A fast solution method and system for the dynamic partition model of power grid based on single-stage robustness

Abstract

The present invention provides a method and system for quickly solving a dynamic partitioning model of a power grid based on single-stage robustness, the method comprising: obtaining and constructing an objective function of dynamic partitioning of a power grid according to uncertainty information of new energy output, and obtaining a single-stage robust optimization model; converting the single-stage robust optimization model into a two-layer optimization model; using a constraint generation method CG to obtain and solve a partition optimization robust upper-layer model of a preset partition recovery time, decomposing the partition optimization robust upper-layer model, and obtaining a main problem and a sub-problem; using a column and constraint generation method C&CG to solve a partition approximate recovery lower-layer model, and obtaining an applicable partitioning scheme; obtaining and constructing a partition evaluation index system according to structural information and functional information, and evaluating the effectiveness of an applicable partitioning scheme. The present invention solves the technical problems of using a black start power supply as a partition starting power supply for power grid partitioning, such as low feasibility, slow solution speed, and low applicability to scenarios with large system scales.

Description

Technical Field

[0001] The present invention relates to the technical field of solving power grid partition optimization models, and in particular to a method and system for quickly solving a power grid dynamic partition model based on a single-stage robust method. Background Art

[0002] With the increase in the number of grid-connected new energy and energy storage stations, the number of power sources with black start capability in the power grid is increasing. If these power sources are used as black start power sources, the power grid can be restored in parallel after a large-scale power outage, which can quickly improve the recovery speed of the power grid, thereby greatly reducing the losses caused by large-scale power outages. Since the rationality of the partition directly affects the recovery process of the power grid, accordingly, how to use the black start power source as the starting power source for the partition to achieve reasonable partitioning of the power grid has become an important research direction.

[0003] Commonly used methods include: partitioning method based on network theory; partitioning method based on intelligent evolution and optimization algorithm; partitioning method based on 0-1 mixed integer programming. Among the methods based on network theory, for example, the existing invention patent application document "A Capacity Feasible Domain Prediction Method, System and Medium for Inter-provincial Power Trading" with publication number CN117745339A, the existing method includes: constructing an objective function based on the operation data of the inter-provincial power system with the goal of minimizing the power dispatching cost in the first two days of power trading; constructing a first constraint and a second constraint based on the operation data of the inter-provincial power system; solving the objective function based on the actual operation mode of the inter-provincial power system and the first constraint to obtain a first calculation result, and then judging the first calculation result in combination with the second constraint to obtain the capacity feasible domain of the inter-provincial power system; adjusting the power dispatching of the inter-provincial power system based on the capacity feasible domain. However, most of the above-mentioned existing schemes only consider the power balance constraint but fail to consider the constraints of safe operation of the power grid, such as the zoning regulation capacity constraint, which makes the actual feasibility of the zoning scheme low; the model built by the zoning method based on the intelligent evolutionary optimization algorithm can meet a variety of zoning goals, such as the existing invention patent application document "New Energy Zoning Layout Optimization Method Considering Spatiotemporal Characteristics" with publication number CN116316854A, the existing method includes: establishing a spatiotemporal characteristic characterization model for each zone, taking the average photovoltaic output in each period of a typical month as the equivalent output at the whole point; establishing a new energy zoning layout optimization model considering spatiotemporal characteristics, and its constraints include new energy scale constraints, power abandonment rate constraints, power balance constraints, inter-zone interconnection line power constraints and unit output constraints; the objective function is the maximum function of the equivalent annual investment rate of return of the newly added wind power and photovoltaic power generation installed capacity; solving the new energy zoning layout optimization model considering spatiotemporal characteristics through a genetic algorithm, outputting the newly added zoning photovoltaic capacity, the newly added zoning wind power capacity and the newly added wind and solar space layout. However, the disadvantage of the above existing scheme is that the solution speed is slow and it is only suitable for smaller-scale system partitions; the partition method based on 0-1 mixed integer programming has a relatively complete mathematical programming theory, such as the existing invention patent application document "Distribution network operation optimization method and related device based on optimization acceleration" with publication number CN118554432A. The existing method includes: establishing a distribution network operation optimization model, and converting the nonlinear continuous constraints of the distribution network operation optimization model into a solvable form of mixed integer convex optimization; converting variables that can be solved by a reinforcement learning algorithm from the distribution network operation optimization model; based on variables and constraints, solving the converted variables by a reinforcement learning algorithm, substituting the variable values ​​into the distribution network operation optimization model, and obtaining the optimization results. However, the above existing technology has the disadvantage of only constructing an approximate model, and because the constructed model is a mixed integer nonlinear optimization model, when the system scale is expanded, it cannot be guaranteed that the optimal solution of the system will be obtained within a given time.

[0004] In summary, the existing scheme of partitioning the power grid using the black start power supply as the starting power supply of the partition has technical problems such as low feasibility, slow solution speed and low applicability to scenarios with large system scale. Summary of the invention

[0005] The technical problem to be solved by the present invention is: how to solve the technical problems of low feasibility, slow solution speed and low applicability to scenarios with large system scale in the existing scheme of partitioning the power grid using black start power supply as the starting power supply of the partition.

[0006] The present invention adopts the following technical solutions to solve the above technical problems: A fast solution method based on a single-stage robust power grid dynamic partition model includes:

[0007] S1. Obtain and construct the objective function of dynamic grid partitioning based on the uncertainty information of renewable energy output, and obtain a single-stage robust optimization model;

[0008] S2. Convert the single-stage robust optimization model into a two-layer optimization model, wherein the two-layer optimization model includes: an upper model and a lower model, and use the upper model to process the preset node outage time. The partition optimization problem is solved by using the lower-level model to process the sub-partition approximate recovery problem of a given partition, and the load node recovery time is obtained by processing. The partition system recovery robust optimization model is defined, and the system recovery process is divided into time period;

[0009] S3. Using the constraint generation method CG (Constraint Generate), according to the partition system recovery robust optimization model, obtain and solve the partition optimization robust upper model with preset partition recovery time, decompose the partition optimization robust upper model, and obtain the main problem and sub-problems;

[0010] S4. Use the Column and Constraint Generate (C&CG) method to solve the partition of the established partition scheme according to the main problem and sub-problems to approximate and restore the underlying model and obtain an applicable partition scheme.

[0011] S5. Obtain and construct a zoning evaluation index system based on structural and functional information to evaluate the effectiveness of applicable zoning schemes.

[0012] The present invention proposes a modeling method, a solution strategy and a partition evaluation index system for a dynamic partition of a power grid. A single-stage robust optimization model is established to minimize power outage losses, partition decoupling degree and balance degree; when solving, the single-stage robust optimization model is converted into a two-layer optimization model, which is solved by the constraint generation method CG and the column and constraint generation method C&CG respectively; and a partition evaluation index system is constructed to quantify the advantages and disadvantages of applicable partition schemes. The partition where each black start power source is located includes the network, load, the started power source, and the new energy station that does not have black start capability to which the partition belongs. The partition operation of the present invention has strong adaptability for promotion.

