Virtual power plant distributed energy storage optimal configuration method for multi-element power service

Abstract

The application discloses a kind of virtual power plant distributed energy storage optimization configuration methods for multi-element electric power service, constructs the calculation model of uncertainty variable and decision variable and obtains the optimization direction of maximum profit in preset time, and then constructs distributed robust optimization configuration model;Obtain the historical data sample set of second uncertainty variable, construct the fuzzy set of second uncertainty variable using Wasserstein metric, and obtain the min-sup shape objective function and data-driven uncertainty set boundary of reconstruction using dual theory, convert uncertainty error probability constraint into deterministic error probability constraint;Solve distributed robust optimization configuration model, obtain the number set of decision variable and configure electrochemical energy storage equipment.The application realizes that a large number of heterogeneous distributed resources participate in the flexible regulation and control of virtual power plant, improves operation efficiency and active support capability, and at the same time improves the income obtained by each aggregation object participating in the market and the competitiveness of virtual power plant participating in the power market.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of distributed energy storage planning, and particularly relates to a virtual power plant distributed energy storage optimization configuration method for multi-element power services. BACKGROUND

[0002] In response to the national call for "energy saving and emission reduction", achieving the goal of "carbon peak and carbon neutral", and early realization of the task of "green water and green mountains are golden mountains and silver mountains", distributed power generation technology has developed rapidly. Large-scale distributed power sources such as photovoltaic and wind turbines are connected to the distribution network, and non-fossil energy power generation has become the development trend of future power grids. However, distributed power generation is greatly affected by environmental factors, and its output has great randomness and volatility. After large-scale connection to the power grid, it will cause certain impact on power system stability, power quality, and operation planning.

[0003] Virtual power plant (VPP) as a distributed power coordination and management technology provides a new method to solve the above problems. VPP can aggregate distributed generators, battery energy storage systems and controllable loads as a controllable power supplier to participate in grid dispatching and power market transactions, which can better allocate load and direct consumption, improve renewable energy consumption, and improve power system operation stability. Among them, the power storage technology can make the VPP better participate in multi-element power market transactions through its flexible adjustment capability, which can effectively improve the flexibility and reliability of renewable energy consumption and grid operation control. However, the current energy storage system has problems such as high cost, low efficiency, and short service life, which limits its application in power systems. In order to fully play the role of the energy storage system in the VPP and improve the operation income of the VPP, it is necessary to optimize the configuration of the energy storage system in the VPP.

[0004] Defects and deficiencies of prior art:

[0005] Lack of effective energy management means for a large number of heterogeneous distributed resources: VPP contains a large number of distributed generators, distributed energy storage and controllable loads, which are small in capacity, large in number, widely distributed, and may belong to different industrial user interest subjects. In the process of energy management, the effects and interests of different subjects should be considered. How to coordinate a large number of heterogeneous distributed resources to participate in VPP regulation and control, improve operation efficiency and active support ability is a difficult problem.

[0006] The market mechanism for distributed resources to participate in diversified power services is not yet perfect: Since VPPs contain abundant photovoltaic, storage and charging resources, they can participate in diversified power markets such as demand response, peak regulation and frequency regulation to obtain superimposed benefits. However, multiple factors such as cost, risk and synergy will affect the economic benefits of distributed resources participating in the market. There is an unfair distribution of risks and benefits. The market mechanism for multiple entities to jointly participate in diversified power services needs further exploration.

[0007] The study does not involve the optimization configuration model of multiple types of distributed energy storage under the multi-market model: on the one hand, in response to the differentiated needs of different power markets, how to effectively take into account the differences in time scales and resource time domain response performance of different power markets and effectively aggregate and optimize distributed resources with multi-terminal randomness is a major challenge; on the other hand, the time-varying relationship between the energy storage demand characteristics and energy storage benefits of different power services, the coupling of energy storage at different times to assist VPP distributed resources in participating in multiple services, and the aggregation and reuse methods and benefit mechanisms of multiple types of distributed energy storage power and energy have not yet been clarified. Summary of the Invention

[0008] In order to solve the above-mentioned problems in the existing technology, the present invention provides a distributed energy storage optimization configuration method for virtual power plants for diversified power services, and proposes a distributed energy storage refined selection and capacity optimization configuration model for virtual power plants participating in diversified power services based on distributed blue bars, so as to flexibly coordinate a large number of heterogeneous distributed resources to participate in virtual power plants for regulation, improve operating efficiency and active support capabilities, and at the same time improve the benefits obtained by each aggregated object participating in the market and the competitiveness of virtual power plants in participating in the power market.

[0009] To achieve the above-mentioned object of the invention, the present invention provides a method for optimizing the configuration of distributed energy storage in a virtual power plant for diversified power services, comprising the following steps:

[0010] The first uncertainty variable is obtained based on the real-time electricity price, the frequency regulation service price and the load electricity sales price; the second uncertainty variable is obtained based on the output of the distributed wind and solar power generation units and the load of the controllable load;

[0011] Constructing a calculation model of uncertainty variables based on the first uncertainty variable and the second uncertainty variable;

[0012] Constructing decision variables; the virtual power plant includes the distributed wind and photovoltaic generators, the electrochemical energy storage device and the controllable load, and the decision variables are at least used to configure the electrochemical energy storage device;

[0013] Obtaining an optimization direction based on the calculation model of the uncertainty variable and the decision variable, wherein the optimization direction is to maximize the profit of the virtual power plant within a preset time;

[0014] Based on the optimization direction and the calculation model of the uncertainty variables, a distributed energy storage distributed blue-rod optimization configuration model of a virtual power plant for multiple power services is constructed, wherein the distributed blue-rod optimization configuration model includes a min-sup type objective function and constraints, wherein the constraints include deterministic constraints and uncertainty error probability constraints, the virtual power plant includes an electrochemical energy storage device, and the decision variables of the optimization configuration model are used at least to configure the electrochemical energy storage device;

[0015] Acquire a historical data sample set, where the data in the historical data sample set includes historical data of the second uncertainty variable;

[0016] Based on the historical data sample set, a fuzzy set of a second uncertainty variable is constructed using Wasserstein metric;

[0017] Based on the fuzzy set and duality theory of the second uncertainty variable, a reconstructed min-sup form objective function is obtained, and a data-driven uncertainty set boundary is obtained;

[0018] Based on the data-driven uncertainty set boundary, converting the uncertainty error probability constraint into a deterministic error probability constraint;

[0019] The distributionally robust optimization configuration model is solved based on the deterministic constraint, the deterministic error probability constraint and the reconstructed min-sup type objective function to obtain a set of decision variables.

[0020] Furthermore, the steps of constructing the calculation model of the uncertainty variable are:

[0021] Based on market operation information, the preset time is divided according to Formula 1, which is:

[0022]

[0023] Where, π is the total number of time periods divided within the preset time, T is the preset time, and T E is the time interval for market operators to release spot market information, κ is the total number of moments divided into each time period, T FR The time interval for the market operator to issue announcements on the demand for frequency regulation ancillary services;

[0024] Based on the formula 1, constructing a calculation model of the first uncertainty variable and a calculation model of the second uncertainty variable within the preset time respectively;

[0025] The calculation model of the first uncertainty variable includes:

[0026]

[0027]

[0028] The calculation model of the second uncertainty variable includes:

[0029]

[0030] Where, is the real-time electricity price at time t during period i, is the price of frequency modulation service at time t during period i, is the load electricity price at time t during period i, is the real-time electricity price forecast value at time t in period i, is the predicted price of frequency modulation service at time t in period i, is the predicted value of the load electricity price at time t during period i, is the real-time electricity price forecast deviation at time t in period i, is the forecast deviation of the frequency modulation service price at time t in period i, is the load electricity price forecast deviation at time t in period i, N(·,·) is a normal distribution, σ E is the standard deviation of the real-time electricity price forecast error, σ FR is the standard deviation of the frequency modulation service price forecast error, σ load is the standard deviation of the load electricity price forecast error, is the output of the x-th wind turbine at time t during period i, is the output of the yth photovoltaic array at time t during period i, is the load of the jth controllable load in period i and time t, is the output forecast value of the x-th wind turbine at time t in period i, is the output forecast value of the yth photovoltaic array at time t in period i, is the load forecast value of the jth controllable load in period i and time t, is the output prediction deviation of the x-th wind turbine at time t and time period i, is the output prediction deviation of the yth photovoltaic array at time t in period i, is the load forecast deviation of the jth controllable load in period i and time t.

[0031] Furthermore, the steps of constructing the fuzzy set of the second uncertainty variable using Wasserstein metric are as follows:

[0032] Obtain the support set of the second uncertainty variable, where the support set of the second uncertainty variable is:

[0033]

[0034] where Ξ is a support set of the second uncertainty variable, is a support matrix, ω is a second uncertainty variable vector, is a support vector;

[0035] constructing an empirical probability distribution model of the second uncertainty variable based on the historical data sample set, the empirical probability distribution model of the second uncertainty variable being:

[0036]

[0037] where is is an empirical probability distribution of the second uncertainty variable, M is a number of historical data samples in Ξ, l is a serial number of the lth historical data sample in Ξ, is a Dirac measure of the lth historical data sample in Ξ, is the lth historical data sample in Ξ, is the historical data sample set;

[0038] constructing a Wasserstein metric-based distance measurement model according to the empirical probability distribution model of the second uncertainty variable, the Wasserstein metric-based distance measurement model being:

[0039]

[0040] where, is a distance between and is a true probability distribution of the second uncertainty variable, inf represents a lower bound function, and Π is a joint probability distribution of the second uncertainty variable sample in Ξ and is an edge probability distribution, k is a serial number of the second uncertainty variable sample in Ξ, is a number of second uncertainty variable samples in Ξ, ω k is the kth second uncertainty variable sample in Ξ, Π kl is a probability value of Π taking ω k is a probability value of taking ω k ; s.t. represents a constraint.

[0041] ​​​​​​Obtain a confidence level, and obtain a Wasserstein sphere radius according to the confidence level; the Wasserstein sphere radius is:

[0042]

[0043] Where r(M,ε) is the radius of the Wasserstein sphere, ε is the confidence level, and D is the diameter of the support of the uncertain variable;

[0044] The fuzzy set of the second uncertainty variable is obtained according to the Wasserstein sphere radius and the distance measurement model based on the Wasserstein metric; the fuzzy set of the second uncertainty variable is:

[0045]

[0046] Where, is the fuzzy set of the second uncertainty variable.

[0047] Furthermore, the min-sup objective function is:

[0048]

[0049] Where min is the minimization function, and min is the outer function of the min-sup objective function, sup is the upper limit function, and sup is the inner function of the min-sup objective function, χ is the decision variable vector, ω is the second uncertainty variable vector, is the true probability distribution of the second uncertainty variable, is the fuzzy set of the second uncertainty variable, represents expectation, G is the function of the opposite number of returns;

[0050] The steps for obtaining the reconstructed min-sup objective function are:

[0051] The first process function is obtained by rewriting the inner function of the min-sup objective function based on the fuzzy set of the second uncertainty variable; the first process function is:

[0052]

[0053] The first process function is transformed based on the duality theory to obtain the dual function of the first process function; the dual function of the first process function is:

[0054]

[0055] Where λ, s l is the dual variable;

[0056] The function of the maximum benefit inverse number is rewritten as a piecewise linear convex function, which is:

[0057]

[0058] In the formula, J is a finite set, q is the sequence number of function segmentation, max is the maximum function, a q (χ) and b q (χ) are affine function coefficients.

[0059] An auxiliary vector and a dual norm of the auxiliary vector are obtained;

[0060] A second process function is obtained based on the auxiliary vector, the dual norm of the auxiliary vector and the piecewise linear convex function rewriting the first process function dual function; the second process function is:

[0061]

[0062] In the formula, is an auxiliary vector, is a dual norm of ;

[0063] A third process function is obtained based on the dual theory transformation of the constraints of the second process function; the third process function is:

[0064]

[0065] In the formula, r lq is a dual variable, and the superscript represents transposition;

[0066] The reconstructed min-sup type objective function is obtained according to the third process function and the outer function of the min-sup type objective function; the reconstructed min-sup type objective function is:

[0067]

[0068] Further, the optimization direction is:

[0069] maxS=S E +S FR +S load -C curt -C re -C ba -C loss -C om -C ex ;

[0070] In the formula, max represents maximization, S is benefit, S Eis the revenue from electricity sales to the virtual power plant grid, S FR The revenue of virtual power plants participating in the frequency regulation auxiliary service market, S load is the revenue from selling electricity to the internal load of the virtual power plant, C curt is the demand response cost, C re is the standby cost, C ba is the energy storage regulation cost, C loss is the life loss cost, C om is the operation and maintenance cost, C EX To compensate for the cost of incentives;

[0071] The revenue from electricity sales of the virtual power plant grid is:

[0072]

[0073] Where S E is the electricity sales revenue of the virtual power plant grid, π is the total number of time periods divided within the preset time, κ is the total number of moments divided within each time period, i is the time period number, t is the moment number, is the real-time electricity price at time t during period i, is the real-time electricity price in time period i and time 1~κ, is the amount of electricity sold by the virtual power plant to the spot market at time t during period i;

[0074] The benefits of the virtual power plant participating in the frequency regulation ancillary service market are:

[0075]

[0076] Where S FR To benefit from the participation of virtual power plants in the frequency regulation auxiliary service market, is the price of frequency modulation service at time t during period i, The frequency regulation capacity provided by the virtual power plant to the frequency regulation ancillary service market at time t during period i;

[0077] The revenue from electricity sales for the internal load of the virtual power plant is:

[0078]

[0079] Where S load is the revenue from electricity sales to the internal load of the virtual power plant, is the load electricity price at time t during period i, N load is the number of controllable loads aggregated in the virtual power plant, j is the sequence number of the controllable loads in the virtual power plant, is the load of the jth controllable load in period i and time t;

[0080] The demand response cost is:

[0081]

[0082] Where C curt is the demand response cost, is the interruption load compensation price of the jth controllable load in period i and time t, is the load interruption of the jth controllable load in period i and time t;

[0083] The standby cost is:

[0084]

[0085] Where C re is the standby cost, N Ti is the number of lithium titanate batteries aggregated in the virtual power plant, N Fe is the number of lithium iron phosphate batteries aggregated in the virtual power plant, m is the serial number of the lithium titanate battery in the virtual power plant, and n is the serial number of the lithium iron phosphate battery in the virtual power plant. is the upper reserve capacity price of the mth lithium titanate battery, is the lower reserve capacity price of the mth lithium titanate battery, is the upper reserve capacity price of the nth lithium iron phosphate battery, is the lower reserve capacity price of the nth lithium iron phosphate battery; is the maximum reserve capacity of the mth lithium titanate battery at time t in period i, r i,t,m is the reserve capacity of the mth lithium titanate battery at time t in period i; is the maximum reserve capacity of the nth lithium iron phosphate battery at time t in period i, r i,t,n is the reserve capacity of the nth lithium iron phosphate battery at time t during period i;

[0086] The energy storage regulation cost is:

[0087]

[0088] Where C ba Adjusting costs for energy storage, is the participation factor of the mth lithium titanate battery at time t in period i, is the participation factor of the nth lithium iron phosphate battery at time t in period i, is the unit adjustment cost of the mth lithium titanate battery, is the unit adjustment cost of the nth lithium iron phosphate battery; i,t is the total uncertainty of source and load within the virtual power plant at time i and time t; N W is the number of wind turbines aggregated in the virtual power plant, NPV is the number of photovoltaic arrays aggregated in the virtual power plant, x is the serial number of the wind turbine in the virtual power plant, y is the serial number of the photovoltaic array in the virtual power plant, is the output prediction deviation of the x-th wind turbine at time t and time period i, is the output prediction deviation of the yth photovoltaic array at time t in period i, is the load forecast deviation of the jth controllable load at time t in period i;

[0089] The life loss cost is:

[0090]

[0091] Where C loss is the life loss cost, is the unit charge and discharge power life loss cost coefficient of the mth lithium titanate battery, is the unit charge and discharge power life loss cost coefficient of the nth lithium iron phosphate battery; is the charging efficiency of the mth lithium titanate battery, is the discharge efficiency of the mth lithium titanate battery, is the charging efficiency of the nth lithium iron phosphate battery, Discharge efficiency of the nth lithium iron phosphate battery; is the charging power of the mth lithium titanate battery at time t during period i, is the discharge power of the mth lithium titanate battery at time t in period i, is the charging power of the nth lithium iron phosphate battery at time t during period i, is the discharge power of the nth lithium iron phosphate battery at time t during period i;

[0092] The operation and maintenance costs are:

[0093]

[0094] Where C om For operation and maintenance costs, is the operation and maintenance cost coefficient of the mth lithium titanate battery, is the operation and maintenance cost coefficient of the nth lithium iron phosphate battery; is the unit capacity investment cost of the mth lithium titanate battery, is the unit capacity investment cost of the nth lithium iron phosphate battery; is the rated capacity of the mth lithium titanate battery, is the rated capacity of the nth lithium iron phosphate battery; is the discount rate of the mth lithium titanate battery, is the discount rate of the nth lithium iron phosphate battery; is the service life of the mth lithium titanate battery, is the service life of the nth lithium iron phosphate battery;

[0095] The incentive compensation cost is:

[0096]

[0097] Where C EX To motivate compensation costs, is the incentive compensation price of the mth lithium titanate battery at time t in period i, is the incentive compensation price of the nth lithium iron phosphate battery at time t in period i, is the rated power of the mth lithium titanate battery, is the rated power of the nth lithium iron phosphate battery, The basic incentive compensation price.

[0098] Furthermore, the deterministic constraints include power balance constraints, affine constraints, demand response constraints, energy storage equipment personalized operation constraints, market access constraints, and energy storage aggregation capacity constraints;

[0099] The power balance constraint is:

[0100]

[0101] The affine constraints are:

[0102]

[0103] The demand response constraints are:

[0104]

[0105] Where, is the load interruption of the jth controllable load in period i and time (t+1), is the load of the jth controllable load in period i and time (t+1), η max is the maximum call rate of controllable load at a single moment, It is the maximum value of the sum of the continuous call rates of the controllable loads at two adjacent moments;

[0106] The personalized operation constraints of the energy storage device are:

[0107]

[0108]

[0109] 2≤Δ Ti <Δ Fe <κ;

[0110] Where, is the state of charge of the mth lithium titanate battery at time t in period i, is the state of charge of the nth lithium iron phosphate battery at time t during period i; is the self-discharge rate of the mth lithium titanate battery, is the self-discharge rate of the nth lithium iron phosphate battery; is the lower limit of the state of charge of the mth lithium titanate battery, The lower limit of the state of charge of the nth lithium iron phosphate battery; is the upper limit of the state of charge of the mth lithium titanate battery, The upper limit of the state of charge of the nth lithium iron phosphate battery; is the Boolean variable of the mth lithium titanate battery at time i and time t, 1 means that the mth lithium titanate battery is in the charging state at time t during period i. 0 means that the mth lithium titanate battery is in the discharge state at time t during period i. is the Boolean variable of the nth lithium iron phosphate battery at time i and time t, 1 means the nth lithium iron phosphate battery is in the charging state at time t during period i. 0 means that the nth lithium iron phosphate battery is in the discharge state at time t during period i, Δ Ti is the maximum continuous high power output time of the power type energy storage device, Δ Fe The maximum continuous high power output time of the energy storage device; The continuous Δ Ti The maximum value of the sum of load factors at any moment; The continuous Δ for the nth lithium iron phosphate battery Fe The minimum sum of load factors at all times;

[0111] The market access constraints are:

[0112]

[0113] Where, Adjust the capacity threshold upward for the aggregate, is the aggregate charging power threshold; Adjust the capacity threshold downward for the aggregation, θ is the aggregate discharge power threshold;

[0114] The energy storage aggregation capacity constraint is:

[0115]

[0116] Where, φTi Configure the capacity limit for a single lithium titanate battery, φ Fe Configure the capacity upper limit for a single lithium iron phosphate battery; φ con It is the upper limit of the total aggregate capacity of energy storage equipment in the virtual power plant.

[0117] Furthermore, the uncertainty error probability constraint is:

[0118]

[0119] In the formula, inf represents the lower bound function, is the true probability distribution of the second uncertainty variable; β is the probability of violating the uncertainty error probability constraint.

[0120] Furthermore, the step of converting the uncertainty error probability constraint into the deterministic error probability constraint is:

[0121] The uncertainty error probability constraint is rewritten based on Formula 2 to obtain a general mathematical expression for the uncertainty error probability constraint. Formula 2 is:

[0122]

[0123] Where g(χ,ω) is the function expression of the event in {·};

[0124] The historical data sample set is processed according to the normalization equation to obtain the random variable sample set; the normalization equation is:

[0125]

[0126] Where, is the lth random variable sample, for The lth historical data sample in , Σ is the variance of the historical data sample set, is the mean of the historical data sample set;

[0127] Based on the random variable sample set, a random variable uncertainty set is constructed, and the random variable uncertainty set is:

[0128]

[0129] Where, is an uncertain set of random variables with τ as the boundary, σ is a random variable, R is a real number, and τ is the boundary of the uncertain set of random variables;

[0130] Based on the fuzzy set of the second uncertainty variable and the random variable uncertainty set, a random variable uncertainty set boundary solution model is constructed. The random variable uncertainty set boundary solution model is:

[0131]

[0132] Based on the duality theory, the random variable uncertainty set boundary solution model is transformed to obtain a reconstructed random variable uncertainty set boundary solution model, and the reconstructed random variable uncertainty set boundary solution model is:

[0133]

[0134] The data-driven uncertainty set boundary solution model is solved according to the reconstructed random variable uncertainty set boundary solution model and Formula 3; Formula 3 is:

[0135]

[0136] Where ψ is the boundary of the data-driven uncertainty set;

[0137] The deterministic error probability constraint is obtained according to the data-driven uncertainty set boundary solution model and the uncertainty error probability constraint. The deterministic error probability constraint is:

[0138] g(χ,ψ)≤0.

[0139] Furthermore, the electrochemical energy storage device includes a power type energy storage device and an energy type energy storage device, and the power type energy storage device includes N Ti The energy storage device includes a lithium titanate battery. Fe Lithium iron phosphate battery, N Ti is the number of lithium titanate batteries aggregated in the virtual power plant, N Fe is the number of lithium iron phosphate batteries aggregated in the virtual power plant;

[0140] The decision variables include the real-time upper spare capacity of the lithium titanate battery, the real-time lower spare capacity of the lithium titanate battery, the real-time upper spare capacity of the lithium iron phosphate battery, the real-time lower spare capacity of the lithium iron phosphate battery, the real-time participation factor of the lithium titanate battery, the real-time participation factor of the lithium iron phosphate battery, the real-time charging power of the lithium titanate battery, the real-time discharging power of the lithium titanate battery, the real-time charging power of the lithium iron phosphate battery, the real-time discharging power of the lithium iron phosphate battery, the rated capacity of the lithium titanate battery, the rated capacity of the lithium iron phosphate battery, the rated power of the lithium titanate battery, the rated power of the lithium iron phosphate battery, the real-time state of charge of the lithium titanate battery, the real-time state of charge of the lithium iron phosphate battery, the real-time Boolean variable of the lithium titanate battery, and the real-time Boolean variable of the lithium iron phosphate battery.

[0141] Furthermore, the decision variables also include the real-time electricity sold by the virtual power plant to the spot market, the real-time frequency regulation capacity provided by the virtual power plant to the frequency regulation auxiliary service market, the real-time interruption load compensation price of the controllable load, the real-time load interruption amount of the controllable load, the real-time incentive compensation price of the lithium titanate battery, and the real-time incentive compensation price of the lithium iron phosphate battery.

[0142] The beneficial effects of the present invention are:

[0143] 1. The present invention establishes a virtual power plant bidding framework and its coordinated operation mechanism under a diversified market environment, and clarifies the joint operation mode between distributed wind and solar power generation units, multiple types of electrochemical energy storage equipment and controllable loads within the virtual power plant.

[0144] 2. Aiming at the energy and power requirements of different types of power services, the present invention proposes an aggregation and multiplexing method and incentive mechanism for distributed power-type energy storage devices and energy-type energy storage devices, and constructs personalized operation models and cost models for multiple types of distributed electrochemical energy storage devices.

[0145] 3. The distributed energy storage distributed blue-rod optimization configuration model for virtual power plants for diversified power services constructed by the present invention can carry out refined selection and capacity optimization configuration of distributed electrochemical energy storage equipment for the virtual power plant system participating in diversified power services, realize the flexible regulation of a large number of heterogeneous distributed resources participating in the virtual power plant, and at the same time improve the benefits obtained by each aggregated object participating in the market and the competitiveness of the virtual power plant in participating in the power market. BRIEF DESCRIPTION OF THE DRAWINGS

[0146] Figure 1 It is a schematic diagram of the process of the present invention;

[0147] Figure 2 It is a schematic diagram of the principle of the present invention. DETAILED DESCRIPTION

[0148] In order to clearly illustrate the technical features of this solution, this solution is described below through specific implementation methods.

[0149] See also Figure 1 、 2 The embodiment of the present invention provides a method for optimizing distributed energy storage configuration of a virtual power plant for multiple power services, including the following steps:

[0150] The first uncertainty variable is obtained based on the real-time electricity price, the frequency regulation service price and the load electricity sales price; the second uncertainty variable is obtained based on the output of the distributed wind and solar power generation units and the load of the controllable load;

[0151] A computational model for the uncertainty variables is constructed based on the first uncertainty variable and the second uncertainty variable. The virtual power plant participates in multiple electricity markets, namely the spot market and the frequency regulation auxiliary service market. During market transactions, the virtual power plant coordination and control center first predicts the real-time electricity price, frequency regulation service price, load electricity sales price, renewable energy output (distributed wind and solar power generation units include distributed wind turbines and distributed photovoltaic arrays, and renewable energy output includes distributed wind turbine output and distributed photovoltaic array output), and controllable load based on historical information, thus constructing a computational model for the uncertainty variables.

[0152] Constructing decision variables; the virtual power plant includes distributed wind and solar power generation units, electrochemical energy storage devices, and controllable loads. The decision variables are at least used to configure the electrochemical energy storage devices;

[0153] Assume that the time intervals between the market operation agency’s announcement of spot market information and frequency regulation auxiliary service demand are T E (usually 15 minutes) and T FR (usually 5 minutes), the preset time T can be divided into π = T / T assuming that it is 24 hours a day E time periods, each of which can be divided into κ=T E / T FR Based on market operation information, the preset time is divided according to Formula 1, which is:

[0154]

[0155] Based on Formula 1, a calculation model of a first uncertainty variable and a calculation model of a second uncertainty variable within a preset time are respectively constructed;

[0156] The first uncertainty variable includes the real-time electricity price forecast deviation, the frequency regulation service price forecast deviation, and the load electricity sales price forecast deviation. The real-time electricity price, frequency regulation service price, and load electricity sales price forecast accuracy are high, so the corresponding forecast deviations of the three, namely the first uncertainty variable, can be accurately grasped, that is, the first uncertainty variable obeys the normal distribution; the calculation model of the first uncertainty variable includes:

[0157]

[0158] The second uncertainty variable includes the prediction deviation of distributed wind turbine output, distributed photovoltaic array output, and controllable load. The load prediction accuracy of distributed wind turbine output, distributed photovoltaic array output, and controllable load is low. Therefore, the corresponding prediction deviations of the three, namely the second uncertainty variable, cannot be accurately grasped. In other words, it is difficult to obtain an accurate probability distribution of the second uncertainty variable. The calculation model of the second uncertainty variable includes:

[0159]

[0160] Based on the calculation model and decision variables of uncertain variables, the optimization direction is obtained, and the optimization direction is to maximize the benefits of the virtual power plant within a preset time;

[0161] Based on the calculation model of optimization direction and uncertainty variables, a distributed energy storage distributed blue stick optimization configuration model for virtual power plants oriented to multiple power services is constructed. The optimization direction is:

[0162] maxS=S E +S FR +S load -C curt -C re -C ba -C loss -C om -C ex ;

[0163] Among them, the revenue from electricity sales of the virtual power plant grid is:

[0164]

[0165] The benefits of virtual power plants participating in the frequency regulation ancillary service market are:

[0166]

[0167] The revenue from electricity sales of internal loads in the virtual power plant is:

[0168]

[0169] The demand response cost is:

[0170]

[0171] In order not to affect the load comfort, the interruption load compensation price is associated with the load interruption amount. That is, the greater the interrupted load amount, the greater the compensation given by the virtual power plant. Its form is

[0172]

[0173] Electrochemical energy storage devices include power type energy storage devices and energy type energy storage devices. Power type energy storage devices include N Ti Lithium titanate battery, energy storage devices including N Fe Lithium iron phosphate battery, N Ti is the number of lithium titanate batteries aggregated in the virtual power plant, N Fe is the number of lithium iron phosphate batteries aggregated in the virtual power plant; all electrochemical energy storage devices have backup capacity to balance the uncertainty of source and load within the virtual power plant; the backup cost is:

[0174]

[0175] The energy storage regulation cost is the cost incurred by the electrochemical energy storage equipment aggregated in the virtual power plant to balance the uncertainty of source and load:

[0176]

[0177]

[0178] Frequent charging and discharging will cause the service life of electrochemical energy storage equipment to decrease, so this phenomenon is penalized. The life loss cost is:

[0179]

[0180] The virtual power plant needs to bear the operation and maintenance costs of the electrochemical energy storage equipment during the period of calling the electrochemical energy storage equipment. The operation and maintenance costs are:

[0181]

[0182] Each time an electrochemical energy storage device performs a charge or discharge behavior, it can receive incentive compensation from the virtual power plant. The cost of the incentive compensation is:

[0183]

[0184] To encourage energy storage operators to actively participate in the operation and regulation of virtual power plants, the incentive compensation price is linked to the charge and discharge power of the electrochemical energy storage equipment. That is, the greater the charge and discharge power of the electrochemical energy storage equipment, the greater the incentive compensation provided by the virtual power plant. The form is:

[0185]

[0186] The distributed robust optimization configuration model includes a min-sup objective function and constraints. The core of the distributed robust optimization configuration model is to construct a min-sup objective function. The min-sup objective function is:

[0187]

[0188] Where G is the function of the inverse of the return, that is, G(χ,ω) = -S(χ,ω);

[0189] Constraints include deterministic constraints and uncertain error probability constraints. Deterministic constraints include power balance constraints, affine constraints, demand response constraints, personalized operation constraints of energy storage equipment, market access constraints, and energy storage aggregation capacity constraints.

[0190] The power balance constraint is:

[0191]

[0192] At any time and time, the sum of the participation factors of all electrochemical energy storage devices in the virtual power plant that participate in balancing the source-load uncertainty is 1, that is, all electrochemical energy storage devices in the virtual power plant are responsible for balancing the source-load uncertainty. The affine constraint is:

[0193]

[0194] In order to reduce the impact on users within the virtual power plant, the continuous call time of the interruptible load is limited to prevent the load from maintaining a high call rate. The demand response constraint is:

[0195]

[0196] The personalized operation constraints of energy storage equipment constrain the state of charge and output of electrochemical energy storage equipment. The personalized operation constraints of energy storage equipment are:

[0197]

[0198] The ratio of the rated power of an electrochemical energy storage device to its rated capacity is usually 0.5:

[0199]

[0200] At the same time, electrochemical energy storage devices cannot be in the charging and discharging states at the same time:

[0201]

[0202]

[0203] Among them, power-type energy storage devices focus on providing high power output in a short period of time and are suitable for scenarios that require high-power instantaneous output; while energy-type energy storage devices focus on continuously outputting stable electricity for a long time and are suitable for applications that require long working time and sustained power supply. Therefore, in order to reflect the differences in power supply capacity between power-type energy storage devices and energy-type energy storage devices, the continuous high power output time and the sum of the load rate during the continuous output time of power-type energy storage devices and energy-type energy storage devices are distinguished. The form is:

[0204]

[0205] 2≤Δ Ti <Δ Fe <κ;

[0206] Market access constraints require virtual power plants to meet minimum thresholds for aggregated regulation capacity and aggregated charge and discharge power in order to participate in electricity market transactions:

[0207]

[0208] Limited by the land area and the investment amount, the aggregated energy storage capacity in the virtual power plant needs to meet the single capacity constraint and the total capacity constraint, and the energy storage aggregated capacity constraint is:

[0209]

[0210] In any period or time, all electrochemical energy storage devices in the virtual power plant can have sufficient standby capacity to balance the source and load uncertainty under a certain confidence level; the uncertainty error probability constraint is:

[0211]

[0212]

[0213] Since the min-sup-shaped objective function in the distributed robust optimization configuration model and the uncertainty error probability constraint have poor computational complexity and high computational complexity, the min-sup-shaped objective function and the uncertainty error probability constraint need to be transformed to improve the computational complexity and reduce the computational complexity.

[0214] Obtain a historical data sample set, and data in the historical data sample set includes historical data of a second uncertainty variable; if the expectation value of the function G of the reciprocal of the revenue is to be calculated, the probability distribution of the second uncertainty variable is necessary, however, in actual application, the true probability distribution of the second uncertainty variable is unknown, and only some historical data samples that can be obtained are used to construct the historical data sample set

[0215] Based on the historical data sample set, a fuzzy set of the second uncertainty variable is constructed by using the Wasserstein metric; in some preferred embodiments, the step of constructing the fuzzy set of the second uncertainty variable by using the Wasserstein metric is:

[0216] Obtain a support set of the second uncertainty variable, the support set of the second uncertainty variable internally contains second uncertainty variable samples, and the support set of the second uncertainty variable is:

[0217]

[0218] Based on the historical data sample set, an empirical probability distribution model of the second uncertainty variable is constructed, and the empirical probability distribution of the second uncertainty variable can be regarded as constructing the historical data sample set by extracting M historical data samples from the support set Ξ of the second uncertainty variable and the probability of each historical data sample under the empirical probability distribution of the second uncertainty variable is 1 / M, and the The probability of the sample is 0, that is, the empirical probability distribution of the second uncertainty variable is uniformly distributed on the historical data sample set. The empirical probability distribution model of the second uncertainty variable is:

[0219]

[0220] According to the empirical probability distribution model of the second uncertainty variable, a distance measurement model based on Wasserstein metric is constructed. The distance measurement model based on Wasserstein metric is:

[0221]

[0222] Can be seen as The Wasserstein sphere is centered and has a radius of r(M,ε). If the radius of the Wasserstein sphere is too large, Too many distributions are included, which leads to overly pessimistic optimization results; the Wasserstein sphere radius is too small, which may cause Not included in internal.

[0223] Therefore, the Wasserstein sphere is required to convert Included in Internally, obtain the confidence level, and obtain the Wasserstein sphere radius based on the confidence level; the Wasserstein sphere radius is:

[0224]

[0225] The fuzzy set of the second uncertainty variable is obtained according to the Wasserstein sphere radius and the distance measurement model based on the Wasserstein metric; the fuzzy set of the second uncertainty variable is:

[0226]

[0227] Based on the fuzzy set and duality theory of the second uncertainty variable, a reconstructed min-sup objective function is obtained; in some preferred embodiments, the steps of obtaining the reconstructed min-sup objective function are:

[0228] The inner function of the min-sup objective function is rewritten based on the fuzzy set of the second uncertainty variable as follows:

[0229]

[0230] Simplify the above formula to obtain the first process function; the first process function is:

[0231]

[0232] The first process function contains (M+1) constraint conditions, if tends to infinity, the decision variable tends to infinity, and a dual function of the first process function is obtained based on a dual theory transformation of the first process function; the dual function of the first process function is:

[0233]

[0234] The function of the maximum income reciprocal is rewritten as a piecewise linear convex function, in the application, the function of the income reciprocal, that is, G(x, omega) is a linear function about the second uncertain variable omega, and therefore can be rewritten as a piecewise linear convex function, and the piecewise linear convex function is:

[0235]

[0236] An auxiliary vector and a dual norm of the auxiliary vector are obtained;

[0237] A second process function is obtained based on the auxiliary vector, the dual norm of the auxiliary vector and the piecewise linear convex function rewriting the dual function of the first process function; the second process function is:

[0238]

[0239] A third process function is obtained based on a dual theory transformation of the constraint of the second process function; the third process function is:

[0240]

[0241] A reconstructed min-sup-shaped target function is obtained according to the third process function and an outer function of the min-sup-shaped target function; the reconstructed min-sup-shaped target function is:

[0242]

[0243] The reconstructed min-sup-shaped target function is a single-layer min-shaped function, the transformation of the min-sup-shaped target function in the distributed robust optimization configuration model is completed, and the improvement of the computability and the reduction of the calculation complexity are realized.

[0244] The uncertainty error probability constraint in the constraint condition cannot be directly calculated, and a data-driven uncertainty set boundary is obtained based on a fuzzy set of the second uncertain variable and a dual theory;

[0245] Preferably, without losing generality, the uncertainty error probability constraint is rewritten based on formula two to obtain a general mathematical expression of the uncertainty error probability constraint, and the formula two is:

[0246]

[0247] The random variable sample set is obtained by processing the historical data sample set according to the standardized equation; the standardized equation is:

[0248]

[0249] Based on the random variable sample set, the random variable uncertainty set is constructed. The random variable uncertainty set is:

[0250]

[0251] Based on the fuzzy set of the second uncertainty variable and the uncertain set of random variables, a model for solving the boundary of the uncertain set of random variables is constructed. The model for solving the boundary of the uncertain set of random variables is:

[0252]

[0253] Based on the duality theory, the random variable uncertainty set boundary solution model is transformed to obtain the reconstructed random variable uncertainty set boundary solution model. The reconstructed random variable uncertainty set boundary solution model is:

[0254]

[0255] The data-driven uncertainty set boundary solution model is solved based on the reconstructed random variable uncertainty set boundary solution model and formula 3; formula 3 is:

[0256]

[0257] Based on the data-driven uncertainty set boundary, the uncertainty error probability constraint is transformed into a deterministic error probability constraint; the deterministic error probability constraint is:

[0258] g(χ,ψ)≤0.

[0259] Converting the uncertainty error probability constraint into a deterministic error probability constraint can further improve the computability of the distributed robust optimization configuration model and reduce its computational complexity, thereby accelerating the realization of refined selection and capacity optimization configuration of distributed energy storage equipment in virtual power plants.

[0260] Based on deterministic constraints, deterministic error probability constraints and a reconstructed min-sup objective function, the distributionally robust optimization model is solved to obtain a set of decision variables.

[0261] The decision variables include the real-time upper reserve capacity of the lithium titanate battery, the real-time lower reserve capacity of the lithium titanate battery, the real-time upper reserve capacity of the lithium iron phosphate battery, the real-time lower reserve capacity of the lithium iron phosphate battery, the real-time participation factor of the lithium titanate battery, the real-time participation factor of the lithium iron phosphate battery, the real-time charging power of the lithium titanate battery, the real-time discharging power of the lithium titanate battery, the real-time charging power of the lithium iron phosphate battery, the real-time discharging power of the lithium iron phosphate battery, the rated capacity of the lithium titanate battery, the rated capacity of the lithium iron phosphate battery, the rated power of the lithium titanate battery, the rated power of the lithium iron phosphate battery, the real-time state of charge of the lithium titanate battery, the real-time state of charge of the lithium iron phosphate battery, the real-time Boolean variable of the lithium titanate battery, and the real-time Boolean variable of the lithium iron phosphate battery.

[0262] Decision variables are also used to configure controllable loads, which includes optimizing the load interruption amount of controllable loads. Decision variables also include the real-time electricity sold by virtual power plants to the spot market, the real-time frequency regulation capacity provided by virtual power plants to the frequency regulation auxiliary service market, the real-time interruption load compensation price of controllable loads, the real-time load interruption amount of controllable loads, the real-time incentive compensation price of lithium titanate batteries, and the real-time incentive compensation price of lithium iron phosphate batteries.

[0263] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0264] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0265] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A distributed energy storage optimization configuration method for virtual power plants oriented to diversified power services, characterized in that: The following steps are involved: The first uncertainty variable is obtained based on the real-time electricity price, the frequency regulation service price and the load electricity sales price; the second uncertainty variable is obtained based on the output of the distributed wind and solar power generation units and the load of the controllable load; Constructing a calculation model of uncertainty variables based on the first uncertainty variable and the second uncertainty variable; Constructing decision variables; the virtual power plant includes the distributed wind and photovoltaic generators, electrochemical energy storage equipment and the controllable load; The decision variables are at least used to configure the electrochemical energy storage device; Obtaining an optimization direction based on the calculation model of the uncertainty variable and the decision variable, wherein the optimization direction is to maximize the profit of the virtual power plant within a preset time; Based on the optimization direction and the calculation model of the uncertainty variable, a distributed energy storage distributed robust optimization configuration model of a virtual power plant for multiple power services is constructed, wherein the distributed robust optimization configuration model includes a min-sup type objective function and constraints, and the constraints include deterministic constraints and uncertainty error probability constraints; Acquire a historical data sample set, where the data in the historical data sample set includes historical data of the second uncertainty variable; Based on the historical data sample set, a fuzzy set of a second uncertainty variable is constructed using Wasserstein metric; Based on the fuzzy set and duality theory of the second uncertainty variable, a reconstructed min-sup form objective function is obtained, and a data-driven uncertainty set boundary is obtained; Based on the data-driven uncertainty set boundary, converting the uncertainty error probability constraint into a deterministic error probability constraint; Solving the distributional robust optimization model based on the deterministic constraint, the deterministic error probability constraint, and the reconstructed min-sup objective function to obtain a set of decision variables; The steps of constructing the calculation model of the uncertainty variable are: Based on market operation information, the preset time is divided according to Formula 1, which is: Where, π is the total number of time periods divided within the preset time, T is the preset time, and T E is the time interval for market operators to release spot market information, κ is the total number of moments divided into each time period, T FR The time interval for the market operator to issue announcements on the demand for frequency regulation ancillary services; Based on the formula 1, constructing a calculation model of the first uncertainty variable and a calculation model of the second uncertainty variable within the preset time respectively; The calculation model of the first uncertainty variable includes: The calculation model of the second uncertainty variable includes: Where, is the real-time electricity price at time t during period i, is the price of frequency modulation service at time t during period i, is the load electricity price at time t during period i, is the real-time electricity price forecast value at time t in period i, is the predicted price of frequency modulation service at time t in period i, is the predicted value of the load electricity price at time t in period i, is the real-time electricity price forecast deviation at time t in period i, is the forecast deviation of the frequency modulation service price at time t in period i, is the load electricity price forecast deviation at time t in period i, N(·,·) is a normal distribution, σ E is the standard deviation of the real-time electricity price forecast error, σ FR is the standard deviation of the frequency modulation service price forecast error, σ load is the standard deviation of the load electricity price forecast error, is the output of the x-th wind turbine at time t during period i, is the output of the yth photovoltaic array at time t during period i, is the load of the jth controllable load in period i and time t, is the output forecast value of the x-th wind turbine at time t in period i, is the output forecast value of the yth photovoltaic array at time t in period i, is the load forecast value of the jth controllable load in period i and time t, is the output prediction deviation of the x-th wind turbine at time t and time period i, is the output prediction deviation of the yth photovoltaic array at time t in period i, is the load forecast deviation of the jth controllable load at time t in period i; The steps of constructing the fuzzy set of the second uncertainty variable using Wasserstein metric are as follows: Obtain the support set of the second uncertainty variable, where the support set of the second uncertainty variable is: Where Ξ is the support set of the second uncertainty variable, is the support matrix, ω is the second uncertainty variable vector, is the support vector; An empirical probability distribution model of a second uncertainty variable is constructed based on the historical data sample set. The empirical probability distribution model of the second uncertainty variable is: Where, is the empirical probability distribution of the second uncertainty variable, M is The number of historical data samples in , l is The serial number of the historical data sample in , for The Dirac measure of the lth historical data sample in, for The lth historical data sample in is a sample set of historical data; A distance measurement model based on Wasserstein metric is constructed according to the empirical probability distribution model of the second uncertainty variable. The distance measurement model based on Wasserstein metric is: Where, for and distance, is the true probability distribution of the second uncertainty variable, inf represents the lower bound function, Π is the probability distribution of the second uncertainty variable sample in Ξ and The joint probability distribution of historical data samples in , and Π is and is the marginal probability distribution, k is the number of the second uncertainty variable sample in Ξ, is the number of samples of the second uncertainty variable in Ξ, ω k is the kth second uncertainty variable sample in Ξ, ∏ kl Take ω for π k and The probability value when for Take ω k The probability value when ; st represents the constraint; Obtain a confidence level, and obtain a Wasserstein sphere radius according to the confidence level; the Wasserstein sphere radius is: Where r(M,ε) is the radius of the Wasserstein sphere, ε is the confidence level, and D is the diameter of the support of the uncertain variable; The fuzzy set of the second uncertainty variable is obtained according to the Wasserstein sphere radius and the distance measurement model based on the Wasserstein metric; the fuzzy set of the second uncertainty variable is: Where, is the fuzzy set of the second uncertainty variable.

2. The distributed energy storage optimization configuration method for virtual power plants oriented to diversified power services according to claim 1 is characterized in that: The min-sup objective function is: Where min is the minimization function, and min is the outer function of the min-sup objective function, sup is the upper limit function, and sup is the inner function of the min-sup objective function, χ is the decision variable vector, ω is the second uncertainty variable vector, is the true probability distribution of the second uncertainty variable, is the fuzzy set of the second uncertainty variable, represents expectation, G is the function of the opposite number of returns; The steps for obtaining the reconstructed min-sup objective function are: The first process function is obtained by rewriting the inner function of the min-sup objective function based on the fuzzy set of the second uncertainty variable; the first process function is: The first process function is transformed based on the duality theory to obtain the dual function of the first process function; the dual function of the first process function is: Where λ, s l is the dual variable; The function of the opposite number of the maximum benefit is rewritten as a piecewise linear convex function, which is: In the formula, J is a finite set, q is the sequence number of the function segment, max is the maximization function, a q (χ) and b q (χ) is the coefficient of the affine function; Obtain the auxiliary vector and the dual norm of the auxiliary vector; The second process function is obtained by rewriting the dual function of the first process function based on the auxiliary vector, the dual norm of the auxiliary vector and the piecewise linear convex function; the second process function is: Where, is the auxiliary vector, for The dual norm of ; The constraints of the second process function are transformed based on the duality theory to obtain the third process function; the third process function is: Where r lq is the dual variable, the superscript represents transpose; The reconstructed min-sup objective function is obtained according to the third process function and the outer layer function of the min-sup objective function; the reconstructed min-sup objective function is:

3. The distributed energy storage optimization configuration method for virtual power plants oriented to diversified power services according to claim 1 is characterized in that: The optimization direction is: maxS=S E +S FR +S load -C curt -C re -C ba -C loss -C om -C ex ; In the formula, max means maximization, S is the benefit, S E is the revenue from electricity sales to the virtual power plant grid, S FR The revenue of virtual power plants participating in the frequency regulation auxiliary service market, S load is the revenue from selling electricity to the internal load of the virtual power plant, C curt is the demand response cost, C re is the standby cost, C ba is the energy storage regulation cost, C loss is the life loss cost, C om is the operation and maintenance cost, C EX To compensate for the cost of incentives; The revenue from electricity sales of the virtual power plant grid is: Where S E is the electricity sales revenue of the virtual power plant grid, π is the total number of time periods divided within the preset time, κ is the total number of moments divided within each time period, i is the time period number, t is the moment number, is the real-time electricity price at time t during period i, is the real-time electricity price in time period i and time 1~κ, is the amount of electricity sold by the virtual power plant to the spot market at time t during period i; The benefits of the virtual power plant participating in the frequency regulation ancillary service market are: Where S FR To benefit from the participation of virtual power plants in the frequency regulation auxiliary service market, is the price of frequency modulation service at time t during period i, The frequency regulation capacity provided by the virtual power plant to the frequency regulation ancillary service market at time t during period i; The revenue from electricity sales for the internal load of the virtual power plant is: Where S load is the revenue from electricity sales to the internal load of the virtual power plant, is the load electricity price at time t during period i, N load is the number of controllable loads aggregated in the virtual power plant, j is the sequence number of the controllable loads in the virtual power plant, is the load of the jth controllable load in period i and time t; The demand response cost is: Where C curt is the demand response cost, is the interruption load compensation price of the jth controllable load in period i and time t, is the load interruption of the jth controllable load in period i and time t; The standby cost is: Where C re is the standby cost, N Ti is the number of lithium titanate batteries aggregated in the virtual power plant, N Fe is the number of lithium iron phosphate batteries aggregated in the virtual power plant, m is the serial number of the lithium titanate battery in the virtual power plant, and n is the serial number of the lithium iron phosphate battery in the virtual power plant. is the upper reserve capacity price of the mth lithium titanate battery, is the lower reserve capacity price of the mth lithium titanate battery, is the upper reserve capacity price of the nth lithium iron phosphate battery, is the lower reserve capacity price of the nth lithium iron phosphate battery; is the maximum reserve capacity of the mth lithium titanate battery at time t in period i, r i,t,m is the reserve capacity of the mth lithium titanate battery at time t in period i; is the maximum reserve capacity of the nth lithium iron phosphate battery at time t in period i, r i,t,n is the reserve capacity of the nth lithium iron phosphate battery at time t during period i; The energy storage regulation cost is: Where C ba Adjusting costs for energy storage, is the participation factor of the mth lithium titanate battery at time t in period i, is the participation factor of the nth lithium iron phosphate battery at time t in period i, is the unit adjustment cost of the mth lithium titanate battery, is the unit adjustment cost of the nth lithium iron phosphate battery; ξ i,t is the total uncertainty of source and load within the virtual power plant at time i and time t; N W is the number of wind turbines aggregated in the virtual power plant, N PV is the number of photovoltaic arrays aggregated in the virtual power plant, x is the serial number of the wind turbine in the virtual power plant, y is the serial number of the photovoltaic array in the virtual power plant, is the output prediction deviation of the x-th wind turbine at time t and time period i, is the output prediction deviation of the yth photovoltaic array at time t in period i, is the load forecast deviation of the jth controllable load at time t in period i; The life loss cost is: Where C loss is the life loss cost, is the unit charge and discharge power life loss cost coefficient of the mth lithium titanate battery, is the unit charge and discharge power life loss cost coefficient of the nth lithium iron phosphate battery; is the charging efficiency of the mth lithium titanate battery, is the discharge efficiency of the mth lithium titanate battery, is the charging efficiency of the nth lithium iron phosphate battery, Discharge efficiency of the nth lithium iron phosphate battery; is the charging power of the mth lithium titanate battery at time t during period i, is the discharge power of the mth lithium titanate battery at time t in period i, is the charging power of the nth lithium iron phosphate battery at time t during period i, is the discharge power of the nth lithium iron phosphate battery at time t during period i; The operation and maintenance costs are: Where C om For operation and maintenance costs, is the operation and maintenance cost coefficient of the mth lithium titanate battery, is the operation and maintenance cost coefficient of the nth lithium iron phosphate battery; is the unit capacity investment cost of the mth lithium titanate battery, is the unit capacity investment cost of the nth lithium iron phosphate battery; is the rated capacity of the mth lithium titanate battery, is the rated capacity of the nth lithium iron phosphate battery; is the discount rate of the mth lithium titanate battery, is the discount rate of the nth lithium iron phosphate battery; is the service life of the mth lithium titanate battery, is the service life of the nth lithium iron phosphate battery; The incentive compensation cost is: Where C EX To motivate compensation costs, is the incentive compensation price of the mth lithium titanate battery at time t in period i, is the incentive compensation price of the nth lithium iron phosphate battery at time t in period i, is the rated power of the mth lithium titanate battery, is the rated power of the nth lithium iron phosphate battery, The basic incentive compensation price.

4. The distributed energy storage optimization configuration method for virtual power plants oriented to diversified power services according to claim 3 is characterized in that: The deterministic constraints include power balance constraints, affine constraints, demand response constraints, energy storage equipment personalized operation constraints, market access constraints, and energy storage aggregation capacity constraints; The power balance constraint is: The affine constraints are: The demand response constraints are: Where, is the load interruption of the jth controllable load in period i and time (t+1), is the load of the jth controllable load in period i and time (t+1), η max is the maximum call rate of controllable load at a single moment, It is the maximum value of the sum of the continuous call rates of the controllable loads at two adjacent moments; The personalized operation constraints of the energy storage device are: 2≤Δ Ti <D Fe <k; Where, is the state of charge of the mth lithium titanate battery at time t in period i, is the state of charge of the nth lithium iron phosphate battery at time t during period i; is the self-discharge rate of the mth lithium titanate battery, is the self-discharge rate of the nth lithium iron phosphate battery; is the lower limit of the state of charge of the mth lithium titanate battery, The lower limit of the state of charge of the nth lithium iron phosphate battery; is the upper limit of the state of charge of the mth lithium titanate battery, The upper limit of the state of charge of the nth lithium iron phosphate battery; is the Boolean variable of the mth lithium titanate battery at time i and time t, 1 means that the mth lithium titanate battery is in the charging state at time t during period i. 0 means that the mth lithium titanate battery is in the discharge state at time t during period i. is the Boolean variable of the nth lithium iron phosphate battery at time i and time t, 1 means the nth lithium iron phosphate battery is in the charging state at time t during period i. 0 means that the nth lithium iron phosphate battery is in the discharge state at time t during period i, Δ Ti is the maximum continuous high power output time of the power type energy storage device, Δ Fe The maximum continuous high power output time of the energy storage device; The continuous Δ Ti The maximum value of the sum of load factors at any moment; The continuous Δ for the nth lithium iron phosphate battery Fe The minimum sum of load factors at all times; The market access constraints are: Where, Adjust the capacity threshold upward for the aggregate, is the aggregate charging power threshold; Adjust the capacity threshold downward for the aggregation, θ is the aggregate discharge power threshold; The energy storage aggregation capacity constraint is: Where, φ Ti Configure the capacity limit for a single lithium titanate battery, φ Fe Configure the capacity upper limit for a single lithium iron phosphate battery; φ con It is the upper limit of the total aggregate capacity of energy storage equipment in the virtual power plant.

5. The distributed energy storage optimization configuration method for virtual power plants oriented to diversified power services according to claim 3 is characterized in that: The uncertainty error probability constraint is: In the formula, inf represents the lower bound function, is the true probability distribution of the second uncertainty variable; β is the probability of violating the uncertainty error probability constraint.

6. The distributed energy storage optimization configuration method for virtual power plants oriented to diversified power services according to claim 5 is characterized in that: The steps of converting the uncertainty error probability constraint into the deterministic error probability constraint are: The uncertainty error probability constraint is rewritten based on Formula 2 to obtain a general mathematical expression for the uncertainty error probability constraint. Formula 2 is: Where g(χ,ω) is the function expression of the event in {·}; The historical data sample set is processed according to the normalization equation to obtain the random variable sample set; the normalization equation is: Where, is the lth random variable sample, for The lth historical data sample in , Σ is the variance of the historical data sample set, is the mean of the historical data sample set; Based on the random variable sample set, a random variable uncertainty set is constructed, and the random variable uncertainty set is: Where, is an uncertain set of random variables with τ as the boundary, σ is a random variable, R is a real number, and τ is the boundary of the uncertain set of random variables; Based on the fuzzy set of the second uncertainty variable and the random variable uncertainty set, a random variable uncertainty set boundary solution model is constructed. The random variable uncertainty set boundary solution model is: Based on the duality theory, the random variable uncertainty set boundary solution model is transformed to obtain a reconstructed random variable uncertainty set boundary solution model, and the reconstructed random variable uncertainty set boundary solution model is: Where min is the minimization function and λ is the dual variable; The data-driven uncertainty set boundary solution model is solved according to the reconstructed random variable uncertainty set boundary solution model and Formula 3; Formula 3 is: Where ψ is the boundary of the data-driven uncertainty set; The deterministic error probability constraint is obtained according to the data-driven uncertainty set boundary solution model and the uncertainty error probability constraint. The deterministic error probability constraint is: g(χ,ψ)≤0.

7. The distributed energy storage optimization configuration method for virtual power plants oriented to diversified power services according to any one of claims 1 to 6, characterized in that: The electrochemical energy storage device includes a power type energy storage device and an energy type energy storage device. The power type energy storage device includes N Ti The energy storage device includes a lithium titanate battery. Fe Lithium iron phosphate battery, N Ti is the number of lithium titanate batteries aggregated in the virtual power plant, N Fe is the number of lithium iron phosphate batteries aggregated in the virtual power plant; The decision variables include the real-time upper spare capacity of the lithium titanate battery, the real-time lower spare capacity of the lithium titanate battery, the real-time upper spare capacity of the lithium iron phosphate battery, the real-time lower spare capacity of the lithium iron phosphate battery, the real-time participation factor of the lithium titanate battery, the real-time participation factor of the lithium iron phosphate battery, the real-time charging power of the lithium titanate battery, the real-time discharging power of the lithium titanate battery, the real-time charging power of the lithium iron phosphate battery, the real-time discharging power of the lithium iron phosphate battery, the rated capacity of the lithium titanate battery, the rated capacity of the lithium iron phosphate battery, the rated power of the lithium titanate battery, the rated power of the lithium iron phosphate battery, the real-time state of charge of the lithium titanate battery, the real-time state of charge of the lithium iron phosphate battery, the real-time Boolean variable of the lithium titanate battery, and the real-time Boolean variable of the lithium iron phosphate battery.

8. The distributed energy storage optimization configuration method for virtual power plants oriented to diversified power services according to claim 7 is characterized in that: The decision variables also include the real-time electricity sold by the virtual power plant to the spot market, the real-time frequency regulation capacity provided by the virtual power plant to the frequency regulation auxiliary service market, the real-time interruption load compensation price of the controllable load, the real-time load interruption amount of the controllable load, the real-time incentive compensation price of the lithium titanate battery, and the real-time incentive compensation price of the lithium iron phosphate battery.

Images