Method for constructing risk characterization model of primary frequency modulation cost function of virtual control unit

By constructing a risk characterization model for the one-time frequency modulation cost function of virtual control unit, the problem of difficult to characterize the cost structure when distributed control resource aggregation provides one-time frequency modulation services is solved, and the effective characterization and risk management of the relationship between frequency modulation service volume and cost is realized, and the optimization of the Economic benefits and service decomposition.

CN118970886APending Publication Date: 2024-11-15DALIAN UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410957472.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-17
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

At this stage, there is a lack of a cost function representation method for virtual control unit aggregation for primary frequency modulation auxiliary services, which is difficult to effectively reflect the cost structure of distributed control resource aggregation when providing primary frequency modulation services, and considers the distributed resource power and frequency modulation capabilities of new energy and load side. strong uncertainty.

Method used

Build a risk characterization model of the cost function of the cost of the virtual control unit, analyze the operation characteristics of the distributed control resources on the substation side, establish a frequency modulation operation model, and use the conditional risk value method to quantify uncontrollable distributed resources to provide frequency modulation services The aggregation cost loss risk caused by fluctuations in quantity is established to establish a cost function model that identifies risks, and characterizes the relationship between the aggregation frequency modulation service volume and cost of the virtual control unit in the form of a time-division and segmented linear curve cluster.

Benefits of technology

The economic benefits and risk management of the primary frequency modulation service are optimized, and the cost of providing primary frequency modulation services can be intuitively described by the aggregation of distributed control resources, and the frequency modulation service volume is reasonably decomposed to each distributed resource for execution after receiving the scheduling signal of the superior power grid.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118970886A_ABST
    Figure CN118970886A_ABST
Patent Text Reader

Abstract

The invention discloses a method for constructing a risk characterization model of a primary frequency modulation cost function of a virtual control unit, and belongs to the field of a primary frequency modulation auxiliary service market in which virtual control units participate in aggregation. The method comprises the following steps of: analyzing operation characteristics of various distributed control resources at a local dispatching primary variable grid side, constructing a dynamic aggregation model of the primary frequency modulation capability of the distributed resources of a virtual control unit, and converting missing or surplus fluctuation of partial distributed control resource frequency modulation service quantity into an aggregation cost loss risk; establishing a primary frequency modulation service quantity risk description model of the virtual control unit; and establishing a cost function model for providing the primary frequency modulation service by the virtual control unit considering the risk, and representing the relationship between the aggregation primary frequency modulation service quantity and the cost of the virtual control unit in a time-phased piecewise linear curve cluster form. According to the method, a distributed control resource aggregation frequency modulation cost function under the risk of partial distributed control resource frequency modulation capability deficiency or surplus can be intuitively described, and a scheduling instruction of a superior power grid to a virtual control unit can be effectively decomposed to each distributed control resource for execution through the proposed method.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of virtual control unit aggregation participating in the primary frequency regulation auxiliary service market, and relates to a method for constructing a virtual control unit primary frequency regulation cost function risk characterization model. For a large number of distributed control resources with primary frequency regulation capabilities on the local primary substation grid side, a virtual control unit is aggregated to provide frequency regulation auxiliary services for the provincial power grid. The operating characteristics of various types of distributed control resources on the local primary substation grid side are analyzed, and primary frequency regulation operation models are established for distributed control resources with different operating characteristics; based on the distributed resource primary frequency regulation operation model, distributed resources with strong power randomness are defined as uncontrollable distributed resources, and the conditional risk value method is used to quantify the aggregate cost loss risk caused by the fluctuation of the frequency regulation service volume provided by the uncontrollable distributed resources, and a virtual control unit primary frequency regulation service volume risk characterization model is established; on this basis, a cost function model for the virtual control unit to provide primary frequency regulation services taking into account risks is established, and the relationship between the aggregated primary frequency regulation service volume and the cost of the virtual control unit is characterized in the form of a time-divided piecewise linear curve cluster. Background Art

[0002] Against the backdrop of an increasing proportion of new energy, the power system faces the challenge of transforming traditional frequency regulation resources to clean energy. Distributed Control Resource (DCR), with its flexibility and heterogeneity, has become an effective supplement to frequency regulation resources. Policies such as the "Regulations on the Management of Power Grid Operation" and the "Management Measures for Power Auxiliary Services" encourage DCR to reach market access thresholds through aggregation and participate in the ancillary service market. It promotes the realization of economies of scale and ensures compensation based on frequency regulation performance, thereby improving the primary frequency regulation capability and stability of the system. Therefore, the aggregation of primary frequency regulation resources of DCR is crucial to improving the stability and market responsiveness of the power system.

[0003] Regarding the potential of DCR in primary frequency regulation, although they have problems such as small single capacity, scattered distribution and strong uncertainty, the regulation potential of these resources can be effectively tapped by aggregating and participating in the primary frequency regulation auxiliary service market through the virtual control unit (VCU). In the present invention, the VCU is formed with the local primary substation as the basic unit, and the existing control and communication system is used to realize the direct dispatch of the provincial dispatch to the VCU, avoiding additional investment. Through centralized management, the safe and optimized operation of the system is ensured, while striving for the maximum benefit of the DCR. In addition, the VCU framework optimizes energy utilization and improves the efficiency and scale effect of the power grid by unifying and coordinating heterogeneous DCRs.

[0004] Under this management framework, it is critical to study the representation method of the cost function. The model of the cost function should help VCU clarify the relationship between cost and service volume, make reasonable bidding decisions, and effectively decompose the dispatch instructions of the provincial power grid's primary frequency regulation service. At the same time, considering the uncertainty of DCR's frequency regulation capacity, the model should also evaluate the impact of this uncertainty on the cost function to manage related risks.

[0005] At present, the research on DCR aggregation to provide ancillary services for the power grid mainly focuses on peak load ancillary services, while the research on DCR aggregation to provide frequency regulation services for the power grid mainly focuses on the joint research of secondary frequency regulation or primary frequency regulation services and other markets. There are relatively few studies on the participation of distributed control resources in the paid primary frequency regulation ancillary service market. Research on DCR aggregation mainly focuses on how DCRs with a certain type of characteristics can be aggregated to form scale effects (Cui Yang, Zhou Feijie, Zhao Yuting, et al. Market clearing method for electric vehicle loads participating in frequency regulation ancillary services considering substitution effects [J]. Electric Power Automation Equipment, 2022, 42(07): 111-119.), as well as the description of external characteristics after heterogeneous DCR aggregation. However, the cost function of heterogeneous DCR participating in primary frequency regulation services has not been fully studied in existing studies. The study of cost function is crucial for subsequent research on DCR aggregation participating in the market. At the same time, when heterogeneous DCRs are aggregated, the impact of the uncertainty of their service capabilities on costs cannot be ignored.

