CN115764938A - Microgrid group-containing shared energy storage optimal scheduling method considering new energy power generation uncertainty - Google Patents
Microgrid group-containing shared energy storage optimal scheduling method considering new energy power generation uncertainty Download PDFInfo
- Publication number
- CN115764938A CN115764938A CN202211361933.4A CN202211361933A CN115764938A CN 115764938 A CN115764938 A CN 115764938A CN 202211361933 A CN202211361933 A CN 202211361933A CN 115764938 A CN115764938 A CN 115764938A
- Authority
- CN
- China
- Prior art keywords
- power
- energy storage
- scene
- microgrid
- formula
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The microgrid group-containing shared energy storage optimal scheduling method considering the uncertainty of new energy power generation comprises the following steps: obtaining a daily state type transfer process set according to historical data; generating a confrontation network by using conditions to generate a mass daily scene; transferring the day field Jing Dairu day state type to obtain a Zhou Changjing set; obtaining a typical scene set through scene reduction; establishing a master multi-slave game model by taking the generated typical scene set as operation simulation data and taking each microgrid as a follower by taking a shared energy storage operator as a leader; and solving a Nash equilibrium strategy for the established master-slave game model by using a particle swarm algorithm nested solver to realize the optimal scheduling of the shared energy storage and the micro-grid. The invention improves the adaptability of the micro-grid-contained shared energy storage system to the uncertainty of the output of new energy, and improves the benefits of each micro-grid and the shared energy storage operator.
Description
Technical Field
The invention relates to the technical field of optimized operation of a power distribution network, in particular to a microgrid-containing group sharing energy storage optimized dispatching method considering uncertainty of new energy power generation.
Background
In recent years, our country has remarkable development effect of new energy represented by wind power and photovoltaic power generation, and micro-grids also face a more severe problem of new energy consumption: because the generated energy of high-proportion new energy has uncertainty, the demand of local users on high-quality power supply is difficult to meet, and the power distribution network is also difficult to accept the high-fluctuation electric energy to surf the internet, so that the problem of wind and light abandonment is serious. The method for configuring the shared energy storage for the micro-grid group is an effective means for stabilizing wind and light fluctuation, and can solve the problem that the high energy storage cost brings restriction to the micro-grid group energy storage construction.
In the related research of sharing energy storage optimization scheduling of a microgrid group, for example, documents [1]: wu Chengjun, lie group, and the like, a cooling, heating and power multi-microgrid system double-layer optimization configuration [ J/OL ] based on energy storage power station service is adopted, and a power grid technology comprises the following steps: 1-16 and 2021.08, a typical day is generally selected as a scene of operation simulation, and the lease price of shared energy storage is mostly set in an energy storage capacity configuration stage, that is, the optimal energy storage capacity and the corresponding energy storage lease price are set at the same time in the initial stage of shared energy storage construction.
However, with the improvement of the permeability of new energy, the prior art has the following defects: the existing shared energy storage optimization scheduling strategy adopts a typical day as an operation scene, and the uncertainty of new energy output cannot be fully considered, so that the adaptability of the scheduling strategy to each new energy output scene is insufficient, and the benefit maximization of a shared energy storage system cannot be ensured;
at present, the shared energy storage adopts fixed energy storage lease prices, such as documents [2]: tang Xiafei, wu Xianxiang, niqing, and the like, wind farm energy storage capacity optimization configuration by using cloud energy storage lease service [ J ], electric power science and technology report, 2020, 35 (1): 90-95. And the energy storage lease price can be flexibly adjusted to guide each micro-grid to reasonably plan an energy storage lease capacity plan, so that the energy storage utilization rate is improved, and the cost of energy storage use is reduced.
Disclosure of Invention
Aiming at the problems of reducing grid-connected power fluctuation of a microgrid group and improving income of each microgrid and a shared energy storage operator, the invention provides a microgrid group-containing shared energy storage optimal scheduling method considering new energy power generation uncertainty, which can effectively reduce the grid-connected power fluctuation of the microgrid group, improve the income of each microgrid and the shared energy storage operator by adjusting energy storage lease electricity prices, and relieve the problem of a large amount of 'wind and light abandonment' caused by overhigh grid-connected power fluctuation of the microgrid group; and the power consumption cost of each micro-grid is reduced, and the income of a shared energy storage operator is improved.
The technical scheme adopted by the invention is as follows:
the microgrid group-containing shared energy storage optimal scheduling method considering the uncertainty of new energy power generation comprises the following steps:
step1: obtaining a daily state type transfer process set according to historical power generation data and load data of the micro-grid;
and 2, step: generating a day scene with a day state type label by using the condition generation countermeasure network;
and 3, step3: substituting the day scene in the step2 into a day state type transfer process according to the day state type label to obtain a Zhou Changjing set;
and 4, step4: obtaining a small amount of typical scene sets which furthest reserve scene information through scene reduction;
and 5: taking the typical scene set generated in the step4 as operation simulation data, and taking a shared energy storage operator as a leader and each microgrid as a follower to construct a master multi-slave game model;
step 6: solving a Nash equilibrium strategy for the one-master-multi-slave game model established in the step5, and the optimal scheduling of the shared energy storage and the micro-grid is realized.
In the step1, according to the daily state and frequency of the clustering, the wind and light output p of the micro-grid w 、p pv And load p load All with uncertainty, net generated power p of the microgrid G Represented by the formula (1):
p G =p w +p pv -p load (1);
in the formula (1), p w 、p pv 、p load Respectively representing wind power, photovoltaic power generation and load power of the micro-grid.
In order to reveal the transition rule of daily net generating power states of the microgrid, scenes in a historical net generating power scene set are clustered into K state types by adopting a K-means clustering algorithm, and a Markov state transition probability matrix P is calculated according to the daily states and frequency thereof obtained by clustering r :
In the formula (2), the element p ij Represents the probability of a day state transitioning from state K to state j (K, l =1,2, …, K);
in the formula (3), n ij And n ik Respectively representing the transition times from the state i to the state j and the state k in the historical data;
then, a set of day-of-week state transition processes is randomly generated by using a Markov Chain Monte Carlo (MCMC) algorithm based on the Markov state transition probability matrix. The specific process is as follows:
1) Initializing i =1, and randomly generating an integer in a range of [1,K ] to represent a scene state of day 1;
2) i = i +1, generating a random number u in the range of [0,1], and obtaining an integer representing the scene state of the day i;
3) If i is larger than or equal to 7, outputting a one-week scene state transition sample, otherwise returning to the process 2);
repeat the above procedure N S Then N is obtained S A set of individual net electrical power generation day state transition processes.
