A frequency regulation control method for energy storage considering uncertainty of renewable energy generation and load

Abstract

The present invention relates to an energy storage frequency regulation control method that takes into account the uncertainty of renewable energy power generation and load. First, an economic dispatch model that takes into account the minimization of the comprehensive cost of AGC is established; secondly, the characteristics of the AGC signal prediction data are obtained and its probability distribution function is established; then, a fuzzy set of the uncertainty of the AGC signal is constructed and solved using a distributed robust optimization model to obtain the basic dispatching point of the frequency regulation resource; then, according to the high-frequency component of the AGC power demand and the energy storage operation constraint, a battery energy storage AGC control model is established considering the battery energy storage SOC balance and the optimization of the frequency regulation cost; finally, the AGC control model is solved using a distributed consistency algorithm to obtain the real-time power output of the energy storage cluster and maintain the stability of the power grid frequency. The present invention not only realizes the dispatching and control of the frequency regulation resources in the distribution network by always minimizing the economic cost by comprehensively considering economic dispatch and frequency control, but also takes into account the SOC balance of energy storage to achieve the sustainability of frequency regulation control.

Description

Technical Field

[0001] The invention relates to an energy storage frequency modulation control method taking into account the uncertainty of new energy power generation and load, and belongs to the technical field of power systems. Background Art

[0002] In recent years, in response to the huge challenges brought about by energy supply, environmental protection and climate change, the energy revolution has gradually emerged worldwide, and the world's energy structure has undergone major changes. The proposal of the "dual carbon" strategy requires that my country's energy structure needs to accelerate the transformation to diversification and cleanliness, and at the same time, the development focus should shift from traditional energy to new energy. New energy power generation represented by photovoltaics and wind power not only optimizes the energy structure, but also has many advantages such as clean, low-carbon and renewable. Therefore, building a new power system with new energy as the main body is of great significance to achieving the "dual carbon" goal and promoting my country's high-quality development.

[0003] However, with the large-scale grid connection of renewable energy, the pressure on backbone power grids to absorb renewable energy has gradually emerged, and the phenomenon of wind and solar power abandonment has occurred frequently. In order to maintain the stability of the backbone power grid, the distribution network, as a key link in energy production, conversion and consumption, is also a supporting platform for renewable energy consumption. Under the guidance of the dual-carbon strategy, the distribution network has gradually become the core of the power system. Traditional distribution networks maintain the balance of active power through economic dispatch and frequency control. However, with the increase in the penetration rate of renewable energy in the distribution network, frequency regulation has changed from only responding to load uncertainty to requiring the dual uncertainty of renewable energy generation and load at the same time. Not only has the capacity and ramp rate of frequency regulation resources become higher, but the cost of frequency regulation ancillary services has also increased, especially the addition of fast frequency regulation resources such as energy storage, making the control of frequency regulation resources more complicated.

[0004] Most current studies focus on improving the economic performance of frequency regulation resources by improving the sustainability of frequency regulation resource operation, ignoring the impact of a reasonable and comprehensive optimization objective function on the economic efficiency of frequency control.

[0005] The invention patent application with application number 201910862624.7 and titled "A real-time economic dispatch method considering the flexible climbing capacity provided by wind, solar and storage" (hereinafter referred to as "Patent One") proposes a method considering the flexible climbing capacity provided by wind, solar and storage, which greatly reduces the pressure on conventional units to provide FRC. At the same time, by providing a combination of multiple "flexible sources" that provide FRC, real-time economic dispatch is performed with high flexibility.

[0006] The invention patent application with application number 201910294381.1 and titled "An economic dispatch method for microgrids based on dynamic constraints of SOC of energy storage devices" (hereinafter referred to as "Patent 2") proposes an economic dispatch method that takes into account the dynamic constraints of energy storage SOC, which solves the problem of insufficient reserved capacity of energy storage devices to ensure important loads, and ensures that the power supply can be uninterrupted at any time when the grid is disconnected, thereby reducing losses and improving economic benefits.

[0007] The above-mentioned Patents 1 and 2 start from the constraints of economic dispatch and ensure the continuity of operation to improve the economic benefits of economic dispatch, but ignore the frequent frequency adjustments under high penetration of new energy sources, which lead to a significant increase in AGC mileage costs, and do not consider the frequency regulation costs in the objective function, which leads to a decline in economic benefits.

[0008] The invention patent application with application number 202010333512.5 and titled "A two-layer control method for large-scale energy storage participating in frequency regulation on the distribution network side" (hereinafter referred to as "Patent Three") proposes a two-layer control method for energy storage frequency regulation, in which the upper-layer control considers economic efficiency and the lower-layer control considers frequency regulation performance, thereby reducing costs while maintaining system frequency stability.

[0009] Patent three takes economic performance into consideration but ignores the SOC balance of energy storage and does not take into account the sustainability of energy storage operation.

[0010] In summary, these methods do not comprehensively consider the coordination between economic dispatch and frequency control to maintain the SOC balance of energy storage while improving economy. Summary of the invention

[0011] The technical problem to be solved by the present invention is that the existing energy storage frequency control method does not comprehensively consider the coordination of economic dispatch and frequency control, and maintains the SOC balance of energy storage while improving the economy.

[0012] In order to solve the above technical problems, the technical solution of the present invention is to provide an energy storage frequency regulation control method considering the dual uncertainty of new energy and load, which is characterized by comprising the following steps:

[0013] Step S1, obtaining the constraint conditions of the economic dispatch model, and establishing an economic dispatch optimization model with the minimum comprehensive cost considering the automatic power generation control cost;

[0014] Step S2: Based on the historical AGC signal data, the probability distribution of the AGC signal characteristic value is established and the fuzzy set is constructed, the random optimization model is converted into a distributed robust optimization model and solved, and the basic power dispatch point of the frequency modulation resource is obtained, where AGC represents the automatic generation control cost;

[0015] Step S3: Based on the high-frequency component of the AGC power demand and the energy storage operation constraints, a battery energy storage AGC control model is established by considering the battery energy storage SOC balance and frequency modulation cost optimization, as shown in the following formula:

[0016]

[0017] Where: I is the effective operation range of energy storage; P Sref (t) represents the total power demand that the battery energy storage needs to meet at time t; D S,i (t) represents the frequency modulation consumption of the i-th battery storage at time t; P i S (t) represents the real-time charging and discharging power of the i-th battery storage; P i SL (t) and P i SU (t) represents the upper and lower limits of the charging and discharging power for the energy storage to meet its operating constraints;

[0018] Step S4: Solve the battery energy storage AGC control model using a consistent distributed algorithm, and continuously iterate the consistent variables until the optimal solution is obtained as the power output value of the energy storage to achieve grid frequency control.

[0019] Preferably, in step S1, the objective function of the economic dispatch optimization model with the minimum comprehensive cost considering the automatic power generation control cost is as shown in the following formula:

[0020]

[0021] Where: represents the degradation cost of battery energy storage; and They represent the charging power and discharging power of the i-th battery storage at the n-th moment respectively; P It means to find the mathematical expectation; represents the power generation cost of the jth traditional motor; is the power generated by the jth motor, represents the average output power value of the jth traditional motor at time n; represents the mileage cost of frequency regulation resources; Mn represents the regulation mileage; PF k,n represents the participation coefficient of the kth device at the nth time; S represents battery energy storage; G represents traditional motors; {G, S} represents all frequency modulation resources, namely traditional motors and battery energy storage.

[0022] Preferably, in step S1, the constraints include constraints of real-time economic dispatch, opportunity constraints considering AGC power demand, and SOC constraints of energy storage.

[0023] Preferably, it is characterized in that the constraints of the real-time economic dispatch include:

[0024] Power balance constraints:

[0025]

[0026] Power upper and lower limit constraints:

[0027]

[0028] Power ramp constraint:

[0029]

[0030] Line capacity constraints:

[0031]

[0032] In the formula, is the output power of the i-th battery storage at the nth moment; is the load power of node b at time n, where b represents a node; P k,min and P k,max They are the upper and lower limits of the output power of a traditional motor or energy storage, respectively; represents the output power of the kth traditional unit or battery energy storage at time n; rr k is the ramp rate of frequency regulation resources; T represents an economic dispatch cycle; f l is the capacity of line l; SF l,b is the transfer factor of node b on line l; Respectively represent the vector values ​​of traditional units, battery energy storage and new energy power generation in node b; Respectively represent the output power of traditional motors, battery energy storage and renewable energy power generation;

[0033] The opportunity constraints for AGC power requirements are:

[0034] In response to the AGC signal requirement, the power output of the frequency modulation resource is within its own constraints:

[0035]

[0036] Where: PF k,n is the participation coefficient of the frequency modulation resource labeled k; P(·) represents the probability of the event occurring; is the upper quantile of the AGC signal characteristic value in an economic dispatch cycle, is the lower quantile of the AGC signal characteristic value in an economic dispatch cycle; β represents the probability value of satisfying the specified power output opportunity constraint;

[0037] At the same time, the ramp rate of the frequency modulation resource is higher than the regulation rate of the AGC power demand:

[0038]

[0039] In the formula, represents the upper quantile of the regulation rate, represents the lower quantile of the regulation rate; P(·) represents the probability of the event occurring, and β represents the probability value of satisfying the specified climbing rate opportunity constraint

[0040] The participation coefficients of all frequency modulation resources should be between 0 and 1, and the sum of the participation coefficients should be 1, that is,

[0041] The SOC constraints of energy storage are:

[0042]

[0043] In the formula, SOC i,n-1 Represents the battery energy storage SOC value of the i-th battery at the n-1th moment; It is the cumulative value of the AGC signal upward adjustment within a cycle; The CE is the cumulative value of the AGC signal downward adjustment within a cycle; i represents the energy storage capacity of the i-th battery; and They represent the charging efficiency and discharging efficiency of energy storage respectively; and They respectively represent the upper and lower limits of the battery energy storage SOC value.

[0044] Preferably, the step S2 comprises:

[0045] Step S201: Process the AGC signal history data to extract the accumulated regulation energy Cumulative adjustment mileage M n , adjustment amplitude quantile and the adjustment rate quantile Then use the LSTM neural network to make predictions based on historical data and obtain its probability distribution;

[0046] Step S202, using Wasserstein metric to construct a fuzzy set from the predicted AGC signal sample set, so as to find a decision that minimizes the worst case;

[0047] Step S203: Based on the objective function and its constraints established in step S1, the optimization model is restated as follows:

[0048]

[0049]

[0050] Where: Expressed as the total degradation cost of all energy storage participating in economic dispatch; represents the total electricity generation cost of all conventional motors, E Q represents the expectation of the worst distribution that satisfies the fuzzy set, Β ki represents the FM participation coefficient matrix, The eigenvalue vector representing the AGC signal, D ki represents the cost of electricity generation from conventional motors; represents the frequency modulation mileage cost of all frequency modulation resources, M represents the participation coefficient matrix of frequency regulation mileage cost; m represents the coefficient matrix in the chance constraint, N m Represents the function matrix related to the output power of frequency modulation resources, β m represents the probability value of the mth opportunity constraint being satisfied, and J represents the set of all opportunity constraints;

[0051] Step S204, converting the opportunity constraint of the AGC power demand into a conditional risk value constraint, and then multiplying the constraint condition expressed by the conditional risk value constraint by a risk aversion parameter and adding the constraint condition to the objective function;

[0052] Step S205: Use the solver in MATLAB to solve and obtain the basic scheduling point of the frequency modulation resources.

[0053] Preferably, the step S3 comprises:.

[0054] The single energy storage frequency regulation consumption function is established as follows:

[0055]

[0056]

[0057] Where: D S,i (t) represents the frequency modulation consumption of the i-th battery energy storage at time t, which is the weighted square sum of the real-time charging and discharging power and capacity deviation, P i S (t) represents the real-time charging and discharging power of the i-th battery energy storage, represents the capacity deviation of the i-th battery energy storage, CE i (t) represents the energy storage capacity of the ith battery at time t, represents the rated capacity of the i-th battery energy storage, a S,i and b S,i are coefficients respectively, Δt represents the frequency modulation control interval time;

[0058] Energy storage operation is limited by rated power, rated capacity, ramp rate and power balance constraints as shown below:

[0059] P i,min ≤P i S ≤P i,max

[0060]

[0061]

[0062] P i S (t)∈Ω P,i (t)∈{P i S (t)|P i SL (t)≤P i S (t)≤P i SU (t)}

[0063] Where: and Indicates the charging and discharging ramp rate range of energy storage; Ω P,i (t) represents the feasible range of battery energy storage output power; P i SL (t) and P i SU (t) represents the upper and lower limits of the charging and discharging power for the energy storage to meet its operating constraints;

[0064] Considering that the real-time capacity of energy storage is affected by its charging and discharging power, the energy storage frequency regulation consumption function is transformed into a function form with the energy storage charging and discharging power as a single variable, as shown below:

[0065] D S,i (t) = α S,i [P i S (t)] 2 +β S,i P i S (t)+γ S,i

[0066] Where: α S,i , β S,i and γ S,i They are the coefficients of the quadratic term, linear term and constant term of the frequency modulation consumption function respectively;

[0067] The optimization model of AGC is established as follows:

[0068]

[0069] Where: I is the effective operation range of energy storage; P Sref (t) represents the total power demand that the battery energy storage needs to meet at time t.

[0070] Preferably, the step S4 comprises:

[0071] Step S401: introduce the Lagrange multiplier λ and describe the original optimization problem as a new optimization problem as shown below:

[0072]

[0073] Step S402: When the energy storage charging and discharging power is within its upper and lower limits, the marginal cost of frequency regulation of each energy storage in the cluster tends to be consistent and equal to λ * When , the optimization problem obtains the optimal solution, and the corresponding formula is as follows:

[0074]

[0075]

[0076] Step S403: continuously iterate the consistency variables between adjacent energy storage clusters until they reach consistency, and obtain the optimal marginal cost and optimal output power as follows:

[0077]

[0078]

[0079] In the formula, β S,i (t) represents the linear coefficient of the frequency modulation loss function of battery energy storage, α S,i Represents the quadratic coefficient of the battery energy storage frequency modulation loss function, Ω i (t) represents the feasible range of battery energy storage output power.

[0080] Compared with the prior art, the present invention has the following beneficial effects:

[0081] (1) The present invention comprehensively considers economic dispatch and frequency control, realizes the power allocation of energy storage at the optimal economic cost all the time, and considers the cost of automatic generation control (AGC) in the real-time economic dispatch model, thereby improving the economy of real-time economic dispatch;

[0082] (2) The present invention adopts a distributed robust optimization model based on the Wasserstein metric, which has better economic benefits and robustness than the robust optimization model.

[0083] (3) The present invention takes into account the SOC balance of energy storage in the AGC control model, thereby achieving SOC balance of energy storage while reducing the cost of energy storage frequency regulation, thereby realizing sustainable operation of energy storage frequency regulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1 It is a schematic diagram of the implementation process of the present invention;

[0085] Figure 2 It is the flow chart of real-time economic dispatch model;

[0086] Figure 3 It is a frequency tracking curve. DETAILED DESCRIPTION

[0087] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0088] The present invention discloses a method for energy storage frequency regulation control considering the dual uncertainty of new energy and load, which can be applied to distributed battery energy storage clusters. First, the present invention establishes an economic dispatch model that minimizes the comprehensive cost considering the AGC cost; secondly, the present invention obtains the characteristics of the AGC signal prediction data and establishes its probability distribution function based on several characteristic values ​​of the AGC signal historical data; then, the present invention measures the distance between the true probability distribution of the predicted value of the AGC signal characteristic quantity and the reference probability distribution, constructs a fuzzy set of the AGC signal uncertainty quantity and solves it with a distributed robust optimization model to obtain the basic dispatch point of the frequency regulation resource; subsequently, the present invention establishes a battery energy storage AGC control model based on the high-frequency component of the AGC power demand and the energy storage operation constraints, considering the battery energy storage SOC balance and the frequency regulation cost optimization; finally, the present invention uses a distributed consistency algorithm to solve the AGC control model, obtains the real-time power output of the energy storage cluster, and maintains the stability of the power grid frequency.

[0089] Specifically, if Figure 1 As shown, the present invention discloses a method for energy storage frequency regulation control considering dual uncertainties of new energy and load, comprising the following steps:

[0090] Step S1, obtaining the constraints of the economic dispatch model, and establishing an economic dispatch optimization model with the minimum comprehensive cost considering the AGC cost.

[0091] As a preferred implementation, step S1 further includes the following contents:

[0092] The real-time economic dispatch objective function of the economic dispatch model is established as shown in the following formula:

[0093]

[0094] Where: represents the degradation cost of battery energy storage; and They represent the charging power and discharging power of the i-th battery storage at the n-th moment respectively; P It means to find the mathematical expectation; represents the power generation cost of the jth traditional motor; is the power generated by the jth motor, represents the average output power of the jth conventional motor at time n; represents the mileage cost of frequency regulation resources; Mn represents the regulation mileage; PF k,n represents the participation coefficient of the kth device at the nth time; S represents battery energy storage; G represents traditional motors; {G, S} represents all frequency modulation resources, namely traditional motors and battery energy storage.

[0095] The constraints of the economic dispatch model include the constraints of real-time economic dispatch, the opportunity constraints of AGC power demand, and the SOC constraints of energy storage, where:

[0096] The constraints of real-time economic dispatch include power balance constraint, power upper and lower limit constraint, power ramp constraint, and line capacity constraint, which are expressed as

[0097] Power balance constraints:

[0098]

[0099] Power upper and lower limit constraints:

[0100]

[0101] Power ramp constraint:

[0102] nLine capacity constraints:

[0103]

[0104] In the formula, is the output power of the i-th battery storage at the nth moment; is the load power of node b at time n, where b represents a node; P k,min and P k,max They are the upper and lower limits of the output power of a traditional motor or energy storage, respectively; represents the output power of the kth traditional unit or battery energy storage at time n; rr kis the ramp rate of frequency regulation resources; T represents an economic dispatch cycle; f l is the capacity of line l; SF l,b is the transfer factor of node b on line l; Respectively represent the vector values ​​of traditional units, battery energy storage and new energy power generation in node b; They represent the output power of traditional motors, battery energy storage and new energy power generation respectively.

[0105] The opportunity constraints for AGC power requirements are:

[0106] In response to the AGC signal requirement, the power output of the frequency modulation resource should be within its own constraints:

[0107]

[0108] Where: PF k,n is the participation coefficient of the frequency modulation resource labeled k; P(·) represents the probability of the event occurring; is the upper quantile of the AGC signal characteristic value in an economic dispatch cycle, is the lower quantile of the AGC signal characteristic value in an economic dispatch cycle; β represents the probability value that the specified power opportunity constraint should be met.

[0109] At the same time, the ramp rate of the frequency modulation resource should be higher than the regulation rate of the AGC power demand:

[0110]

[0111] In the formula, represents the upper quantile of the regulation rate, represents the lower quantile of the regulation rate, P(·) represents the probability of the event occurring, and β represents the probability value that the specified ramp rate opportunity constraint should meet.

[0112] The participation coefficients of all frequency modulation resources should be between 0 and 1, and the sum of the participation coefficients should be 1, that is,

[0113]

[0114] The SOC constraints of energy storage are:

[0115]

[0116] In the formula, SOC i,n-1 Represents the battery energy storage SOC value of the i-th battery at the n-1th moment; It is the cumulative value of the AGC signal upward adjustment within a cycle; The CE is the cumulative value of the AGC signal downward adjustment within a cycle; i represents the energy storage capacity of the i-th battery; and They represent the charging efficiency and discharging efficiency of energy storage respectively; and They respectively represent the upper and lower limits of the battery energy storage SOC value.

[0117] Step S2: According to the historical AGC signal data, the probability distribution of the AGC signal characteristic value is established and the fuzzy set is constructed, the random optimization model is converted into a distributed robust optimization model and solved, and the basic power scheduling point of the frequency modulation resource is obtained.

[0118] As a preferred implementation, step S2 further includes the following contents:

[0119] Step S201, process the AGC signal history data and extract the following characteristic values: cumulative regulation energy Cumulative adjustment mileage M n , adjustment amplitude quantile and the adjustment rate quantile Then use the LSTM neural network to make predictions based on historical data and obtain its probability distribution.

[0120] Step S202, using Wasserstein metric to construct fuzzy sets from the predicted AGC signal sample set, so as to find a decision that minimizes the worst case. The robust optimization algorithm based on Wasserstein distance needs to roughly set the set of random variables first, generally taking the polyhedral set Ξ={ξ|Cξ≤d}. At the same time, the radius quantization confidence upper limit provided by the Wasserstein sphere can achieve better performance than other methods, and its specific definition is as follows:

[0121]

[0122] Based on the Wasserstein metric, the fuzzy set of the AGC signal data is determined as follows:

[0123]

[0124] Where: Q represents the probability distribution of all supported AGCs on the polyhedral set; is the empirical distribution of the AGC signal based on the Copula function prediction; represents the Wasserstein distance between two distributions; S(Ξ) represents the Wasserstein sphere that meets the radius requirement; ε is the radius of the Wasserstein sphere; It is a distribution set with the predicted distribution as the center and ε as the radius, that is, the fuzzy set of AGC signal distribution.

[0125] Step S203: Based on the objective function and its constraints established in step S1, the optimization model is restated as follows (in addition, the real-time economic dispatch constraints mentioned above must also be met):

[0126]

[0127]

[0128] Where: Expressed as the total degradation cost of all energy storage participating in economic dispatch; represents the total electricity generation cost of all conventional motors, E Q represents the expected value of the worst distribution that satisfies the fuzzy set, Β ki represents the FM participation coefficient matrix, The eigenvalue vector representing the AGC signal, D ki Represents the cost matrix form of electricity generation of traditional motors; represents the frequency modulation mileage cost of all frequency modulation resources, represents the participation coefficient matrix of frequency regulation mileage cost; the constraint condition is the abbreviation of the above AGC power demand and energy storage SOC opportunity constraint, M m The coefficient matrix representing the chance constraint, N m Represents the function matrix related to the output power of frequency modulation resources, β m represents the probability value of the mth chance constraint being satisfied, and J represents the set of all chance constraints.

[0129] Step S204: convert the opportunity constraint of the AGC power demand into a conditional value at risk (CVaR) constraint, where the constraint condition is expressed as follows:

[0130]

[0131] In the formula, represents the risk value of the mth opportunity constraint. The constraint condition expressed by CvaR is then multiplied by the risk aversion parameter and added to the objective function, which is expressed as follows:

[0132]

[0133] Q represents the probability distribution of all supported AGCs on the polyhedral set; is the empirical distribution of the AGC signal; ρ is the risk aversion parameter; represents the risk value of the mth opportunity constraint.

[0134] Step S205: Use the solver in MATLAB to solve and obtain the basic scheduling point of the frequency modulation resources.

[0135] Step S3: According to the high-frequency component of the AGC power demand and the energy storage operation constraints, a battery energy storage AGC control model is established taking into account the battery energy storage SOC balance and the frequency modulation cost optimization.

[0136] As a preferred implementation, step S3 further includes the following contents:

[0137] Obtain the high-frequency component of the AGC signal and establish the objective function of energy storage cluster balanced control for frequency modulation requirements, including the following steps:

[0138] Step S301: Establish a single energy storage frequency regulation consumption function as follows:

[0139]

[0140]

[0141] Where: D S,i (t) represents the frequency modulation consumption of the i-th battery energy storage at time t, which is the weighted square sum of the real-time charging and discharging power and capacity deviation, P i S (t) represents the real-time charging and discharging power of the i-th battery energy storage, represents the capacity deviation of the i-th battery energy storage, CE i (t) represents the energy storage capacity of the ith battery at time t, represents the rated capacity of the i-th battery energy storage, a S,i and b S,i are coefficients respectively, and Δt represents the frequency modulation control interval time.

[0142] Furthermore, the operation of energy storage is limited by the rated power, rated capacity, ramp rate and power balance constraints as shown below:

[0143] P i,min ≤P i S ≤P i,max

[0144]

[0145]

[0146] P i S (t)∈Ω P,i (t)∈{P i S (t)|P i SL (t)≤P i S (t)≤P i SU(t)}

[0147] Where: and Indicates the charging and discharging ramp rate range of energy storage; Ω P,i (t) represents the feasible domain of battery energy storage output power; P i SL (t) and P i SU (t) represents the upper and lower limits of the charging and discharging power required for the energy storage to meet its operating constraints.

[0148] Furthermore, considering that the real-time capacity of energy storage is affected by its charging and discharging power, the energy storage frequency regulation consumption function can be transformed into a function form with the energy storage charging and discharging power as a single variable, as shown below:

[0149] D S,i (t) = α S,i [P i S (t)] 2 +β S,i P i S (t)+γ S,i

[0150] Where: α S,i , β S,i and γ S,i They are the coefficients of the quadratic term, linear term and constant term of the frequency modulation consumption function respectively.

[0151] The above formula is combined to establish the optimization model of AGC as follows:

[0152]

[0153] Where: I is the effective operation range of energy storage; P Sref (t) represents the total power demand that the battery energy storage needs to meet at time t.

[0154] Step S4: Solve the battery energy storage AGC optimization model using a consistent distributed algorithm, and continuously iterate the consistent variables until the optimal solution is obtained as the power output value of the energy storage to achieve grid frequency control.

[0155] As a preferred implementation, step S4 further includes the following contents:

[0156] Solving with a consistent distributed algorithm includes the following steps:

[0157] Step S401, introducing the Lagrange multiplier λ, the original optimization problem can be described as a new optimization problem as shown below, and both have the same optimal solution.

[0158]

[0159] Step S402: According to the Lagrange extreme value condition, when the energy storage charging and discharging power is within its upper and lower limits, the marginal cost of frequency regulation of each energy storage in the cluster tends to be consistent and equal to λ * When , the optimization problem obtains the optimal solution, and the corresponding formula is as follows:

[0160]

[0161]

[0162] Step S403: continuously iterate the consistency variables between adjacent energy storage clusters until they reach consistency, and obtain the optimal marginal cost and optimal output power as follows:

[0163]

[0164]

[0165] In the formula, β S,i (t) represents the linear coefficient of the frequency modulation loss function of battery energy storage, α S,i Represents the quadratic coefficient of the battery energy storage frequency modulation loss function, Ω i (t) represents the feasible domain of battery energy storage output power.

[0166] In summary, the output power distribution scheme of energy storage is obtained.

[0167] The method disclosed in the present invention not only comprehensively considers economic dispatch and frequency control to achieve the dispatch and control of frequency regulation resources in the distribution network at the minimum economic cost, but also takes into account the SOC balance of energy storage to achieve the sustainability of frequency regulation control.

Claims

1. A method for energy storage frequency regulation control considering the dual uncertainty of new energy and load, characterized in that: The following steps are involved: Step S1, obtaining the constraint conditions of the economic dispatch model, and establishing an economic dispatch optimization model with the minimum comprehensive cost considering the automatic power generation control cost, wherein the objective function of the economic dispatch optimization model with the minimum comprehensive cost considering the automatic power generation control cost is as shown in the following formula: Where: represents the degradation cost of battery energy storage; and They represent the charging power and discharging power of the i-th battery storage at the n-th moment respectively; P It means to find the mathematical expectation; represents the cost of electricity generation from conventional motors; is the power generated by the jth motor, represents the average output power value of the jth traditional motor at time n; represents the mileage cost of frequency regulation resources; Mn represents the regulation mileage; PF k,n represents the participation coefficient of the kth device at the nth time; S represents battery energy storage; G represents traditional motors; {G, S} represents all frequency modulation resources, i.e. traditional motors and battery energy storage; Step S2: Based on the historical AGC signal data, the probability distribution of the AGC signal characteristic value is established and the fuzzy set is constructed. The economic dispatch optimization model constructed in step 1 is converted into a distributed robust optimization model and solved to obtain the basic power dispatch point of the frequency modulation resource, where AGC represents the automatic generation control cost. Step S3: Based on the high-frequency component of the AGC power demand and the energy storage operation constraints, a battery energy storage AGC control model is established by considering the battery energy storage SOC balance and frequency modulation cost optimization, as shown in the following formula: Where: I is the effective operation range of energy storage; P Sref (t) represents the total power demand that the battery energy storage needs to meet; D S,i (t) represents the frequency modulation consumption of the i-th battery storage at time t; P i S (t) represents the real-time charging and discharging power of the i-th battery storage; P i SL (t) and P i SU (t) represents the upper and lower limits of the charging and discharging power for the energy storage to meet its operating constraints; Step S4: Solve the battery energy storage AGC control model using a consistent distributed algorithm, and continuously iterate the consistent variables until the optimal solution is obtained as the power output value of the energy storage to achieve grid frequency control.

2. The energy storage frequency modulation control method considering the dual uncertainty of new energy and load as claimed in claim 1, characterized in that: In step S1, the constraints include constraints of real-time economic dispatch, opportunity constraints considering AGC power demand, and SOC constraints of energy storage.

3. The energy storage frequency modulation control method considering the dual uncertainty of new energy and load as claimed in claim 2, characterized in that: The constraints of the real-time economic dispatch include: Power balance constraints: Power upper and lower limit constraints: Power ramp constraint: Line capacity constraints: In the formula, is the output power of the i-th battery storage at the nth moment; is the load power of node b at time n, where b represents a node; P k,min and P k,max They are the upper and lower limits of the output power of a traditional motor or energy storage, respectively; represents the output power of the kth traditional unit or battery energy storage at time n; rr k is the ramp rate of frequency regulation resources; T represents an economic dispatch cycle; f l is the capacity of line l; SF l,b is the transfer factor of node b on line l; Respectively represent the vector values ​​of traditional units, battery energy storage and new energy power generation in node b; Respectively represent the output power of traditional motors, battery energy storage and renewable energy power generation; The opportunity constraints for AGC power requirements are: In response to the AGC signal requirement, the power output of the frequency modulation resource is within its own constraints: Where: PF k,n is the participation coefficient of the frequency modulation resource labeled k; Ρ(·) represents the probability of the event occurring; is the upper quantile of the AGC signal characteristic value in an economic dispatch cycle, is the lower quantile of the characteristic value of the AGC signal in an economic dispatch cycle; β represents the probability value that the specified power output opportunity constraint should be met; At the same time, the ramp rate of the frequency modulation resource is higher than the regulation rate of the AGC power demand: In the formula, represents the upper quantile of the regulation rate, represents the lower quantile of the regulation rate, P(·) represents the probability of the event occurring, and β represents the probability value that the specified ramp rate opportunity constraint should meet; The participation coefficients of all frequency modulation resources should be between 0 and 1, and the sum of the participation coefficients should be 1, that is, The SOC constraints of energy storage are: In the formula, SOC i,n-1 Represents the battery energy storage SOC value of the i-th battery at the n-1th moment; It is the cumulative value of the AGC signal upward adjustment within a cycle; The CE is the cumulative value of the AGC signal downward adjustment within a cycle; i represents the energy storage capacity of the i-th battery; and They represent the charging efficiency and discharging efficiency of energy storage respectively; and They respectively represent the upper and lower limits of the battery energy storage SOC value.