[0006] In summary, a method for characterizing the aggregated cost function of virtual control units for primary frequency regulation auxiliary services is urgently needed at this stage. This method can reflect the cost structure when distributed control resources are aggregated to provide primary frequency regulation services, and considering the strong uncertainty of the power and frequency regulation capabilities of distributed resources on the new energy and load sides, the cost function should include a reasonable characterization of risks. In addition, this characterization method must also have the function of reasonably decomposing the frequency regulation service volume to the execution of each distributed resource after receiving the dispatch signal from the upper power grid. Summary of the invention

[0007] The present invention is aimed at a large number of distributed control resources on the local primary transformer grid side, and aggregates to form a virtual control unit to provide frequency regulation auxiliary services for the provincial power grid. In order to realize the aggregation of virtual control units to participate in the primary frequency regulation auxiliary service market, it is necessary to clarify the relationship between the aggregated service volume and cost. At the same time, due to the strong uncertainty of the frequency regulation capability of some distributed control resources, it is necessary to consider the risk of aggregated frequency regulation cost loss caused by the uncertainty of frequency regulation capability.

[0008] The technical solution adopted by the present invention to achieve the above-mentioned purpose is:

[0009] A method for constructing a risk characterization model of the primary frequency regulation cost function of a virtual control unit. For a large number of distributed control resources with primary frequency regulation capabilities on the local primary substation grid side, a virtual control unit is aggregated to provide frequency regulation auxiliary services for the provincial power grid. The operating characteristics of various types of distributed control resources on the local primary substation grid side are analyzed, and primary frequency regulation operation models are established for distributed control resources with different operating characteristics; based on the distributed resource primary frequency regulation operation model, distributed resources with strong power randomness are defined as uncontrollable distributed resources, and the conditional risk value method is used to quantify the aggregated cost loss risk caused by the fluctuation of the frequency regulation service volume provided by uncontrollable distributed resources, and a virtual control unit primary frequency regulation service volume risk characterization model is established; on this basis, a cost function model for the primary frequency regulation service provided by a virtual control unit taking into account risks is established, and the relationship between the aggregated primary frequency regulation service volume and the cost of the virtual control unit is characterized in the form of a time-divided piecewise linear curve cluster. The following steps are included:

[0010] S1: Analyze the operating characteristics of various distributed control resources on the primary substation grid side, and establish primary frequency regulation operation models for distributed control resources with different operating characteristics; the details are as follows:

[0011] For the large number of distributed control resources with primary frequency regulation capability on the grid side of the local primary substation, they are divided into three categories: power supply, load and energy storage according to their operating characteristics. Primary frequency regulation operation models are established for each of them as the basis for constructing subsequent models, including: distributed control resources with power supply characteristics, distributed control resources with load characteristics, and distributed control resources with energy storage characteristics.

[0012] The primary frequency modulation operation model of the distributed control resource with power supply characteristics is as follows:

[0013]

[0014] Where: P D,t is the output of distributed control resources in period t; α are the upper and lower limit coefficients of the unit's operating power respectively; r D is the upper limit of the climbing rate of the unit; P Dr P is the primary frequency regulation reserve capacity of the unit; D is the rated power of the unit; R D is the adjustment coefficient of the unit; f N is the rated value of the system frequency; Δf m is the maximum deviation of the system frequency.

[0015] The primary frequency modulation operation model of the distributed control resource with load characteristics is as follows:

[0016]

[0017] Where: P A,t is the output of distributed control resources in period t, β are the upper and lower limit coefficients of the unit’s operating power respectively; P Ar The primary frequency regulation reserve capacity of the unit; P A is the rated power of the unit; R A is the adjustment coefficient of the unit.

[0018] The primary frequency modulation operation model of the distributed control resource with energy storage characteristics is as follows:

[0019]

[0020] SOC min ≤SOC t ≤SOC max (7)

[0021]

[0022] Where: P c,ES,t , P d,ES,t are the charging power and discharging power of the distributed control resources in period t respectively; P c,ES,max , P d,ES,max They are the upper limits of the device’s charging power and discharging power respectively; SOC t SOC is the state of charge of the device during period t; max , SOC min are the upper and lower limits of the device’s state of charge respectively; η c , η d E are the charging efficiency and discharging efficiency of the device respectively; t-1 is the capacity of the device at time t-1; P E is the rated capacity of the equipment; P Er The primary frequency regulation reserve capacity of the device; R E is the adjustment coefficient of the device.

[0023] The three types of distributed resources with primary frequency regulation capability on the local primary substation side are aggregated into a virtual control unit. The present invention uses an equivalent unit to replace each distributed resource under its jurisdiction. The equivalent regulation coefficient after equalization is:

[0024]

[0025] Where: R eq,t is the equivalent adjustment coefficient of the virtual control unit during period t; ΔP Σ,t P is the primary frequency modulation capacity of the virtual control unit in period t;ΣN,t is the rated power of each distributed resource aggregated by the virtual control unit during period t.

[0026] S2: Based on the distributed resource primary frequency regulation operation model obtained in S1, the distributed resources with strong power randomness are defined as uncontrollable distributed resources. The conditional risk value method is used to quantify the aggregate cost loss risk caused by the fluctuation of the frequency regulation service volume provided by uncontrollable distributed resources, and a virtual control unit primary frequency regulation service volume risk characterization model is established; the details are as follows:

[0027] It is known that the distribution method of the service volume provided by each distributed control resource under the provision of a frequency modulation service volume is:

[0028] x T =(x1,...,x i ,...,x N ) (11)

[0029] Where: x T It is a method for allocating the service volume provided by each distributed control resource under the condition that the service volume provided by the primary frequency modulation service is known; i The proportion of service volume provided by the i-th distributed control resource providing primary frequency modulation service; N is the number of all distributed control resources providing primary frequency modulation service.

[0030] The distributed control resource combination frequency modulation cost vector is:

[0031] y=(y1,...y i ,..y N ) (12)

[0032] Where: y is the frequency modulation cost vector of the distributed control resource combination; y i The cost of providing frequency modulation service volume for the i-th distributed control resource providing primary frequency modulation service; N is the number of all distributed control resources providing primary frequency modulation service.

[0033] The profit function of the distributed control resource combination is:

[0034] R(x,y)=x T y (13)

[0035] Where: R(x,y) is the revenue function of the distributed control resource combination under the known primary frequency regulation service volume; x T is the allocation method of the service volume provided by each distributed control resource under the known frequency modulation service volume; y is the frequency modulation cost vector of the distributed control resource combination.

