A smart power distribution network fast island division method based on a look-ahead greedy algorithm

Abstract

The application discloses a power distribution network island division method based on a heuristic forward-looking greedy algorithm and containing energy storage equipment; Step 1: a day-ahead economic dispatch model is established to calculate the initial electric quantity of the energy storage; Step 2: an island division model is constructed; Step 3: the decoupling process of the island division model; including: the island division model is processed by an electric quantity abundance information algorithm to obtain a maximum island division unit; the island division model is processed by a power balance algorithm to obtain an island checking unit; Step 5: the island checking unit is processed by a CPLEX commercial solver to obtain an optimal island division scheme; the application realizes the fast search of the island scheme based on the forward-looking greedy algorithm, and can be converted into a mixed integer linear programming model which is easy to solve, so that the two sub-problems of the power distribution network are alternately solved, the island division scheme is quickly formulated, and the stability of the power distribution network is improved.

Description

Technical Field

[0001] This invention belongs to the technical field of joint operation of distributed power sources and energy storage in active distribution networks and analysis of the consequences of distribution network faults, and particularly relates to a fast islanding method for smart distribution networks based on a look-ahead greedy algorithm. Background technology:

[0002] Distribution network failures directly impact social production and residents' daily power supply, leading to a growing emphasis on methods for rapidly restoring power to critical loads after a failure. In recent years, the penetration rate of distributed generation (DG) in distribution networks has increased annually. A high proportion of distributed wind and solar power connected to the distribution network can significantly improve power supply reliability, maintaining uninterrupted power supply to critical loads during failures. However, DG is characterized by its dispersed nature and fluctuating output. Developing emergency power restoration plans based on these characteristics—i.e., implementing islanding—has become a research hotspot and challenge. To enable DG to better fulfill its power supply role, the IEEE 1547.4-2011 standard encourages conscious islanding operation, specifying that planned islanding should take into account the dynamic characteristics of micro-sources and loads, as well as switching operation strategies.

[0003] The construction of islanding models is fundamental to formulating distributed power supply restoration schemes during faults. Islanding models are essentially mixed-integer programming models with 0-1 variables. Their solutions require comprehensive consideration of key factors such as electrical safety, source-load balance, and connectivity constraints. The solution process is complex, and the combinations of islanding solutions are highly variable, making islanding a typical NP-hard problem. In existing islanding modeling research, some literature models the islanding problem as a complex mixed-integer nonlinear model. Modeling islanding as a nonlinear model can improve computational accuracy, but solving this problem using nonlinear programming methods is complex and lacks versatility. For some scenarios involving solving nonlinear models, second-order cone optimization can transform complex nonlinear models into cone models, representing complex relationships between variables with specially structured cone sets, greatly simplifying the solution of the original model and improving convergence speed. However, research shows that as the dimensionality of decision variables in the optimization model increases, cone programming faces low solution efficiency or even non-convergence.

[0004] Existing methods for solving island partitioning models can be categorized into graph-based partitioning methods, intelligent algorithms, and heuristic algorithms. Graph-based island partitioning methods can be further divided into undirected and directed graph-based methods. The former typically transforms the island partitioning problem into a minimum spanning tree problem, a typical NP-hard problem. Current graph-based partitioning methods often struggle to balance computational speed and accuracy. Intelligent algorithms, represented by genetic algorithms, differential evolution algorithms, and particle swarm optimization, are highly versatile and can solve nonlinear island partitioning models through iterative computation. However, they suffer from slow computational speed and a tendency to get trapped in local optima. Furthermore, the local search of intelligent algorithms is stochastic, often failing to achieve the desired results when dealing with island partitioning problems in distribution networks with radial constraints. Heuristic algorithms can overcome the computational time problem caused by repeated iterations in intelligent algorithms and can specifically solve complex, constrained island partitioning models.