In the step2, the scene day state type is used as a label c and the historical scene x is respectively used as a condition and a real sample input CGAN for training, and the generator G outputs a generated scene x' = G (z | c). The discriminator D of the CGAN needs to not only judge the degree of similarity between the generated scene distribution p (x ') and the historical scene distribution p (x), but also judge whether the generated scene x' belongs to the type label c. Loss function L of generator and discriminator G And L D The following were used:
L G =-E x′~p(x′) [D(x′∣c)] (4);
in the formula (4), E x′~p(x′) Represents the expected value of the corresponding x 'under the p (x') distribution; d (-) represents a discriminator function;
L D =-E x~p(x) [D(x∣c)]+E x′~p(x′) [D(x′∣c)] (5);
in the formula (5), E x~p(x) Representing the expected value of the corresponding x under the p (x) distribution; e represents the expected value of the corresponding distribution;
CGAN trained objective function using Wasserstein distance:
in the formula (6), λ represents a regular term coefficient;
in step2, the day scene with the day state type label is a net generating power sequence including each time period in one day of the microgrid. In the invention, a day is divided into 24 time intervals, namely a daily scene is a net generating power sequence containing 24 time intervals in a day of a microgrid.
In the step3, the daily status type transfer process is substituted according to the daily status type label, and the specific flow is as follows:
1) Generating a model according to the daily state transition process to obtain the state of each scene and a daily state transition process set;
2) Taking the day state as a label history scene as a training sample to train the CGAN;
3) Taking states in the daily state transition process set as labels, and taking noise z as a drive to generate a daily net discharge power scene;
4) And sequentially connecting the generated daily net discharge power curves according to the sequence in the daily state transition process set to obtain a Zhou Jing discharge power scene set.
In the step4, scenes are reduced by using a K-means algorithm, wherein the ith scene X i And the jth scene X j Calculation of inter-distance D ij The formula is as follows
In the formula (7), the reaction mixture is,andrespectively representing values of a scene average power, a scene power mean square error and a scene peak-valley difference time interval i; norm (-to) represents normalized calculation, and a calculation formula is shown as a formula (8);
in the formula (8), the reaction mixture is,andindexes y respectively representing the i-th periods i Where i belongs to [1,Ns ]]Maximum and minimum values within the range, wherein y i (may be)AndN s the number of scenes in a scene set;
in the formula (9), the reaction mixture is,representing the scene average power;represents the net electrical power generated at time t in the ith scenario; t represents the number of the scheduled time periods, each time period is 1 hour, and the scheduling cycle is one week, namely T =168;
in the formula (10), the compound represented by the formula (10),representing the scene power mean square error;
in the formula (11), the reaction mixture is,representing a scene peak-to-valley difference;are respectively provided withTo representBelongs to [1,T ] at t]The maximum and minimum values in the range;
the probability pi of the occurrence of the kth typical scene after reduction by using a clustering algorithm k The calculation is as follows:
in the formula (12), N k Is the number of scenes belonging to the kth class.
In the step5, firstly, according to an energy storage lease charging scheme formulated by a shared energy storage operator, a microgrid typical scene operation simulation electricity consumption cost expectation C is established MG The lowest model is as follows:
C MG =C zl +C bs +C yw +C bd (13);
in the formula (13), C zl Energy storage lease cost; c bs The cost of purchasing and selling electricity for the micro-grid; c yw Charging and discharging service fee for energy storage; c bd Punishment is carried out on the power fluctuation of the tie line;
C zl =α·Q E +β·Q p (14);
in the formula (14), α and β are unit energy capacity and power capacity lease price, respectively; q E 、Q p Energy storage capacity and power capacity of micro-grid leasing are respectively;
in the formula (15), T represents the number of scheduled time segments, and the scheduling period of the present invention takes one week, i.e., T =168;the power distribution network purchase and sale prices are respectively t time periods every day;purchasing and selling power of the microgrid at the t time period of the kth typical scene respectively, and taking a non-negative value; pi k Representing the probability of occurrence of the kth typical scene;
in the formula (16), γ is a charge and discharge service price;respectively taking the charge and discharge power of the microgrid at the t time period of the kth typical scene, and taking the non-negative value;
in the formula (17), epsilon is a power fluctuation penalty factor;the net generating power average value of the kth typical scene of the micro-grid is obtained;
the charging and discharging power of the microgrid cannot exceed the rented power capacity, and the microgrid cannot be charged and discharged at the same time:
the energy storage capacity used by the microgrid meets the charge constraint under the energy capacity leased by the microgrid, and the initial and final charge states are equal:
in the formula (21), the compound represented by the formula,the state of charge of the microgrid at the end of t of the kth typical scene;
in the formula (22), the reaction mixture is,the state of charge of the microgrid at the end of t-1 of the kth typical scene; Δ t is the time period duration; eta is the energy storage charge-discharge efficiency.
In the formula (23), the compound represented by the formula,for initial state of charge of stored energy, take here
The micro-grid meets self power balance constraint:
in the formula (24), the reaction mixture is,net generated power for the microgrid during period t of the kth typical scenario.
The purchase and sale power of the micro-grid is less than the maximum allowed tie line power between the micro-grid and the distribution networkp MG,bs And can not buy and sell electricity simultaneously:
according to the microgrid optimization models of the formulas (13) to (26), the charging and discharging power scene set S of the ith microgrid can be obtained i ={1,2,...,K i },K i Is its number of scenes; in the presence of a catalyst composed of N M In the micro-grid group consisting of the micro-grids, each micro-grid group charge and discharge power scene is formed by respectively taking out one scene from each micro-grid and combining the scenes, and then the charge and discharge scenes of the micro-grid group are collected into a set S ∑ ={1,2,...,K ∑ Number of scenes thereofThe charging micro-grid and the discharging micro-grid compensate each other to obtain the charging and discharging power of the kth scene in the period of tAnd probability of occurrence theta k Comprises the following steps:
in the formula (27), the reaction mixture is,respectively taking the charge and discharge power of the microgrid i at the t time period of the kth typical scene, and taking the non-negative value;
in formula (29), n i,ni Representing the probability of the occurrence of the nth scene in the microgrid i, wherein the ni value is [1,K i ]Within the range.