[0036] The loss function of the distributed control resource combination is:

[0037] f(x,y)=-(x1y1+...+x i y i +...+x N y N )=-x T y (14)

[0038] Where: f(x,y) is the loss function of the distributed control resource combination under the known primary frequency modulation service volume; x i The proportion of service volume provided by the distributed control resource that provides primary frequency modulation service for the i-th time; y i The cost of providing frequency modulation service volume for the i-th distributed control resource providing primary frequency modulation service; x T is an allocation method for the service volume provided by each distributed control resource under the condition that the amount of primary frequency modulation service provided is known; y is the frequency modulation cost vector of the distributed control resource combination; N is the number of all distributed control resources providing primary frequency modulation service.

[0039] Based on the conditional risk value theory, the risk characterization model of the primary frequency regulation service volume of the virtual control unit is established as follows:

[0040]

[0041] Where: α is the value of risk value, which means the maximum possible loss of the aggregated frequency modulation cost within a specific period of time under the confidence level β; F β is the value of conditional risk value, which means the average loss of aggregated frequency modulation under risk value; β is the confidence level; x T is the allocation method of the service volume provided by each distributed control resource under the known frequency modulation service volume; y is the frequency modulation cost vector of the distributed control resource combination; ρ(y) is the probability density function of the random variable y.

[0042] In general, since the analytical expression of the probability density function ρ(y) is difficult to obtain, the present invention uses the historical data of the random variable y and obtains sample data through a clustering algorithm to give an estimate of the integral:

[0043]

[0044] Where: α is the value of risk value, which means the maximum possible loss of the aggregated frequency modulation cost within a specific period of time under the confidence level β; F β is the value of conditional risk value, which means the average loss of aggregated frequency modulation under risk value; β is the confidence level; x Tis an allocation method for the service volume provided by each distributed control resource under the condition that the amount of primary frequency modulation service provided is known; y is the frequency modulation cost vector of the distributed control resource combination; N is the number of all distributed control resources providing primary frequency modulation service.

[0045] S3: For the distributed resources with strong uncertainty in primary frequency regulation capability of uncontrollable distributed control resources, use their annual frequency regulation service volume data as a scenario set to reduce the scenarios to obtain the service volume and corresponding probability of each scenario, with the goal of optimizing the economic efficiency of primary frequency regulation services provided by the aggregation of each distributed resource, and use the conditional risk value to describe the cost loss risk brought by the uncontrollable distributed control resources for the aggregated frequency regulation cost, and obtain the cost curve of different primary frequency regulation service volumes taking into account the risk. The details are as follows:

[0046] First, for uncontrollable distributed resources, the full-year data is used as a scenario set to perform scenario reduction and obtain the power and response probability of each scenario.

[0047] Secondly, based on the obtained scenarios and probabilities after the reduction of uncontrollable distributed control resources, and the primary frequency regulation operation models of various distributed control resources obtained in step S1, on the basis of the risk characterization model of the primary frequency regulation service volume of the virtual control unit obtained in step S2, several calculation points are selected within the aggregation boundary of the virtual control unit of the primary frequency regulation of the distributed control resources, and the cost loss risk brought by the uncertainty of the primary frequency regulation capacity of the uncontrollable distributed control resources to the aggregated frequency regulation cost is described by using the conditional risk value with the goal of optimizing the economic efficiency of distributed resource aggregation, and the virtual control unit cost when aggregating different spare capacities is calculated; the objective function is:

[0048]

[0049] Where: ψ is the set of all scheduling times; S is the set of all scenarios; π s is the probability corresponding to scene s; c(p i,t ) is the operating cost function of device i; c t,p is the electricity purchase price at time t; p t,tie is the actual power of the tie line at time t; c r (p i,t ) is the cost function for providing primary frequency regulation backup for device i; L is the risk coefficient; δ is the value of conditional risk value;

[0050] The constraints include:

[0051] (1) Active power dynamic balance constraint:

[0052]

[0053] Where: f(pi,t ) is the active power of device i in period t; S E is a collection of devices with energy storage characteristics; S A is a collection of equipment with load characteristics; ΔP u,t,PFR , ΔP d,t,PFR c are the primary frequency regulation up / down capacity provided by the virtual control unit to the upper power grid at time t; u 、c d They are the steps for reporting the up-frequency modulation capacity and down-frequency modulation capacity of the virtual aggregation unit, expressed as percentages; are the primary frequency regulation up / down capacities of power supply device i, energy storage device j, and load device k in period t; P es,j,t is the power of energy storage j in period t, with charging as the positive direction; P Ln,t is the non-adjustable load power during period t; P a,k,t is the power of load device k in period t;

[0054] (2) Equivalent adjustment coefficient constraint

[0055] In steady state, the frequency change in the entire system is the same. Each distributed resource aggregated into a virtual control unit can be replaced by an equivalent unit. The calculation of the equivalent adjustment coefficient after equalization is the same as formula (10), which will not be repeated here.

[0056] The equivalent adjustment coefficient must satisfy the constraints:

[0057]

[0058] Where: R eq They are the upper and lower limits of the equivalent adjustment coefficient respectively;

[0059] (3) Conditional Value at Risk Constraints

[0060] Conditional risk value is used to measure the risk of cost changes brought by uncontrollable distributed control resources to aggregate frequency modulation. For the defined discrete cost distribution, when the confidence level is β, the conditional risk value is approximately the expected cost corresponding to the set of scenarios with a probability of (1-α)×100% of higher cost. The relevant constraints of the conditional risk value are:

[0061]

[0062] 0≤L≤1 (22)

[0063]

[0064] x i ≥0 (24)

[0065] z s ≥(-x T y-α) (25)

[0066] z s ≥0 (26)

[0067] Where: δ is the value of conditional risk value; α is the value of risk value, which is the boundary value of conditional risk value calculation; π s is the probability of scene s; β is the confidence level; z s is the dummy variable of the conditional risk value hypothesis; L is the risk coefficient, the larger its value is, the more risk-averse the virtual control unit is; x i The proportion of service volume provided by the distributed control resource that provides primary frequency modulation service for the i-th time; y i The cost of providing frequency modulation service volume for the i-th distributed control resource providing primary frequency modulation service; x T is an allocation method for the service volume provided by each distributed control resource under the condition that the amount of primary frequency modulation service provided is known; y is the frequency modulation cost vector of the distributed control resource combination; N is the number of all distributed control resources providing primary frequency modulation service.

[0068] The other constraints are the same as those in equations (1), (3), (5)-(8), and will not be elaborated here.

[0069] The present invention aggregates the distributed control resources on the primary transformer grid side to form a virtual control unit, providing primary frequency regulation auxiliary services for the provincial power grid. The invention constructs a primary frequency regulation operation model by analyzing the operating characteristics of various distributed control resources, and uses the conditional risk value method to quantify the fluctuation risk of frequency regulation service volume for uncontrollable resources. Furthermore, the invention establishes a primary frequency regulation service cost function model that takes risk into account, characterizes the relationship between service volume and cost through a cluster of time-divided and piecewise linear curves, and optimizes the economic benefits and risk management of primary frequency regulation services.

[0070] The beneficial effects of the present invention are:

[0071] The present invention aggregates a large number of distributed control resources with primary frequency regulation capabilities on the local primary substation side to form a virtual control unit to provide frequency regulation auxiliary services for the provincial power grid. The operating characteristics of various types of distributed control resources on the local primary substation side are analyzed, and primary frequency regulation operation models are established for distributed control resources with different operating characteristics; based on the distributed resource primary frequency regulation operation model, distributed resources with strong power randomness are defined as uncontrollable distributed resources, and the conditional risk value method is used to quantify the aggregated cost loss risk caused by the fluctuation of the frequency regulation service volume provided by uncontrollable distributed resources, and a virtual control unit primary frequency regulation service volume risk characterization model is established; on this basis, a cost function model for the virtual control unit to provide primary frequency regulation services taking into account risks is established, and the relationship between the aggregated primary frequency regulation service volume and the cost of the virtual control unit is characterized in the form of a time-divided piecewise linear curve cluster. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 Schematic diagram of the virtual control unit architecture.

[0073] Figure 2 Solve the flow chart for the virtual control unit cost function.

[0074] Figure 3 is the total load of the example system.

[0075] Figure 4 It is a time-of-use electricity price.

[0076] Figure 5(a) shows a typical scenario of wind power;

[0077] Figure 5(b) shows a typical scenario of flexibility load A;

[0078] Figure 5(c) shows a typical scenario of flexibility load B.

[0079] Figure 6 Aggregate primary frequency regulation reserve capacity cost curve for the virtual control unit.

[0080] Figure 7 Aggregate cost-effective frontier curve for virtual control units considering risk.

[0081] Figure 8 (a) is the frequency modulation capacity ratio allocation when the risk preference coefficient L = 0.3;

[0082] Figure 8 (b) is the frequency modulation capacity ratio allocation when the risk preference coefficient L = 0.5;

[0083] Figure 8 (c) is the frequency modulation capacity ratio allocation when the risk preference coefficient L = 0.6;

[0084] Figure 8(d) is the frequency modulation capacity ratio allocation when the risk preference coefficient L = 0.7. DETAILED DESCRIPTION

[0085] The present invention is described in detail below with reference to the accompanying drawings and embodiments, but the present invention is not limited to the specific embodiments.

[0086] A method for constructing a risk characterization model of a primary frequency modulation cost function of a virtual control unit comprises the following steps:

[0087] S1: Analyze the operating characteristics of various distributed control resources on the primary substation grid side, and establish primary frequency regulation operation models for distributed control resources with different operating characteristics; the details are as follows:

[0088] For the large number of distributed control resources with primary frequency regulation capability on the grid side of the local primary substation, they are divided into three categories: power supply, load and energy storage according to their operating characteristics. Primary frequency regulation operation models are established for each of them as the basis for constructing subsequent models, including: distributed control resources with power supply characteristics, distributed control resources with load characteristics, and distributed control resources with energy storage characteristics.

[0089] The primary frequency modulation operation model of the distributed control resource with power supply characteristics is as follows:

[0090]

[0091] Where: P D,t is the output of distributed control resources in period t; α are the upper and lower limit coefficients of the unit's operating power respectively; r D is the upper limit of the climbing rate of the unit; P Dr The primary frequency regulation reserve capacity of the unit; P D is the rated power of the unit; R D is the adjustment coefficient of the unit; f N is the rated value of the system frequency; Δf m is the maximum deviation of the system frequency.

[0092] The primary frequency modulation operation model of the distributed control resource with load characteristics is as follows:

[0093]

[0094] Where: P A,t is the output of distributed control resources in period t, β are the upper and lower limit coefficients of the unit’s operating power respectively; P Ar The primary frequency regulation reserve capacity of the unit; P Ais the rated power of the unit; R A is the adjustment coefficient of the unit.

[0095] The primary frequency modulation operation model of the distributed control resource with energy storage characteristics is as follows:

[0096]

[0097]

[0098] SOC min ≤SOC t ≤SOC max (7)

[0099]

[0100] Where: P c,ES,t , P d,ES,t are the charging power and discharging power of the distributed control resources in period t respectively; P c,ES,max , P d,ES,max They are the upper limits of the device’s charging power and discharging power respectively; SOC t SOC is the state of charge of the device during period t; max , SOC min are the upper and lower limits of the device’s state of charge respectively; η c , η d E are the charging efficiency and discharging efficiency of the device respectively; t-1 is the capacity of the device at time t-1; P E is the rated capacity of the equipment; P Er The primary frequency regulation reserve capacity of the device; R E is the adjustment coefficient of the device.

[0101] The distributed resources with primary frequency regulation capability on the local primary substation side are aggregated into a virtual control unit. An equivalent unit can be used to replace each distributed resource under its jurisdiction. The equivalent regulation coefficient after equalization is:

[0102]

[0103] Where: R eq,t is the equivalent adjustment coefficient of the virtual control unit during period t; ΔP Σ,t P is the primary frequency modulation capacity of the virtual control unit in period t; ΣN,t is the rated power of each distributed resource aggregated by the virtual control unit during period t.

[0104] S2: Based on the distributed resource primary frequency regulation operation model obtained in S1, the distributed resources with strong power randomness are defined as uncontrollable distributed resources. The conditional risk value method is used to quantify the aggregate cost loss risk caused by the fluctuation of the frequency regulation service volume provided by uncontrollable distributed resources, and a virtual control unit primary frequency regulation service volume risk characterization model is established; the details are as follows:

[0105] It is known that the distribution method of the service volume provided by each distributed control resource under the provision of a frequency modulation service volume is:

[0106] x T =(x1,...,x i ,...,x N ) (11)

[0107] Where: x T It is a method for allocating the service volume provided by each distributed control resource under the condition that the service volume provided by the primary frequency modulation service is known; i The proportion of service volume provided by the i-th distributed control resource providing primary frequency modulation service; N is the number of all distributed control resources providing primary frequency modulation service.

[0108] The distributed control resource combination frequency modulation cost vector is:

[0109] y=(y1,...y i ,..y N )

[0110] (12) where: y is the frequency modulation cost vector of the distributed control resource combination; y i The cost of providing frequency modulation service volume for the i-th distributed control resource providing primary frequency modulation service; N is the number of all distributed control resources providing primary frequency modulation service.

[0111] The profit function of the distributed control resource combination is:

[0112] R(x,y)=x T y (13)

[0113] Where: R(x,y) is the revenue function of the distributed control resource combination under the known primary frequency regulation service volume; x T is the allocation method of the service volume provided by each distributed control resource under the known frequency modulation service volume; y is the frequency modulation cost vector of the distributed control resource combination.

[0114] The loss function of the distributed control resource combination is:

[0115] f(x,y)=-(x1y1+...+x i y i +...+xN y N )=-x T y (14)

[0116] Where: f(x,y) is the loss function of the distributed control resource combination under the known primary frequency modulation service volume; x i The proportion of service volume provided by the distributed control resource that provides primary frequency modulation service for the i-th time; y i The cost of providing frequency modulation service volume for the i-th distributed control resource providing primary frequency modulation service; x T is an allocation method for the service volume provided by each distributed control resource under the condition that the amount of primary frequency modulation service provided is known; y is the frequency modulation cost vector of the distributed control resource combination; N is the number of all distributed control resources providing primary frequency modulation service.

[0117] The risk characterization model of the primary frequency modulation service volume of the virtual control unit established based on the conditional risk value theory is:

[0118]

[0119] Where: α is the value of risk value, which means the maximum possible loss of the aggregated frequency modulation cost within a specific period of time under the confidence level β; F β is the value of conditional risk value, which means the average loss of aggregated frequency modulation under risk value; β is the confidence level; x T is the allocation method of the service volume provided by each distributed control resource under the known frequency modulation service volume; y is the frequency modulation cost vector of the distributed control resource combination; ρ(y) is the probability density function of the random variable y.

[0120] Usually, since the analytical expression of the probability density function ρ(y) is difficult to obtain, we can use the historical data of the random variable y and obtain sample data through a clustering algorithm to give an estimate of the integral:

[0121]

[0122] Where: α is the value of risk value, which means the maximum possible loss of the aggregated frequency modulation cost within a specific period of time under the confidence level β; F β is the value of conditional risk value, which means the average loss of aggregated frequency modulation under risk value; β is the confidence level; x T is an allocation method for the service volume provided by each distributed control resource under the condition that the amount of primary frequency modulation service provided is known; y is the frequency modulation cost vector of the distributed control resource combination; N is the number of all distributed control resources providing primary frequency modulation service.

[0123] S3: For the distributed resources with strong uncertainty in primary frequency regulation capability of uncontrollable distributed control resources, use their annual frequency regulation service volume data as a scenario set to reduce the scenarios to obtain the service volume and corresponding probability of each scenario, with the goal of optimizing the economic efficiency of primary frequency regulation services provided by the aggregation of each distributed resource, and use the conditional risk value to describe the cost loss risk brought by the uncontrollable distributed control resources for the aggregated frequency regulation cost, and obtain the cost curve of different primary frequency regulation service volumes taking into account the risk. The details are as follows:

[0124] First, for uncontrollable distributed resources, the full-year data is used as a scenario set to perform scenario reduction and obtain the power and response probability of each scenario.

[0125] Secondly, based on the obtained scenarios and probabilities after the reduction of uncontrollable distributed control resources, and the primary frequency regulation operation models of various distributed control resources obtained in step S1, on the basis of the risk characterization model of the primary frequency regulation service volume of the virtual control unit obtained in step S2, several calculation points are selected within the aggregation boundary of the virtual control unit of the primary frequency regulation of the distributed control resources, and the cost loss risk brought by the uncertainty of the primary frequency regulation capacity of the uncontrollable distributed control resources to the aggregated frequency regulation cost is described by using the conditional risk value with the goal of optimizing the economic efficiency of distributed resource aggregation, and the virtual control unit cost when aggregating different spare capacities is calculated; the objective function is:

[0126]

[0127] Where: ψ is the set of all scheduling times; S is the set of all scenarios; π s is the probability corresponding to scene s; c(p i,t ) is the operating cost function of device i; c t,p is the electricity purchase price at time t; p t,tie is the actual power of the tie line at time t; c r (p i,t ) is the cost function for providing primary frequency regulation backup for device i; L is the risk coefficient; δ is the value of conditional risk value;

[0128] The constraints include:

[0129] (1) Active power dynamic balance constraint:

[0130]

[0131] Where: f(p i,t ) is the active power of device i in period t; S E is a collection of devices with energy storage characteristics; S A is a collection of equipment with load characteristics; ΔP u,t,PFR , ΔP d,t,PFRc are the primary frequency regulation up / down capacity provided by the virtual control unit to the upper power grid at time t; u 、c d They are the steps for reporting the up-frequency modulation capacity and down-frequency modulation capacity of the virtual aggregation unit, expressed as percentages; are the primary frequency regulation up / down capacities of power supply device i, energy storage device j, and load device k in period t; P es,j,t is the power of energy storage j in period t, with charging as the positive direction; P Ln,t is the non-adjustable load power during period t; P a,k,t is the power of load device k in period t;

[0132] (2) Equivalent adjustment coefficient constraint

[0133] In steady state, the frequency change in the entire system is the same. Each distributed resource aggregated into a virtual control unit can be replaced by an equivalent unit. The calculation of the equivalent adjustment coefficient after equalization is the same as formula (10), which will not be repeated here.

[0134] The equivalent adjustment coefficient must satisfy the constraints:

[0135]

[0136] Where: R eq They are the upper and lower limits of the equivalent adjustment coefficient respectively;

[0137] (3) Conditional Value at Risk Constraints

[0138] Conditional risk value is used to measure the risk of cost changes brought by uncontrollable distributed control resources to aggregate frequency modulation. For the defined discrete cost distribution, when the confidence level is β, the conditional risk value is approximately the expected cost corresponding to the set of scenarios with a probability of (1-α)×100% of higher cost. The relevant constraints of the conditional risk value are:

[0139]

[0140] 0≤L≤1 (22)

[0141]

[0142] x i ≥0 (24)

[0143] z s ≥(-x T y-α) (25)

[0144] z s ≥0 (26)

[0145] Where: δ is the value of conditional risk value; α is the value of risk value, which is the boundary value of conditional risk value calculation; π s is the probability of scene s; β is the confidence level; z s is the dummy variable of the conditional risk value hypothesis; L is the risk coefficient, the larger its value is, the more risk-averse the virtual control unit is; x i The proportion of service volume provided by the distributed control resource that provides primary frequency modulation service for the i-th time; y i The cost of providing frequency modulation service volume for the i-th distributed control resource providing primary frequency modulation service; x T is an allocation method for the service volume provided by each distributed control resource under the condition that the amount of primary frequency modulation service provided is known; y is the frequency modulation cost vector of the distributed control resource combination; N is the number of all distributed control resources providing primary frequency modulation service.

[0146] The other constraints are the same as those in equations (1), (3), (5)-(8), and will not be elaborated here.

[0147] The specific implementation of the present invention is described in detail by taking the aggregation of multiple distributed control resources with three types of characteristics, namely power supply, energy storage and load, into a virtual control unit to provide a frequency regulation service for a provincial power grid as an example. The parameters of each device are shown in Table 1. The scheduling time interval is 1h, and the total load is as follows: Figure 3 As shown in the figure, the time-of-use electricity price cost is calculated according to the time-of-use electricity price policy of Liaoning Province. The time-of-use electricity price is as follows: Figure 4 The interior point method is used to optimize the operation and aggregation of distributed control resources in the system within 24 hours to verify the correctness of the established model.

[0148] The specific steps of this embodiment are:

[0149] Step 1: Diesel generators and energy storage systems are set as controllable DER, while wind power and some flexible loads are regarded as uncontrollable DER due to the randomness of their output and frequency regulation capabilities. The randomness of wind speed is characterized by Weibull distribution, and the frequency regulation reserve capacity of flexible loads A and B is assumed to be normally distributed. In order to enhance diversity, flexible load C is regarded as controllable DER. Monte Carlo simulation is used to generate scenario sets of wind power output and flexible loads A and B, and the K-means algorithm is used to cluster each group of data to obtain 4 scenarios. The typical wind speed scenario is shown in Figure 5(a), the typical scenario of flexible load A is shown in Figure 5(b), and the typical scenario of flexible load B is shown in Figure 5(c). The scenario probabilities of wind power and flexible loads are shown in Table 2. Considering the uncertainty of wind power output and flexible loads, a total of 64 scenarios are defined, and the final probability of each scenario is the product of the probability of wind power scenario and the probability of flexible loads A and B scenarios.

[0150] Table 1 Parameters of various DCR devices

[0151]

[0152] Table 2 Probability of wind power and load scenarios

[0153]

[0154] Step 2: Since the frequency modulation capability of VCU aggregation has time series characteristics, the specific spare capacity of each period is not a fixed value. Therefore, when calculating the cost curve, a step calculation is performed according to the percentage of spare capacity provided in each period to obtain a cost curve with both time series characteristics and different frequency modulation spare capacities, as shown in the figure below: Figure 6 shown. Figure 6 Four typical cost curve calculation results were selected for analysis, corresponding to t=8, t=12, t=17 and t=21, respectively. This set of curves was calculated under the assumption that the confidence level β=0.95 and the risk preference coefficient L=0.5, and that the VCU decision makers were risk neutral. At t=8, the range of frequency regulation reserve that VCU can provide is from 1.18MW upward to 3.43MW downward, and the upward frequency regulation cost has a significant turning point at 40%. This is because the cost function calculation takes into account the economic efficiency of providing frequency regulation services. When the reserve capacity provided is less than 40%, the DCR combination that provides frequency regulation services is the diesel generator B and energy storage system with lower frequency regulation costs; when the reserve capacity that VCU aggregation needs to provide is greater than 40%, the diesel generator B and energy storage system reach their operating boundaries, and at this time, flexible loads with higher frequency regulation reserve costs are required to participate in frequency regulation services. Therefore, the cost curve shows the characteristics of a concave function as the frequency regulation reserve capacity increases.

[0155] According to the VCU aggregation decision model proposed in the present invention, the cost curve obtained can intuitively reflect the cost of VCU providing different capacity aggregation services, and the subsequent market bidding research can be carried out based on the cost curve. At the same time, the cost curve also has the function of VCU decomposing the dispatching instructions to each DCR for execution according to different dispatching instructions of the upper power grid. Figure 6 The displayed curves and calculation results are calculated under the assumption that the VCU decision maker has a neutral risk level. When DER has randomness and volatility, the decision model proposed in the present invention can also screen DER combinations according to different risk preferences.

[0156] Step 3: Taking t = 8, when the downward frequency modulation service is 80% of the maximum capacity, as an example, the impact of the risk preference coefficient L in the CVaR model on the VCU aggregation frequency modulation cost is analyzed when the confidence level β = 0.95. The results are as follows: Figure 7As shown in the figure. Under the same frequency modulation cost calculation conditions, the effective frontier curve has a certain range. Since the risk and cost cannot be infinitely large or infinitely small in actual situations, the risk preference coefficient L should be taken within a certain range in the model. When L is lower than 0.3, the model has no effective solution; when L is higher than 0.8, the calculated frequency modulation cost is a certain extreme value. When L is large, VCU decision makers are conservative and tend to reduce the CVaR value to reduce the potential risk brought by the uncertainty of DCR frequency modulation capability. At this time, the frequency modulation cost of VCU is high. When L is small, VCU decision makers are radical and tend to increase the CVaR value, so that the frequency modulation cost of VCU is low. As L increases, the attitude of VCU decision makers becomes more conservative, which is reflected in the continuous increase of VCU frequency modulation cost and the continuous decrease of CVaR value. The effective frontier curve of VCU aggregate frequency modulation cost and CVaR value can realize the quantification of cost and risk. The VCU aggregation points that meet the model conditions are all on the effective frontier curve. VCU decision makers can determine the risk preference of decision-making based on psychological expectations and this curve.

[0157] Step 4: After the day-ahead market ends, the provincial dispatching agency will send the winning service to the VCU for execution. At this time, the VCU decision model plays a role in decomposing the frequency regulation service volume to each DCR for execution. Taking t = 8, when the downward frequency regulation service is 80% of the maximum capacity, as an example, under the same frequency regulation cost calculation conditions, the aggregation results of VCU under different risk preference coefficients L are analyzed, that is, the proportion of frequency regulation services provided by VCU among different DERs. The results are as follows: Figure 8 As shown. Generators A and B operate stably, but their operating and environmental costs during frequency regulation are high; wind power and flexible load B have low frequency regulation standby costs, but their frequency regulation capabilities have high randomness and volatility; flexible load A has high frequency regulation standby costs, but its frequency regulation capabilities are less volatile than those of wind power and flexible load B. When the risk preference coefficient L is low and VCU decision makers are in an aggressive state of risk, they tend to choose wind power and flexible load B, which have higher risks but lower frequency regulation standby costs; when the risk preference coefficient L is high and VCU decision makers are in a conservative state, they tend to choose generators with low risks but high standby costs. Due to its flexible adjustment capabilities and low standby costs, the energy storage system plays a stable role in the VCU aggregated frequency regulation capabilities.

[0158] The results of the example verify the effectiveness and superiority of the method of the present invention in solving the cost function risk characterization method of the external characteristics of the distributed control resource aggregation primary frequency modulation. The relevant research conclusions can provide a reference for the virtual control unit aggregation risk decision-making problem.

[0159] A risk characterization model of the primary frequency regulation cost function of a virtual control unit of the present invention is used to aggregate and form a virtual control unit for a large number of distributed control resources with primary frequency regulation capabilities on the local primary substation side to provide frequency regulation auxiliary services for the provincial power grid. The operating characteristics of various types of distributed control resources on the local primary substation side are analyzed to construct a dynamic aggregation model of the primary frequency regulation capabilities of distributed resources of the virtual control unit; based on the dynamic aggregation model of the primary frequency regulation capabilities of distributed resources, the lack or surplus fluctuation of the frequency regulation service volume of some distributed control resources is converted into the risk of aggregated cost loss, and a risk characterization model of the primary frequency regulation service volume of the virtual control unit is established based on the conditional risk value theory; on this basis, a cost function model of the primary frequency regulation service provided by the virtual control unit taking into account the risk is established, and the relationship between the aggregated primary frequency regulation service volume and the cost of the virtual control unit is characterized in the form of a time-divided piecewise linear curve cluster. The method proposed in the present invention can intuitively describe the aggregated frequency regulation cost function of distributed control resources under the risk of lack or surplus of the frequency regulation capabilities of some distributed control resources, and the dispatching instructions of the upper power grid to the virtual control unit can also be effectively decomposed to each distributed control resource for execution through the proposed method.

[0160] The above contents are further detailed descriptions of the present invention in combination with the preferred technical solutions, and the specific implementation of the invention cannot be limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, simple deductions and substitutions can be made without departing from the concept of the present invention, which should be regarded as the protection scope of the present invention.

Claims

1. A method for constructing a risk characterization model of a primary frequency modulation cost function of a virtual control unit, characterized in that: The construction method described is as follows: first, the operating characteristics of various types of distributed control resources on the primary substation side are analyzed, and primary frequency regulation operation models are established for distributed control resources with different operating characteristics; second, based on the primary frequency regulation operation model of distributed resources, distributed resources with strong power randomness are defined as uncontrollable distributed resources, and the conditional risk value method is used to quantify the aggregated cost loss risk caused by the fluctuation of the frequency regulation service volume provided by the uncontrollable distributed resources, and a virtual control unit primary frequency regulation service volume risk characterization model is established; finally, a cost function model for the primary frequency regulation service provided by a virtual control unit taking into account the risk is established, and the relationship between the aggregated primary frequency regulation service volume and the cost of the virtual control unit is characterized in the form of a time-divided piecewise linear curve cluster.

2. The method for constructing a risk characterization model of a virtual control unit primary frequency modulation cost function according to claim 1, characterized in that: The following steps are involved: S1: Analyze the operating characteristics of various distributed control resources on the primary substation grid side, and establish primary frequency regulation operation models for distributed control resources with different operating characteristics; the details are as follows: For the large number of distributed control resources with primary frequency regulation capability on the primary substation side, they are divided into three categories according to their operating characteristics: power supply, load and energy storage. Primary frequency regulation operation models are established for each of them as the basis for building subsequent models, including: distributed control resources with power supply characteristics, distributed control resources with load characteristics, and distributed control resources with energy storage characteristics. The three types of distributed resources with primary frequency regulation capability on the local primary substation side are aggregated into a virtual control unit, and an equivalent unit is used to replace each distributed resource under its jurisdiction. The equivalent regulation coefficient after equalization is: Where: R eq,t is the equivalent adjustment coefficient of the virtual control unit during period t; ΔP Σ,t P is the primary frequency modulation capacity of the virtual control unit in period t; ΣN,t is the rated power of each distributed resource aggregated by the virtual control unit during period t; f N is the rated value of the system frequency; Δf m is the maximum deviation of the system frequency; S2: Based on the distributed resource primary frequency regulation operation model obtained in S1, the distributed resources with strong power randomness are defined as uncontrollable distributed resources. The conditional risk value method is used to quantify the aggregate cost loss risk caused by the fluctuation of the frequency regulation service volume provided by uncontrollable distributed resources, and a virtual control unit primary frequency regulation service volume risk characterization model is established; specifically, it is known that the distribution method of the service volume provided by each distributed control resource under the primary frequency regulation service volume is: x T =(x1,...,x i ,...,x N ) (11) Where: x T It is a method for allocating the service volume provided by each distributed control resource under the condition that the service volume provided by the primary frequency modulation service is known; i The proportion of service volume provided by the i-th distributed control resource providing primary frequency modulation service; N is the number of all distributed control resources providing primary frequency modulation service; The distributed control resource combination frequency modulation cost vector is: y=(y1,...y i ,..y N ) (12) Where: y is the frequency modulation cost vector of the distributed control resource combination; y i The cost of providing frequency modulation service volume for the i-th distributed control resource providing primary frequency modulation service; N is the number of all distributed control resources providing primary frequency modulation service; The profit function of the distributed control resource combination is: R(x,y)=x T y (13) Where: R(x,y) is the revenue function of the distributed control resource combination under the known primary frequency regulation service volume; x T is the allocation method of the service volume provided by each distributed control resource under the known frequency modulation service volume; y is the frequency modulation cost vector of the distributed control resource combination; The loss function of the distributed control resource combination is: f(x,y)=-(x1y1+...+x i and i +...+x N and N )=-x T and (14) Where: f(x,y) is the loss function of the distributed control resource combination under the known primary frequency modulation service volume; x i The proportion of service volume provided by the distributed control resource that provides primary frequency modulation service for the i-th time; y i The cost of providing frequency modulation service volume for the i-th distributed control resource providing primary frequency modulation service; x T is a method of allocating the service volume provided by each distributed control resource under the condition that the primary frequency modulation service volume is known; y is the frequency modulation cost vector of the distributed control resource combination; N is the number of all distributed control resources providing primary frequency modulation services; Based on the conditional risk value theory, the risk characterization model of the primary frequency regulation service volume of the virtual control unit is established as follows: Where: α is the value of risk value, which means the maximum possible loss of the aggregated frequency modulation cost within a specific period of time under the confidence level β; F β is the value of conditional risk value, which means the average loss of aggregated frequency modulation under risk value; β is the confidence level; x T is the allocation method of the service volume provided by each distributed control resource under the known frequency modulation service volume; y is the frequency modulation cost vector of the distributed control resource combination; ρ(y) is the probability density function of the random variable y; Using the historical data of the random variable y, we obtain sample data through a clustering algorithm to give an estimate of the integral: Where: α is the value of risk value, which means the maximum possible loss of the aggregated frequency modulation cost within a specific period of time under the confidence level β; F β is the value of conditional risk value, which means the average loss of aggregated frequency modulation under risk value; β is the confidence level; x T is a method of allocating the service volume provided by each distributed control resource under the condition that the primary frequency modulation service volume is known; y is the frequency modulation cost vector of the distributed control resource combination; N is the number of all distributed control resources providing primary frequency modulation services; S3: For the distributed resources with strong uncertainty in primary frequency regulation capability of uncontrollable distributed control resources, the annual frequency regulation service volume data is used as the scenario set to reduce the scenarios to obtain the service volume and corresponding probability of each scenario. The goal is to aggregate the distributed resources to provide the most economical primary frequency regulation service. The conditional risk value is used to describe the cost loss risk brought by the uncontrollable distributed control resources for the aggregated frequency regulation cost, and the cost curve of different primary frequency regulation service volumes taking into account the risk is obtained; the details are as follows: First, for uncontrollable distributed resources, the full-year data is used as a scenario set to perform scenario reduction and obtain the power and response probability of each scenario. Secondly, based on the obtained scenarios and probabilities after the reduction of uncontrollable distributed control resources, and the primary frequency regulation operation models of various distributed control resources obtained in step S1, on the basis of the risk characterization model of the primary frequency regulation service volume of the virtual control unit obtained in step S2, several calculation points are selected within the aggregation boundary of the virtual control unit of the primary frequency regulation of the distributed control resources, and the cost loss risk brought by the uncertainty of the primary frequency regulation capacity of the uncontrollable distributed control resources to the aggregated frequency regulation cost is described by using the conditional risk value with the goal of optimizing the economic efficiency of distributed resource aggregation, and the virtual control unit cost when aggregating different spare capacities is calculated; the objective function is: Where: ψ is the set of all scheduling times; S is the set of all scenarios; π s is the probability corresponding to scene s; c(p i,t ) is the operating cost function of device i; c t,p is the electricity purchase price at time t; p t,tie is the actual power of the tie line at time t; c r (p i,t ) is the cost function for providing primary frequency regulation backup for device i; L is the risk coefficient; δ is the value of conditional risk value; The constraints include active power dynamic balance constraint, equivalent regulation coefficient constraint, and conditional risk value constraint. Other constraints are the same as those in equations (1), (3), (5)-(8).

3. The method for constructing a risk characterization model of a virtual control unit primary frequency modulation cost function according to claim 2, characterized in that: The three types of control resources in S1 are: The primary frequency modulation operation model of the distributed control resource with power supply characteristics is as follows: Where: P D,t is the output of distributed control resources in period t; α are the upper and lower limit coefficients of the unit's operating power; r D is the upper limit of the climbing rate of the unit; P Dr The primary frequency regulation reserve capacity of the unit; P D is the rated power of the unit; R D is the adjustment coefficient of the unit; f N is the rated value of the system frequency; Δf m is the maximum deviation of the system frequency; The primary frequency modulation operation model of the distributed control resource with load characteristics is as follows: Where: P A,t is the output of distributed control resources in period t, β are the upper and lower limit coefficients of the unit's operating power; P Ar The primary frequency regulation reserve capacity of the unit; P A is the rated power of the unit; R A is the adjustment coefficient of the unit; The primary frequency modulation operation model of the distributed control resource with energy storage characteristics is as follows: SOC min ≤SOC t ≤SOC max (7) Where: P c,ES,t , P d,ES,t are the charging power and discharging power of the distributed control resources in period t respectively; P c,ES,max , P d,ES,max They are the upper limits of the charging power and discharging power of the device respectively; SOC t SOC is the state of charge of the device during period t; max , SOC min They are the upper and lower limits of the state of charge of the device respectively; η c , η d E are the charging efficiency and discharging efficiency of the device respectively; t-1 is the capacity of the device at time t-1; P E is the rated capacity of the equipment; P Er The primary frequency regulation reserve capacity of the device; R E is the adjustment coefficient of the device.

4. The method for constructing a risk characterization model of a virtual control unit primary frequency modulation cost function according to claim 2, characterized in that: In S3, there is active dynamic balance constraint: Where: f(p i,t ) is the active power of device i in period t; S E is a collection of devices with energy storage characteristics; S A is a collection of equipment with load characteristics; ΔP u,t,PFR , ΔP d,t,PFR are the primary frequency regulation up / down capacity provided by the virtual control unit to the upper power grid at time t; c u 、c d They are the steps for reporting the up-frequency modulation capacity and down-frequency modulation capacity of the virtual aggregation unit, expressed as percentages; are the primary frequency regulation up / down capacities of power supply device i, energy storage device j, and load device k in period t; P es,j,t is the power of energy storage j in period t, with charging as the positive direction; P Ln,t is the non-adjustable load power during period t; P a,k,t is the power of load device k in period t.

5. The method for constructing a risk characterization model of a virtual control unit primary frequency modulation cost function according to claim 2, characterized in that: In S3, the equivalent adjustment coefficient constraint is: In steady state, the frequency change in the whole system is the same. Each distributed resource aggregated into a virtual control unit can be replaced by an equivalent unit. The equivalent adjustment coefficient after equalization is calculated in the same way as formula (10); The equivalent adjustment coefficient must satisfy the constraints: Where: R eq They are the upper and lower limits of the equivalent adjustment coefficient respectively.

6. The method for constructing a risk characterization model of a virtual control unit primary frequency modulation cost function according to claim 2, characterized in that: In S3, the conditional risk value constraint is: Conditional risk value is used to measure the risk of cost changes brought by uncontrollable distributed control resources to aggregate frequency modulation. For the defined discrete cost distribution, when the confidence level is β, the conditional risk value is approximately the expected cost corresponding to the set of scenarios with a probability of (1-α)×100% of higher cost; the relevant constraints of the conditional risk value are: 0≤L≤1 (22) x i ≥0 (24) z s ≥(−x T y − α) ( 25 z s ≥0 (26) Where: δ is the value of conditional risk value; α is the value of risk value, which is the boundary value of conditional risk value calculation; π s is the probability of scene s; β is the confidence level; z s is the dummy variable of the conditional risk value hypothesis; L is the risk coefficient, the larger its value is, the more risk-averse the virtual control unit is; x i The proportion of service volume provided by the distributed control resource that provides primary frequency modulation service for the i-th time; y i The cost of providing frequency modulation service volume for the i-th distributed control resource providing primary frequency modulation service; x T is an allocation method for the service volume provided by each distributed control resource under the condition that the amount of primary frequency modulation service provided is known; y is the frequency modulation cost vector of the distributed control resource combination; N is the number of all distributed control resources providing primary frequency modulation service.