[0005] With the large-scale integration of distributed generation (DG) into power distribution networks, configuring appropriate energy storage devices has gradually become an important solution for mitigating DG output fluctuations and providing emergency load support. When system components fail, the location and duration of the fault, as well as the DG output, are random. Conducting islanding operations with energy storage during fault periods can significantly mitigate the impact of DG output fluctuations on islanding effectiveness, significantly improving islanding efficiency and serving as a crucial means to enhance power distribution network reliability. However, energy storage operation requires consideration of sequential characteristics. Fully taking into account the impact of source-load timing fluctuations and energy storage charging / discharging strategies on islanding schemes significantly increases the complexity of islanding. Therefore, it is urgent to propose a modeling and solution method for islanding that, while considering the sequential characteristics of energy storage, fully leverages the power supply potential of DG and quickly obtains the final islanding scheme.

[0006] To address the shortcomings of existing research, this invention proposes a rapid islanding method for distribution networks containing energy storage devices based on a heuristic look-ahead greedy algorithm. Its key features are: first, it formulates a maximum islanding scheme based on power sufficiency information; second, it verifies and optimizes the islanding scheme based on power sufficiency information. These two alternating steps achieve rapid formulation of the distribution network islanding scheme. First, it fully considers the significant impact of sequential operation of energy storage and secondary power outage constraints, establishing a distribution network islanding model oriented towards source-load time-series fluctuations. Then, considering the role of distribution network tie switches in extending power restoration paths, it alternately performs maximum islanding based on look-ahead greedy algorithms and Prim's algorithm, followed by islanding verification based on a power sufficiency verification model, resulting in a final islanding scheme with a radial operating structure. Compared with commonly used intelligent algorithms, this method overcomes the long convergence time problem caused by iterative calculations, improving the formulation speed of the islanding scheme while ensuring islanding benefits. Summary of the Invention

[0007] This invention proposes a fast islanding method for distribution networks containing energy storage devices based on a heuristic look-ahead greedy algorithm. According to the proposed method, a physically meaningful islanding scheme can be quickly derived using given power sufficiency and power balance information. A mixed-integer knapsack model for islanding, considering time-series fluctuations on both the source and load sides, is established and decoupled into two sub-problems: maximum islanding based on power information and islanding optimization verification based on power balance information. The former can be rapidly searched for islanding schemes using a look-ahead greedy algorithm, while the latter can be transformed into an easily solvable mixed-integer linear programming model. The rapid formulation of the islanding scheme can be achieved by alternately solving the two sub-problems. Specifically, this invention achieves rapid formulation of the islanding scheme through the following steps:

[0008] A fast islanding method for smart distribution networks based on a look-ahead greedy algorithm includes the following steps:

[0009] Step 1: Establish a day-ahead economic dispatch model and calculate the initial energy storage capacity; including:

[0010] The operating costs of the distribution network are divided into electricity purchase costs and electricity sales revenue, distribution network loss costs, and energy storage loss costs. With cost minimization as the objective function, a day-ahead economic dispatch model is constructed taking into account the operating constraints of the distribution network. The model is then solved using the commercial software CPLEX to obtain the initial energy storage capacity.

[0011] Step 2: Constructing the Islanding Model: Taking the load benefits of being islanded as the optimization objective, an islanding model is constructed based on distribution network operation constraints, distributed generation, and energy storage-related operation constraints; that is:

[0012]

[0013]

[0014]

[0015]

[0016] Among them, ST i With st i,t A 0-1 binary variable used to account for secondary power outage constraints; if node i is included in the island at time t, then st i,t The value is 1 if node i is in the fault period [t1,t2], otherwise it is 0; i,t If both are 1, then ST i B is 1 if it is 1, otherwise it is 0. i,t This represents the revenue value of node i at time t, and is the optimization objective of the island partitioning model. This variable is determined by the active power of the load within the island. and priority weight PR i The product of ; set N represents the set of distribution network load points; t1 represents the initial time of the fault, t2 represents the end time of the fault; A represents the number of islands;

[0017] Step 3: Decoupling process of the island partitioning model:

[0018] The maximum island partitioning unit is obtained by processing the island partitioning model using a power sufficiency information algorithm.