1): the energy storage operator adopts a business mode of providing lease service for the unit of year and participating in power grid peak regulation auxiliary service, and the energy storage operator obtains F net income within a dispatching cycle bat The maximum is the target.
F bat =F zl +F tf -F yw (30)
In the formula, F zl Charging the energy storage rental fee; f tf The peak regulation is carried out for the auxiliary power distribution network to obtain a profit; f yw Operating and maintaining costs for shared energy storage; c MGi,yw Charging and discharging service cost paid for the ith microgrid;respectively purchasing and selling electric quantity at the shared energy storage t moment;respectively sharing the charge and discharge amount of the stored energy at the moment t; q E,i 、Q p,i Energy storage energy capacity and power capacity leased for the ith micro-grid respectively; delta is the operation and maintenance cost of the energy storage unit power; n is a radical of hydrogen M The number of the micro-grids in the power distribution network.
2): shared energy storage charging and discharging power priority meeting micro-grid groupThe charge and discharge requirements can not exceed the total power capacity Q p,bat And the charge and the discharge can not be simultaneously carried out:
in the formula (34), the reaction mixture is,representing the total charging power of the microgrid group in a t period in a kth scene;
in the formula (35), the reaction mixture is,representing the total discharge power of the microgrid group in a t period in a kth scene;
3): the energy storage capacity used by the shared energy storage should meet the charge constraint under the energy capacity leased by the shared energy storage, and the initial and final charge states are equal:
in the formula (I), the compound is shown in the specification,the state of charge of the shared energy storage at the end t of the kth typical scene;an energy storage initial state of charge; q E,bat To share the total power capacity of the energy storage.
4): the shared energy storage system should satisfy its own power balance constraint:
5): the power for buying and selling the shared energy storage is smaller than the maximum connecting line power p allowed between the shared energy storage and the power distribution network bat,bs And can not buy and sell the electricity simultaneously:
in the step 6, a particle swarm algorithm is used for nesting a solver, a Nash equilibrium strategy is solved for the master-slave game model established in the step5, and optimal scheduling of shared energy storage and a micro-grid is realized;
the game model comprises three elements of a participant set, a strategy set and a profit set, and the established master-slave game model comprises the following steps:
(1) and a participant set: { SESO uegi } denotes the set of participants, with the shared energy storage operator SESO as leader and each microgrid MGi denotes a follower.
(2) And a strategy set: the strategy of the leader is to adjust the prices of the rental power capacity and the energy capacity, which are expressed as { alpha, beta }; the follower's strategy is to continually adjust the rental capacity scheme, denoted as { Q } MGi,E ,Q MGi,p ,p MGi,C ,p MGi,D }。
(3) And benefit set: the profit for each participant is their objective function, denoted F bat And { C MGi }。
When any party can not obtain larger profit for the party by changing the strategy, the game is considered to reach nash balance, and the strategy { alpha, beta; q MGi,E ,Q MGi,p ,p MGi,C ,p MGi,D Is the equilibrium solution.
The concrete solving steps are as follows:
and Step1, generating a net generating power scene around each microgrid, establishing a decision model of each individual, setting the number of particles and the maximum iteration times, and initializing the particles containing the energy storage capacity lease price information.
And Step2, calling the optimization models of the micro grids, solving the optimal energy storage leasing capacity including the energy capacity and the power capacity on the upper layer based on a particle swarm algorithm, and solving the charge and discharge plan of each micro grid in each time period on the lower layer based on a CPLEX solver.
And Step3, calling an SESO optimization model, and optimizing a charging and discharging scheme based on a CPLEX solver.
And Step4, calculating the fitness value of each particle, updating the speed and the position of the particle, and updating the local optimum of each particle and the global optimum of the particle swarm.
And Step5, judging whether the iteration times reach the maximum iteration times, if not, returning to Step2, otherwise, outputting the operation plan of each microgrid and the SESO, the lease energy storage capacity of each microgrid and the optimal energy storage lease price.
The invention relates to a micro-grid group-containing shared energy storage optimal scheduling method considering new energy power generation uncertainty, which has the following beneficial effects:
1) The method generates the operation simulation scene of the countermeasure network generating long scheduling period by using the conditions, so that the scheduling problem can fully consider the energy storage day scheduling and day scheduling situations, meanwhile, the assumption that the new energy power generation obeys certain probability distribution is avoided, and the capacity of the shared energy storage system containing the microgrid group to absorb new energy in different scenes is improved.
2) The invention utilizes the method for reducing scenes, reduces the number of scenes concentrated in typical scenes, simultaneously furthest retains the information which influences the configuration result of the energy storage leasing capacity, and improves the solving efficiency of the random optimization problem.
3) According to the method, a master-slave game optimization framework taking the shared energy storage operator as a main body and each microgrid as a slave body is constructed, and the energy storage lease pricing optimal strategy and the microgrid lease energy storage capacity optimal strategy of the shared energy storage operator are obtained by solving the Nash equilibrium solution of the game model, so that the benefits of each microgrid and the shared energy storage operator are improved.
Drawings
FIG. 1 is a block diagram of an optimized scheduling strategy according to the present invention.
FIG. 2 is a flow chart for solving the present invention.
FIG. 3 is a block diagram of the weekly scenario generation of the present invention.
FIG. 4 is a time-of-use electricity price chart according to the present invention.