[0013] In a more specific technical solution, S1 includes:

[0014] S11. Establish an objective function. According to the uncertainty information of new energy output, in the scenario of abnormal fluctuation of new energy, adjust the node and line ownership, and minimize the power outage loss, interconnection line transmission power and restoration time difference of each sub-area. The objective function is constructed using the following logic:

[0015] (1)

[0016] In the formula, Indicates i Bus node pair sub-partition k An integer variable representing the attributed state of Indicates line l For k A continuous variable for the belonging status of each sub-division; For the i The active load at each bus node is: Indicates line l The active load, d Indicates load; For new energy units r The predicted output of the new energy units connected to the corresponding nodes, Indicates i Bus nodes in sub-partitions k The power outage time, Indicates k The time required for the sub-partition to complete the system recovery; Indicates m The time required for the sub-partition to complete the system recovery; Indicates the total number of nodes; and represents the total number of lines; Indicates the number of black start power supplies; is the weight of the first objective function; is the weight of the second objective function; is the weight of the third objective function;

[0017] S12. Define partition power balance constraints, reactive power regulation reserve capacity constraints, active power reserve capacity constraints, unit start-up power constraints, node and branch partition constraints, subsystem network connectivity constraints and uncertainty sets.

[0018] The present invention establishes an objective function for dynamic grid partitioning, taking into account the uncertainty of renewable energy output, and minimizes the power outage losses of each sub-division, the transmission power of the interconnection lines, and the recovery time difference of each division by adjusting the node and line ownership in the worst renewable energy fluctuation scenario.

[0019] In a more specific technical solution, S12 also includes:

[0020] S121. Use the following logic to express the partition power balance constraint and constrain the power imbalance of each partition to a preset threshold range:

[0021] (2)

[0022] In the formula, is the maximum power imbalance allowed for each partition, Indicates i The unit is connected to the bus node g The power rating of For new energy units r The predicted output of the new energy units connected to the corresponding nodes, is a double line item, Indicates i The busbar node is k An integer variable containing the ownership status of the subpartitions;

[0023] Adopting the Big M method and introducing new energy units r The corresponding node is k Auxiliary continuous variables corresponding to sub-partitions , transform equation (2) into the following constraint:

[0024] (3)

[0025] (4)

[0026] (5)

[0027] In the formula, M represents the large number in the big-M method;

[0028] S122. Use the following logic to express the reactive regulation reserve capacity constraint, provide reactive regulation reserve capacity for the preset system, and perform voltage control:

[0029] (6)

[0030] (7)

[0031] In the formula, Busbar node i Access unit g The reactive power output limit, Busbar node i Access unit g The lower limit of reactive power output; For the i The upper limit of reactive power output of each bus node connected to reactive power source; For the i The lower limit of reactive power output when a bus node is connected to a reactive power source; For the k Reactive power of sub-areas Q positive spare capacity requirements; For the k Reactive power of sub-areas Q negative spare capacity requirements; Indicates i The reactive load of a bus node, d Indicates load;

[0032] S123, expressing active reserve capacity constraints, and providing active regulation reserve capacity for each partition;

[0033] S124, express the unit startup power constraint, in the k A preset number of non-black start units are reserved in each sub-area j , for starting operation;

[0034] S125, construct and use node and branch partition constraints to make the node subordinate to the k sub-partitions;

[0035] S126. Based on the network flow theory, the subsystem network connectivity constraints are constructed, wherein the black start unit node is defined as the system source point, the unit to be started and the load node are defined as the sink point, and when the preset network meets the network flow constraint conditions, the connectivity of the obtained sub-area is guaranteed; wherein, the integer variable is defined using the following logic For the k Lines in sub-areas l The virtual flow on the line is set to flow from the node with a smaller node number to the node with a larger node number as the positive direction of the line virtual flow:

[0036] (15)

[0037] (16)

[0038] (17)

[0039] (18)

[0040] (19)

[0041] In the formula, l(m,i) Indicates the end point is i The branch of a bus node, m The preceding node of the endpoint. l(i,n) Indicates the starting point is i The branch of a bus node, n represents the subsequent node of the starting point, Indicates the complete set, Represents the set of nodes connected to the black start unit, Indicates the set of nodes that are not connected to the black start unit. NB Indicates that no black start unit is connected. B Indicates black start unit, Indicates line l For k A continuous variable for the belonging status of each sub-division;

[0042] S127. According to the uncertainty information of new energy output, an uncertainty set is expressed, wherein the uncertainty information of new energy output satisfies the following uncertainty set U :

[0043] (20)

[0044] In the formula, NW is the total number of new energy stations, Indicates the capacity demand of new energy stations; For new energy units r Expected output value; Represents new energy units r Output fluctuation limit; Indicates uncertain budget value, Represents the node set of new energy units, WT Represents a wind farm.

[0045] In the present invention, for k For each sub-area, the starting power of at least one started unit should be less than 70% of the upper limit of the black start unit output in the sub-area, so as to ensure that at least one non-black start unit in the sub-area can be successfully started.

[0046] The present invention constructs subsystem network connectivity constraints based on network flow theory. Traffic is injected from the system source point, and the traffic reaches each node through the line. All nodes consume unit traffic. When each node has traffic arriving, the connectivity of the network can be ensured.

[0047] In a more specific technical solution, in S123, the active reserve capacity constraint of the active regulation reserve capacity is expressed using the following logic:

[0048] (8)

[0049] (9)

[0050]

[0051] In the formula, Indicates i Units with bus nodes g The power, Indicates i Units with bus nodes g The power limit, Indicates i Units with bus nodes g The lower power limit; Indicates the units in the system G the total number of Indicates the total number of loads in the system; Indicates k Units in sub-areas g Active positive reserve factor; Indicates k Units in sub-areas g Active negative reserve factor; Indicates k Sub-partition load d The demand positive reserve factor; Indicates k Sub-partition load d Negative reserve factor of demand; Indicates the number of black start power supplies; For the i The active load at each bus node is: Indicates i Bus nodes in sub-partitions k The corresponding integer variable.

[0052] In a more specific technical solution, in S124, the following logic is used to express the unit startup power constraint:

[0053] (10)

[0054] In the formula, Indicates j Starting power of the started units, st Indicates startup. Indicates the first unit that is not connected to the black start unit. j The unit started is k Integer variables corresponding to subpartitions, Indicates i Units with bus nodes g The power limit, Indicates existence.

[0055] In a more specific technical solution, in S125, the following formula is used to express the node and branch partition constraints:

[0056] (11)

[0057] (12)

[0058] (13)

[0059] (14)

[0060] In the formula, represents the number of black start power supplies, where (12) indicates that each node belongs to one partition; (13) indicates that each branch belongs to at most one partition. When a branch does not belong to any partition, it means that the branch is a tie line branch; (14) indicates that there is a black start power supply node in each area.

[0061] In a more specific technical solution, S2 includes:

[0062] S21. Use the following matrix logic to express and construct the upper-level model:

[0063] (twenty one)

[0064] (twenty two)

[0065] (twenty three)

[0066] In the formula, x is a 0-1 integer variable; y is a continuous variable, u represents a random variable; is the 0-1 integer variable corresponding to the objective function x The coefficient matrix of Represents a 0-1 integer variablex , continuous variables y and random variables u The inequality constraint corresponds to equation (3): is the first corresponding coefficient matrix, is the second corresponding coefficient matrix, is the third corresponding coefficient matrix; Indicates a variable containing 0-1 integers x , continuous variables y The inequality constraints correspond to equations (4) to (5). is the fourth corresponding coefficient matrix, is the fifth corresponding coefficient matrix, d For load; Represents a 0-1 integer variable x The equality constraints correspond to equations (11), (13), (18) to (19); Represents a 0-1 integer variable x The inequality constraints correspond to equations (6) to (10), (12), (14) to (17). is the sixth corresponding coefficient matrix, h represents the thermal parameters, Represents new energy units r The corresponding node is k Auxiliary continuous variables corresponding to sub-partitions, Indicates black start unit b The corresponding node is k Auxiliary continuous variables corresponding to the subpartitions.