[0019] The island partitioning model is processed by a power balance algorithm to obtain island verification units;

[0020] Step 5: Process the islanding verification units using the CPLEX commercial solver to obtain the optimal islanding partitioning scheme. Further, in Step 3, the islanding partitioning model is processed using a power sufficiency information algorithm to obtain the maximum islanding partitioning units, including:

[0021] 301. Construct a power sufficiency constraint according to the following formula:

[0022]

[0023] Among them: the power sufficiency constraint means that during the fault period [t1,t2], the islanding scheme should satisfy the requirement that the sum of the total power of DG and ES is greater than the total power of the load in the island;

[0024] 302. Decoupling the island partitioning model to obtain the largest island partitioning unit:

[0025] OF1: Eq.(8)

[0026]

[0027] Furthermore, in Step 3, the islanding model is processed using a power balance algorithm to obtain islanding verification units:

[0028]

[0029]

[0030] Furthermore, Step 4: obtaining a preliminary islanding scheme that satisfies the power sufficiency constraint of the distribution network by solving the maximum islanding partitioning unit using a heuristic look-ahead greedy algorithm, includes the following steps:

[0031] 401. Based on the fault update of the distribution network topology at time t1, determine the downstream power outage area of ​​the distribution network caused by this fault, and denote the topology map corresponding to the downstream power outage area of ​​the distribution network as graph G.

[0032] 402. Select the distributed power sources to be restored in descending order of their access capacity; if all distributed power sources are marked, proceed to step (7).

[0033] Selected distributed power sources and energy storage devices are connected to the node set Λ a Establish the initial set of points for the island partitioning problem;

[0034] Select a set of node loads V for the isolated island, and calculate the sum P of the active power loads of all nodes in the set of node loads V during the fault period. V Island revenue B V The maximum capacity of distributed power sources and energy storage is calculated using the following formula:

[0035]

[0036]

[0037]

[0038] 403. Obtain the set of load nodes NE for the power restoration path by searching the node load set V of the island using a look-ahead greedy algorithm. 1 The neighborhood point NE of the load node set of the power supply restoration path is calculated using the following formula. 1 The value of (m) is greater than Va 1 (m):

[0039]

[0040] In the formula: C R P represents the maximum remaining battery capacity; V (i) represents the active load of node i, B V (i) represents the load revenue of node i, which is the active load P. V (i) with load weight PR i The product of; NE 1 (m) represents the neighborhood set NE 1 The neighborhood set NE of the m-th vertex. 1 The number of mid-vertices is N0;

[0041] 404. Using a look-ahead greedy algorithm to analyze the load node set NE of the power supply path. 1 Search to obtain the set of forward neighboring points. The power restoration path neighborhood point NE is calculated using the following formula. 1 (m) and the lookahead neighbor points The value of the combination is Va m (n):

[0042]

[0043] In the formula: Represents NE 1 The nth neighbor of (m), the lookahead neighborhood set Let N be the number of mid-vertex. m ;

[0044] 405. Calculate the optimal value ratio for island partitioning using the following formula, and include the corresponding load nodes in the partitioned islands;

[0045] 406. Determine the optimal value ratio Va for the partitioning. max If the value is zero, the search for island pre-solutions starting from this DG node ends, the DG node is marked, and the process returns to Step 2; otherwise, Va is set to zero. max The corresponding solution is included in the island set V. In cases where the value ratios are the same, nodes with larger active power loads are prioritized for inclusion, and the remaining power C is updated according to the following formula. R Return to Step 3:

[0046]

[0047] 407. The Prim algorithm for minimum spanning trees is used to process the maximum island partitioning scheme to obtain the final island partitioning scheme with radial structure;

[0048] 408. Verify the final islanding scheme by using node voltage and branch power flow constraints (Eq.(12)~(13)).