Detailed Description
The microgrid group-sharing energy storage optimal scheduling method considering the uncertainty of new energy power generation is characterized in that an optimal scheduling strategy framework is shown in figure 1 and comprises the following steps:
step I: obtaining a daily state type transfer process set according to historical power generation data and load data of the micro-grid;
step II: generating a mass day scene with a day state type label by using a condition generation countermeasure network;
step III: substituting the day scene in the step II into a day state type transfer process according to the day state type label to obtain a Zhou Changjing set;
step IV: obtaining a small amount of typical scene sets which furthest retain scene information through scene reduction;
step V: taking the typical scene set generated in the step IV as operation simulation data, and establishing a master multi-slave game model by taking a shared energy storage operator as a leader and each micro-grid as a follower;
step VI: and D, solving the Nash equilibrium strategy for the model established in the step V by using a particle swarm algorithm nested solver, and realizing the optimal scheduling of the shared energy storage and the micro-grid.
Step VII: an example simulation analysis was performed.
The concrete description is as follows:
step I: obtaining a daily state type transfer process set according to historical power generation data and load data of the micro-grid; according to the daily state and frequency of the clustered state, the wind and light output p of the micro-grid w 、p pv And load p load All with uncertainty, net generated power p of the microgrid G Represented by the formula:
p G =p w +p pv -p load (45)
in the formula, p w 、p pv 、p load Respectively representing the wind power, the photovoltaic power generation and the load power of the micro-grid.
In order to reveal the transition rule of the daily power generation state of the wind and light field station, a K-means clustering algorithm is adopted to cluster the scenes in the historical net power generation power scene set into K state types. Calculating a Markov state transition probability matrix P according to the daily state and frequency of the daily state obtained by clustering r The definition is as follows:
in the formula, the element p ij Represents the probability of a day state transitioning from state i to state j (K, l =1,2, …, K), n ij And n ik Respectively representing the transition times from the state i to the state j and the state k in the historical data;
then, a set of day-of-week state transition processes is randomly generated by using a Markov Chain Monte Carlo (MCMC) algorithm based on the Markov state transition probability matrix. The specific process is as follows:
1) Initializing i =1, and randomly generating an integer in a range of [1,K ] to represent a scene state of day 1;
2) i = i +1, generating a random number u in the range of [0,1], and obtaining an integer representing the scene state of the day i;
3) And if i is larger than or equal to 7, outputting a one-week scene state transition sample, and otherwise, returning to the process 2.
Repeat the above procedure N S Then N is obtained S A set of individual net electrical power generation day state transition processes.
Step II: generating a mass day scene with a day state type label by using a condition generation countermeasure network; and training by taking the scene day state type as a label c and the historical scene x as conditions and real sample input CGAN respectively, and outputting and generating a scene x' = G (z | c) through a generator G. The discriminator D of the CGAN needs to not only judge the degree of similarity between the generated scene distribution p (x ') and the historical scene distribution p (x), but also judge whether the generated scene x' belongs to the type label c. The loss function of the generator and the arbiter is as follows:
L G =-E x′~p(x′) [D(x′∣c)] (48)
L D =-E x~p(x) [D(x∣c)]+E x′~p(x′) [D(x′∣c)] (49)
in the formula, E represents an expected value of the corresponding distribution; d (-) denotes the discriminator function.
CGAN trained objective function using Wasserstein distance:
in the formula, λ represents a regular term coefficient.
Step III: substituting the day scene generated in the step II into a day state type transfer process according to the day state type label to obtain a Zhou Changjing set; the flow framework is shown in fig. 3, and the specific flow is as follows:
1) Generating a model according to the daily state transition process to obtain the state of each scene and a daily state transition process set;
2) Training the CGAN by taking a historical scene with a daily state as a label as a training sample;
3) Taking states in the daily state transition process set as labels, and taking noise z as a drive to generate a daily net discharge power scene;
4) And sequentially connecting the generated daily net discharge power curves according to the sequence in the daily state transition process set to obtain a Zhou Jing discharge power scene set.
Step IV: obtaining a small amount of typical scene sets which furthest retain scene information through scene reduction; the scene reduction improves the solving efficiency of the stochastic optimization problem, but simultaneously causes the loss of the uncertainty information of power generation, thereby influencing the result of the stochastic optimization. To preserve to the maximum extent the information that has an influence on the outcome, the average power is chosenMean square error of powerDifference between peak and valley 3 pieces of feature information are attributes of the ith scene. Scene reduction using K-means algorithm, scene X i 、X j The inter-distance calculation formula is as follows:
in the formula, N s The number of scenes in a scene set;represents the net electrical power generated at time t in the ith scenario; t represents the number of the scheduled time segments, and the scheduling period of the method takes one week, namely T =168;respectively representBelongs to [1,T ] at t]The maximum and minimum values in the range; y is i (may be)Andindex in 3.
The probability pi of the occurrence of the kth typical scene after reduction by using a clustering algorithm k The calculation is as follows:
in the formula, N k Is the number of scenes belonging to the kth class.
Step V: taking the typical scene set generated in the step IV as operation simulation data, and establishing a master multi-slave game model by taking a shared energy storage operator as a leader and each micro-grid as a follower; first according to a shared storeAn energy storage lease charging scheme made by an operator is used for establishing a microgrid typical scene operation simulation electricity utilization cost expectation C MG The lowest model is as follows:
C MG =C zl +C bs +C yw +C bd (57)
C zl =α·Q E +β·Q p (58)
in the formula, C zl Energy storage lease cost; c bs The cost of purchasing and selling electricity for the micro-grid; c yw Charging and discharging service fee for energy storage; c bd Punishment is carried out on the power fluctuation of the tie line; alpha, beta and gamma are unit energy capacity, power capacity lease price and charge and discharge service price; q E 、Q p Energy storage capacity and power capacity for microgrid leasing;the power distribution network purchases the electricity selling price every day in a period of t;purchasing power for the microgrid at the t time period of the kth typical scene, wherein the value is non-negative;for sharing stored energy at the secondCharging and discharging power of k typical scenes in the t period is non-negative in value; epsilon is a power fluctuation penalty factor;and (4) the net generating power average value of the kth typical scene of the microgrid.
The charging and discharging power of the microgrid cannot exceed the rented power capacity and cannot be charged and discharged simultaneously:
the energy storage capacity used by the microgrid should meet the charge constraint under the energy capacity leased by the microgrid, and the initial and final charge states are equal:
in the formula (I), the compound is shown in the specification,the state of charge of the microgrid at the end of the kth typical scene is shown;for initial state of charge of stored energy, take hereΔ t is the time period duration; eta is the energy storage charge-discharge efficiency.
