Power grid planning method considering risk constraint in uncertain scene

Abstract

The invention discloses a power grid planning method considering risk constraints in an uncertain scene. According to the method, statistical historical data is divided into two parts of historical data of a limit scene and a normal scene, the historical data of the normal scene is divided into a plurality of characteristic partitions, then corresponding subsets are generated respectively, then a multi-state model is constructed, and a Markov model and a Markov chain Monte Carlo simulation method are adopted to generate a power grid uncertainty operation scene. And reducing the scene by adopting a kmeans clustering method to respectively obtain typical operation scene subsets of the power grid in a normal state and a limit state. And then establishing a model constraint and a target function, establishing a model according to the model constraint and the target function, performing power grid optimization planning by the model according to the power grid typical operation scene subset, and outputting a power grid planning result. According to the method, optimization planning of the power grid transmission line, the generator set and the renewable energy unit with known parameters is realized, and a reference basis can be provided for power grid planning decision under uncertainty.

Description

Technical Field

[0001] The present invention relates to the field of power system optimization planning, and in particular to a power grid planning method considering risk constraints under uncertain scenarios. Background Art

[0002] The proposal of the "carbon peak and carbon neutrality" goals has put forward new requirements for the construction and development of the power system. The proportion of renewable energy in the current power system continues to increase, and the load growth continues to accelerate. The uncertainty of renewable energy output and power load fluctuations brings many operational risks to the safe operation of the power grid. In order to achieve safe operation and efficient planning of the power grid, it is necessary to analyze the uncertainty characteristics of renewable energy output and power load at the corresponding time scale, and on this basis, obtain typical scenarios close to their random fluctuation laws, and then optimize the transmission network planning considering the system operation under typical scenarios. Traditional planning methods usually do not consider the seasonal differences in renewable energy output fluctuations and power load fluctuations on medium and long-term time scales, but directly perform cluster analysis on the data of the whole year to obtain typical scenarios. At the same time, the serious operation risks of the power grid are mostly in extreme scenarios such as high load state and low renewable energy output, low load state and high renewable energy output. In this context, it is also necessary to incorporate extreme scenarios into the planning model. In this context, it is necessary to take into account the time series partitioning of normal scenarios and extreme scenario analysis, and study the transmission network optimization planning method considering the analysis of uncertain operation scenarios. Summary of the invention

[0003] In view of the problems in the above-mentioned background technology, the present invention provides a power grid planning method considering risk constraints in uncertain scenarios.

[0004] The technical solution adopted by the present invention is:

[0005] The risk-constrained power grid planning method under uncertain scenarios of the present invention includes:

[0006] Step 1: According to the extreme scenario judgment criteria, extreme scenario historical data are screened out from the annual statistical historical data of renewable energy output and power load, and the remaining data are used as normal scenario historical data;

[0007] Step 2: Obtain characteristic parameters of renewable energy output and power load in different periods according to the historical data of normal scenarios, and then divide the historical data of normal scenarios into monthly units according to the characteristic parameters to obtain multiple characteristic partitions;

[0008] Step 3: According to the characteristic partitions obtained in step 2, a subset of the annual statistical historical data of renewable energy output and power load of each characteristic partition is constructed, and then a multi-state model of renewable energy output and power load of each characteristic partition is constructed, and the state transition probability between each state in each multi-state model is obtained by using the Markov model;

[0009] Step 4: Based on the multi-state model and state transition probability obtained in step 3, the Markov chain Monte Carlo simulation method is used to simulate and generate the grid uncertainty operation scenarios under each characteristic partition;

[0010] Step 5: Use the kmeans clustering method to reduce the uncertainty operation scenarios of the power grid under each characteristic partition obtained in step 4, generate a subset of typical operation scenarios of the power grid in each characteristic partition, and then obtain a subset of typical operation scenarios of the power grid in normal operation;

[0011] Step 6: Based on the extreme scenario historical data obtained in step 1, a subset of renewable energy output historical data and a subset of power load historical data under extreme scenarios are constructed respectively, and then multi-state models of renewable energy output and power load under extreme scenarios are constructed respectively, and the state transition probabilities between various states in the multi-state models of renewable energy output and power load are obtained by combining with Markov model processing;

[0012] Step 7: Based on the multi-state model of renewable energy output and power load under extreme scenarios in step 6 and the state transition probability, the Markov chain Monte Carlo simulation method is used to simulate and generate the grid uncertainty operation scenarios under extreme scenarios, and the kmeans clustering method is used to reduce the grid uncertainty operation scenarios under extreme scenarios to obtain a subset of typical grid operation scenarios under extreme operating states;

[0013] Step 8: Establish the model constraints consisting of grid operation constraints, component operation constraints, grid load shedding risk constraints and renewable energy curtailment risk constraints, construct the objective function with minimizing the total grid cost as the optimization goal, and then construct the grid optimization planning model;

[0014] Step 9: Based on the subsets of typical grid operation scenarios of typical normal operation and extreme operation obtained in steps 5 and 7, respectively, and the model constraints and objective function in step 8, a grid optimization planning model is used to perform grid optimization planning to obtain grid planning results, which include physical parameters of transmission lines, non-renewable energy generators, and renewable energy generators.