4. The energy storage frequency modulation control method considering the dual uncertainty of new energy and load as claimed in claim 3 is characterized in that: The step S2 comprises: Step S201: Process the AGC signal history data to extract the accumulated regulation energy Cumulative adjustment mileage M n , adjustment amplitude quantile and the adjustment rate quantile Then use the LSTM neural network to make predictions based on historical data and obtain its probability distribution; Step S202, using Wasserstein metric to construct a fuzzy set from the predicted AGC signal sample set, so as to find a decision that minimizes the worst case; Step S203: Based on the objective function and its constraints established in step S1, the optimization model is restated as follows: Where: Z(P c ,P d ) is represented by the total degradation cost of all energy storage participating in economic dispatch; represents the total electricity generation cost of all conventional motors, E Q It means the expectation of the worst distribution that satisfies the fuzzy set, Β ki represents the FM participation coefficient matrix, The eigenvalue vector representing the AGC signal, D ki represents the cost of electricity generation from conventional motors; represents the frequency regulation mileage cost of all frequency regulation resources, L(PF) represents the participation coefficient matrix of frequency regulation mileage cost; M m represents the opportunity constraint coefficient matrix, N m Represents the function matrix related to the output power of frequency modulation resources, β m represents the probability value that the mth chance constraint formula should satisfy, J represents the set of chance constraints; Q represents the probability distribution of all supported AGCs on the polyhedral set; is the empirical distribution of the AGC signal; Step S204, converting the opportunity constraint of the AGC power demand into a conditional risk value constraint, and then multiplying the constraint condition expressed by the conditional risk value constraint by a risk aversion parameter and adding the constraint condition to the objective function; Step S205: Use the solver in MATLAB to solve and obtain the basic scheduling point of the frequency modulation resources.

5. The energy storage frequency modulation control method considering the dual uncertainty of new energy and load as claimed in claim 3, characterized in that: The step S3 comprises: The single energy storage frequency regulation consumption function is established as follows: Where: D S,i (t) represents the frequency modulation consumption of the i-th battery energy storage at time t, which is the weighted square sum of the real-time charging and discharging power and capacity deviation, P i S (t) represents the real-time charging and discharging power of the i-th battery energy storage, represents the capacity deviation of the i-th battery storage, CE i (t) represents the battery energy storage capacity at time t, Indicates the rated capacity of the battery energy storage, a S,i and b S,i are coefficients respectively, Δt represents the frequency modulation control interval time; Energy storage operation is limited by rated power, rated capacity, ramp rate and power balance constraints as shown below: P i,min ≤P i S ≤P i,max P i S (t)∈Ω P,i (t)∈{P i S (t)|P i SL (t)≤P i S (t)≤P i SU (t)} Where: and Indicates the charging and discharging ramp rate range of energy storage; Ω P,i (t) represents the feasible range of battery energy storage output; P i SL (t) and P i SU (t) represents the upper and lower limits of the charging and discharging power for the energy storage to meet its operating constraints; Considering that the real-time capacity of energy storage is affected by its charging and discharging power, the energy storage frequency regulation consumption function is transformed into a function form with the energy storage charging and discharging power as a single variable, as shown below: D S,i (t)=a S,i [P i S (t)] 2 +b S,i P i S (t)+γ S,i Where: α S,i , β S,i and γ S,i They are the coefficients of the quadratic term, linear term and constant term of the frequency modulation consumption function respectively; The optimization model of AGC is established as follows: Where: I is the effective operation range of energy storage; P Sref (t) represents the total power demand that the battery energy storage needs to meet at time t.

6. The energy storage frequency modulation control method considering the dual uncertainty of new energy and load as claimed in claim 5, characterized in that: The step S4 comprises: Step S401: introduce the Lagrange multiplier λ and describe the original optimization problem as a new optimization problem as shown below: Step S402: When the energy storage charging and discharging power is within its upper and lower limits, the marginal cost of frequency regulation of each energy storage in the cluster tends to be consistent and equal to λ * When , the optimization problem obtains the optimal solution, and the corresponding formula is as follows: Step S403: continuously iterate the consistency variables between adjacent energy storage clusters until they reach consistency, and obtain the optimal marginal cost and optimal output power as follows: In the formula, β S,i (t) represents the linear coefficient of the frequency modulation loss function of battery energy storage, α S,i Represents the quadratic coefficient of the battery energy storage frequency modulation loss function, Ω i (t) represents the feasible domain of battery energy storage output power.

Images