The micro-grid should satisfy self power balance constraints:
in the formula (I), the compound is shown in the specification,net generated power for the microgrid during period t of the kth typical scenario.
The purchase and sale power of the micro-grid is less than the maximum allowable tie line power p between the micro-grid and the distribution network MG,bs And electricity cannot be purchased at the same time:
then, the charging and discharging power scene set S of the ith microgrid can be obtained according to the expressions (13) to (26) i ={1,2,...,K i },K i Is its number of scenes. From N M In the micro-grid group formed by the micro-grids, each micro-grid group charge and discharge power scene is formed by respectively taking out one scene from each micro-grid and combining the scenes, and then the charge and discharge scenes of the micro-grid group are collected into a set S ∑ ={1,2,...,K ∑ Number of scenes thereofThe charging micro-grid and the discharging micro-grid compensate each other to obtain the charging and discharging power of the kth sceneAnd probability of occurrence theta k Comprises the following steps:
in the formula, pi i,ni Representing the probability of the occurrence of the nth scene in the microgrid i, wherein the ni value is [1,K i ]Within the range;and respectively, the charging and discharging power of the microgrid i in the t time period of the kth typical scene is non-negative.
The energy storage operator adopts a business mode of providing lease service for the unit of year and participating in power grid peak regulation auxiliary service, and the energy storage operator obtains a net income F in a dispatching cycle bat The maximum is the target.
F bat =F zl +F tf -F yw (74)
In the formula, F zl Charging the energy storage rental fee; f tf The peak regulation is carried out for the auxiliary power distribution network to obtain a profit; f yw Operating maintenance costs for shared energy storage; c MGi,yw Charging and discharging service cost paid for the ith microgrid;the power is purchased and sold at the time t for sharing the stored energy;sharing the charge and discharge amount of the stored energy at the t moment; q E,i 、Q p,i Energy storage capacity and power capacity leased for the ith microgrid; delta is the operation and maintenance cost of the energy storage unit power; n is a radical of M The number of the micro-grids in the power distribution network.
The shared energy storage charging and discharging power should preferably meet the charging and discharging requirements of the micro-grid group and cannot exceed the total power capacity Q thereof p,bat And the charge and the discharge can not be simultaneously carried out:
the energy storage capacity used by the shared energy storage should meet the charge constraint under the energy capacity leased by the shared energy storage and the initial and final charge states are equal:
in the formula (I), the compound is shown in the specification,the state of charge of the shared energy storage at the end t of the kth typical scene;an energy storage initial state of charge; q E,bat To share the total power capacity of the energy storage.
The shared energy storage system should satisfy its own power balance constraint:
the power for buying and selling the shared energy storage is smaller than the maximum connecting line power p allowed between the shared energy storage and the power distribution network bat,bs And can not buy and sell the electricity simultaneously:
step VI: solving a Nash equilibrium strategy for the model established in the step II by using a particle swarm algorithm nested solver to realize the optimal scheduling of the shared energy storage and the micro-grid; the game model comprises three elements of a participant set, a strategy set and a profit set, and the established master-slave game model comprises the following steps:
1) The participant set is as follows: { SESO uegomgi } represents a set of participants, with the shared energy storage operator SESO being the leader and each microgrid MGi representing a follower.
2) The strategy set is as follows: the strategy of the leader is to adjust the prices of the rental power capacity and the energy capacity, which are expressed as { alpha, beta }; the follower's strategy is to continually adjust the rental capacity scheme, denoted as { Q } MGi,E ,Q MGi,p ,p MGi,C ,p MGi,D }。
3) Benefit set: the revenue for each participant is their objective function, which may be expressed as F bat And { C MGi }。
And when any party cannot obtain larger profit for the party by changing the strategy, the game is considered to reach nash balance. The policy at this time { α, β; q MGi,E ,Q MGi,p ,p MGi,C ,p MGi,D Is the equilibrium solution. And solving a master-slave game optimization model optimization sharing energy storage lease electricity price scheme by adopting a particle swarm algorithm, solving an auxiliary peak regulation scheme by calling a CPLEX solver, and calculating lease profit by combining a micro-grid energy storage lease optimization model. The micro-grid energy storage leasing optimization model adopts a particle swarm algorithm nested CPLEX solver to solve the micro-grid leasing optimization model. The upper-layer particle swarm algorithm optimizes the power capacity and the energy capacity of the micro-grid lease, and the lower layer utilizes a CPLEX solver to solve the micro-grid operation plan.
The scenarios used in the above-described operational simulations were generated using the scenario generation method herein taking into account the net generated power uncertainty, and the overall solution flow is shown in fig. 2.
Step VII: example simulation analysis was performed:
a research main body consists of a shared energy storage system and three micro-grids, actual measurement data of wind-solar power generation power and load power in a year of 3 micro-grids in a certain region in China are taken as historical scene data sets, and the sampling interval of the data sets is 1 hour. The maximum transmission power of the micro-grids 1 to 3 and the junctor of the shared energy storage and distribution network is 500kW, and the centralized energy storage capacity Q of the shared energy storage configuration E,bat 4000kWh, power capacity Q p,bat 800kW, the energy storage charge-discharge efficiency eta is 95%, the energy storage unit power operation and maintenance cost delta is 0.1542 yuan/kWh, the established unit power charge-discharge service price gamma is 0.1542 yuan/kWh, the power fluctuation penalty factor epsilon is 0.15,the daily time-of-use electricity price is shown in fig. 3, the net generated power shortage of the microgrid group is severe in the high electricity price period in fig. 3, and the net generated power surplus of the microgrid group is excessive in the low electricity price period.
The profit-making results of the subjects before and after the game optimization are shown in table 1, and the capacity requirements of the microgrid are shown in table 2.
Table 1 comparison of profit between sharing energy storage operator and microgrid user before and after optimization of energy storage lease price
As can be seen from table 1, the profit of the sharing energy storage operator and the micro grid after game optimization are improved to different degrees.
Table 2 comparison of energy storage demand of each microgrid user before and after optimization of energy storage lease price
As can be seen from tables 1 and 2, the benefits of each party are improved because the energy storage lease price schemes of the shared energy storage providers stimulate each microgrid to lease more energy storage capacity in a price reduction manner, and the free capacity of each microgrid is reduced, so that the profit of each microgrid is improved. Each microgrid also increases its own profit due to a lower energy storage lease price.