[0049] Beneficial effects

[0050] This invention proposes a fast islanding method for distribution networks containing energy storage devices based on a heuristic look-ahead greedy algorithm. Its key features are: formulating a maximum islanding scheme based on power sufficiency information, and verifying and optimizing the islanding scheme based on power sufficiency information. By alternating between these two steps, a distribution network islanding scheme is rapidly formulated, resulting in an islanding plan. Compared to existing research on islanding modeling and solving, this invention offers the following advantages:

[0051] (1) This invention overcomes the limitation of some studies that usually only consider a single time segment when carrying out island partitioning, and fully takes into account the source-load time-series fluctuations during system failures to obtain a more realistic and accurate island partitioning scheme.

[0052] (2) In the island partitioning modeling, the important impact of load priority, sequential characteristics of energy storage and secondary power outage constraints on the island partitioning scheme should be fully considered.

[0053] (3) Introduce tie switches in the topology, fully consider the role of tie switches in extending the power supply restoration path and the flexible change of the distribution network topology, and disconnect the part containing the ring network in the islanding scheme based on the Prim algorithm to obtain the radial islanding scheme.

[0054] (4) Maximum islanding partitioning and islanding optimization verification are carried out alternately based on power sufficiency information and power availability information. Compared with commonly used intelligent algorithms, it overcomes the problem of long convergence time caused by iterative calculation and improves the speed of islanding partitioning scheme formulation while ensuring islanding benefits. Attached Figure Description

[0055] Figure 1 This is a complete flowchart of the island partitioning process of this invention. Detailed Implementation

[0056] The following is in conjunction with the appendix Figure 1 The present invention is described as follows:

[0057] This invention proposes a fast islanding method for distribution networks containing energy storage devices based on a heuristic look-ahead greedy algorithm.

[0058] According to the proposed method, specifically, this invention will achieve rapid formulation of an island partitioning scheme through the following steps:

[0059] Step 1: Establish a day-ahead economic dispatch model and calculate the initial energy storage capacity.

[0060] The operating costs of the distribution network are divided into electricity purchase costs and electricity sales revenue, distribution network loss costs, and energy storage loss costs. Among them, the day-ahead economic dispatch model is constructed with cost minimization as the objective function and distribution network operation constraints are taken into account. The model is solved with the commercial software CPLEX to obtain the initial energy storage capacity.

[0061] Step 2: Construct an island partitioning model

[0062] Taking the load benefits of being islanded as the optimization objective, an islanding model is constructed based on the operational constraints of the distribution network, distributed power sources, and energy storage. This model is a 0-1 mixed-integer nonlinear programming model that comprehensively considers many factors such as source-load fluctuations, load priorities, and secondary power outage constraints.

[0063] Step 3: Decoupling the Island Partitioning Model

[0064] The complex island partitioning model in Step 2 is decoupled into a maximum island partitioning based on power sufficiency information and an island verification unit based on power balance, thereby accelerating the solution of the island partitioning model.

[0065] Step 4: Solve for the maximum island partitioning unit based on a heuristic look-ahead greedy algorithm

[0066] A preliminary islanding solution that satisfies the power sufficiency constraint is quickly obtained based on a look-ahead greedy algorithm.

[0067] Step 5: Solve the isolated check cells using the CPLEX commercial solver

[0068] The islanding power verification model is solved using the commercial software CPLEX to obtain the system power shortage results. The verified results are the final islanding scheme. The overall solution process for the islanding model is as follows: Figure 1 As shown.

[0069] 1. Establish a day-ahead economic dispatch model to calculate the initial energy storage capacity.

[0070] (1) Construct the objective function that optimizes the distribution network operation cost.

[0071] Distribution network operating costs include the cost of purchasing electricity and the revenue from selling electricity, the network loss cost, and the energy storage loss cost. The system operating cost is calculated as follows:

[0072]

[0073] Among them, C t,buy and C t,sell These represent the unit prices for purchasing and selling electricity between the distribution network and the upstream power grid at time t, respectively. and C represents the power that the distribution network operator purchases and sells from the upstream grid at time t; t The cost represents the unit network loss; E represents the set of all branches in the distribution network; R ij and I ij,t C represents the branch resistance starting at node i and ending at node j, and the branch current at time t, respectively; ES This represents the cost of charging / discharging 1 kWh of energy storage; N ES This indicates the number of nodes where energy storage is installed.

[0074] (2) Constructing power flow constraints

[0075]

[0076]

[0077] Among them, P ji,t and Q ji,t x represents the active and reactive power of the branch flowing from node j to node i at time t, respectively; ij H(i) represents the branch reactance between node i and node j; H(i) represents the set of branches associated with node i. This represents the reactive load of the user at node i at time t.

[0078] (3) Construct voltage drop constraints between adjacent nodes:

[0079]

[0080]

[0081] U i,t and U j,t These represent the voltages at the beginning and end nodes of a branch with node i as the first segment and node j as the last segment, respectively.

[0082] (4) Establish branch current and node voltage constraints

[0083]

[0084] U i,min ≤U i,t ≤U i,max (7)

[0085] U i,min and U i,max Represent the lower and upper voltage limits of node i, respectively; I ij,max This indicates the upper limit of current for each branch.

[0086] (5) Solve the day-ahead economic dispatch model constructed in step 1 (1)-(4) using commercial software CPLEX to obtain the initial energy storage capacity.

[0087] 2. Construct an island partitioning model

[0088] (1) Constructing the objective function of the island model

[0089] Considering the significant impact of load priority on the power restoration sequence, the total benefit of loads assigned to islands (the product of priority and load quantity) is used as the optimization objective. The objective function of the islanding model is shown in the following equation:

[0090]

[0091]

[0092]

[0093]

[0094] Among them, ST i With st i,t A 0-1 binary variable used to account for secondary power outage constraints. If node i is included in the island at time t, then st i,tThe value is 1 if node i is in the fault range [t1,t2], and 0 otherwise. i,t If both are 1, then ST i If the value is 1, then the value is 0; otherwise, the value is 0. B i,t This represents the revenue value of node i at time t, and is the optimization objective of the island partitioning model. This variable is determined by the active power of the load within the island. and priority weight PR i The product of these elements constitutes the set. Set N represents the set of load points in the distribution network. t1 represents the initial time of the fault, and t2 represents the end time of the fault. A represents the number of islands.

[0095] (2) Constructing distribution network operation constraints

[0096] ① Node voltage constraints

[0097]

[0098] Among them, U min and U max U represents the lower and upper limits of the voltage at each node in the distribution network; i,t This represents the voltage amplitude at node i at time t.

[0099] ② Branch flow constraints

[0100]

[0101] Among them, I ij,t This represents the current in the line between nodes i and j at time t; Represents the neighborhood set of node i; This represents the upper limit of the branch current between nodes i and j at time t.

[0102] ③ Constraints of radial operation of distribution network

[0103]

[0104]

[0105]

[0106]

[0107]

[0108] Where, β ij and β ji ω represents a 0-1 variable representing the line between nodes i and j, used to help determine whether the distribution network maintains a radial operation. ij ω is a 0-1 variable representing the switch state of the line between nodes i and j. When the line is closed, ω ijω is 1, otherwise ω ij It is 0. Ω a Λ represents the set of load points of the a-th isolated island; a This represents the set of load points connected to distributed power sources and energy storage within the a-th island.

[0109] ④ The constraint that the scope of the isolated islands does not overlap

[0110]

[0111]

[0112] Equations (12) and (13) indicate that the two islands cannot contain the same distributed power sources and load nodes. That is, the islands do not contain each other.

[0113] ⑤ Secondary power outage constraints

[0114] The secondary power outage constraint is an important constraint related to the islanding control strategy, expressed by equations (2) and (3). This constraint can be described as follows: the islanding scheme must remain consistent at all times during the fault, and there cannot be any load spikes. i,t State change, i.e., the final island partitioning scheme ST i For each power supply restoration scheme during the fault period [t1,t2], st i,t The intersection of.

[0115] (3) Constructing relevant constraints for the operation of distributed power sources and energy storage

[0116] ① Output constraints of distributed power sources

[0117]

[0118]

[0119] in, and These represent the charge and discharge amounts of the energy storage at time t. The output power of the DG at different times in different islands is PG. i,t Cannot exceed the upper limit The power of the island remains balanced at each fault point.

[0120] ② Operational constraints of energy storage devices

[0121]

[0122]

[0123]

[0124]

[0125]

[0126]

[0127] In equation (23), E i,t E represents the remaining energy storage capacity of node i at time t; min and E max Equation (24) represents the lower and upper limits of the energy storage capacity state, respectively; Equation (25) represents the remaining energy storage capacity state at the initial moment of the fault; Equation (26) represents the recursive formula for energy storage charge and discharge capacity, respectively; Equation (27) restricts the charge and discharge processes from occurring simultaneously, where... Let be a 0-1 variable representing the charging state of node i at time t, which is 1 when charging and 0 otherwise. Let β be a 0-1 variable representing the discharge state of node i at time t, equal to 1 during discharge and 0 otherwise. Equations (27)-(28) represent the single charge-discharge power constraints of the energy storage device. Where β min and β max These are the relevant parameters for the upper and lower limits of energy storage charging and discharging.

[0128] 3. Decoupling of the island partitioning model

[0129] (1) Establish the largest isolated island partitioning unit

[0130] ① Construct a power sufficiency constraint

[0131] The solution for the largest island partitioning unit is a pre-scheme for island partitioning that satisfies the power sufficiency constraint, as shown in equation (29):

[0132]

[0133] The power sufficiency constraint states that during the fault period [t1, t2], the islanding scheme must satisfy the condition that the sum of the total power of DG and ES is greater than the total power of the loads within the island; otherwise, the islanding scheme will inevitably fail to meet the power constraint. This scheme aims to determine the optimal range of islanding to ensure that more critical loads can be restored during the fault period.

[0134] ② Decouple the island partitioning model obtained in step 2 to obtain the largest island partitioning unit.

[0135] OF1: Eq.(8)

[0136]

[0137] (2) Establish an islanded power verification model

[0138] The island partitioning model obtained in step 2 is decoupled to obtain the island power verification model. The island power verification model based on power balance information is shown in equation (31).

[0139]

[0140]

[0141] The islanding solution optimized by the objective function comes from the solution of the largest islanding unit, rather than from all load nodes in the distribution network.

[0142] 4. Solving for the maximum island partitioning unit based on a heuristic look-ahead greedy algorithm

[0143] Assuming the system fails during the period [t1, t2], the process for formulating the maximum island partitioning scheme based on the heuristic look-ahead greedy algorithm is as follows:

[0144] (1) Based on the fault situation at time t1, update the distribution network topology, determine the downstream power outage area caused by this fault, and denote the topology map corresponding to this area as graph G.

[0145] (2) Select the distributed power sources to be restored in descending order of their access capacity. If all distributed power sources are marked, proceed to step (7).

[0146] For selected distributed power sources and energy storage devices, their access nodes are included in set Λ. a In this context, this node serves as the initial point for the islanding problem. The set of nodes to be islanded is denoted as V. The sum of the active power loads P of all nodes in set V during the fault period is calculated. V Island revenue B V The maximum capacity of distributed power sources and energy storage is calculated using the following formula:

[0147]

[0148]

[0149]

[0150] (3) Search for neighboring points and look-ahead neighboring points, and calculate their value ratio.

[0151] The search for power restoration paths is based on a look-ahead greedy algorithm to maximize the islanding benefit. The specific search scheme is implemented by comparing the load value ratio between the first-neighbor and the look-ahead neighbor: for the first-neighbor set NE... 1 The load points in the set V have a topological connection with at least one load point in the island set V, but a neighborhood set NE 1The intersection with the isolated island set V is an empty set. For the lookahead neighborhood set... The load points in the [database], and their neighborhood set NE 1 At least one load point in the network has a topological connection, but a neighborhood set NE 1 With forward-looking neighborhood set The intersection of the island set V and the lookahead neighborhood set is the empty set. The intersection of these sets is also an empty set. A neighboring point NE 1 The value of a neighborhood of (m) is greater than Va 1 (m) can be calculated using the following formula:

[0152]

[0153] C R P represents the maximum remaining battery capacity; V (i) represents the active load of node i, B V (i) represents the load revenue of node i, which is the active load P. V (i) with load weight PR i The accumulation of. NE 1 (m) represents a neighborhood set NE 1 The m-th vertex in the set NE has a neighborhood set. 1 The number of vertices in the middle is N0.

[0154] One neighboring point NE 1 (m) and the lookahead neighbor points The value of the combination is Va m (n) can be calculated as follows:

[0155]

[0156] Represents NE 1 The nth neighbor of (m), the lookahead neighborhood set Let N be the number of mid-vertex. m .

[0157] (4) Select the solution with the highest value ratio and include the corresponding load node in the island. If Va is found... max If the value is 0, proceed to step (6).

[0158] The optimal value ratio Va of the island set V max The maximum value of the neighborhood value ratio and the forward value ratio can be calculated by the following formula:

[0159]

[0160] (5) Update the remaining power of distributed power sources and energy storage.

[0161] If Va max If it is not zero, Va max The corresponding solution is incorporated into the island set V. In cases where the value ratios are the same, nodes with higher active power loads are prioritized for inclusion. The remaining power C is updated. R Return to step (3).

[0162]

[0163] (6) If Va is found max If the value is 0, the search for the island solution starting from this DG node ends. Mark the DG node and return to step (2).

[0164] (7) Compress all nodes in the island set V into a new node s. a

[0165] For compressed node s a The interior belongs to set Λ a The load point will transfer the remaining distributed power C. R Merge into a new distributed power access compression node s a Disconnect set Ω a Search for and compress node s based on the original topological connections of the medium-load point. a A neighborhood set, connecting the load points in the set with the compression nodes s a Establish topology connections one by one and update the distribution network topology diagram G.

[0166] (8) Determine whether each isolated island satisfies the radial operation constraint. If a ring network exists, apply Prim's algorithm for minimum spanning trees to break the ring network and obtain the maximum island partitioning scheme with a radial structure.

[0167] (9) Node voltage and branch power flow verification.

[0168] The final islanding scheme also needs to satisfy node voltage and branch power flow constraints (Eq.(12)~(13)) to ensure the safe and stable operation of the island during faults.

[0169] 5. Solving isolated check cells using the CPLEX commercial solver

[0170] Using the commercial software CPLEX, the islanded power verification model shown in formula (31) is solved to obtain the system power shortage result. The verified result is the final islanding scheme. The overall solution process of the islanding model is as follows: Figure 1 As shown.

[0171] Although the present invention has been described above, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many modifications under the guidance of the present invention without departing from the spirit of the present invention, and these modifications are all within the protection scope of the present invention.

Claims

1. A fast islanding method for smart distribution networks based on a look-ahead greedy algorithm, characterized in that, Includes the following steps: Step 1: Establish a day-ahead economic dispatch model and calculate the initial energy storage capacity; including: The operating costs of the distribution network are divided into electricity purchase costs and electricity sales revenue, distribution network loss costs, and energy storage loss costs. With cost minimization as the objective function, a day-ahead economic dispatch model is constructed taking into account the operating constraints of the distribution network. The model is then solved using the commercial software CPLEX to obtain the initial energy storage capacity. Step 2: Constructing the Islanding Model: Taking the load benefits of being islanded as the optimization objective, an islanding model is constructed based on distribution network operation constraints, distributed generation, and energy storage-related operation constraints; that is: in, ST i and st i,t A 0-1 binary variable used to account for secondary power outage constraints; if the node i At any moment t If it is incorporated into an isolated island, then st i,t If the node is 1, otherwise it is 0; i During the fault [ t 1, t 2] st i,t If both are 1, then ST i It is 1 if it is true, otherwise it is 0; B i,t Represents a node i exist t The revenue value at any given time is the optimization objective of the island partitioning model. This variable is determined by the active power of the loads within the island. and priority weight PR i The product of; sets N Represents the set of load points in the distribution network; t 1 indicates the initial moment of the fault. t 2 indicates the time when the fault ended; A Indicates the number of isolated islands; Step 3: Decoupling process of the island partitioning model: The islanding model is processed using a power sufficiency information algorithm to obtain the largest islanding unit; including:

301. Construct a power sufficiency constraint according to the following formula: Where: power adequacy constraint indicates that during a fault [ t 1, t Within [2], the islanding scheme should satisfy the requirement that the sum of the total power of DG and ES is greater than the total power of the load within the island; 302. Decoupling the island partitioning model to obtain the largest island partitioning unit: The island partitioning model is processed by a power balance algorithm to obtain island verification units; Step 4: Obtain a preliminary islanding scheme that satisfies the power sufficiency constraint of the distribution network by solving the maximum islanding partitioning unit using a heuristic look-ahead greedy algorithm. Step 5: Process the island check cells using the CPLEX commercial solver to obtain the optimal island partitioning scheme.

2. The method for fast islanding partitioning of smart distribution networks based on a look-ahead greedy algorithm according to claim 1, characterized in that, In Step 3, the island partitioning model is processed using a power balance algorithm to obtain the island verification unit: 。 3. The method for fast islanding partitioning of smart distribution networks based on a look-ahead greedy algorithm according to claim 1, characterized in that, Step 4 involves using a heuristic look-ahead greedy algorithm to solve for the maximum islanding unit and obtain a pre-scheme for islanding that satisfies the power sufficiency constraint of the distribution network. This includes the following steps:

401. According to the distribution network t At time 1, the fault update of the distribution network topology is performed, the downstream power outage area caused by the fault is determined, and the topology corresponding to the downstream power outage area is denoted as Figure 1. G； 402. Select the distributed power sources to restore power in descending order of their access capacity; if all distributed power sources are marked, proceed to step (7); Selected distributed power sources and energy storage devices are connected to the node set. Establish the initial set of points for the island partitioning problem; Select the set of node loads on the island V Calculate the node load set during the fault period respectively. V The sum of the active loads of all nodes in the process P V Island revenue B V The maximum capacity of distributed power sources and energy storage is calculated using the following formula:

403. Using a look-ahead greedy algorithm to process the node load set of the island. V Search to obtain the set of load nodes along the power restoration path. NE 1 ; The neighborhood points of the load node set of the power supply path are calculated using the following formula. NE 1 ( m The value of ) is higher than Va 1 ( m ): (35) In the formula ：C R Indicates the maximum remaining battery power; P V ( i ) represents a node i Active load, B V ( i ) represents a node i The load benefit is the active load. P V ( i ) and load weight PR i The accumulation of; NE 1 ( m ) represents the neighborhood set NE 1 The Middle m vertex, neighborhood set NE 1 The number of mid vertices is N 0; 404. Using a look-ahead greedy algorithm to analyze the set of load nodes along the power supply path. NE 1 Search to obtain the set of forward neighboring points. NE m 2 ; The power restoration path area point is calculated using the following formula. NE 1 ( m ) and look-ahead neighborhood points NE m 2 ( n The value ratio of the combination Va m ( n ): In the formula: NE m 2 ( n )represent NE 1 ( m ) n 1 neighboring point, lookahead neighborhood set NE m 2 The number of mid-vertices is denoted as N m ; 405. Calculate the optimal value ratio for island partitioning using the following formula, and include the corresponding load nodes in the partitioned islands; 406. Determine the optimal value ratio for the partitioning. Va max If the value is zero, the search for the island pre-solution starting from the DG node ends, the DG node is marked, and the process returns to Step 2; otherwise, ... Va max The corresponding solutions are included in the island set. V In cases where the value ratios are the same, nodes with larger active power loads will be prioritized and their remaining power will be updated according to the following formula. C R Return to Step 3:

407. The Prim algorithm for minimum spanning trees is used to process the maximum island partitioning scheme to obtain the final island partitioning scheme with radial structure; 408. Constraints based on node voltage and branch power flow ( Eq Verify the final island partitioning scheme. (12)~(13))

Images