time period t.

[0037] Furthermore, the steps for calculating the modularity index reflecting the structural characteristics of the virtual power plant in step S3 are as follows:

[0038] S31. Obtain active and reactive voltage sensitivity matrices through the relationship between voltage amplitude and power:

[0039] S PV =ΔP / ΔV

[0040] S QV =ΔQ / ΔV

[0041] S32. Define the electrical distance between nodes i and j based on active and reactive voltage sensitivity:

[0042]

[0043]

[0044] The electrical distance of the above two is obtained:

[0045]

[0046] S33. Based on the definition of modularity index, this paper uses the electrical distance between nodes to determine the edge weight e between network node i and node j ij . Set the edge weights between nodes to:

[0047] e ij =1-L ij / max(L)

[0048] S34. Introduce the modularity index first proposed by Newman:

[0049]

[0050] Among them, Ω represents the set of all nodes in the distribution network; represents the sum of the edge weights of all edges in the distribution network; It represents the sum of the weights of all edges connected to node i; when node i and node j are in the same virtual power plant, δ(i,j) = 1, otherwise δ(i,j) = 0. Modularity structural index The larger the value, the better the structure.

[0051] Furthermore, the virtual power plant optimization model in step S4 is as follows:

[0052] S41. The maximum value of the weighted sum of the peak-shaving benefit index and the modularity index is taken as the objective function.

[0053] The objective function of the optimization model is:

[0054]

[0055] Among them, ω1, ω2∈[0,1] represent the weights of the modularity structure index and the deep peak-shaving benefit index respectively; the two weight coefficients need to be gradually adjusted in the example to find the appropriate values ​​so that the structure and benefit average of each VPP are good. ω1+ω2=1.

[0056] The objective function is the fitness function, and the virtual power plant establishment plan with the highest function value is saved.

[0057] S42. Determine whether the fitness function of the best virtual power plant assembly plan in this iteration is equal to that in the previous 10 iterations. If not, select the best grouping plan to enter the next generation of grouping clusters, cross and mutate them, generate the next generation of grouping clusters and return to step S2; if so, output the optimized virtual power plant assembly plan.

[0058] The present invention adopts the above technical solution, which has the following beneficial effects: the obtained virtual power plant optimization assembly solution has a structure that makes the electrical connection between the internal nodes of each VPP close and the connection between each VPP loose, which is more convenient for the operation and management of the VPP. Functionally, on the premise of providing deep peak load auxiliary services, the income of each VPP is averaged, ensuring orderly and fair competition between each VPP. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 A flowchart of a virtual power plant optimization assembly method for power grid deep peak load auxiliary services according to the present invention;

[0060] Figure 2 This is a flow chart of the sequential deep peak load regulation auxiliary service market clearing in the present invention. DETAILED DESCRIPTION

[0061] The present invention is further explained below with reference to the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0062] like Figure 1 As shown, the virtual power plant optimization assembly method for power grid deep peak load auxiliary service proposed in the present invention includes the following steps:

[0063] Step S1: Input the original data of the distribution network including load, line parameters, etc., as well as parameters such as renewable energy power generation under typical daily scenarios and the capacity of demand-side flexibility resources; use genetic algorithm to randomly generate an initial VPP formation plan population;

[0064] Step S2: simulate each virtual power plant participating in the peak load ancillary service market based on a typical day scenario, solve the clearing results using the linear programming solver ILOG CPLEX under a unified marginal clearing method, and calculate the peak load revenue index;

[0065] Step S3: Calculate the electrical distance between nodes of the distribution network based on power and voltage sensitivity to obtain a modularity index reflecting the structural characteristics of the virtual power plant;

[0066] Step S4: Calculate the fitness function of the virtual power plant optimization model and iterate and solve it according to the genetic algorithm.

[0067] Specifically, step S2 includes:

[0068] S21. Calculate the deep peak-shaving power of the VPP in period t, which is the sum of the regulation capabilities of each internal flexibility resource:

[0069]

[0070] Among them, P VPP,t,dpr Indicates the load power that can be increased during the deep peak load period tVPP; P fl,m,t,dpr P represents the load power increased by the flexible load during a certain deep peak load period t; ess,n,t,dpr Indicates that the nth energy storage device increases charging or decreases discharging power during a certain deep peak load period t; M represents the total number of flexible loads in the VPP; N represents the total number of energy storage devices in the VPP; φ dpr represents the set of deep peak load periods;

[0071] S22. Calculate the declared electricity price of each VPP in period t, i.e. the marginal cost electricity price λ t Based on the marginal utility and the cost price c of each flexible resource in the period t t,i get:

[0072] λ t =max(c t,1 ,c t,2 ,c t,3 ,...,c t,n ,)

[0073] S23. Each VPP and thermal power unit simulates participating in the deep peak load regulation auxiliary service market based on a typical day scenario. After the electricity spot market based on a typical day scenario is cleared, if there is wind and solar power abandonment, the sequential deep peak load regulation market is started and the linear programming solver ILOG CPLEX is used to solve the clearing results.

[0074] S24. Analyze the market clearing results to obtain the benefits of each VPP. VPP benefits refer to the net benefits obtained by each VPP from selling its own flexible resource load increment based on the deep peak-shaving auxiliary service market clearing electricity price and its own declared electricity price, minus the benefits obtained after reducing the output cost of purchasing thermal power during non-deep peak-shaving periods due to the transfer characteristics. Then the benefits of VPPj are:

[0075]

[0076] Among them, T dpr Indicates the number of deep peak load regulation periods; dpr,t represents the clearing price of the deep peak load ancillary service market in period t; j,t P represents the declared electricity price of VPPj in period t; VPP,j,t,zdpr It represents the deep peak load regulation power of VPPj in period t;

[0077] S25. Calculate the deep peak load benefit index:

[0078]

[0079] Among them, N VPP Indicates the number of VPP groups; I max =max{I j} represents the maximum peak-shaving benefit among all VPPs. Peak-shaving benefit index

[0080] like Figure 2 As shown in the figure, the steps of clearing the sequential deep peak load ancillary service market in step S23 include:

[0081] S231. The dispatch center releases the deep peak-shaving demand. If only thermal power units apply for deep peak-shaving capacity, the dispatch center will clear it. If all wind and solar power can be absorbed, the next step of deep peak-shaving market will not be started. If there is still abandonment of wind and solar power, the next step of deep peak-shaving market will be started.

[0082] S232. Each VPP declares that the deep peak-shaving capacity that cannot be met even after thermal power participates in deep peak-shaving during the deep peak-shaving period will be cleared by the dispatching center.

[0083] (1) The objective function of this clearing model is to minimize cost:

[0084]

[0085] Among them, λ j,t,ex and P VPP,j,t,ex are the thermal power reduction price and purchase quantity of VPPj during the non-deep peak load period t; φ ndpr Represents a collection of non-deep peak-shaving periods.

[0086] (2) The clearing model must also meet the following constraints:

[0087] ① Two types of flexibility resources and VPP constraints after establishment:

[0088]

[0089] Among them, T ndpr Indicates the number of non-deep peak-shaving periods; P fl,m,t,ndpr P represents the load power reduced by the flexible load for power balance during a non-deep peak load period t; fl ' ,m,t represents the actual power of the mth flexible load in period t; P fl,m,t,base represents the baseline power of the mth flexible load in period t; M j represents the flexible load set of the jth VPP; P fl,m,max , P fl,m,min They represent the upper and lower limits of the mth flexible load power respectively; φ represents the set of 24 time periods in a day; ΔP fl,m,U , ΔP fl,m,D They represent the maximum upward and downward climbing rates of the mth flexible load respectively; P e ' ss,n,t represents the actual power of the nth energy storage device in time period t; P ess,n,t,base represents the baseline power of the nth energy storage device in period t; P ess,n,t,ndpr Indicates that in a non-deep peak load period t, the nth energy storage device increases the discharge power or reduces the charging power for power balance; N j represents the energy storage device set of the jth VPP; P ess,n,t,out , P ess,n,t,in They represent the discharge and charging power of the nth energy storage device in time period t respectively; E ess,n,t represents the capacity state of the nth energy storage device at the initial time of period t; η n,out , η n,in Respectively represent the discharge and charging efficiency of the nth energy storage device; T out,t , T in,t They represent the total number of time periods in which the energy storage device is in the discharging and charging states before time period t; T out , T in They represent the total number of time periods during which the energy storage device is in the discharging and charging states in one day; Eess,n,max represents the maximum capacity of the nth energy storage device; P ess,n,out,max , P ess,n,in,max Respectively represent the maximum discharge and charge power of the nth energy storage device; ΔP ess,n,U , ΔP ess,n,D Respectively represent the maximum upward and downward climbing rates of the nth energy storage device; P VPP,min Indicates the minimum deep peak-shaving capacity of the VPP.

[0090] ②Balance of deep peak load regulation demand and network security constraints:

[0091]

[0092] Among them, R t It indicates the deep peak load regulation demand that cannot be met even after the thermal power units participate in the deep peak load regulation during period t; G l-n G represents the power transfer distribution factor of the node where the energy storage device n is located to the line l; l-m G represents the power transfer distribution factor of the node where the flexible load m is located to the line l; l-d N represents the power transfer distribution factor of node d to line l; DP Indicates the number of nodes on the distribution network side; D d,t represents the rigid load of node d in time period t.

[0093] Specifically, the S3 steps are as follows:

[0094] S31. Obtain active and reactive voltage sensitivity matrices through the relationship between voltage amplitude and power:

[0095] S PV =ΔP / ΔV

[0096] S QV =ΔQ / ΔV

[0097] S32. Define the electrical distance between nodes i and j based on active and reactive voltage sensitivity:

[0098]

[0099]

[0100] The electrical distance of the above two is obtained:

[0101]

[0102] S33. Based on the definition of modularity index, this paper uses the electrical distance between nodes to determine the edge weight e between network node i and node j ij . Set the edge weights between nodes to:

[0103] e ij =1-L ij / max(L)

[0104] S34. Introduce the modularity index first proposed by Newman:

[0105]

[0106] Among them, Ω represents the set of all nodes in the distribution network; represents the sum of the edge weights of all edges in the distribution network; It represents the sum of the weights of all edges connected to node i; when node i and node j are in the same virtual power plant, δ(i,j) = 1, otherwise δ(i,j) = 0. Modularity structural index The larger the value, the better the structure.

[0107] Specifically, step S4 includes:

[0108] S41. The maximum value of the weighted sum of the peak-shaving benefit index and the modularity index is taken as the objective function.

[0109] The objective function of the optimization model is:

[0110]

[0111] Among them, ω1, ω2∈[0,1] represent the weights of the modularity structure index and the deep peak-shaving benefit index respectively; the two weight coefficients need to be gradually adjusted in the example to find the appropriate values ​​so that the structure and benefit average of each VPP are good. ω1+ω2=1.

[0112] The objective function is the fitness function, and the virtual power plant establishment plan with the highest function value is saved.

[0113] S42. Determine whether the fitness function of the best virtual power plant assembly plan in this iteration is equal to that in the previous 10 iterations. If not, select the best grouping plan to enter the next generation of grouping clusters, cross and mutate them, generate the next generation of grouping clusters and return to step S2; if so, output the optimized virtual power plant assembly plan.

[0114] The above embodiments are only for explaining the technical concept of the present invention and are not intended to limit the present invention. Any non-substantial modification, conversion and improvement made within the basic concept and framework of the method proposed by the present invention shall be included in the protection scope of the present invention.

Claims

1. A virtual power plant optimization construction method for power grid deep peak load auxiliary service, characterized in that: The steps include: Step S1: Input the original data of the distribution network including load, line parameters, renewable energy generation under typical daily scenarios, and the capacity of demand-side flexibility resources; use a genetic algorithm to randomly generate an initial VPP formation scheme population; Step S2: simulate each virtual power plant participating in the peak load ancillary service market based on a typical day scenario, solve the clearing results using the linear programming solver ILOG CPLEX under a unified marginal clearing method, and calculate the peak load revenue index; Step S3: Calculate the electrical distance between nodes of the distribution network based on power and voltage sensitivity to obtain a modularity index reflecting the structural characteristics of the virtual power plant; Step S4: Calculate the fitness function of the virtual power plant optimization model and iterate and solve it according to the genetic algorithm; The peak load benefit index is calculated in step S2 as follows: S21. Calculate the deep peak-shaving power of the VPP in period t, which is the sum of the regulation capabilities of each internal flexibility resource: Among them, P VPP,t,dpr P represents the load power that VPP can increase during the deep peak load regulation period t; fl,m,t,dpr P represents the load power increased by the flexible load during a certain deep peak load period t; ess,n,t,dpr Indicates that the nth energy storage device increases charging or decreases discharging power during a certain deep peak load period t; M represents the total number of flexible loads in the VPP; N represents the total number of energy storage devices in the VPP; φ dpr represents the set of deep peak load periods; S22. Calculate the declared electricity price of each VPP in period t, i.e. the marginal cost electricity price λ t Based on the marginal utility and the cost price c of each flexible resource in the period t t,i get: λ t =max(c t,1 ,c t,2 ,c t,3 ,...,c t,n ,) S23. Each VPP and thermal power unit simulates participating in the deep peak load regulation auxiliary service market based on a typical day scenario. After the electricity spot market based on a typical day scenario is cleared, if there is wind and solar power abandonment, the sequential deep peak load regulation market is started and the linear programming solver ILOG CPLEX is used to solve the clearing results; S24. Analyze the market clearing results to obtain the benefits of each VPP; VPP benefits refer to the net benefits obtained by each VPP from selling its own flexible resource load increment based on the deep peak-shaving auxiliary service market clearing electricity price and its own declared electricity price, minus the benefits obtained from the cost of reducing output by purchasing thermal power during non-deep peak-shaving periods due to the transfer characteristics; the benefits of VPPj are: Among them, T dpr Indicates the number of deep peak load regulation periods; dpr,t represents the clearing price of the deep peak load ancillary service market in period t; j,t P represents the declared electricity price of VPPj in period t; VPP,j,t,zdpr It represents the deep peak load regulation power of VPPj in period t; S25. Calculate the deep peak load benefit index: Among them, N VPP Indicates the number of VPP groups; I max =max{I j } represents the maximum peak-shaving benefit among all VPPs, the peak-shaving benefit index The steps of clearing the sequential deep peak load ancillary service market in step S23 are as follows: S231. The dispatch center releases the deep peak-shaving demand. If only thermal power units apply for deep peak-shaving capacity, the dispatch center will clear it. If all wind and solar power can be absorbed, the next step of deep peak-shaving market will not be started. If there is still wind and solar power abandonment, the next step of deep peak-shaving market will be started. S232. Each VPP declares that the deep peak-shaving capacity that cannot be met after thermal power participates in deep peak-shaving during the deep peak-shaving period shall be cleared by the dispatching center: (1) The objective function of the clearing model is to minimize cost: Among them, λ j,t,ex and P VPP,j,t,ndpr are the thermal power reduction price and purchase quantity of VPPj during the non-deep peak load period t; φ ndpr represents the set of non-deep peak-shaving periods; (2) The clearing model must also meet the following constraints: ① Two types of flexibility resources and VPP constraints after establishment: Among them, T ndpr Indicates the number of non-deep peak-shaving periods; P fl,m,t,ndpr Indicates the load power reduced by the flexible load for power balance during a non-deep peak load period t; P′ fl,m,t represents the actual power of the mth flexible load in period t; P fl,m,t,base represents the baseline power of the mth flexible load in period t; M j represents the flexible load set of the jth VPP; P fl,m,max , P fl,m,min They represent the upper and lower limits of the mth flexible load power respectively; φ represents the set of 24 time periods in a day; ΔP fl,m,U , ΔP fl,m,D They represent the maximum upward and downward climbing rates of the mth flexible load respectively; P e ' ss,n,t represents the actual power of the nth energy storage device in time period t; P ess,n,t,base represents the baseline power of the nth energy storage device in period t; P ess,n,t,ndpr Indicates that in a non-deep peak load period t, the nth energy storage device increases the discharge power or reduces the charging power for power balance; N j represents the energy storage device set of the jth VPP; P ess,n,t,out , P ess,n,t,in They represent the discharge and charging power of the nth energy storage device in time period t respectively; E ess,n,t represents the capacity state of the nth energy storage device at the initial time of period t; η n,out , η n,in Respectively represent the discharge and charging efficiency of the nth energy storage device; T out,t , T in,t They represent the total number of time periods in which the energy storage device is in the discharging and charging states before time period t; T out , T in They represent the total number of time periods during which the energy storage device is in the discharging and charging states in one day; E ess,n,max represents the maximum capacity of the nth energy storage device; P ess,n,out,max , P ess,n,in,max Respectively represent the maximum discharge and charge power of the nth energy storage device; ΔP ess,n,U , ΔP ess,n,D Respectively represent the maximum upward and downward climbing rates of the nth energy storage device; P VPP,min Indicates the minimum deep peak-shaving capacity of the VPP; ②Balance of deep peak load regulation demand and network security constraints: Among them, R t It indicates the deep peak load regulation demand that cannot be met even after the thermal power units participate in the deep peak load regulation during period t; G l-n G represents the power transfer distribution factor of the node where the energy storage device n is located to the line l; l-m G represents the power transfer distribution factor of the node where the flexible load m is located to the line l; l-d N represents the power transfer distribution factor of node d to line l; DP Indicates the number of nodes on the distribution network side; D d,t represents the rigid load of node d in time period t; The step S4 virtual power plant optimization model is as follows: S41. The maximum value of the weighted sum of the peak-shaving benefit index and the modularity index is used as the objective function. The objective function of the optimization model is: Among them, ω1, ω2∈[0,1] represent the weights of the modularity structure index and the deep peak-shaving benefit index respectively; the two weight coefficients need to be gradually adjusted in the example to find the appropriate values ​​so that the structure and benefit average of each VPP are good, ω1+ω2=1; The objective function is the fitness function, and the virtual power plant formation plan with the highest function value is saved; S42. Determine whether the fitness function of the best virtual power plant assembly plan in this iteration is equal to that in the previous 10 iterations. If not, select the best grouping plan to enter the next generation of grouping clusters, cross and mutate them, generate the next generation of grouping clusters and return to step S2; if so, output the optimized virtual power plant assembly plan.

2. The method for optimizing and establishing a virtual power plant for power grid deep peak load auxiliary services according to claim 1 is characterized in that: The modularity index reflecting the structural characteristics of the virtual power plant is calculated in step S3 as follows: S31. Obtain active and reactive voltage sensitivity matrices through the relationship between voltage amplitude and power: S PV =ΔP / ΔV S QV =ΔQ / ΔV S32. Define the electrical distance between nodes i and j based on active and reactive voltage sensitivity: The electrical distance of the above two is obtained: S33. Based on the definition of modularity index, this paper uses the electrical distance between nodes to determine the edge weight e between network node i and node j ij , set the edge weights between nodes to: e ij =1-L ij / max(L) S34. Introduce the modularity index first proposed by Newman: Among them, Ω represents the set of all nodes in the distribution network; represents the sum of the edge weights of all edges in the distribution network; Represents the sum of the weights of all edges connected to node i; when node i and node j are in the same virtual power plant, δ(i,j) = 1, otherwise δ(i,j) = 0; modularity structural index The larger the value, the better the structure.

Images