The comparison of the leasing energy storage and shared energy storage idle energy storage peak shaving effects of the micro-grid groups before and after game optimization is shown in table 3.
TABLE 3 comparison of peak shaving effects before and after optimization of energy storage lease price
According to table 3, after the game optimization of the charging and discharging of the microgrid group, the mean square error expectation of the net generated power is reduced, because more energy storage capacity is leased, a better net discharging power stabilizing effect is obtained; the mean square error expectation of the net generating power is further reduced after the shared energy storage operator stabilizes the net generating power in the microgrid group, because the mean square error expectation of the network power in the microgrid group is reduced by utilizing idle energy storage, and the utilization rate of the stored energy is improved.
In summary, the provided optimized scheduling scheme can improve profits of all parties, improve the utilization rate of stored energy, and reduce the fluctuation of net generated power.
Claims (10)
1. The microgrid group-containing shared energy storage optimal scheduling method considering the uncertainty of new energy power generation is characterized by comprising the following steps of:
step1: obtaining a daily state type transfer process set according to historical power generation data and load data of the micro-grid;
and 2, step: generating a day scene with a day state type label by using a conditional generation countermeasure network;
and step3: substituting the day scene in the step2 into a day state type transfer process according to the day state type label to obtain a Zhou Changjing set;
and 4, step4: obtaining a small amount of typical scene sets which furthest reserve scene information through scene reduction;
and 5: taking the typical scene set generated in the step4 as operation simulation data, and taking a shared energy storage operator as a leader and each microgrid as a follower to construct a master multi-slave game model;
step 6: and (5) solving a Nash equilibrium strategy for the one-master multi-slave game model established in the step (5) to realize the optimal scheduling of the shared energy storage and the micro-grid.
2. The microgrid-containing group shared energy storage optimal scheduling method considering uncertainty of new energy power generation according to claim 1, characterized in that: in step1, the net generated power p of the microgrid G Represented by the formula (1):
p G =p w +p pv -p load (1);
in the formula (1), p w 、p pv 、p load Respectively representing wind power, photovoltaic power generation power and load power of the microgrid;
clustering the scenes in the historical net generating power scene set into K state types by adopting a K-means clustering algorithm, and calculating a Markov state transition probability matrix P according to the daily state and frequency of the clustered scenes r :
In the formula (2), the element p ij Represents the probability of a day state transitioning from state K to state j (K, l =1,2, …, K);
in the formula (3), n ij And n ik Respectively representing the transition times from the state i to the state j and the state k in the historical data;
then, based on a Markov state transition probability matrix, randomly generating a set of a day-of-week state transition process by adopting a Markov Chain Monte Carlo (MCMC) algorithm; the specific process is as follows:
1) Initializing i =1, and randomly generating an integer in a range of [1,K ] to represent a scene state of day 1;
2) i = i +1, generating a random number u in the range of [0,1], and obtaining an integer representing the scene state of the day i;
3) If i is larger than or equal to 7, outputting a one-week scene state transition sample, otherwise returning to the process 2);
repeat the above procedure N S Then N is obtained S A set of individual net electrical power generation day state transition processes.
3. The microgrid-containing group shared energy storage optimization considering new energy power generation uncertainty according to claim 1The method for scheduling is characterized in that: in the step2, the scene day state type is used as a label c and the historical scene x is respectively used as a condition and input into the CGAN for training, and the generator G outputs a generated scene x' = G (z | c); loss function L of generator and discriminator G And L D The following were used:
L G =-E x′~p(x′) [D(x′∣c)] (4);
in the formula (4), E x′~p(x′) Represents the expected value of the corresponding x 'under the p (x') distribution; d (-) represents a discriminator function;
L D =-E x~p(x) [D(x∣c)]+E x′~p(x′) [D(x′∣c)] (5);
in the formula (5), E x~p(x) Represents the expected value of the corresponding x under the p (x) distribution; e represents the expected value of the corresponding distribution;
CGAN trained objective function using Wasserstein distance:
in equation (6), λ represents a regular term coefficient.
4. The optimal scheduling method for the shared energy storage of the microgrid-containing group with the new energy generation uncertainty considered in claim 1 is characterized in that: in step2, the day scene with the day state type label is a net generated power sequence including each time period in one day of the microgrid.
5. The microgrid-containing group shared energy storage optimal scheduling method considering uncertainty of new energy power generation according to claim 1, characterized in that: in the step3, the daily status type transfer process is substituted according to the daily status type label, and the specific flow is as follows:
1) Generating a model according to the daily state transition process to obtain the state of each scene and a daily state transition process set;
2) Taking the day state as a label history scene as a training sample to train the CGAN;
3) Taking states in the daily state transition process set as labels, and taking noise z as a drive to generate a daily net discharge power scene;
4) And sequentially connecting the generated daily net discharge power curves according to the sequence in the daily state transition process set to obtain a Zhou Jing discharge power scene set.
6. The microgrid-containing group shared energy storage optimal scheduling method considering uncertainty of new energy power generation according to claim 1, characterized in that: in the step4, scenes are reduced by using a K-means algorithm, wherein the ith scene X i And the jth scene X j Calculation of inter-distance D ij The formula is as follows
In the formula (7), the reaction mixture is,andrespectively representing values of a scene average power, a scene power mean square error and a scene peak-valley difference time interval i; norm (-to) represents normalized calculation, and a calculation formula is shown as a formula (8);
in the formula (8), the reaction mixture is,andindexes y respectively representing the i-th periods i Where i belongs to [1,Ns ]]Within the range ofMaximum and minimum values, wherein y i (may be)AndN s the number of scenes in a scene set;
in the formula (9), the reaction mixture is,representing the scene average power;represents the net electrical power generated at time t in the ith scenario; t represents the number of scheduled time periods;
in the formula (10), the compound represented by the formula (10),representing the scene power mean square error;
in the formula (11), the reaction mixture is,representing a scene peak-to-valley difference;respectively representBelongs to [1,T ] at t]The maximum and minimum values in the range;
the probability pi of the occurrence of the kth typical scene after reduction by using a clustering algorithm k The calculation is as follows:
in the formula (12), N k Is the number of scenes belonging to the kth class.