[0067] S22, let the length of each period be The total recovery time is , set the black start unit in the partition and the objective function of the node where the black start unit is located to recover at time 0:

[0068] (twenty four)

[0069] (25)

[0070] In formula (24), Indicates k In the sub-partition i The state of a bus node in time period s, Indicates k In the sub-partition i The state of a load node in time period s, Represents a partition k Middle i The state of each unit node in time period s; Indicates j The units started and i Lines l The recovery state in period s; Indicates the expected scenario k In the sub-partition i Unit g The active output of the corresponding node; Indicates the expected scenario k In the sub-partition i Unit g The reactive power output of the corresponding node; Indicates the expected scenario k In the sub-partition i New energy units r The actual active output; Indicates k of the subpartitions i Load d The amount of load that the corresponding node needs to recover; Indicates i Load d The weight coefficient corresponding to the corresponding node; Indicates the expected scenario k In the sub-partition i Unit g The active output adjustment amount of the corresponding node; Indicates the expected scenario k In the sub-partition i Unit g The reactive power output adjustment amount of the corresponding node; Indicates the expected scenario k In the sub-partition i New energy units r The actual active output size adjustment amount; Indicates k of the subpartitions i Load d The corresponding node recovery load size adjustment; Indicates k The number of busbar nodes in the sub-partitions, represents the time state set, Indicates k of the subpartitions i The time corresponding to each node, Indicates time T The corresponding total recovery period.

[0071] In a more specific technical solution, in S3, the following logic is used to express the main problem:

[0072] (26)

[0073] (27)

[0074] (28)

[0075] (29)

[0076] (30)

[0077] In the formula, x is a 0-1 integer variable; y is a continuous variable, u represents a random variable; is the 0-1 integer variable corresponding to the objective function x The coefficient matrix of is the first corresponding coefficient matrix, is the second corresponding coefficient matrix, is the third corresponding coefficient matrix; Indicates a variable containing 0-1 integers x , continuous variables y The inequality constraint, is the fourth corresponding coefficient matrix, is the fifth corresponding coefficient matrix, d For load; g Indicates the unit; Indicates that it only contains 0-1 integer variables x The inequality constraint, is the sixth corresponding coefficient matrix, h represents the thermal parameters, represents the set of scenarios that violate constraint (27) to the highest degree;

[0078] Using the following logic, we can express the sub-problems and obtain the constraint violation identification scenario model:

[0079] (31)

[0080] (32)

[0081] In the formula, and The decision variable solution is obtained by solving the main problem; by solving the constraint violation identification scenario model, the current scenario with the highest degree of constraint violation in formula (3) is identified , U Represents an uncertain set.

[0082] In a more specific technical solution, S4 includes:

[0083] S41. Make a decision based on the main problem, find the minimum adjustment amount, and obtain the recovery plan for the most unsuitable scenario;

[0084] S42. Solve the main problem MP and obtain a set of decision variable solutions And the objective function value corresponding to the main problem MP , update the iteration lower limit LB :

[0085] ;

[0086] S43. Solve the decision variables Substitute into the subproblem, solve the subproblem SP, and obtain a set of optimal solutions to the subproblems and the corresponding objective function values , and the least suitable scenario set , update the iteration upper limit UB :

[0087] .

[0088] S44, according to the iteration lower limit LB , Update iteration lower limit LB , and make a decision on stopping the iteration operation.

[0089] The present invention decomposes the upper-level model into a main problem and sub-problems, wherein the main problem determines the component recovery order and load recovery amount under the expected scenario of new energy output, and the sub-problem seeks the minimum adjustment amount based on the decision of the main problem under the worst scenario of new energy output, so that the adjusted recovery plan is feasible under the worst scenario.

[0090] In a more specific technical solution, S5 includes:

[0091] S51, constructing a coupling index within the region to measure the coupling strength between nodes in each partition in the applicable partitioning scheme;

[0092] S52, construct an inter-regional coupling index to measure the coupling strength between each partition in the applicable partition scheme;

[0093] S53, constructing an active power balance index to measure the active power distribution balance of the applicable partitioning scheme;

[0094] S54. Construct a reactive power balance index to measure the reactive power distribution balance applicable to the zoning scheme.

[0095] The present invention proposes a complete zoning evaluation index system, evaluates applicable zoning schemes from structural and functional perspectives, and analyzes the advantages and disadvantages of various zoning schemes.

[0096] In a more specific technical solution, the fast solution system of the power grid dynamic partition model based on a single-stage robust system includes:

[0097] A single-stage robust optimization model building module is used to obtain and construct the objective function of the dynamic partitioning of the power grid based on the uncertainty information of the renewable energy output, so as to obtain a single-stage robust optimization model;

[0098] The two-layer optimization module is used to convert the single-stage robust optimization model into a two-layer optimization model, where the two-layer optimization model includes: an upper model and a lower model. The upper model is used to process the power outage time of a given node. The partition optimization problem is solved by using the lower-level model to process the sub-partition approximate recovery problem of a given partition, and the load node recovery time is obtained by processing. The partition system recovery robust optimization model is defined, and the system recovery process is divided into In each period, the two-layer optimization module is connected with the single-stage robust optimization model building module;

[0099] The main problem and sub-problem solving module is used to adopt the constraint generation method CG, recover the robust optimization model according to the partition system, obtain and solve the partition optimization robust upper model of the given partition recovery time, decompose the partition optimization robust upper model, and obtain the main problem and sub-problem. The main problem and sub-problem solving module is connected with the double-layer optimization module;

[0100] The approximate recovery lower model solving module is used to adopt the column and constraint generation method C&CG to solve the partition approximate recovery lower model of the established partitioning scheme according to the main problem and the sub-problem, and obtain the applicable partitioning scheme. The approximate recovery lower model solving module is connected with the main problem and the sub-problem obtaining module;

[0101] The scheme evaluation module is used to obtain and construct a partition evaluation index system based on structural information and functional information to evaluate the effectiveness of the partition scheme. The scheme evaluation module is connected to the approximate recovery lower model solution module.

[0102] Compared with the prior art, the present invention has the following advantages:

[0103] The present invention proposes a modeling method, a solution strategy and a partition evaluation index system for a dynamic partition of a power grid. A single-stage robust optimization model is established to minimize power outage losses, partition decoupling degree and balance degree; when solving, the single-stage robust optimization model is converted into a two-layer optimization model, which is solved by the constraint generation method CG and the column and constraint generation method C&CG respectively; and a partition evaluation index system is constructed to quantify the advantages and disadvantages of applicable partition schemes. The partition where each black start power source is located includes the network, load, the started power source, and the new energy station that does not have black start capability to which the partition belongs. The partition operation of the present invention has strong adaptability for promotion.

[0104] The present invention establishes an objective function for dynamic grid partitioning, taking into account the uncertainty of renewable energy output, and minimizes the power outage losses of each sub-division, the transmission power of the interconnection lines, and the recovery time difference of each division by adjusting the node and line ownership in the worst renewable energy fluctuation scenario.

[0105] In the present invention, for k For each sub-area, the starting power of at least one started unit should be less than 70% of the upper limit of the black start unit output in the sub-area, so as to ensure that at least one non-black start unit in the sub-area can be successfully started.

[0106] The present invention constructs subsystem network connectivity constraints based on network flow theory. Traffic is injected from the system source point, and the traffic reaches each node through the line. All nodes consume unit traffic. When each node has traffic arriving, the connectivity of the network can be ensured.

[0107] The present invention decomposes the upper-level model into a main problem and sub-problems, wherein the main problem determines the component recovery order and load recovery amount under the expected scenario of new energy output, and the sub-problem seeks the minimum adjustment amount based on the decision of the main problem under the worst scenario of new energy output, so that the adjusted recovery plan is feasible under the worst scenario.

[0108] The present invention proposes a complete zoning evaluation index system, evaluates applicable zoning schemes from structural and functional perspectives, and analyzes the advantages and disadvantages of various zoning schemes.

[0109] The present invention solves the technical problems of low feasibility, slow solution speed and low applicability to scenarios with large system scale in the existing scheme of partitioning the power grid using the black start power supply as the starting power supply of the partition. BRIEF DESCRIPTION OF THE DRAWINGS

[0110] Figure 1 This is a schematic diagram of the basic steps of a method for quickly solving a dynamic partitioning model of a power grid based on a single-stage robust method according to Embodiment 1 of the present invention;

[0111] Figure 2 This is a schematic diagram of specific steps for obtaining a single-stage robust optimization model according to Embodiment 1 of the present invention;

[0112] Figure 3 This is a schematic diagram of the specific steps of converting a two-layer optimization model in Example 1 of the present invention;

[0113] Figure 4 A schematic diagram of specific steps of solving a lower-layer model for partition approximate recovery of a given partitioning scheme according to Embodiment 1 of the present invention;

[0114] Figure 5This is a schematic diagram of obtaining IEEE (Institute of Electrical and Electronics Engineers, Institute of Electrical and Electronics Engineers, USA) 39-node partitioning according to Embodiment 2 of the present invention;

[0115] Figure 6 This is a schematic diagram of the comparison of IEEE118 node system partitions according to Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0116] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0117] Example 1

[0118] like Figure 1 As shown, the method for quickly solving the single-stage robust power grid dynamic partition model provided by the present invention includes the following basic steps:

[0119] S1. Obtain and construct the objective function of dynamic grid partitioning based on the uncertainty information of renewable energy output, and obtain a single-stage robust optimization model;

[0120] like Figure 2 As shown, in this embodiment, step S1 of obtaining a single-stage robust optimization model further includes the following specific steps:

[0121] S11. Establish an objective function, adjust the node and line ownership according to the uncertainty information of new energy output, and minimize the power outage loss, interconnection line transmission power and restoration time difference of each sub-area under the scenario of abnormal fluctuation of new energy;

[0122] In this embodiment, the specific objective function is as follows (1):

[0123] (1)

[0124] in, Indicates i Bus node pair sub-partition k An integer variable representing the attributed state of Indicates line l For subpartitions k The continuous variable of the belonging status of , A value of 1 indicatesi Bus nodes, lines l Belong to k sub-partitions, if , A value of 0 means i Bus nodes, lines l Not in k sub-partitions; in this embodiment, if: , it means the line l For the contact line; For the i Active load at each bus node; For new energy units r The predicted output of the new energy units connected to the corresponding nodes, Indicates i Bus nodes in sub-partitions k The power outage time, Indicates k The time required for the sub-partition to complete the system recovery; Indicates m The time required for the sub-partition to complete the system recovery; Indicates the total number of nodes; and represents the total number of lines; Indicates the number of black start power supplies; the number of black start power supplies is consistent with the number of partitions; is the weight of the first objective function; is the weight of the second objective function; is the weight of the third objective function, where .

[0125] S12. Define partition power balance constraints, reactive power regulation reserve capacity constraints, active power reserve capacity constraints, unit start-up power constraints, node and branch partition constraints, subsystem network connectivity constraints and uncertainty sets.

[0126] In this embodiment, the partition power balance constraint is expressed using the following formula (2):

[0127] (2)

[0128] The above formula (2) indicates that the power imbalance of each partition must be within the preset threshold range. For a dual-line project, the Big M method can be used to treat it, and new energy units can be introduced. r The corresponding node is k Auxiliary continuous variables corresponding to sub-partitions , transform equation (2) into the following constraint:

[0129] (3)

[0130] (4)

[0131] (5)

[0132] Wherein, M represents the large number in the big-M method, and M can be, for example, 1000.

[0133] In this embodiment, the following formulas (6) and (7) are used to express the reactive power regulation reserve capacity constraint:

[0134] (6)

[0135] (7)

[0136] in, Busbar node i Access unit g The reactive power output limit, Busbar node i Access unit g The lower limit of reactive power output; For the i The upper limit of reactive power output of each bus node connected to reactive power source; For the i The lower limit of reactive power output of each bus node connected to the reactive power source, the reactive power source includes: integrated capacitors and reactors; For the k Reactive power of sub-areas Q positive spare capacity requirements; For the k Reactive power of sub-areas Q negative spare capacity requirements; Indicates i The reactive load of each bus node. Equations (6) to (7) indicate that the system should have reactive reserve regulation capacity for voltage control.

[0137] In this embodiment, the active reserve capacity constraint is expressed using the following formula:

[0138] Specifically, each partition has an active regulation reserve capacity, and the active regulation reserve capacity corresponding to the active reserve capacity is constrained as follows:

[0139] (8)

[0140] (9)

[0141]

[0142] in, Indicatesk Units in sub-areas g Active positive reserve factor; Indicates k Units in sub-areas g The active negative reserve factor takes into account the influence of the unit output adjustment rate on the reserve capacity, and the difference in the adjustment amount can be, for example, 0.1~0.3, and the values ​​can be, for example, 0.3 and 0.1, respectively. Indicates i Units with bus nodes g Power; Indicates k Units in sub-areas g Active negative reserve factor; Indicates k Sub-partition load d The demand positive reserve factor can be set to 8% of the load, and the active negative reserve factor can be set to 2% of the load. The values ​​can be 0.08 and 0.02 respectively.

[0143] In this embodiment, the following formula is used to express the unit startup power constraint:

[0144] Specifically, for the k sub-partitions, j The starting power of the started unit should be less than 70% of the upper limit of the black start unit output in the current partition, so as to ensure that at least one non-black start unit in the partition can be successfully started, as shown in the following formula (10):

[0145] (10)

[0146] in, Represents the set of nodes connected to the black start unit, Indicates the set of nodes that are not connected to the black start unit. Indicates j The starting power of the started unit.

[0147] In this embodiment, the node and branch partition constraints are expressed using the following formula:

[0148] (11)

[0149] (12)

[0150] (13)

[0151] (14)

[0152] The above formula (12) indicates that each node belongs to only one partition; the above formula (13) indicates that each branch can belong to at most one partition. When a branch does not belong to any partition, it means that the current branch is a tie line branch; the above formula (14) indicates that each area must have a black start power supply node. The above formula (15) indicates that each partition has at least one started unit.

[0153] In this embodiment, the subsystem network connectivity constraints are constructed based on network flow theory. Specifically, flow is injected from the system source point, and the flow reaches each node through the line. All nodes consume unit flow. When each node has flow arriving, the connectivity of the preset network can be ensured. In this embodiment, the black start unit node is defined as the system source point, and the unit to be started and the load node are defined as the sink. When the preset network meets the network flow constraint conditions, the connectivity of the resulting sub-area can be guaranteed. Define integer variables For the k Lines in sub-areas l The virtual flow on the line is set to flow from the node with a smaller node number to the node with a larger node number as the positive direction of the line virtual flow:

[0154] (15)

[0155] (16)

[0156] (17)

[0157] (18)

[0158] (19)

[0159] Among them, M represents the large number in the big M method, l(m,i) Indicates the end point is i The branch of a bus node, l(i,n) Indicates the starting point is i The above formula (15) indicates that the virtual flow only flows on the internal lines of the subsystem, and there is no flow on the section tie line. is a variable of 0 or 1. =1 means F lk 1, that is, there is traffic on the line, =0 then F lk =0, that is, line lThere is no flow on the node. Formula (16) indicates that the flow value of other nodes except the black start power supply node is not less than 1 unit. Formula (17) indicates the flow of other nodes in the subsystem except the black start power supply. Formula (18) indicates that the flow out of the black start power supply node is equal to the total number of nodes in the area minus 1, and formula (19) indicates that the flow into the black start power supply node is 0.

[0160] In this embodiment, the following formula is used to express the uncertainty set:

[0161] Specifically, the uncertainty information of new energy output satisfies the following budget uncertainty set:

[0162] (20)

[0163] in, is the total number of new energy stations; Indicates an uncertain budget value, which is an integer. represents the node set of new energy units. The aforementioned formula (15) indicates that the virtual flow is limited to flow on the internal lines of the subsystem, and there is no flow on the section interconnection line. The aforementioned formula (16) indicates that except for the black start power supply node, the flow value consumed by the remaining nodes is not less than 1 unit. The aforementioned formula (17) indicates the inflow flow of the remaining nodes in the subsystem except for the black start power supply node. The aforementioned formula (18) indicates that the outflow flow of the black start power supply node is equal to the total number of nodes in the current area minus 1. The aforementioned formula (19) indicates that the inflow flow of the black start power supply node is 0.

[0164] S2, converting the single-stage robust optimization model into a two-layer optimization model; in this embodiment, the two-layer optimization model includes: an upper model and a lower model, and the upper model is used to process the preset node power outage time The partition optimization problem is solved by using the lower-level model to process the sub-partition approximate recovery problem of a given partition, and the load node recovery time is obtained by processing. The partition system recovery robust optimization model is defined, and the system recovery process is divided into time period;

[0165] like Figure 3 As shown, in this embodiment, the step S2 of converting to obtain a two-layer optimization model also includes the following specific steps:

[0166] S21, expression construction upper model;

[0167] In this embodiment, the upper layer model is used as a partition optimization model for a given node power outage time and is represented in the following matrix form:

[0168] (twenty one)

[0169] in, xis a 0-1 integer variable, and its specific form is as follows (22); y is a continuous variable, and its specific form is as follows (23); u represents a random variable; is the 0-1 integer variable corresponding to the objective function x Specifically, in the aforementioned coefficient matrix In i Bus nodes in sub-partitions k Power outage time It is obtained by solving the lower model. i Bus nodes in sub-partitions k Power outage time Considered as a fixed value.

[0170] (twenty two)

[0171] (twenty three)

[0172] In the above formula (21), Represents a 0-1 integer variable x , continuous variables y and random variables u The inequality constraint corresponds to the above formula (3); Indicates a variable containing 0-1 integers x , continuous variables y The inequality constraints correspond to equations (4) to (5); Represents a 0-1 integer variable x The equality constraints correspond to equations (11), (13), (18) to (19);

[0173] Represents a 0-1 integer variable x The inequality constraints correspond to equations (6) to (10), (12), (14) to (17).

[0174] S22, let the length of each period be The total recovery time is , set the objective function of the black start unit in the partition and the node where the black start unit is located to recover at time 0; in this embodiment, the lower layer model is constructed as the sub-partition approximate recovery problem of a given partition to obtain the load node recovery time;

[0175] In this embodiment, it is assumed that the black start unit and the node in the partition recover the corresponding objective function at time 0:

[0176] (twenty four)

[0177] (25)

[0178] In formula (24), Indicates k In the sub-partition i The state of a bus node in time period s, Indicates k In the sub-partition i The state of a load node in time period s, Represents a partition k Middle i The state of a unit node in time period s, , , A value of 1 indicates that the node has been restored, and a value other than 1 indicates that it has not been restored; Indicates j The units started and i Lines l The recovery state in period s; Indicates the expected scenario k In the sub-partition i New energy units r The actual active output; Indicates k of the subpartitions i Load d The amount of load that the corresponding node needs to recover; Indicates i Load d The corresponding weight coefficient of the corresponding node is based on the load d The importance of loads is divided into 4 categories: Class 1, Class 2, Class 3, and Class 4. The weight coefficient of each type of load They are 100, 50, 30 and 10 respectively; specifically, the generator plant load is Class 1 load; Indicates the expected scenario k In the sub-partition i Unit g The active output adjustment amount of the corresponding node; Indicates the expected scenario k In the sub-partition i Unit g The reactive power output adjustment amount of the corresponding node; Indicates the expected scenario k In the sub-partition i New energy units r The actual active output size adjustment amount; Indicates k of the subpartitions i Loadd The corresponding node recovery load size adjustment; Indicates k The number of bus nodes in a sub-partition.

[0179] In this embodiment, the constraints for partition recovery include but are not limited to: conventional unit output and ramp constraints; new energy unit output constraints; cold and hot start time limit constraints; safe operation constraints; transformer ratio and capacitor input group upper and lower limit constraints; node power balance constraints taking into account AC constraints; equipment recovery state constraints; new energy output uncertainty set; cold load recovery constraints; overvoltage constraints; frequency constraints; and adjustment amount constraints.

[0180] S3. Using the constraint generation method CG, according to the partition system recovery robust optimization model, obtain and solve the partition optimization robust upper model with preset partition recovery time, decompose the partition optimization robust upper model, and obtain the main problem and sub-problems;

[0181] In this embodiment, in the process of obtaining the partition optimized robust upper model operation, the upper model is decomposed to obtain the main problem:

[0182] (26)

[0183] (27)

[0184] (28)

[0185] (29)

[0186] (30)

[0187] in, Represents the set of scenarios that violate constraint (27) to the highest degree.

[0188] In this embodiment, the upper model is decomposed to obtain sub-problems:

[0189] (31)

[0190] (32)

[0191] in, and The decision variable solution obtained for solving the main problem.

[0192] In this embodiment, by solving the above-mentioned formula (31) and formula (32) to form a model, the current scene with the highest degree of violation of the constraint of formula (3) is identified. .

[0193] In this embodiment, if the function value of the sub-problem objective function is less than 0, it means that the current solution does not violate the constraint, the current solution is the optimal solution, and the iteration ends. Otherwise, the identified current scene is fed back to the main problem, that is, , and then solve the main problem again.

[0194] S4, using the column and constraint generation method C&CG, according to the main problem and sub-problems, solve the partition of the established partition scheme to approximate the recovery of the lower model and obtain the applicable partition scheme;

[0195] like Figure 4 As shown, in this embodiment, step S4 of solving the lower layer model of partition approximate recovery of the established partition scheme also includes the following specific steps:

[0196] S41. Make a decision based on the main problem, find the minimum adjustment amount, and obtain the recovery plan for the most unsuitable scenario;

[0197] In this embodiment, the lower model is decomposed into a main problem and a sub-problem, wherein the main problem determines the component recovery order and load recovery amount under the expected scenario of new energy output, and the sub-problem seeks the minimum adjustment amount based on the decision of the main problem under the worst scenario of new energy output, so that the adjusted recovery plan is feasible under the worst scenario;

[0198] S42. Solve the main problem MP and obtain a set of decision variable solutions And the objective function value corresponding to the main problem MP , update the iteration lower limit LB :

[0199] ;

[0200] S43. Solve the decision variables Substitute into the subproblem, solve the subproblem SP, and obtain a set of optimal solutions to the subproblems and the corresponding objective function values , and the least suitable scenario set , update the iteration upper limit UB :

[0201] .

[0202] S44, according to the iteration lower limit LB , Update iteration lower limit LB , and make a decision on stopping the iteration operation.

[0203] In this embodiment, the iteration lower limit and the iteration upper limit are compared, and if the following conditions are met: , then the iteration is stopped, otherwise the decision variable solution and the least applicable scenario set , and go to step S42 to continue iterative solution. is the iteration deviation.

[0204] S5. Obtain and construct a zoning evaluation index system based on structural and functional information to evaluate the effectiveness of applicable zoning schemes.

[0205] Example 2

[0206] In this embodiment, specific parameter settings of the present invention are given, see Table 1 below.

[0207] Table 1. Basic parameters of the example

[0208]

[0209] In this embodiment, the IEEE39 node partitioning is obtained as follows: Figure 5 As shown, and compared with the partitioning method that only considers active power balance and the method using Dijkstra algorithm, the first node 1, the thirty-first node 31 and the thirty-fourth node 34 are selected to access the black start unit, and the third node 3, the fifth node 5, the fourteenth node 14, the sixteenth node 16, and the seventeenth node 17 are connected to the wind farm. The capacity of each wind farm is 100MW, and the uncertainty budget parameter is 5.

[0210] In this embodiment, in order to evaluate the rationality of the zoning scheme, a zoning evaluation index system including structural and functional aspects is constructed to evaluate the effectiveness of the zoning scheme.

[0211] Intra-region coupling index: This index is used to measure the coupling strength between nodes in each partition in the partitioning scheme. The specific expression is:

[0212] (33)

[0213] (34)

[0214] in, Indicates k The degree of coupling within the sub-regions; Indicates k The set of load nodes in the sub-partitions, Indicates the total number of nodes. Indicates the number of black start power supplies; Indicates i Node, j The electrical distance between nodes; Indicates k The number of nodes in a sub-partition; Represents the average intra-regional coupling degree in the first partition.

[0215] Inter-regional coupling index: This index is used to measure the coupling strength between regions. The specific expression is:

[0216] (35)

[0217] (36)

[0218] in, Indicates k The degree of coupling within the sub-regions; For the p The set of nodes in each sub-partition; Represents the average intra-regional coupling degree in the second partition.

[0219] Based on this, the average coupling degree within the first partition is taken and the average coupling degree within the second partition The weighted sum of the values ​​is used as the first structural evaluation index value. , specifically expressed as:

[0220] (37)

[0221] Active power balance index: This index is used to measure the active power distribution balance of the partition scheme. The specific expression is:

[0222] (38)

[0223] (39)

[0224] in, For the k Active power balance of each partition, Indicates k In the partition i Conventional units g Active output rating, Indicates k New energy units in each zone r Output forecast value; Indicates k Partition No. i The node load, Indicates the average active power balance of the first partition.

[0225] Reactive power balance index, which is used to measure the reactive power distribution balance of the partition scheme. The specific expression is:

[0226] (40)

[0227] (41)

[0228] in, For the k The reactive power balance of each partition. Indicates k Reactive reserve capacity of each zone; Indicates k Partition No. i The reactive load of each node, Indicates the average reactive power balance of the second partition.

[0229] Based on this, take the average active balance of the first partition and the average reactive power balance of the second partition The weighted sum of the values ​​is used as the second structural evaluation index value. , specifically expressed as:

[0230] (42)

[0231] Table 2 gives the comparison results of the evaluation indicators obtained by the three methods of the IEEE39 node system. It can be seen that the system structural and functional indicators of the method of the present invention are both lower than those of the partition method that only considers active power balance and the scheme using the Dijkstra algorithm. The reason is that the scheme using the Dijkstra algorithm attributes nodes with close electrical distances together, so the nodes in the region are closely coupled, but the balance of power distribution is not considered, resulting in a large degree of active and reactive balance. The partition method that only considers active power balance considers the minimum electrical distance between units, the power of the tie line and the network loss in the objective function, but does not consider the electrical distance between the unit and the load node. In addition, when adjusting the partition scheme, the scheme is adjusted only according to the active power imbalance, without considering the reactive power balance, resulting in a large degree of reactive balance. The scheme of the present invention considers the partition power balance constraint, active and reactive reserve capacity constraint from a functional perspective in the constraints, and considers the connectivity constraint based on flow and the node branch partition constraint from a structural perspective, so that all indicators of the partition scheme can be met.

[0232] Table 2. Comparison of evaluation indicators of different partitioning schemes of node systems

[0233]

[0234] For a comparison of IEEE118 node system partitions, see Figure 5 In the example, the generators connected to the first node 1, the fifty-fifth node 55 and the sixty-ninth node 69 are used as black start units, and wind turbines are connected to the fourth node 4, the fifty-fourth node 54 and the one hundred and fourth node 104, and the uncertain parameter is 5.

[0235] Table 3 shows the comparison results of the evaluation indicators obtained by the three methods of the IEEE118 node system. It can be seen that the structural indicators and functional indicators of the scheme of the present application are better than the partition method that only considers active power balance and the scheme using the Dijkstra algorithm. The internal coupling of the region of the present invention is tight, the coupling between regions is sparse, and the power distribution is uniform, which reflects the superiority of the model and algorithm of the present application.

[0236] Table 3. Comparison of evaluation indicators of different partitioning schemes of node systems

[0237]

[0238] like Figure 6 As shown, in this embodiment, an actual power grid example is further used to verify the applicability of the method proposed in the present invention, and the ownership of nodes in each partition is analyzed; specifically, the actual power grid has 419 plants, 134 generators, 275 loads and 559 lines, and the voltage levels include 230 kV, 525 kV and 1050 kV; there are 98 new energy plants.

[0239] In this embodiment, the following table 4 shows the partition scheme evaluation index results obtained by the solution of the present invention. It can be seen that all the indicators of the partition scheme can meet the partition recovery requirements:

[0240] Table 4. Evaluation indicators of actual system partitioning schemes

[0241]

[0242] In summary, the present invention proposes a modeling method, solution strategy and partition evaluation index system for dynamic partitioning of power grids. A single-stage robust optimization model that minimizes power outage losses, partition decoupling degree and balance degree is established; when solving, the single-stage robust optimization model is converted into a two-layer optimization model, and the constraint generation method CG and the column and constraint generation method C&CG are used to solve them respectively; and a partition evaluation index system is constructed to quantify the advantages and disadvantages of applicable partitioning schemes. The partition where each black start power source is located includes the network, load, the started power source, and the new energy station that does not have black start capability to which the partition belongs. The partition operation of the present invention has strong adaptability for promotion.

[0243] The present invention establishes an objective function for dynamic grid partitioning, taking into account the uncertainty of renewable energy output, and minimizes the power outage losses of each sub-division, the transmission power of the interconnection lines, and the recovery time difference of each division by adjusting the node and line ownership in the worst renewable energy fluctuation scenario.

[0244] In the present invention, for k For each sub-area, the starting power of at least one started unit should be less than 70% of the upper limit of the black start unit output in the sub-area, so as to ensure that at least one non-black start unit in the sub-area can be successfully started.

[0245] The present invention constructs subsystem network connectivity constraints based on network flow theory. Traffic is injected from the system source point, and the traffic reaches each node through the line. All nodes consume unit traffic. When each node has traffic arriving, the connectivity of the network can be ensured.

[0246] The present invention decomposes the upper-level model into a main problem and sub-problems, wherein the main problem determines the component recovery order and load recovery amount under the expected scenario of new energy output, and the sub-problem seeks the minimum adjustment amount based on the decision of the main problem under the worst scenario of new energy output, so that the adjusted recovery plan is feasible under the worst scenario.

[0247] The present invention proposes a complete zoning evaluation index system, evaluates applicable zoning schemes from structural and functional perspectives, and analyzes the advantages and disadvantages of various zoning schemes.

[0248] The present invention solves the technical problems of low feasibility, slow solution speed and low applicability to scenarios with large system scale in the existing scheme of partitioning the power grid using the black start power supply as the starting power supply of the partition.

[0249] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A fast solution method for the single-stage robust power grid dynamic partition model, characterized in that: The method comprises: S1. Obtain and construct the objective function of dynamic grid partitioning based on the uncertainty information of renewable energy output to obtain a single-stage robust optimization model; S2, converting the single-stage robust optimization model into a two-layer optimization model, wherein the two-layer optimization model includes: an upper layer model and a lower layer model, using the upper layer model to process the preset node power outage time The partition optimization problem is solved by using the lower layer model to process the sub-partition approximate recovery problem of a given partition, and the load node recovery time is obtained by processing. The partition system recovery robust optimization model is defined, and the system recovery process is divided into time period; S3, using constraint generation method CG, according to the partition system recovery robust optimization model, obtaining and solving the partition optimization robust upper model with preset partition recovery time, decomposing the partition optimization robust upper model, and obtaining the main problem and sub-problems; S4, using the column and constraint generation method C&CG, according to the main problem and the sub-problem, solving the partition approximation of the established partition scheme to restore the lower model and obtain an applicable partition scheme; S5. Obtain and construct a zoning evaluation index system based on structural information and functional information to evaluate the effectiveness of the applicable zoning scheme.

2. The method for quickly solving the dynamic partitioning model of a power grid based on a single-stage robustness according to claim 1 is characterized in that: The S1 includes: S11, establishing the objective function, adjusting the node and line ownership in the scenario of abnormal fluctuation of new energy according to the uncertainty information of the new energy output, minimizing the power outage loss, the transmission power of the tie line and the restoration time difference of each sub-area, wherein the objective function is constructed using the following logic: (1) In the formula, Indicates i The busbar node is k an integer variable representing the belonging status of the sub-partitions, Indicates line l For k A continuous variable representing the belonging status of each of the sub-partitions; For the i The active load at the bus node is Indicates the line l The active load, d Indicates load; For new energy units r The predicted output of the new energy units connected to the corresponding nodes, Indicates i The bus nodes in the sub-partition k The power outage time, Indicates k The time required for the sub-partition to complete system recovery; Indicates m The time required for the sub-partition to complete system recovery; Indicates the total number of nodes; and represents the total number of lines; Indicates the number of black start power supplies; is the first weight coefficient; is the second weight coefficient; is the third weight coefficient; S12. Define partition power balance constraints, reactive power regulation reserve capacity constraints, active power reserve capacity constraints, unit start-up power constraints, node and branch partition constraints, subsystem network connectivity constraints and uncertainty sets.

3. The method for quickly solving the dynamic partitioning model of a power grid based on a single-stage robustness according to claim 2 is characterized in that: The S12 further includes: S121. Use the following logic to express the partition power balance constraint, and constrain the power imbalance of each partition to a preset threshold range: (2) In the formula, is the maximum power imbalance allowed for each partition, Indicates i The bus node is connected to the unit g The power rating of For new energy units r The predicted output of the new energy units connected to the corresponding nodes, is a double line item, Indicates i The bus node is connected to the k an integer variable representing the belonging status of each of the sub-partitions; Adopting the Big M method to introduce the new energy unit r The corresponding node is k Auxiliary continuous variables corresponding to the sub-partitions , transform the formula (2) into the following constraint: (3) (4) (5) Wherein, M represents the large number in the large-M method; S122. Using the following logic, the reactive power regulation reserve capacity constraint is expressed to provide reactive power regulation reserve capacity for the preset system and perform voltage control: (6) (7) In the formula, The bus node i Access to the unit g The reactive power output limit, The bus node i Access to the unit g The lower limit of reactive power output; For the i The upper limit of reactive power output of the bus node connected to the reactive power source; For the i The lower limit of reactive power output by the bus node connected to the reactive power source; For the k The reactive power of the sub-area Q positive spare capacity requirements; For the k The reactive power of the sub-areas Q negative spare capacity requirements; Indicates i The reactive load of the bus node, d Indicates load; S123, expressing the active reserve capacity constraint to provide active regulation reserve capacity for each partition; S124, expressing the starting power constraint of the unit, in the k The sub-areas reserve a preset number of non-black start units j , for starting operation; S125, construct and use the node and branch partition constraints to make the node subordinate to the k said sub-partitions; S126. Based on the network flow theory, the subsystem network connectivity constraint is constructed, wherein the black start unit node is defined as the system source point, the unit to be started and the load node are defined as the sink point, and when the preset network meets the network flow constraint condition, the connectivity of the obtained sub-area is guaranteed; wherein, the integer variable is defined using the following logic For the k the lines in the sub-partition l The virtual flow on the line is set to flow from the node with a smaller node number to the node with a larger node number as the positive direction of the line virtual flow: (15) (16) (17) (18) (19) In the formula, l(m,i) Indicates the end point is i branches of the busbar node, m represents the predecessor node of the end point, l(i, n) Indicates the starting point is i branches of the busbar node, n represents the subsequent node of the starting point, Indicates the complete set, Represents the set of nodes connected to the black start unit, represents the set of nodes that are not connected to the black start unit, NB Indicates that the black start unit is not connected. B represents the black start unit, Indicates the line l For k A continuous variable representing the belonging status of each of the sub-partitions; S127. Express the uncertainty set according to the uncertainty information of the new energy output, wherein the uncertainty information of the new energy output satisfies the following uncertainty set: U : (20) In the formula, NW is the total number of new energy stations, Indicates the capacity demand of new energy stations; For the new energy unit r Expected output value; Indicates the new energy unit r Output fluctuation limit; Indicates an uncertain budget value. Represents the node set of new energy units, WT Represents a wind farm.

4. The method for quickly solving the dynamic partitioning model of a power grid based on a single-stage robustness according to claim 3 is characterized in that: In S123, the active reserve capacity constraint of the active regulation reserve capacity is expressed by the following logic: (8) (9) In the formula, Indicates i The units of the bus node g The power, Indicates i The units of the bus node g The power limit, Indicates i The units of the bus node g The lower power limit; Indicates the units in the system G the total number of Indicates the total number of loads in the system; Indicates k The units in the sub-areas g Active positive reserve factor; Indicates k The units in the sub-areas g Active negative reserve factor; Indicates k The load in the sub-partition d The demand positive reserve factor; Indicates k The load in the sub-partition d Negative reserve factor of demand; Indicates the number of black start power supplies; For the i The active load at the bus node is Indicates i The bus nodes in the sub-partition k The corresponding integer variable.

5. The method for quickly solving the dynamic partitioning model of power grid based on single-stage robustness according to claim 3 is characterized in that: In S124, the following logic is used to express the unit startup power constraint: (10) In the formula, Indicates j Starting power of the started units, st Indicates startup. Indicates the first unit that is not connected to the black start unit. j The started unit is k integer variables corresponding to the subpartitions, Indicates i The units of the bus node g The power limit, Indicates existence.

6. The method for quickly solving the dynamic partition model of power grid based on single-stage robustness according to claim 3 is characterized in that: In S125, the node and branch partition constraints are expressed using the following formula: (11) (12) (13) (14) In the formula, represents the number of black start power supplies, wherein (12) indicates that each node belongs to one partition; (13) indicates that each branch belongs to at most one partition, and when the branch does not belong to any partition, it indicates that the branch is a tie line branch; and (14) indicates that there is one black start power supply node in each area.

7. The method for quickly solving the dynamic partition model of power grid based on single-stage robustness according to claim 1 is characterized in that: The S2 includes: S21. Use the following matrix logic to express and construct the upper model: (21) (22) (23) In the formula, x is a 0-1 integer variable; y is a continuous variable, u represents a random variable; is the 0-1 integer variable corresponding to the objective function x The coefficient matrix of Represents the 0-1 integer variable x , the continuous variable y and the random variable u The inequality constraint corresponds to equation (3): is the first corresponding coefficient matrix, is the second corresponding coefficient matrix, is the third corresponding coefficient matrix; Indicates a 0-1 integer variable containing the x , the continuous variable y The inequality constraints correspond to equations (4) to (5). is the fourth corresponding coefficient matrix, is the fifth corresponding coefficient matrix, d For load; Represents the 0-1 integer variable x The equality constraints correspond to equations (11), (13), (18) to (19); Represents the 0-1 integer variable x The inequality constraints correspond to equations (6) to (10), (12), (14) to (17). is the sixth corresponding coefficient matrix, h represents the thermal parameters, Represents new energy units r The corresponding node is k Auxiliary continuous variables corresponding to sub-partitions, Indicates black start unit b The corresponding node is k Auxiliary continuous variables corresponding to the sub-partitions; S22. Assume that the length of each period is The total recovery time is , set the objective function of the black start unit in the partition and the node where the black start unit is located to recover at time 0: (24) (25) In the formula (24), Indicates k In the sub-partition i The state of a bus node in time period s, Indicates k of the subpartitions i The state of each load node in the period s is: Indicates the partition k Middle i The state of each unit node in the time period s; Indicates j The units started and i Lines l the recovery state during said period s; Indicates the expected scenario k of the subpartitions i Unit g The active output of the corresponding node; Indicates the expected scenario k of the subpartitions i Unit g The reactive power output of the corresponding node; Indicates the expected scenario k of the subpartitions i New energy units r The actual active output; Indicates k of the subpartitions i Load d The amount of load that the corresponding node needs to recover; Indicates i Load d The weight coefficient corresponding to the corresponding node; Indicates the expected scenario k of the subpartitions i Unit g The active output adjustment amount of the corresponding node; Indicates the expected scenario k of the subpartitions i Unit g The reactive power output adjustment amount of the corresponding node; Indicates the expected scenario k of the subpartitions i New energy units r The actual active output size adjustment amount; Indicates k of the subpartitions i Load d The corresponding node recovery load size adjustment; Indicates k The number of busbar nodes in the sub-partition, represents the time state set, Indicates k of the subpartitions i The time corresponding to each node, Indicates time T The corresponding total recovery period.

8. The method for quickly solving the dynamic partition model of power grid based on single-stage robustness according to claim 3 is characterized in that: The main problem is expressed using the following logic: (26) (27) (28) (29) (30) In the formula, x is a 0-1 integer variable; y is a continuous variable, u represents a random variable; is the 0-1 integer variable corresponding to the objective function x The coefficient matrix of is the first corresponding coefficient matrix, is the second corresponding coefficient matrix, is the third corresponding coefficient matrix; Indicates a 0-1 integer variable containing the x , the continuous variable y The inequality constraint, is the fourth corresponding coefficient matrix, is the fifth corresponding coefficient matrix, d For load; g Indicates the unit; Indicates that it contains only the 0-1 integer variables x The inequality constraint, h represents the thermal parameters, is the sixth corresponding coefficient matrix, represents the set of scenarios that violate the constraint (27) to the highest degree; The sub-problems are expressed using the following logic to obtain the constraint violation identification scenario model: (31) (32) In the formula, and The decision variable solution is obtained by solving the main problem; by solving the constraint violation identification scenario model, the current scenario with the highest degree of constraint violation of formula (3) is identified. , U Represents an uncertain set.

9. The method for quickly solving the dynamic partition model of power grid based on single-stage robustness according to claim 1 is characterized in that: The S4 includes: S41, making a decision based on the main problem, finding the minimum adjustment amount, and obtaining a recovery plan for the most unsuitable scenario; S42, solve the main problem MP to obtain a set of decision variable solutions And the objective function value corresponding to the master problem MP , update the iteration lower limit LB : ； S43, solving the decision variables Substitute into the subproblem, solve the subproblem SP, and obtain a set of optimal solutions to the subproblems and corresponding objective function values , and the least suitable scenario set , update the iteration upper limit UB : ； S44, according to the iteration lower limit LB , the update iteration lower limit LB , and make a decision on stopping the iteration operation.

10. The method for quickly solving the dynamic partition model of power grid based on single-stage robustness according to claim 1 is characterized in that: The S5 includes: S51, constructing a coupling index within the region to measure the coupling strength between nodes in each partition in the applicable partitioning scheme; S52, constructing an inter-region coupling index to measure the coupling strength between each of the partitions in the applicable partition scheme; S53, constructing an active power balance index to measure the active power distribution balance of the applicable partition scheme; S54. Construct a reactive power balance index to measure the reactive power distribution balance of the applicable partitioning scheme.

11. A fast solution system for the dynamic partition model of power grid based on a single-stage robust system, characterized in that: The system comprises: A single-stage robust optimization model building module is used to obtain and construct the objective function of the dynamic partitioning of the power grid based on the uncertainty information of the renewable energy output, so as to obtain a single-stage robust optimization model; A two-layer optimization module is used to convert the single-stage robust optimization model into a two-layer optimization model, wherein the two-layer optimization model includes: an upper layer model and a lower layer model, and the upper layer model is used to process the power outage time of a given node. The partition optimization problem is solved by using the lower layer model to process the sub-partition approximate recovery problem of a given partition, and the load node recovery time is obtained by processing. The partition system recovery robust optimization model is defined, and the system recovery process is divided into During a period of time, the two-layer optimization module is connected to the single-stage robust optimization model building module; A main problem and sub-problem obtaining module is used to adopt the constraint generation method CG, according to the partition system recovery robust optimization model, obtain and solve the partition optimization robust upper layer model of the predetermined partition recovery time, decompose the partition optimization robust upper layer model, obtain the main problem and sub-problem, and the main problem and sub-problem obtaining module is connected with the double-layer optimization module; An approximate recovery lower model solving module is used to adopt the column and constraint generation method C&CG to solve the partition approximate recovery lower model of the established partition scheme according to the main problem and the sub-problem, and obtain an applicable partition scheme. The approximate recovery lower model solving module is connected to the main problem and the sub-problem obtaining module; The scheme evaluation module is used to obtain and construct a partition evaluation index system based on structural information and functional information to evaluate the effectiveness of the partition scheme. The scheme evaluation module is connected to the approximate recovery lower layer model solving module.

Images