[0015] The characteristic parameters in step 2 include the mean and variance of the renewable energy output and power load in each month, and the mean and variance are obtained according to the following formula:

[0016]

[0017] Where m represents the mth natural month, m∈[1,12], D m is the number of days in the mth month, d represents the dth natural day in the mth natural month, d∈[1,D m ], T is the number of operation time periods in each operation scenario, t represents the tth time period in the dth natural day, t∈[1,T], represents the average value of renewable energy output / power load in the mth month, represents the variance of renewable energy output / electricity load in the mth month, Represents the data of renewable energy output / power load in the tth period of the dth day of the mth month.

[0018] The division in step 2 adopts the kmeans clustering method.

[0019] The multi-state model in step 3 is a state set of renewable energy output / power load, in, represents the state of the i-th renewable energy output / power load, K RES and K PL are the number of renewable energy output and power load states, respectively;

[0020] The state collection is set up according to the following steps:

[0021] In the subset, select the minimum and maximum values ​​of the annual statistical historical data of renewable energy output / power load to construct the interval Then the interval is divided into K RES / PL subintervals, state Indicates that the renewable energy output level / power load level is within the i-th interval, where Represents a subset of historical data on renewable energy output and power load in a subset.

[0022] The Markov chain Monte Carlo simulation method is used in step 4 and step 7 to simulate and generate transmission network uncertainty operation scenarios that take into account renewable energy output and power load fluctuations. Each grid uncertainty operation scenario includes a renewable energy output sequence and a power load sequence for 24 hours a day.

[0023] The kmeans clustering steps in step 4 and step 7 are:

[0024] 1) Set the number of target reduction scenarios k as the number of kmeans target clustering clusters;

[0025] 2) Select k grid uncertainty operation scenarios {C1, C2, C3, ..., C k} is used as the initial center point of clustering (initial center scene of clustering), and the distance from each scene to the initial center scene is obtained according to the following formula:

[0026]

[0027] In the formula, dis(X i ,C j ) is scene X i To the center scene C j The distance, X i =[X i,1 ,…,X i,t ,…,X i,T ], C j =[C j,1 ,…,C j,t ,…,C j,T ], where X i,t For scene X i The data of the tth dimension in C j,t Center scene C j The data of the tth dimension in , T is the total data dimension of the scene data;

[0028] 3) According to the distance from each scene to the central scene, the first clustering is performed according to the central scene closest to obtain the initial clustering result;

[0029] 4) According to the clustering results, the centers of each cluster are obtained as new clustering center scenes;

[0030] 5) Repeated iterations are performed until the changes in the cluster center scenarios meet the convergence conditions, and the final cluster center scenarios are obtained, thereby forming a subset of typical power grid operation scenarios.

[0031] The extreme scenarios in step 1 include two extreme scenarios: low power load level and high renewable energy output level, and high power load level and low renewable energy output level. The extreme scenario judgment basis is set according to the following formula:

[0032]

[0033] In the formula, and μ RES The thresholds of the mean values ​​of renewable energy output periods corresponding to high renewable energy output levels and low renewable energy output levels, and μ PL The thresholds for the load period means corresponding to high load level and low load level respectively.

[0034] In step 9, the total grid cost includes the planning cost F inc and system operating cost F ope , the objective function is set according to the following formula:

[0035] minF=min(F inc +F ope )

[0036]

[0037] In the formula, minF represents the minimum total cost, and the system planning cost F inc It mainly includes the planning costs of new transmission lines, new non-renewable energy generating units, and new renewable energy generating units. and are the planning costs of candidate non-renewable energy generator i, candidate transmission line l and candidate renewable energy generator r, respectively, g ,Ω l and Ω r represent the sets of candidate non-renewable energy generating units, candidate transmission lines and candidate renewable energy generating units respectively, and They represent the optimization decision variables of whether the candidate non-renewable energy generator g, the candidate transmission line l and the candidate renewable energy generator r are planned, η is the annual discount rate, y is the planning year, Φ is the set of characteristic partitions, is the number of running days included in the sth feature partition, is the number of grid operation scenarios in normal operation contained in the sth characteristic partition, c RC , and They represent the unit renewable energy curtailment cost of the system, the unit load removal cost of node i, and the unit power generation cost of non-renewable energy unit g, respectively. r ,Ξ b and g are the number of renewable energy units, grid nodes and non-renewable energy generating units in the grid, ρ s,ω The scene probability of the ω-th scene for the s-th feature partition.

[0038] In step 9, the model constraints include grid operation constraints, component operation constraints, grid load shedding risk constraints, and renewable energy curtailment risk constraints:

[0039] 1) Grid operation constraints include grid power flow balance constraints and grid operation upper and lower limit constraints:

[0040] 1.1) The power grid flow balance constraint is set according to the following formula:

[0041]

[0042] 1.2) The upper and lower limits of power grid operation are set according to the following formula:

[0043]

[0044] Where: l represents the set of existing lines in the power grid, P ω,ij,t is the active power flowing through line ij at time t under scenario ω, θ ω,i,t represents the power angle of node i at time t in scenario ω, x ij represents the reactance of line ij, M l is the parameter introduced when applying the big M method to route l planning decision-making. It is actually a very large number. GB ,Λ RB and Λ BL N b ×N g The non-renewable energy unit-node association matrix of dimension N b ×N r Dimensional renewable energy unit-node association matrix and N b ×N l The transmission line-node association matrix of dimension, are the output sets of non-renewable energy units and renewable energy units at time t under scenario ω, among which, represents the output power of non-renewable energy unit g at time t under scenario ω, represents the output power of renewable energy unit r at time t under scenario ω, and are the power load and power load shedding set at time t under scenario ω, is the power flow set of the transmission lines at time t under scenario ω, θ i and Respectively represent the upper and lower limits of the power angle of node i, represents the maximum allowable power flow of line ij, is the set of grid operation scenarios under typical normal operation conditions throughout the year, N extr A collection of grid operation scenarios under extreme operating conditions;

[0045] 2) Grid risk constraints include load shedding risk constraints and renewable energy curtailment risk constraints:

[0046] 2.1) Load shedding risk constraint is set according to the following formula:

[0047]

[0048]

[0049] 2.2) The risk constraint of renewable energy curtailment is set according to the following formula:

[0050]

[0051]

[0052] Where: represents the amount of power load removal at node i at time t in scenario ω, is the upper limit of the tolerable load shedding rate of node i in any scenario, is the upper limit of the expected load shedding rate that the system can tolerate in characteristic partition s under normal conditions, is the new energy curtailment power of renewable energy unit r at time t under scenario ω, is the maximum output value of renewable energy unit r at time t under scenario ω, is the upper limit of the expected renewable energy curtailment rate that the whole system can tolerate in the characteristic partition s under normal conditions;

[0053] 3) The component operation constraints are set according to the following formula:

[0054]

[0055]

[0056] Where: is the start / stop decision variable of the non-renewable energy unit g at time t under scenario ω, Indicates that the non-renewable energy unit g is in operation at time t under scenario ω, Indicates that the non-renewable energy unit g is in operation at time t+τ under scenario ω, is the start-stop decision variable of renewable energy unit r at time t under scenario ω, Indicates that the renewable energy unit r is in operation at time t under scenario ω, and are the lower and upper limits of the output of non-renewable energy unit g, and are the downward and upward ramping constraints of the non-renewable energy unit g, respectively. and are the minimum on / off time of non-renewable energy unit g respectively.

[0057] The beneficial effects of the present invention are:

[0058] 1) The present invention patent takes into account typical normal operating conditions and low-probability extreme operating conditions. For typical normal operating conditions, a characteristic analysis is proposed for the uncertainty of renewable energy output and power load on an annual time scale, and a quarterly differentiation method for time series data based on characteristic indicators is proposed. The obtained partitioning results are helpful for analyzing each quarter separately; for low-probability extreme operating conditions, corresponding judgment basis for extreme scenario historical data is proposed, and uncertain scenario analysis is performed based on extreme scenario historical data. The relevant methods can improve the precision of uncertain scenario analysis.

[0059] 2) The patent of this invention proposes an uncertainty analysis model for renewable energy output and power load based on a multi-state model and a Markov model, and further uses the Markov Monte Carlo method and the kmeans scenario clustering method to obtain a typical grid operation scenario that is close to its random fluctuation law.

[0060] 3) The patent of this invention combines economic benefits and risk management requirements, and proposes a two-stage "planning-operation" power grid optimization planning model based on a stochastic optimization method. It takes into account the planning costs related to planning and the system operation costs related to the operating status under various uncertain operating scenarios, as well as multiple constraints including system operation, component operation and system risk constraints. It can obtain the power grid optimization planning results of transmission lines and generator sets that take into account both system economy and risk management effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 is a flow chart of the method of the present invention; DETAILED DESCRIPTION

[0062] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0063] like Figure 1 As shown, the specific embodiment of the power grid planning method of the present invention includes the following steps:

[0064] Step 1: In a computer, extreme scenario historical data are screened out from the annual statistical historical data of renewable energy output and power load according to the extreme scenario judgment criteria, and the remaining data are used as normal scenario historical data;

[0065] In the present invention, output and load generally refer to power.

[0066] The extreme scenarios include two extreme scenarios: low power load level and high renewable energy output level, and high power load level and low renewable energy output level. The judgment basis of extreme scenarios is constructed according to the time period average characteristics of historical data and is set according to the following formula:

[0067] Low electricity load levels and high renewable energy output levels,

[0068] High electricity load levels and low renewable energy output levels,

[0069] In the formula, and μ RES The thresholds of the mean values ​​of renewable energy output periods corresponding to high renewable energy output levels and low renewable energy output levels, and μ PL The thresholds for the load period means corresponding to high load level and low load level respectively.

[0070] The above formula shows that when the average value of renewable energy output during a certain day is higher than And the load level period average is lower than μ PL , this day can be identified as an extreme scenario of "low load level and high renewable energy output level"; similarly, when the average renewable energy output period of a day is lower than μ RES , and the average load level during the period is higher than At this time, this day can be identified as an extreme scenario of "high load level and low renewable energy output level".

[0071] Step 2: Obtain characteristic parameters of renewable energy output and power load in different months of the year based on the historical data of normal scenarios, and then divide the historical data of normal scenarios into monthly units based on the characteristic parameters to obtain multiple characteristic partitions with large differences in distribution characteristics, each of which includes historical data of renewable energy output and power load;

[0072] The kmeans clustering method is used for the division.

[0073] The characteristic parameters include the mean and variance of renewable energy output and power load in each month. The mean and variance are obtained according to the following formula:

[0074]

[0075] Where m represents the mth natural month, m∈[1,12], D m is the number of days in the mth month, d represents the dth natural day in the mth natural month, d∈[1,D m ], T is the number of operation time periods in each operation scenario, t represents the tth time period in the dth natural day, t∈[1,T], represents the average value of renewable energy output / power load in the mth month, represents the variance of renewable energy output / electricity load in the mth month, Represents the data of renewable energy output / power load in the tth period of the dth day of the mth month.

[0076] Step 3: According to the characteristic partitions obtained in step 2, a subset of annual statistical historical data of renewable energy output and power load of each characteristic partition is constructed. The historical data of renewable energy output and power load in one characteristic partition can be divided into two subsets, namely, a subset of historical statistical data of renewable energy output in the month corresponding to the characteristic partition and a subset of historical statistical data of power load in the corresponding month. Then, a multi-state model of renewable energy output and power load of each characteristic partition is constructed, and the state transition probability between each state in each multi-state model is obtained by using the Markov model.

[0077] In the present invention, a subset constructs a multi-state model, and the multi-state model defines the renewable energy output and power load state of each characteristic partition.

[0078] The multi-state model in step 3 is a state set of renewable energy output / power load. in, represents the state of the i-th renewable energy output / power load, K RES and K PL are the number of states to which the renewable energy output and the power load can be transferred, respectively;

[0079] The state collection is set up according to the following steps:

[0080] In the subset, select the minimum and maximum values ​​of the annual statistical historical data of renewable energy output / power load to construct the interval Then the interval is divided into K RES / PL subintervals, each subinterval corresponds to a state, state Indicates that the renewable energy output level / power load level is within the i-th interval, where Represents a subset of historical data on renewable energy output and power load in a subset.

[0081] in represents the output state performance of the i-th renewable energy source) and (have in represents the i-th power load state) to represent the state performance;

[0082] Based on historical data of renewable energy output / power load power in the subset in Represent the sth renewable energy / power load historical data subset, and select the minimum and maximum values ​​in the historical data to construct the interval That is, the fluctuation range of renewable energy output / power load is divided into K RES / PL subintervals, the interval length is The i-th interval is On this basis, the i-th state in the multi-state model indicates that the renewable energy output level / power load level is within the i-th interval.

[0083] The Markov model uses Respectively represent the state probabilities of renewable energy output and power load at time t, where represents the probability that the renewable energy output at the tth moment is in the i-th state, that is, represents the probability that the power load at the tth moment is in the i-th state, that is, Further adopting the state transfer matrix represents the probability of renewable energy output / power load state transition at time t, where It represents the transition probability of renewable energy output / power load from state i to state j at time t.

[0084] The state transition satisfies the following formula:

[0085]

[0086] in, Represents the state of renewable energy output / power load at time t, Represents the state of renewable energy output / power load at time t, Indicates the state of renewable energy output / power load at time t+1;

[0087] The state transition probability in step 3 is evaluated by counting the state transition frequency and expressing the probability by frequency:

[0088]

[0089] in, represents the state transition probability of renewable energy output / power load transferring from state i to state j in period t under the sth characteristic partition, represents the frequency of the transfer of renewable energy output / power load from state i to state j in period t under the sth characteristic partition, K represents the frequency of the transfer of renewable energy output / power load from state i to state k in period t under the sth characteristic partition. RES and K PLThe number of states that the renewable energy output and the power load can be transferred to, that is, the total number of states of the renewable energy output multi-state model and the power load multi-state model

[0090] Step 4: Based on the multi-state model and state transition probability obtained in step 3, the Markov chain Monte Carlo simulation method is used to simulate and generate the transmission network uncertainty operation scenarios reflecting the fluctuations of renewable energy output and power load under various characteristic partitions;

[0091] Step 5: Use the kmeans clustering method to reduce the uncertainty operation scenarios of the transmission network under each characteristic partition obtained in step 4, generate a subset of typical operation scenarios of the transmission network for each characteristic partition, and then obtain a subset of typical operation scenarios of the power grid in a typical normal operating state;

[0092] Step 6: Based on the extreme scenario historical data obtained in step 1, a subset of renewable energy output historical data and a subset of power load historical data under extreme scenarios are constructed respectively, and then multi-state models of renewable energy output and power load under extreme scenarios are constructed respectively, and the state transition probabilities between various states in the multi-state models of renewable energy output and power load are obtained by combining with Markov model processing;

[0093] Step 7: Based on the multi-state model of renewable energy output and power load under extreme scenarios in step 6 and the state transition probability, the Markov chain Monte Carlo simulation method is used to simulate and generate grid uncertainty operation scenarios under extreme scenarios, and the kmeans clustering method is used to reduce the grid uncertainty operation scenarios under extreme scenarios to obtain a subset of typical grid operation scenarios under extreme operating states. The probability of grid uncertainty operation scenarios under extreme scenarios being selected by the model is extremely small.

[0094] The Markov Chain Monte Carlo simulation method is used in steps 4 and 7 to simulate and generate transmission network uncertainty operation scenarios that take into account renewable energy output and power load fluctuations. Each grid uncertainty operation scenario includes a renewable energy output sequence and a power load sequence for 24 hours a day.

[0095] And the kmeans clustering method steps in step 4 and step 7 are as follows:

[0096] 1) Set the number of target reduction scenarios k as the number of kmeans target clustering clusters;

[0097] 2) Select k grid uncertainty operation scenarios {C1, C2, C3, ..., C k} is used as the initial center point of clustering (initial center scene of clustering), and the distance from each scene to the initial center scene is obtained according to the following formula:

[0098]

[0099] In the formula, dis(X i ,C j ) is scene X i To the center scene C j The distance, X i =[X i,1 ,…,X i,t ,…,X i,T ], C j =[C j,1 ,…,C j,t ,…,C j,T ], where X i,t For scene X i The data of the tth dimension in C j,t Center scene C j The data of the tth dimension in , T is the total data dimension of the scene data;

[0100] 3) According to the distance from each scene to the central scene, the first clustering is performed according to the central scene closest to obtain the initial clustering result;

[0101] 4) According to the clustering results, the centers of each cluster are obtained as new clustering center scenes;

[0102] 5) Repeated iterations are performed until the changes in the cluster center scenarios meet the convergence conditions, and the final cluster center scenarios are obtained, thereby forming a subset of typical power grid operation scenarios.

[0103] The s-th feature partition is generated typical scenes, the probability of each typical scene is equal to the number of original scenes contained in the class corresponding to it as the cluster center, and the sum of the scene probabilities is 1.

[0104]

[0105] Where: s,ω is the probability of the ωth typical scene in the sth feature partition.

[0106] Step 8: Establish the model constraints consisting of grid operation constraints, component operation constraints, grid load shedding risk constraints and renewable energy curtailment risk constraints, and construct the objective function with the optimization goal of minimizing the total grid cost including operation and planning. Then, construct a two-stage "planning-operation" grid optimization planning model based on stochastic optimization, and use the grid optimization planning model to obtain the corresponding planned grid result when the objective function is minimized.

[0107] Step 9: Based on the subsets of typical grid operation scenarios of typical normal operation status and extreme operation status obtained in steps 5 and 7 respectively, and the model constraints and objective function in step 8, a grid optimization planning model is used to perform grid optimization planning to obtain a transmission network planning result including transmission lines and generator sets. As a grid planning method, the grid planning result includes physical parameters including transmission lines and generator sets.

[0108] The total grid cost includes planning cost F inc and system operating cost F ope The objective function is to minimize the total cost in a typical operation scenario. The objective function is set according to the following formula, where the optimization variables include: the decision variable of whether the candidate renewable energy unit is planned The decision variables for whether the candidate non-renewable energy generator is planned Decision variables for whether a candidate transmission line is planned And the grid operation state variables under each operation scenario, where the objective function and is the optimization variable to be determined, and the other parameters are known parameters:

[0109] minF=min(F inc +F ope )

[0110]

[0111] In the formula, minF represents the minimum total cost, and the system planning cost F inc It mainly includes the planning costs of new transmission lines, new non-renewable energy generating units, and new renewable energy generating units. and are the planning costs of candidate non-renewable energy generator i, candidate transmission line l and candidate renewable energy generator r, respectively, g ,Ω l and Ω r represent the sets of candidate non-renewable energy generating units, candidate transmission lines and candidate renewable energy generating units respectively, and They represent the optimization decision variables of whether the candidate non-renewable energy generator g, the candidate transmission line l and the candidate renewable energy generator r are planned, and they are all 0-1 variables, that is, Indicates that the candidate non-renewable energy generator g is planned, Indicates that the candidate transmission line l is planned, Indicates that the candidate renewable energy unit r is planned. The candidate renewable energy unit is the renewable energy unit to be planned, the newly added renewable energy unit is the renewable energy unit planned in the optimization decision of the candidate renewable energy unit, and the newly added transmission line, newly added non-renewable energy generation unit, candidate transmission line and candidate non-renewable energy generation unit are explained in the same way. η is the annual discount rate, y is the planning year, Φ is the set of characteristic partitions, is the number of running days included in the sth feature partition, is the number of grid operation scenarios in normal operation contained in the sth characteristic partition, c RC , and They represent the unit renewable energy curtailment cost of the system, the unit load removal cost of node i, and the unit power generation cost of non-renewable energy unit g, respectively. r ,Ξ b and g are the number of renewable energy units, grid nodes and non-renewable energy generating units in the grid, ρ s,ω The scene probability of the ω-th scene for the s-th feature partition.

[0112] The model constraints include grid operation constraints, component operation constraints, grid load shedding risk constraints, and renewable energy curtailment risk constraints:

[0113] 1) Grid operation constraints include grid power flow balance constraints and grid operation upper and lower limit constraints:

[0114] 1.1) The power grid flow balance constraint is set according to the following formula:

[0115]

[0116] 1.2) The upper and lower limits of power grid operation are set according to the following formula:

[0117]

[0118] Where: l represents the set of existing lines in the power grid, P ω,ij,t is the active power flowing through line ij at time t under scenario ω, θ ω,i,t represents the power angle of node i at time t in scenario ω, x ij represents the reactance of line ij, M l is the parameter introduced when applying the big M method to route l planning decision-making. It is actually a very large number. GB ,Λ RB and Λ BL N b ×N g The non-renewable energy unit-node association matrix of dimension Nb ×N r Dimensional renewable energy unit-node association matrix and N b ×N l The transmission line-node association matrix of dimension, are the output sets of non-renewable energy units and renewable energy units at time t under scenario ω, among which, represents the output power of non-renewable energy unit g at time t under scenario ω, represents the output power of renewable energy unit r at time t under scenario ω, and are the power load and power load shedding set at time t under scenario ω, P ω,t =[P ω,1i,t ,…,P ω,ji,t ] is the power flow set of the transmission lines at time t under scenario ω, θ i and Respectively represent the upper and lower limits of the power angle of node i, represents the maximum allowable power flow of line ij, is the set of grid operation scenarios under typical normal operation conditions throughout the year, N extr A collection of grid operation scenarios under extreme operating conditions;

[0119] 2) Grid risk constraints include load shedding risk constraints and renewable energy curtailment risk constraints:

[0120] 2.1) Load shedding risk constraint is set according to the following formula:

[0121]

[0122] 2.2) The risk constraint of renewable energy curtailment is set according to the following formula:

[0123]

[0124]

[0125] Where: represents the amount of power load removal at node i at time t in scenario ω, is the upper limit of the tolerable load shedding rate of node i in any scenario, is the upper limit of the expected load shedding rate that the system can tolerate in characteristic partition s under normal conditions, is the new energy curtailment power of renewable energy unit r at time t under scenario ω, is the maximum output value of renewable energy unit r at time t under scenario ω, is the upper limit of the expected renewable energy curtailment rate that the whole system can tolerate in the characteristic partition s under normal conditions;

[0126] 3) The component operation constraints are set according to the following formula:

[0127]

[0128] Where: is the start / stop decision variable of the non-renewable energy unit g at time t under scenario ω, which is a 0-1 variable. Indicates that the non-renewable energy unit g is in operation at time t under scenario ω, Indicates that the non-renewable energy unit g is in operation at time t+τ under scenario ω, is the start / stop decision variable of renewable energy unit r at time t under scenario ω, which is a 0-1 variable. Indicates that the renewable energy unit r is in operation at time t under scenario ω, and are the lower and upper limits of the output of non-renewable energy unit g, and are the downward and upward ramping constraints of the non-renewable energy unit g, respectively. and are the minimum on / off time of non-renewable energy unit g respectively.

[0129] The above specific implementation modes are used to explain the present invention rather than to limit the present invention. Any modification and change made to the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.

[0130] The above description is only a preferred embodiment of the present invention, so all equivalent changes or modifications made according to the structure, characteristics and principles described in the scope of the patent application of the present invention are included in the scope of the patent application of the present invention.

Claims

1. A power grid planning method considering risk constraints under uncertain scenarios, characterized in that: The steps include: Step 1: According to the extreme scenario judgment criteria, extreme scenario historical data are screened out from the annual statistical historical data of renewable energy output and power load, and the remaining data are used as normal scenario historical data; Step 2: Obtain characteristic parameters of renewable energy output and power load in different periods according to the historical data of normal scenarios, and then divide the historical data of normal scenarios into monthly units according to the characteristic parameters to obtain multiple characteristic partitions; Step 3: According to the characteristic partitions obtained in step 2, a subset of the annual statistical historical data of renewable energy output and power load of each characteristic partition is constructed, and then a multi-state model of renewable energy output and power load of each characteristic partition is constructed, and the state transition probability between each state in each multi-state model is obtained by using the Markov model; Step 4: Based on the multi-state model and state transition probability obtained in step 3, the Markov chain Monte Carlo simulation method is used to simulate and generate the grid uncertainty operation scenarios under each characteristic partition; Step 5: Use the kmeans clustering method to reduce the uncertainty operation scenarios of the power grid under each characteristic partition obtained in step 4, generate a subset of typical operation scenarios of the power grid in each characteristic partition, and then obtain a subset of typical operation scenarios of the power grid in normal operation; Step 6: Based on the extreme scenario historical data obtained in step 1, a subset of renewable energy output historical data and a subset of power load historical data under extreme scenarios are constructed respectively, and then multi-state models of renewable energy output and power load under extreme scenarios are constructed respectively, and the state transition probabilities between various states in the multi-state models of renewable energy output and power load are obtained by combining with Markov model processing; Step 7: Based on the multi-state model of renewable energy output and power load under extreme scenarios in step 6 and the state transition probability, the Markov chain Monte Carlo simulation method is used to simulate and generate the grid uncertainty operation scenarios under extreme scenarios, and the kmeans clustering method is used to reduce the grid uncertainty operation scenarios under extreme scenarios to obtain a subset of typical grid operation scenarios under extreme operating states; Step 8: Establish the model constraints consisting of grid operation constraints, component operation constraints, grid load shedding risk constraints and renewable energy curtailment risk constraints, construct the objective function with minimizing the total grid cost as the optimization goal, and then construct the grid optimization planning model; Step 9: Based on the subsets of typical grid operation scenarios of typical normal operation and extreme operation obtained in steps 5 and 7, respectively, and the model constraints and objective function in step 8, a grid optimization planning model is used to perform grid optimization planning to obtain grid planning results, which include physical parameters of transmission lines, non-renewable energy generators, and renewable energy generators.

2. A power grid planning method considering risk constraints under uncertain scenarios according to claim 1, characterized in that: The characteristic parameters in step 2 include the mean and variance of the renewable energy output and power load in each month, and the mean and variance are obtained according to the following formula: Where m represents the mth natural month, m∈[1,12], D m is the number of days in the mth month, d represents the dth natural day in the mth natural month, d∈[1,D m ], T is the number of operation time periods in each operation scenario, t represents the tth time period in the dth natural day, t∈[1,T], represents the average value of renewable energy output / power load in the mth month, represents the variance of renewable energy output / electricity load in the mth month, Represents the data of renewable energy output / power load in the tth period of the dth day of the mth month.

3. The method for power grid planning considering risk constraints under uncertain scenarios according to claim 1, characterized in that: The division in step 2 adopts the kmeans clustering method.

4. The method for power grid planning considering risk constraints in uncertain scenarios according to claim 1, characterized in that: The multi-state model in step 3 is a state set of renewable energy output / power load, in, represents the state of the i-th renewable energy output / power load, K RES and K PL are the number of renewable energy output and power load states, respectively; The state collection is set up according to the following steps: In the subset, select the minimum and maximum values ​​of the annual statistical historical data of renewable energy output / power load to construct the interval Then the interval is divided into K RES / PL subintervals, state Indicates that the renewable energy output level / power load level is within the i-th interval, where Represents a subset of historical data on renewable energy output and power load in a subset.

5. The method for power grid planning considering risk constraints in uncertain scenarios according to claim 1, characterized in that: The Markov chain Monte Carlo simulation method is used in step 4 and step 7 to simulate and generate transmission network uncertainty operation scenarios that take into account renewable energy output and power load fluctuations. Each grid uncertainty operation scenario includes a renewable energy output sequence and a power load sequence for 24 hours a day.

6. The method for power grid planning considering risk constraints under uncertain scenarios according to claim 1, characterized in that: The kmeans clustering steps in step 4 and step 7 are: 1) Set the number of target reduction scenarios k as the number of kmeans target clustering clusters; 2) Select k grid uncertainty operation scenarios {C1, C2, C3, ..., C k } is used as the initial center point of clustering (initial center scene of clustering), and the distance from each scene to the initial center scene is obtained according to the following formula: In the formula, dis(X i ,C j ) is scene X i To the center scene C j The distance, X i =[X i,1 ,…,X i,t, ,X i,T ], C j =[C j,1 ,…,C j,t ,…,C j,T ], where X i,t For scene X i The data of the tth dimension in C j,t Center scene C j The data of the tth dimension in , T is the total data dimension of the scene data; 3) According to the distance from each scene to the central scene, the first clustering is performed according to the central scene closest to obtain the initial clustering result; 4) According to the clustering results, the centers of each cluster are obtained as new clustering center scenes; 5) Repeated iterations are performed until the changes in the cluster center scenarios meet the convergence conditions, and the final cluster center scenarios are obtained, thereby forming a subset of typical power grid operation scenarios.

7. The method for power grid planning considering risk constraints in uncertain scenarios according to claim 1, characterized in that: The extreme scenarios in step 1 include two extreme scenarios: low power load level and high renewable energy output level, and high power load level and low renewable energy output level. The extreme scenario judgment basis is set according to the following formula: In the formula, and μ RES The thresholds of the mean values ​​of renewable energy output periods corresponding to high renewable energy output levels and low renewable energy output levels, and μ PL The thresholds for the load period means corresponding to high load level and low load level respectively.

8. The method for power grid planning considering risk constraints in uncertain scenarios according to claim 1, characterized in that: In step 9, the total grid cost includes the planning cost F inc and system operating cost F ope , the objective function is set according to the following formula: minF=min(F inc +F ope ) In the formula, minF represents the minimum total cost, and the system planning cost F inc It mainly includes the planning costs of new transmission lines, new non-renewable energy generating units, and new renewable energy generating units. and are the planning costs of candidate non-renewable energy generator i, candidate transmission line l and candidate renewable energy generator r, respectively, g ,Ω l and Ω r represent the sets of candidate non-renewable energy generating units, candidate transmission lines and candidate renewable energy generating units respectively, and They represent the optimization decision variables of whether the candidate non-renewable energy generator g, the candidate transmission line l and the candidate renewable energy generator r are planned, η is the annual discount rate, y is the planning year, Φ is the set of characteristic partitions, is the number of running days included in the sth feature partition, is the number of grid operation scenarios in normal operation contained in the sth characteristic partition, c RC , and They represent the unit renewable energy curtailment cost of the system, the unit load removal cost of node i, and the unit power generation cost of non-renewable energy unit g, respectively. r ,Ξ b and g are the number of renewable energy units, grid nodes and non-renewable energy generating units in the grid, ρ s,ω The scene probability of the ω-th scene for the s-th feature partition.

9. The method for power grid planning considering risk constraints in uncertain scenarios according to claim 1, characterized in that: In step 9, the model constraints include grid operation constraints, component operation constraints, grid load shedding risk constraints, and renewable energy curtailment risk constraints: 1) Grid operation constraints include grid power flow balance constraints and grid operation upper and lower limit constraints: 1.1) The power grid flow balance constraint is set according to the following formula: 1.2) The upper and lower limits of power grid operation are set according to the following formula: Where: l represents the set of existing lines in the power grid, P ω,ij,t is the active power flowing through line ij at time t under scenario ω, θ ω,i,t represents the power angle of node i at time t in scenario ω, x ij represents the reactance of line ij, M l is the parameter introduced when applying the big M method to route l planning decision-making. It is actually a very large number. GB ,Λ RB and Λ BL N b ×N g The non-renewable energy unit-node association matrix of dimension N b ×N r Dimensional renewable energy unit-node association matrix and N b ×N l The transmission line-node association matrix of dimension, are the output sets of non-renewable energy units and renewable energy units at time t under scenario ω, among which, represents the output power of non-renewable energy unit g at time t under scenario ω, represents the output power of renewable energy unit r at time t under scenario ω, and are the power load and power load shedding set at time t under scenario ω, P ω,t =[P ω,1i,t ,…,P ω,ji,t ] is the power flow set of the transmission lines at time t under scenario ω, θ i and Respectively represent the upper and lower limits of the power angle of node i, represents the maximum allowable power flow of line ij, is the set of grid operation scenarios under typical normal operation conditions throughout the year, N extr A collection of grid operation scenarios under extreme operating conditions; 2) Grid risk constraints include load shedding risk constraints and renewable energy abandonment risk constraints: 2.1) Load shedding risk constraint is set according to the following formula: 2.2) The risk constraint of renewable energy curtailment is set according to the following formula: Where: represents the amount of power load removal at node i at time t in scenario ω, is the upper limit of the tolerable load shedding rate of node i in any scenario, is the upper limit of the expected load shedding rate that the system can tolerate in characteristic partition s under normal conditions, is the new energy curtailment power of renewable energy unit r at time t under scenario ω, is the maximum output value of renewable energy unit r at time t under scenario ω, is the upper limit of the expected renewable energy curtailment rate that the whole system can tolerate in the characteristic partition s under normal conditions; 3) The component operation constraints are set according to the following formula: Where: is the start / stop decision variable of the non-renewable energy unit g at time t under scenario ω, Indicates that the non-renewable energy unit g is in operation at time t under scenario ω, Indicates that the non-renewable energy unit g is in operation at time t+τ under scenario ω, is the start-stop decision variable of renewable energy unit r at time t under scenario ω, Indicates that the renewable energy unit r is in operation at time t under scenario ω, and are the lower and upper limits of the output of non-renewable energy unit g, and are the downward and upward ramping constraints of the non-renewable energy unit g, respectively. and are the minimum on / off time of non-renewable energy unit g respectively.

Images