7. The microgrid-containing group shared energy storage optimal scheduling method considering uncertainty of new energy power generation according to claim 1, characterized in that: in the step5, firstly, according to an energy storage lease charging scheme formulated by a shared energy storage operator, a microgrid typical scene operation simulation electricity utilization cost expectation C is established MG The lowest model:
C MG =C zl +C bs +C yw +C bd (13);
in the formula (13), C zl Energy storage lease cost; c bs The cost of purchasing and selling electricity for the micro-grid; c yw Charging and discharging service fee for energy storage; c bd Punishment is carried out on the power fluctuation of the tie line;
C zl =α·Q E +β·Q p (14);
in the formula (14), α and β are unit energy capacity and power capacity lease price, respectively; q E 、Q p Energy storage capacity and power capacity of micro-grid leasing are respectively;
in the formula (15), T represents the number of scheduled time periods;the power distribution network purchase and sale prices are respectively t time periods every day; purchasing and selling power of the microgrid at the t time period of the kth typical scene respectively, and taking a non-negative value; pi k Representing the probability of occurrence of the kth typical scene;
in the formula (16), γ is a charge and discharge service price;respectively taking the charge and discharge power of the microgrid at the t time period of the kth typical scene, and taking the non-negative value;
in the formula (17), epsilon is a power fluctuation penalty factor;the net generating power average value of the kth typical scene of the micro-grid is obtained;
the charging and discharging power of the microgrid cannot exceed the rented power capacity, and the microgrid cannot be charged and discharged at the same time:
the energy storage capacity used by the microgrid meets the charge constraint under the energy capacity leased by the microgrid, and the initial and final charge states are equal:
in the formula (21), the reaction mixture is,the state of charge of the microgrid at the end of t of the kth typical scene;
in the formula (22), the reaction mixture is,the state of charge of the microgrid at the end of t-1 of the kth typical scene; Δ t is the time period duration; eta is the energy storage charge-discharge efficiency;
in the formula (23), the reaction mixture is,for initial state of charge of stored energy, take here
The micro-grid meets self power balance constraint:
in the formula (24), the reaction mixture is,net generated power for the microgrid during time t of a kth typical scenario;
the purchase and sale power of the micro-grid is less than the maximum tie line power p allowed between the micro-grid and the distribution network MG,bs And electricity cannot be purchased at the same time:
according to the microgrid optimization models of the formulas (13) to (26), the charging and discharging power scene set S of the ith microgrid can be obtained i ={1,2,...,K i },K i Is its number of scenes; in the presence of a catalyst composed of N M In the micro-grid group consisting of the micro-grids, each micro-grid group charge and discharge power scene is formed by respectively taking out one scene from each micro-grid and combining the scenes, and then the charge and discharge scenes of the micro-grid group are collected into a set S ∑ ={1,2,...,K ∑ Number of scenes thereofThe charging micro-grid and the discharging micro-grid compensate each other to obtain the charging and discharging power of the kth scene in the period of tAnd probability of occurrence theta k Comprises the following steps:
in the formula (27), the reaction mixture is,respectively taking the charge and discharge power of the microgrid i at the t time period of the kth typical scene, and taking the non-negative value;
in the formula (29), n i,ni The probability of the occurrence of the nth scene in the microgrid i is represented, and the ni value is [1,K i ]Within the range.
8. The microgrid-containing group shared energy storage optimal scheduling method considering uncertainty of new energy power generation according to claim 7, characterized in that:
1): the energy storage operator adopts a business mode of providing lease service for the unit of year and participating in power grid peak regulation auxiliary service, and the energy storage operator obtains a net income F in a dispatching cycle bat Maximum target;
F bat =F zl +F tf -F yw (30)
in the formula, F zl Charging the energy storage lease fee; f tf The peak regulation is carried out for the auxiliary power distribution network to obtain a profit; f yw Operating maintenance costs for shared energy storage; c MGi,yw Charging and discharging service cost paid for the ith microgrid;respectively purchasing and selling electric quantity at the shared energy storage t moment;respectively sharing the charge and discharge amount of the stored energy at the moment t; q E,i 、Q p,i Energy storage energy capacity and power capacity leased for the ith micro-grid respectively; delta is the operation and maintenance cost of the energy storage unit power; n is a radical of M The number of micro-grids in the power distribution network;
2): the shared energy storage charging and discharging power preferentially meets the charging and discharging requirements of the micro-grid group, and the total power capacity Q of the micro-grid group cannot be exceeded p,bat And the charge and the discharge can not be simultaneously carried out:
in the formula (34), the reaction mixture is,representing the total charging power of the microgrid group in a t period in a kth scene;
in the formula (35), the reaction mixture is,representing the total discharge power of the microgrid group in a t period in a kth scene;
3): the energy storage capacity used by the shared energy storage should meet the charge constraint under the energy capacity leased by the shared energy storage, and the initial and final charge states are equal:
in the formula (I), the compound is shown in the specification,the state of charge of the shared energy storage at the end of t of the kth typical scene;an energy storage initial state of charge; q E,bat Total power capacity for shared energy storage;
4): the shared energy storage system should satisfy its own power balance constraint:
5): the power of the electricity purchased and sold by sharing the stored energy should be less than the maximum junctor power p allowed between the power purchasing and selling power and the power distribution network bat,bs And can not buy and sell the electricity simultaneously:
9. the optimal scheduling method for the shared energy storage of the microgrid-containing group with the new energy generation uncertainty considered in claim 1 is characterized in that: in the step 6, a particle swarm algorithm is used for nesting a solver, a Nash equilibrium strategy is solved for the master-slave game model established in the step5, and optimal scheduling of shared energy storage and a micro-grid is realized;
the game model comprises three elements of a participant set, a strategy set and a profit set, and the established master-slave game model comprises the following steps:
(1) and a participant set: { SESO $ MGi } represents a set of participants, wherein the shared energy storage operator SESO is a leader, and each microgrid MGi represents a follower;
(2) and a strategy set: the strategy of the leader is to adjust the prices of the rental power capacity and the energy capacity, which are expressed as { alpha, beta }; the follower's strategy is to continually adjust the rental capacity scheme, denoted as { Q } MGi,E ,Q MGi,p ,p MGi,C ,p MGi,D };
(3) And benefit set: the profit for each participant is their objective function, denoted F bat And { C MGi };
When any party can not obtain larger profit for the party by changing the strategy, the game is considered to reach nash balance, and the strategy { alpha, beta; q MGi,E ,Q MGi,p ,p MGi,C ,p MGi,D Is the equilibrium solution.
