Battery energy storage frequency regulation method based on chance constraint under energy arbitrage constraint

Abstract

The application discloses a battery energy storage frequency regulation method based on an opportunity constraint under energy arbitrage constraints, characterized in that: when energy storage participates in power grid frequency modulation and energy arbitrage, the energy arbitrage income is preferentially ensured; a dynamic programming algorithm is used to take the energy storage capacity and electricity price as state variables, and solve the power of energy arbitrage under a long time scale with the maximization of energy storage income as a target; the energy arbitrage decision result, i.e., the arbitrage power and the node capacity, is added to the frequency modulation decision of a short time scale in the form of a constraint condition; a forward-backward bias-based method for searching for a random variable set is used to rewrite the opportunity constraint condition into a second-order cone constraint to overcome the randomness of the model, and the frequency modulation decision problem is rewritten into a second-order cone problem for solving.

Description

TECHNICAL FIELD

[0001] The application discloses a battery energy storage frequency regulation method based on an opportunity constraint under an energy arbitrage constraint, and belongs to the technical field of power generation with energy storage participating in grid frequency regulation. BACKGROUND

[0002] In recent years, due to the rapid growth of electricity demand and the promotion of various new energy power generation, the demand for energy storage equipment is also increasing, and the role of user-side energy storage is gradually valued. As of the end of 2018, the cumulative installed capacity of user-side electrochemical energy storage in China was 1583MWh. As of the end of May 2020, 71 user-side energy storage power stations had been built in Jiangsu Province, with a total capacity of 787MWh. In addition to serving as a backup power supply to avoid unexpected power outages, the main operating strategy of the currently built user-side energy storage is to combine power market price fluctuations to carry out energy arbitrage and reduce electricity costs.

[0003] However, under the current cost of battery energy storage technology, it is difficult to ensure that enterprises can recover the investment cost solely through energy arbitrage. Participating in the auxiliary service market is adopted as a major way to increase energy storage revenue in numerous studies and a large number of foreign practices.

[0004] Currently, energy arbitrage and frequency regulation have been proven to be the most profitable applications in the electricity market, and many scholars combine the two to provide joint coordination strategies for different needs of the electricity market and energy storage users in order to maximize the revenue of energy storage units.

[0005] Currently, the following patent documents are listed for energy storage participating in frequency regulation services:

[0006] Patent document one "A control method and system for optimizing energy storage participating in frequency regulation auxiliary services" (patent application number CN202110337216.7) proposes: obtaining historical frequency regulation signal data of the frequency regulation market, wherein the historical frequency regulation signal data includes a training data set and a test data set; constructing an energy storage participating in frequency regulation auxiliary service model based on a deep reinforcement learning algorithm; establishing an experience replay pool of deep reinforcement learning according to the current state of energy storage, the current action, the immediate return of the current action, the state of the next moment, and the scheduling period end flag; training the energy storage participating in frequency regulation auxiliary service model according to the training data set and the experience replay pool; detecting the energy storage participating in frequency regulation auxiliary service model that has been trained according to the test data set; and optimizing and controlling the energy storage according to the energy storage participating in frequency regulation auxiliary service model.

[0007] Patent document two "a method for formulating a peak regulation and frequency regulation strategy of an energy storage power station in a receiving end power grid" (patent application number CN202110251524.8) proposes: using spot energy market price and frequency regulation auxiliary service market price data as input, considering the characteristics of a typical receiving end power grid, analyzing the feasibility of the application scenario of the energy storage power station, considering the economy of the energy storage power station under multiple application scenarios, considering the state of charge constraint and power constraint of the energy storage power station, and establishing an optimization scheduling model for the energy storage power station to maximize the revenue of participating in peak regulation and frequency regulation services.

[0008] However, patent document one only considers a single application scenario of energy storage, and cannot fully utilize energy storage resources at different time scales. Patent document two does not consider ensuring that energy storage participates in peak regulation and frequency regulation when the frequency regulation market is imperfect and the frequency regulation revenue is unstable when considering joint decision-making of energy storage peak regulation and frequency regulation. SUMMARY

[0009] The purpose of the present application is to overcome the shortcomings of the above background art, and to provide an opportunity constraint-based battery energy storage frequency regulation method under energy arbitrage constraints, which realizes the optimal strategy of battery energy storage participating in frequency regulation when energy arbitrage is prioritized.

[0010] The present application adopts the following technical solutions to achieve the above invention purposes:

[0011] An opportunity constraint-based battery energy storage frequency regulation method under energy arbitrage constraints, characterized in that: when energy storage participates in grid frequency regulation and energy arbitrage at the same time, the energy arbitrage revenue is prioritized; using a dynamic programming algorithm to take the energy storage capacity and the price as state variables, and to maximize the energy storage revenue as the target to solve the energy arbitrage power under long time scale; the energy arbitrage decision result is added to the short time scale frequency regulation decision as a constraint condition; using a forward-backward bias-based method to find a set of random variables to rewrite the opportunity constraint condition as a second-order cone constraint to overcome the randomness of the model, and to convert the frequency regulation decision problem into a second-order cone problem for solving.

[0012] As an opportunity constraint-based battery energy storage frequency regulation method under energy arbitrage constraints, the frequency regulation decision of each stage is the regulation difference coefficient r k following the frequency regulation signal; the decision target is the maximization of the regulation difference coefficient; the constraint conditions include the limits of the energy storage capacity and power: R min ≤R k,i ≤R max ;

[0013] Where k represents the sequence number of the measurement time stage, and the duration of each measurement time stage is Δt minutes. The decision-making process of battery energy storage for one day is divided into K measurement time stages, K = (24 * 60) / Δt, and different stages are represented by subscripts k = 1, 2, ..., K; the frequency modulation signal is updated every τ seconds, and each measurement time stage includes I signal measurement time points, I = (Δt * 60) / τ, and different signal measurement time points are identified by i = 1, 2, ..., I; in summary, the initial signal measurement time point within a day can be represented by the subscript (1, 0), and any other signal measurement time point can be represented by the subscript (k, i). When k > 1, the subscript (k, 0) represents the initial time point of stage k, and this time point is the same signal measurement time point as the time point (k-1, I).

[0014] Among them, R k,i It is the energy storage capacity at time (k,i). R is the output power of the energy storage battery between times (k,i-1) and (k,i). min R max These are the minimum and maximum values ​​of the rated capacity of the energy storage, P. min P max These are the minimum and maximum rated power of the energy storage, respectively. The arbitrage power of the battery within stage k. Energy arbitrage requires energy storage capacity at the end of phase k. and It is calculated from energy arbitrage decisions.

[0015] Furthermore, the energy storage capacity P bat It contains 3 parts, P bat =P E +P D +P ctrl P E For power used in low-storage-high-emission energy arbitrage, P D In response to the frequency modulation signal, the frequency modulation output power, P ctrl This is the control power used to adjust the SOC (State of Charge) of the energy storage battery. SOC is the available state of the remaining charge in the battery, usually expressed as a percentage.

[0016] The relationship between FM power and FM signal is expressed as: P D = r × Δf, where r is the descent coefficient, Δf ∈ [-1, 1], and is the frequency modulation signal;

[0017] The relationship between control power and frequency modulation signal is expressed as follows: P ctrl = D × Δf, where d k,n for The signal Δf at time (k,n)k,n The coefficient (n <i), D is the coefficient d k,n The coefficient matrix is a set of variables that need to be determined by frequency regulation decision;

[0018] The relationship between the frequency modulation signal and the grid frequency is represented as: Where, Δf k,i is the frequency deviation of the grid at (k,i) time, f nom is the standard frequency, Δf max is the maximum frequency deviation, usually set to 0.2 Hz.

[0019] Further, when solving the frequency modulation decision problem, for the random variable Δf, the constraint condition is expressed in the form of chance constraint Pr(AΔf≤b)≥1-∈; where Pr(*) represents the probability of the event, ∈ represents the confidence level, which is a very small number, usually taken as 10 -4 ~10 -6 , A and b are the coefficient vector and variable composed of the transformed constraints in the frequency modulation decision;

[0020] The set of Δf is defined as So that when Pr(AΔf≤b)≥1-∈ is always satisfied; After analyzing and calculating the historical frequency data of the grid, the forward deviation σ fi and the backward deviation σ bi of the set are obtained, and the chance constraint is rewritten as: u≥QAW -1 ; u≥-RAW -1 ;

[0021] Where, is the column vector obtained by taking the average of Δf by row, ‖u‖2 is the 2-norm of the matrix u by row, Q=diag(σ f1 ,σ f2 ,…,σ fK ), R=diag(σ b1 ,σ b2 ,…,σ bK ). Where, is the inverse of the covariance matrix of the matrix Δf.

[0022] Further, the and contained in the constraint condition of the frequency modulation decision are calculated by the energy arbitrage decision; The energy arbitrage decision is a process of obtaining the energy arbitrage power in each stage that makes the energy arbitrage profit of the energy storage highest according to the known electricity price fluctuation curve within a day; The dynamic programming algorithm is used to solve it: define the state variable and the decision variable Profit function State transition equation

[0023] where, is the grid price in stage k, Δt represents the duration of each stage. c represents the charging efficiency of the energy storage, η d represents the discharging efficiency of the energy storage. 1 {*} determines whether an event occurs, taking 1 when it occurs, and 0 otherwise.

[0024] According to the state transition equation, a new battery energy storage profit function is defined as:

[0025]

[0026] where, represents the maximum value of the profit obtained by the battery energy storage participating in grid activities from stage k to the last stage. Through a recursive algorithm, the maximum value of the battery energy storage profit in each stage is ultimately obtained, thereby determining the arbitrage power of the battery energy storage

[0027] The present application has the following beneficial effects by adopting the above technical solutions:

[0028] (1) The present application uses a priority energy arbitrage mode to mobilize energy storage to participate in frequency regulation. For regions where the frequency regulation service market needs to be improved, the energy arbitrage profit of the battery energy storage can be guaranteed to be unaffected.

[0029] (2) The present application uses an opportunity constraint method to consider the uncertainty of the frequency regulation signal, which is more in line with the actual operation scenario, and can enable the battery energy storage to always follow the frequency regulation signal in operation.

[0030] (3) The present method uses a forward-backward bias-based solution to solve the opportunity constraint problem, which can set a minimum limit violation probability while almost not increasing the calculation complexity and calculation time. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 is a method flowchart of the present application.

[0032] Figure 2 (a) is the price curve used in the implementation case of the present application.

[0033] Figure 2 (b) is the frequency regulation signal curve used in the implementation case of the present application.

[0034] Figure 3 is a selected section of the actual power curve of the energy storage in the implementation case of the present application. EAS represents energy arbitrage, and FRS represents frequency regulation. DETAILED DESCRIPTION

[0035] The technical solution of the invention will now be described in detail with reference to the accompanying drawings.

[0036] This invention discloses a battery energy storage frequency regulation method based on opportunity constraints under energy arbitrage constraints. Energy storage resources are given priority to participate in energy arbitrage, and under these constraints, the maximum skew coefficient that can be obtained at each stage is solved.

[0037] In this embodiment, the following scheme is used as the time interval and signal measurement time point, where k represents the sequence number of the measurement time stage. The decision-making process for a day of battery energy storage is divided into 48 half-hour periods, with different stages represented by the subscripts k = 1, 2, ..., 48, and Δt representing 30 minutes. i represents the signal measurement time point of the current measurement time stage. For example, if Δt is 30 minutes and the frequency modulation signal is updated every 4 seconds, then within stage k, a signal measurement time point is defined as 4 seconds, identified by i = 1, 2, ..., 450, and τ representing 4 seconds. In summary, any time point within a day can be represented by the subscript (k, i).

[0038] Figure 1 The implementation steps of the present invention are illustrated below:

[0039] Step 1: Obtain the parameters required for battery energy storage operation: rated capacity and rated power R. min ,R max ,P min ,P max Charge and discharge efficiency η c η d The capacity R at the beginning and end of a day ini R fin Obtain the grid electricity price at each stage. like Figure 2 As shown in (a), historical power grid frequency data f is obtained.

[0040] Step 2: Use dynamic programming to calculate the arbitrage power for each stage of the day. Energy storage capacity before and after the stage

[0041] Step 3: Obtain the historical frequency modulation signal Δf by processing the power grid frequency data f; calculate the random signal set. Relevant parameter matrices: W, Q, R. W, Q, and R are three matrices related to historical frequency data. W describes the whitening transform coefficients of Δf; Q and R describe the forward and backward deviations of the random signal. The methods for obtaining these matrices are described in steps 3.5-3.7.

[0042] Step 4: For stage k, establish constraints and solve for the adjustment coefficient r with the objective of maximizing the adjustment coefficient. kRepeat this step to calculate all the adjustment coefficients r in 48 stages of a day.

[0043] As an implementation step of the present application, the parameters required in step 1 include the following:

[0044] (1) The maximum and minimum values of the capacity R of the battery energy storage that can be used to participate in energy arbitrage and frequency modulation min max .

[0045] (2) The maximum and minimum values of the power P of the battery energy storage that can be used to participate in energy arbitrage and frequency modulation min max .

[0046] (3) The charging and discharging efficiency η of the battery energy storage c d .

[0047] (4) The initial and final capacity R of the battery energy storage in a day ini fin .

[0048] (5) The grid price in each stage of a day

[0049] (6) Historical grid frequency data f.

[0050] As an implementation step of the present application, step 2 can be further divided into the following steps:

[0051] Step 2.1: Define the state variables of the battery energy storage for energy arbitrage

[0052] Step 2.2: Define the decision variables of the battery energy storage for energy arbitrage The decision variables need to satisfy the constraints

[0053] Step 2.3: Determine the revenue function of the battery energy storage participating in energy arbitrage:

[0054]

[0055] where 1 {*} The function determines whether event * occurs, taking 1 if it occurs, otherwise taking 0.

[0056] Step 2.4: Express the sum of the revenue in each stage and all subsequent stages as a Bellman equation:

[0057]

[0058] From k = 48, for each ​​​​traversing all the values, looking for the maximum benefit record the time with

[0059] Step 2.5: Let k = k - 1. Repeat Step 2.5 for each time k ∈ [1, 48] until k = 1. For each k and value, there is a with value and its corresponding.

[0060] Step 2.6: Starting from k = 1, find the optimal power according to the stage and energy storage capacity. For example, when k = 1, value is obtained by index k = 1, in the results obtained in Step 2.5.

[0061] Step 2.7: According to the state transition equation of the battery energy storage capacity the energy storage capacity at the end of stage k is obtained

[0062] Step 2.8: Let k = k + 1. Repeat Step 2.6 and Step 2.7. Until k = 48, end the search and get the power of all stages and the front and back capacity

[0063] As an implementation step of the invention, Step 3 can be further divided into the following steps:

[0064] Step 3.1: Process historical frequency data according to the formula The frequency signal at each time is obtained from the frequency data.

[0065] Step 3.2: Place the energy efficiency loss in the charging and discharging process into the frequency signal according to the following formula to calculate in advance:

[0066]

[0067] Step 3.3: Form a column vector by combining the frequency signals corresponding to the k stage within a day, and form a matrix Δf by combining the signals of different dates. Take the average value by row to get a column vector

[0068] Step 3.4: Calculate the covariance matrix ∑ of matrix Δf Δf ; take the inverse of ∑ Δf matrix to get

[0069] Step 3.5: Place the matrix Carry out the Jolentsky decomposition to obtain the matrix W. Wherein,

[0070] Step 3.6: Calculate the forward and backward deviation σ fi ,σ bi , i∈[1, 450].

[0071]

[0072] Step 3.7: σ fi ,σ bi Diagonal matrix Q = diag(σ f1 ,σ f2 ,…,σ f450 ), R = diag(σ b1 ,σ b2 ,…,σ b450 ).

[0073] As the implementation steps of the present application, wherein step 4 can be further divided into the following steps:

[0074] Step 4.1: According to the energy arbitrage decision obtained in step 2, the constraint condition of the frequency modulation decision is as follows:

[0075]

[0076] Step 4.2: Rewrite the constraint condition as a chance constraint condition by taking the frequency modulation signal as a random variable as follows:

[0077]

[0078] Wherein, Pr(*) represents the probability of the event, ∈ is a very small positive number, usually 10 -4 ~ 10 -6 , represents the confidence of the probability formula, α ∈ (0, 1) is a constant factor, representing the proportion of the allowed capacity deviation to the total capacity. G represents a lower triangular matrix with all elements being 1. H = [1, 1, 1, …1] represents a row vector with all elements being 1. I represents a unit matrix.

[0079] Step 4.3: The above constraint condition can be expressed in the following form:

[0080]

[0081] Step 4.4: Write and b j into a matrix A and a vector b as follows:

[0082]

[0083]

[0084] Step 4.5: Reconstruct the frequency modulation target function and constraints as follows:

[0085] max r k

[0086]

[0087] u≥QAW -1

[0088] u≥-RAW -1

[0089] wherein ‖u‖2 is the 2-norm of the matrix u by row, at this time the target function and the constraints together constitute a second-order cone problem, which can be simply solved by using the existing MATLAB solver.

[0090] Step 4.6: Solve the problem model in 4.5 to obtain r k .

[0091] Step 4.7: Repeat steps 4.1-4.6 to obtain the frequency modulation coefficient r of all stages in a day.

[0092] At this point, the battery energy storage frequency regulation method under the energy arbitrage constraint based on the chance constraint is completed.

[0093] The method of the application is used in simulation, and the energy storage capacity is set to 10 MWh / 7 MW, the charge and discharge efficiency η c = 0.9, and η d = 0.95. The actual running decision results of each stage are as follows:

[0094] Table 1, arbitrage power and frequency modulation coefficient of each stage

[0095]

[0096] The obtained arbitrage power, control strategy and frequency modulation coefficient are applied to the simulation operation, and the frequency modulation signal is randomly extracted from the historical data, as shown in Figure 2 (b). The actual running energy storage power from 13:00 to 14:30 is as shown in Figure 3 . The simulation results show that when the arbitrage power is small in a day, the frequency modulation coefficient is larger. The results show that the energy storage can fully utilize the remaining resources to participate in frequency modulation under the condition of energy arbitrage priority.

[0097] The comparison of the benefits before and after the energy storage participates in frequency modulation is as follows:

[0098] Table 2, energy storage benefits before and after participating in frequency modulation

[0099]

[0100] The simulation results show that participating in frequency modulation can increase the energy storage benefit, and the battery energy storage can ensure that the energy arbitrage benefit is not affected while participating in frequency modulation.

[0101] The embodiments of the present application are illustrated and described in conjunction with the accompanying drawings, but those skilled in the art should understand that as long as the scope defined by the claims of the present application is not exceeded, other embodiments obtained by those skilled in the art without departing from the purpose of the present application and the protection scope of the claims can be obtained, and these all belong to the protection of the present application.

Claims

1. A battery energy storage frequency regulation method based on opportunity constraints under energy arbitrage constraints, characterized in that: When energy storage participates in grid frequency regulation and energy arbitrage simultaneously, priority is given to ensuring energy arbitrage profits. A dynamic programming algorithm is used with energy storage capacity and electricity price as state variables, and the goal of maximizing energy storage profits is to solve for the power of energy arbitrage over a long time scale. The arbitrage power and node capacity of the energy arbitrage decision results are incorporated into the frequency regulation decision on a short time scale as constraints; the chance constraints are rewritten into second-order cone constraints by using the method of finding the set of random variables based on forward and backward bias to overcome the randomness of the model, and the frequency regulation decision problem is transformed into a second-order cone problem to be solved. The frequency modulation decision at each stage is the droop coefficient that follows the frequency modulation signal. The decision objective is to maximize the droop coefficient; constraints include limitations on energy storage capacity and power. ; ; , k >1; ; in, k This indicates the sequence number of the measurement time stage, and the duration of each measurement time stage is... The decision-making process for battery energy storage for a day is divided into minutes. K Each measurement time phase Different stages are labeled below k = 1,2,…, K Indicates; frequency modulation signal per Updated every second, including each measurement time period. I Each signal measurement time point, I Different signal measurement time points are used i = 1,2…, I Identification; In summary, the initial signal measurement time point within a day can be represented by the subscript (1, 0), and any other signal measurement time point can be represented by the subscript (...). k , i ) indicates that when k When >1, use the subscript ( k ,0) indicates stage k The initial time point, which is related to ( k -1, I The time points are the same time points when the signal was measured; It is energy storage ( k , i Capacity at any given time, It is an energy storage battery in ( k , i -1) to ( k , i Output power between times, , These are the minimum and maximum values ​​of the rated capacity of the energy storage, respectively. , These are the minimum and maximum rated power of the energy storage, respectively. For the battery in stage k Arbitrage power within, Energy arbitrage requires energy storage capacity at the end of phase k. and It is calculated from energy arbitrage decisions.

2. A battery energy storage frequency regulation method based on opportunity constraints under energy arbitrage constraints according to claim 1, characterized in that, Energy storage power It consists of 3 parts. , For power used for low-storage, high-emission energy arbitrage, In response to the frequency modulation signal, the frequency modulation output power, For use in adjusting the control power of the energy storage SOC; The relationship between FM power and FM signal is expressed as follows: ,in, This is the adjustment coefficient. This is a frequency modulation signal; The relationship between control power and frequency modulation signal is expressed as follows: , ,in, for Signal at time (k,n) Adjustment coefficient (n) <i), coefficient The coefficient matrix consists of a set of variables that need to be determined through frequency modulation decisions; The relationship between the frequency modulation signal and the power grid frequency is expressed as follows: ,in, For the frequency deviation of the power grid, For standard frequency, This represents the maximum frequency offset.

3. A battery energy storage frequency regulation method based on opportunity constraints under energy arbitrage constraints according to claim 1, characterized in that, When solving the frequency modulation decision problem, for random variables The constraints are expressed in the form of opportunity constraints. ;in, Represents the probability of the * event occurring. This indicates that the confidence level is a very small number. and The coefficient vector and variables are formed after the transformation of various constraints in frequency modulation decision-making. Define about set , so that At that time, always satisfied After analyzing and calculating the historical frequency data of the power grid, the data used to describe the set is obtained. Forward deviation and backward deviation The opportunity constraint is rewritten as follows: ; ; ; in, for The column vector obtained by averaging the rows For the matrix Take the 2-norm of the row. , ;in, , For matrix The inverse of the covariance matrix.

4. A battery energy storage frequency regulation method based on opportunity constraints under energy arbitrage constraints according to claim 1, characterized in that, The constraints included in the frequency modulation decision and The energy arbitrage decision is calculated; the energy arbitrage decision is a process of determining the energy storage arbitrage power in each stage within a day based on the known electricity price fluctuation curve, which maximizes the profit of energy storage participating in energy arbitrage; it is solved using a dynamic programming algorithm: defining state variables. ; Decision variables ; Payoff function State transition equation ;in, Let k be the grid electricity price. Indicates the duration of each stage; Indicates energy storage charging efficiency. Indicates the energy storage discharge efficiency; Determine whether the event * has occurred; return 1 if it has occurred, otherwise return 0.

Images