10. The microgrid-containing group shared energy storage optimal scheduling method considering uncertainty of new energy power generation according to claim 9, characterized in that: the specific solving steps of the master multi-slave game model are as follows:
step1, generating a net generating power scene around each microgrid, establishing a decision model of each individual, setting the number of particles and the maximum iteration times, and initializing the particles containing the energy storage capacity lease price information;
step2, calling each microgrid optimization model, solving the optimal energy storage leasing capacity including energy capacity and power capacity on the upper layer based on a particle swarm algorithm, and solving the charge-discharge plan of each microgrid in each period on the lower layer based on a CPLEX solver;
step3, calling an SESO optimization model, and optimizing a charge-discharge scheme based on a CPLEX solver;
step4, calculating the fitness value of each particle, updating the speed and the position of each particle, and updating the local optimum of each particle and the global optimum of the particle swarm;
and Step5, judging whether the iteration times reach the maximum iteration times, if not, returning to Step2, otherwise, outputting the operation plan of each microgrid and the SESO, the lease energy storage capacity of each microgrid and the optimal energy storage lease price.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211361933.4A CN115764938A (en) | 2022-11-02 | 2022-11-02 | Microgrid group-containing shared energy storage optimal scheduling method considering new energy power generation uncertainty |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211361933.4A CN115764938A (en) | 2022-11-02 | 2022-11-02 | Microgrid group-containing shared energy storage optimal scheduling method considering new energy power generation uncertainty |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115764938A true CN115764938A (en) | 2023-03-07 |
Family
ID=85355378
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211361933.4A Pending CN115764938A (en) | 2022-11-02 | 2022-11-02 | Microgrid group-containing shared energy storage optimal scheduling method considering new energy power generation uncertainty |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115764938A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117060402A (en) * | 2023-10-09 | 2023-11-14 | 山东浪潮数字能源科技有限公司 | Energy internet platform architecture method based on distributed smart grid |
CN117767375A (en) * | 2024-02-22 | 2024-03-26 | 山东理工大学 | shared energy storage fairness allocation strategy based on risk constraint asymmetric cooperative game |
-
2022
- 2022-11-02 CN CN202211361933.4A patent/CN115764938A/en active Pending
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117060402A (en) * | 2023-10-09 | 2023-11-14 | 山东浪潮数字能源科技有限公司 | Energy internet platform architecture method based on distributed smart grid |
CN117060402B (en) * | 2023-10-09 | 2024-01-09 | 山东浪潮数字能源科技有限公司 | Energy internet platform architecture method based on distributed smart grid |
CN117767375A (en) * | 2024-02-22 | 2024-03-26 | 山东理工大学 | shared energy storage fairness allocation strategy based on risk constraint asymmetric cooperative game |
CN117767375B (en) * | 2024-02-22 | 2024-05-14 | 山东理工大学 | Shared energy storage fairness allocation strategy based on risk constraint asymmetric cooperative game |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108599373B (en) | Cascade analysis method for transmission and distribution coordination scheduling target of high-proportion renewable energy power system | |
Chen et al. | Research on day-ahead transactions between multi-microgrid based on cooperative game model | |
CN109713673B (en) | Method for configuring and optimizing operation of grid-connected micro-grid system in electricity selling environment | |
CN115764938A (en) | Microgrid group-containing shared energy storage optimal scheduling method considering new energy power generation uncertainty | |
CN105046371A (en) | Electric vehicle charge-discharge scheduling method based on demand side bidding | |
CN113610311B (en) | Comprehensive energy service provider cooperation operation optimization method considering carbon emission reduction under double-layer cooperation architecture | |
CN112149914A (en) | Method for optimizing and configuring power market resources under multi-constraint condition | |
Zhang et al. | A master-slave game optimization model for electric power companies considering virtual power plant | |
CN115759556A (en) | Multi-virtual power plant optimized operation and distribution method in carbon-electricity market | |
CN117254501A (en) | New energy side centralized shared energy storage capacity planning method and system | |
Kong et al. | Independence enhancement of distributed generation systems by integrating shared energy storage system and energy community with internal market | |
CN110556821B (en) | Multi-microgrid double-layer optimization scheduling method considering interactive power control and bilateral bidding transaction | |
CN116611575A (en) | Multi-VPP shared energy storage capacity optimal configuration method based on double-layer decision game | |
CN117013521A (en) | ARIMA and gray model combined prediction virtual power plant electric energy transaction method and system | |
CN116228461A (en) | P2P transaction-based multi-virtual power plant random optimization scheduling method | |
CN115441494A (en) | Converter station capacity optimal configuration method and device based on flexible direct current interconnection system | |
Singh et al. | Shapley value method and stochastic Dantzig–Wolfe decomposition for decentralized scheduling of multimicrogrid | |
CN114330999A (en) | Virtual power grid resource matching method based on cloud energy storage system | |
CN113177655A (en) | Comprehensive energy system multi-main-body operation optimization method and device based on reinforcement learning | |
Spasova et al. | Energy exchange strategy for local energy markets with heterogenous renewable sources | |
Li et al. | Multiobjective Optimization Model considering Demand Response and Uncertainty of Generation Side of Microgrid | |
Liu | Research on Grid‐Connected Optimal Operation Mode between Renewable Energy Cluster and Shared Energy Storage on Power Supply Side | |
Wang et al. | Spatial-Temporal Power Balance Method for Active Distribution Systems Based on Time Series Operation Simulation with Source-Load-Storage Coordination | |
CN112270432B (en) | Energy management method of comprehensive energy system considering multi-subject benefit balance | |
He et al. | A Review and Outlook on the Participation of High Percentage of Renewable Energy in the Electricity Market |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |