An active distribution network component repair method and system considering network reconstruction and fault repair rolling optimization

Abstract

The present invention discloses a method and system for repairing components of an active distribution network that considers rolling optimization of network reconstruction and fault repair. The method comprehensively considers the correlation between optimized reconstruction and component repair, and includes the following two steps in sequence: first, an active distribution network fault repair model is established with the goal of minimizing repair time and power outage loss, and the state transition probability and evaporation coefficient in the traditional ant colony algorithm are adaptively adjusted to solve the model; secondly, a rolling optimization strategy for component repair and optimized reconstruction is proposed. By coordinating and optimizing the reconstruction and repair plans during the repair process, the shortcomings of the fault repair recovery strategy obtained by using the instantaneous value of the load at the time of the fault are overcome. The method proposed in the present invention fully considers the time-varying demand of the distribution network load and the output fluctuation of distributed power sources. It can shorten the power outage time of users while taking into account the economy of system operation, which helps to improve the fault recovery effect.

Description

Technical Field

[0001] The present invention belongs to the technical field of distribution networks, and in particular relates to an active distribution network component repair method that considers network reconstruction and fault repair rolling optimization. Background Art

[0002] Active distribution networks are platforms that connect power-consuming devices such as photovoltaic energy sources, energy storage, and various AC and DC loads. To prevent a chain reaction that could lead to further expansion of the fault's scope and minimize losses from power outages, it's necessary to consider both indirect restoration of the deprived area through recovery resources like DGs and tie switches, and direct restoration of the deprived loads through manual repair of faulty line components. Active distribution networks rapidly recover from faults immediately after a fault occurs, creating isolated islands. In these situations, maintenance personnel prioritize repairing the faulty line upstream of the island to quickly restore the load in the deprived area. Developing a fault repair strategy that allows maintenance teams to reach the fault point as quickly as possible after a disaster and, through rational decision-making, restore the distribution network to its pre-fault operating state at a limited cost are pressing challenges in the repair of active distribution network components.

[0003] Repairing components after a fault in an active distribution network is a multi-objective, multi-constrained combinatorial optimization problem. Current research on component repair in distribution networks focuses primarily on optimizing repair paths and improving optimization algorithms. However, existing component repair research rarely considers fluctuations in DG output and load demand during the repair process. The resulting repair strategies are not necessarily globally optimal, and traditional fault recovery schemes lack comprehensive consideration of the synergistic effects of network reconstruction and component repair. Consequently, some power-deprived loads must wait until all fault lines are repaired before power can be restored. Therefore, it is urgent to conduct research on component repair based on the above issues, thereby reducing the area of power outage and economic losses after a fault occurs, and providing favorable technical support for improving the resilience of distribution networks. Summary of the Invention

[0004] The technical problem to be solved by the present invention is that the current research on active distribution network component restoration rarely improves the system power supply capacity through network reconstruction operations during emergency repairs. In fact, the DG output fluctuates.

[0005] To solve the above problems, the present invention provides an active distribution network component repair method that considers network reconstruction and fault repair rolling optimization, comprising the following steps:

[0006] A: Collect DG, load output data and topology data after the active distribution network fails, and read the geographical location of each node and line in the active distribution network;

[0007] B: Calculate the required repair time and driving distance between faulty nodes after the active distribution network failure based on the data and geographic location obtained in step A. Perform k-means clustering on the node coordinates to obtain the geographic location of the cluster center as the maintenance center address.

[0008] C: Using the randomly generated maintenance personnel's maintenance path as the decision variable, an active distribution network fault repair model is established considering the maintenance team's repair time and the economic loss of power outages. The model is solved using an adaptive ant colony algorithm to obtain the initial fault repair sequence, and the fault line that comes first in the repair sequence is repaired. The maintenance path is a closed-loop path with the maintenance starting point and end point both being the maintenance center address obtained in step B.

[0009] D: Based on the network topology updated in step C, the active distribution network grid structure and the output of each distributed generation are taken as optimization targets, and the minimization of the current distribution network operating cost is used as the objective function to obtain the corresponding grid topology and DG output results;

[0010] E: Determine whether all fault points have been repaired. If so, output the final repair sequence and the optimized reconstruction strategy formulated after each single line repair. If not, return to step C.

[0011] In the above-mentioned active distribution network component repair method considering network reconstruction and fault repair rolling optimization, the active distribution network fault repair model in step C includes the following parts:

[0012] (1) For the fault repair model, the repair order of maintenance personnel is used as the decision variable, and the repair time of the maintenance team and the economic loss of power outage are considered. The objective function is established as follows:

[0013]

[0014] Where C p is the electricity price sold by the distribution network unit, f is the fault line number or the switch number that can be operated; m is the total number of fault lines; i rp Number the fault repair sequence; Number The power outage time caused by the fault line before repair; w is the load importance; ω w is the weight coefficient corresponding to the load importance; Number The power value that can be recovered after the fault line is repaired; C t is the repair time cost coefficient, Repair number The time required to connect the fault line; For maintenance personnel from the fault point To the fault point Time spent.

[0015] (2) Constraints

[0016] ①Node power balance constraint (power flow constraint) is:

[0017]

[0018] Where, P i,i,t , Q i,i,t are the active power and reactive power injected into node i respectively; P i,i,t , Q i,i,t is the voltage amplitude at nodes i and j at time t; C(i) is the set of nodes connected to node i; G ij 、B ij represents the conductance and susceptance of branch ij; δ ij,t is the voltage phase angle difference between node i and node j at time t.

[0019] ② Distribution networks are generally designed in a closed loop and operated in an open loop. Therefore, the radial structure must always be maintained during the distribution network reconstruction and optimization process. This paper uses a spanning tree model to describe the radial topology of the distribution network and establishes the distribution network radial structure constraints as follows:

[0020]

[0021] Where, E ij,t and Z ij,t Both are Boolean variables; Z ij,t Indicates whether branch ij is connected at time t, E ij,t and E ji,t It is an auxiliary variable related to branch ij, indicating the parent-child relationship. Assuming that the flow direction of branch ij at time t is from node j to node i, then E ij,t = 1. The first item restricts the nodes at both ends of the branch from being each other's parent nodes, ensuring the bidirectionality of the branch ij power flow; the second item restricts the nodes other than the substation node to have only one parent node; the third item indicates that the substation node has no parent node.

[0022] ③Node voltage constraint is:

[0023] V i min ≤V i,t ≤V i max

[0024] Where V i max 、V i min represent the maximum and minimum voltages at node i, respectively.

[0025] ④ Branch current constraint is

[0026]

[0027] Where, is the maximum current allowed to flow through branch ij.

[0028] ⑤ For active distribution networks, the integration of wind power and photovoltaic power improves the system's redundancy and the ability to continuously supply power to critical loads. At the same time, the upper and lower limits of distributed power output will also be affected by typhoon disasters. With the continuous development of related control technologies, solar photovoltaic power stations and wind power stations have corresponding adjustment capabilities. This invention simplifies the distributed power supply into an adjustable PQ model for calculation and analysis:

[0029]

[0030] Where, P iDG,t and Q iDG,t Respectively represent the i DG The active and reactive output of the DG, and They represent the upper and lower limits of the active power output of the i-th DG, and They represent the upper and lower limits of reactive power output of the i-th DG respectively.

[0031] ⑥ During the maintenance process, the amount of materials that the maintenance team can carry in a single trip must comply with the total material capacity limit. The maintenance constraints are:

[0032]

[0033] Where g k is the maintenance material required for the kth faulty equipment, g max The maximum amount of supplies a maintenance team can carry.

[0034] ⑦ It is generally assumed that the fault point will not be damaged again after repair, that is, a single fault point can only be repaired once, and the single repair constraint of the fault point is:

[0035]

[0036] Where r f,i is a 0-1 variable, r f,i 1 means the i-th r The repair point is the fault point f, otherwise it is 0.

[0037] In the above-mentioned active distribution network component repair method considering network reconstruction and fault repair rolling optimization, the active distribution network fault repair model in step C is solved using an adaptive ant colony algorithm.

[0038] During the search process of the traditional ant colony algorithm, the concentration of pheromones will change due to factors such as the time the ants spend traversing the path and the quality of the path. Ants use the roulette method to decide the direction of transfer based on the pheromone concentration on different paths. It represents the selection probability of ant k transferring from fault point i to fault point j at time t:

[0039]

[0040] In the formula, allowed k Represents the set of unrepaired fault points, is the pheromone concentration distributed between fault point i and fault point j, is the heuristic factor, which indicates the expected degree of ants moving from fault site i to fault site j; α and β are the relative importance strengths of pheromone and heuristic factor, respectively. The larger the value, the higher the probability that the ants will move to the roadblock point j in the next stage. When all ants complete a round trip, the pheromone remaining on the path needs to be updated. The pheromone concentration on the path (i, j) is updated by the following formula:

[0041] τ ij (t+n)=(1-ρ)τ ij (t)+Δτ ij (t)

[0042]

[0043]

[0044] Where ρ (0<ρ<1) represents the pheromone evaporation coefficient, and 1-ρ represents the persistence of the pheromone; It represents the amount of pheromone on the path (i, j) after ant k completes one iteration, Q is a constant, L k is the total length of the path traveled by the kth ant in this cycle.

[0045] The present invention improves the state transition rule of ants and uses an adaptive method to adjust the state transition probability to meet the needs of different search stages. Define the new state transition probability after each cycle as follows

[0046]

[0047] Where q is a random number between 0 and 1. The value of q0 determines the relative importance of utilizing prior information versus exploring new paths. When q ≤ q0, the path is selected based on the prior information. Otherwise, a random proportional selection method is used to select a path.

[0048] This section simulates the evaporation status of real-world ant pheromones. When there is no obvious improvement after continuous iteration of the optimal value, the pheromone evaporation coefficient ρ is adaptively adjusted, and ρ is reduced to

[0049] ρ(t) = max[0.95ρ0(t - 1), ρ min

[0050] where ρ0 is the initial value of ρ, and ρ min is the minimum value of ρ, which can prevent ρ from being too small and reducing the convergence speed of the algorithm.

[0051] In the above active distribution network component repair method considering network reconstruction and rolling optimization of fault repair, the specific steps of the adaptive ant colony algorithm to solve the active distribution network fault repair model in step C are as follows:

[0052] Step 1: Initialize the parameters of the ant colony algorithm and the taboo list, select n ant ants and place them on m faulty lines, and initialize the pheromone amount τ ij (0) = c on the distribution network line (i, j), and at the initial moment, Δτ ij (0) = 0; set the initial value ρ0 of the pheromone evaporation coefficient, the importance degree α of the pheromone, the importance degree β of the heuristic factor, and the maximum number of iterations G c,max .

[0053] Step 2: For the set of faulty lines to be repaired, each ant individual selects a faulty line and moves forward according to the probability calculated by the state transition formula of the adaptive ant colony algorithm.

[0054] Step 3: Modify the taboo list pointer and determine whether there exists k < m. If so, jump to Step 3; otherwise, continue to complete a tour of the artificial ant, and finally form a fault repair strategy to update the amount of information on each path.

[0055] Step 4: Calculate the objective function F in formula (4.1) for the artificial ants that meet the constraint conditions b , and store the minimum value F of the objective function at the current iteration bmin .

[0056] Step 5: If the number of iterations G c > G c,max then end the loop and output the program optimization result; otherwise, clear the taboo list and re-initialize the ant colony.

[0057] In the above active distribution network component repair method considering network reconstruction and rolling optimization of fault repair, the active distribution network optimization reconstruction model in step D takes minimizing the current moment's distribution network operation cost as the objective function, which can be expressed as:

[0058] ​

[0059] Where, F a is the operating cost of the distribution network, t is the duration of network optimization and reconstruction, and in this paper, t = 1h; the first item is the cost of purchasing electricity from the transmission network, C1 represents the cost coefficient of the distribution network purchasing electricity from the transmission network, P t represents the active power output of the substation node at time t; the second and third items are the output costs of distributed generation, C PV Represents the photovoltaic electricity sales cost, N PV Represents the total number of PVs connected to the distribution network, represents the i-th PV PV output power, C WT represents the cost of selling wind power, N WT Represents the total number of WTs connected to the distribution network, represents the i-th WT The fourth item represents the economic loss caused by the distribution network loss, C2 is the cost of the distribution network loss, represents the i-th br The fifth item represents the switch operation cost, C3 represents the switch operation cost of the distribution network, k is the switch number, SW is the switch set in the distribution network, sw k,0 and sw k,1 Indicates the switch status before and after network reconstruction.

[0060] In the above-mentioned active distribution network component repair method considering network reconstruction and fault repair rolling optimization, the active distribution network optimization reconstruction model in step D is solved using the YALMIP toolbox in Matlab calling the cplex algorithm package. Therefore, the power flow constraints and branch current constraints need to be transformed into second-order cone relaxations, and the definition is:

[0061]

[0062] Substituting the above equation into the power flow constraint and current constraint, it can be transformed into

[0063]

[0064] Further relax the last term, that is

[0065]

[0066] It can be further deformed into a standard second-order cone form

[0067]

[0068] An active distribution network component repair system considering network reconstruction and fault repair rolling optimization includes the following steps:

[0069] The first module is configured to collect DG, load output data and topology data after the active distribution network fails, and read the geographical location of each node and line of the active distribution network;

[0070] The second module is configured to calculate the required repair time and driving distance between fault nodes after the active distribution network fault based on the data and geographic location obtained by the first module, and perform k-means clustering on the node coordinates to obtain the geographic location of the cluster center as the maintenance center address;

[0071] The third module is configured to use the randomly generated maintenance path of maintenance personnel as a decision variable, establish an active distribution network fault repair model considering the repair team's repair time and the economic loss of power outages, use the adaptive ant colony algorithm to solve the model to obtain an initial fault repair sequence, and repair the fault line that is first in the repair sequence. The maintenance path is a closed-loop path consisting of the maintenance starting point and end point being the maintenance center address obtained by the second module.

[0072] The fourth module is configured to calculate the corresponding grid topology and DG output results based on the network topology updated in the third module, taking the active distribution network grid structure and the output of each distributed generation as optimization targets, and minimizing the current distribution network operating cost as the objective function;

[0073] The fifth module is configured to determine whether all fault points have been repaired. If so, it outputs the final repair sequence and the optimized reconstruction strategy formulated after each repair of a single line. If the repair is not completed, it returns to the third module.

[0074] This invention first establishes an active distribution network fault repair model with the goal of minimizing repair time and power outage losses. It then adaptively adjusts the state transition probability and evaporation coefficient in the traditional ant colony algorithm to solve the model. Secondly, it proposes a rolling optimization strategy for component repair and optimal reconstruction. By coordinating and optimizing the reconstruction and repair plans during the repair process, it overcomes the limitations of fault repair and recovery strategies that use instantaneous load values at the time of the fault. The proposed method fully considers the time-varying demands of distribution network loads and the output fluctuations of distributed power sources. It can shorten user outages while balancing the economic efficiency of system operation, thus improving fault recovery effectiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 Flowchart of the adaptive ant colony algorithm-assisted fault repair model.

[0076] Figure 2 This is the geographical location structure diagram of the IEEE33 node system.

[0077] Figure 3is the 24-hour power distribution curve of DG and load.

[0078] Figure 4 The cost curve for the distribution network to purchase electricity from the upper level.

[0079] Figure 5 This is the distribution network topology at different time periods when the component repair method proposed in the present invention is performed.

[0080] Figure 6 The distribution network topology at different time periods when only single-stage component repair is performed.

[0081] Figure 7 This is the power supply level of the active distribution network during the component repair process using the method proposed in this invention.

[0082] Figure 8 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0083] The following describes an embodiment of the active distribution network component repair method that takes into account network reconfiguration and fault repair rolling optimization, as described in the present invention, with reference to the accompanying drawings and further detailed description. The described embodiments may be modified in various ways or combinations thereof without departing from the spirit and scope of the present invention. Therefore, the drawings and description are illustrative in nature and are not intended to limit the scope of protection of the claims.

[0084] In specific implementation, the active distribution network component repair method considering network reconstruction and fault repair rolling optimization is presented in this paper. Figure 2 The IEEE 33-node power distribution system including distributed power sources is shown as a specific implementation object, and the fault line repair involved in the present invention is described.

[0085] Example 1

[0086] The present invention provides a method for repairing active distribution network components that considers network reconstruction and fault repair rolling optimization, comprising the following steps:

[0087] A: Collect DG, load output data and topology data after the active distribution network fails, and read the geographical location of each node and line in the active distribution network;

[0088] B: Calculate the required repair time and driving distance between faulty nodes after the active distribution network failure based on the data and geographic location obtained in step A. Perform k-means clustering on the node coordinates to obtain the geographic location of the cluster center as the maintenance center address.

[0089] C: Using the randomly generated maintenance personnel's maintenance path as the decision variable, an active distribution network fault repair model is established considering the maintenance team's repair time and the economic loss of power outages. The model is solved using an adaptive ant colony algorithm to obtain the initial fault repair sequence, and the fault line that comes first in the repair sequence is repaired. The maintenance path is a closed-loop path with the maintenance starting point and end point both being the maintenance center address obtained in step B.

[0090] D: Based on the network topology updated in step C, the active distribution network grid structure and the output of each distributed generation are taken as optimization targets, and the minimization of the current distribution network operating cost is used as the objective function to obtain the corresponding grid topology and DG output results;

[0091] E: Determine whether all fault points have been repaired. If so, output the final repair sequence and the optimized reconstruction strategy formulated after each single line repair. If not, return to step C;

[0092] The active distribution network fault repair model in step C includes the following parts:

[0093] (1) For the fault repair model, the repair order of maintenance personnel is used as the decision variable, and the repair time of the maintenance team and the economic loss of power outage are considered. The objective function is established as follows:

[0094]

[0095] Where C p is the electricity price sold by the distribution network unit, f is the fault line number or the switch number that can be operated; m is the total number of fault lines; i rp Number the fault repair sequence; Number The power outage time caused by the fault line before repair; w is the load importance; ω w is the weight coefficient corresponding to the load importance; Number The power value that can be recovered after the fault line is repaired; C t is the repair time cost coefficient, Repair number The time required to connect the fault line; For maintenance personnel from the fault point To the fault point Time spent.

[0096] (2) Constraints

[0097] ①Node power balance constraint (power flow constraint) is:

[0098]

[0099] Where, P i,i,t , Q i,i,t are the active power and reactive power injected into node i respectively; P i,i,t , Q i,i,t is the voltage amplitude at nodes i and j at time t; C(i) is the set of nodes connected to node i; G ij 、B ij represents the conductance and susceptance of branch ij; δ ij,t is the voltage phase angle difference between node i and node j at time t.

[0100] ② Distribution networks are generally designed in a closed loop and operated in an open loop. Therefore, the radial structure must always be maintained during the distribution network reconstruction and optimization process. This paper uses a spanning tree model to describe the radial topology of the distribution network and establishes the distribution network radial structure constraints as follows:

[0101]

[0102] Where, E ij,t and Z ij,t Both are Boolean variables; Z ij,t Indicates whether branch ij is connected at time t, E ij,t and E ji,t It is an auxiliary variable related to branch ij, indicating the parent-child relationship. Assuming that the flow direction of branch ij at time t is from node j to node i, then E ij,t = 1. The first item restricts the nodes at both ends of the branch from being each other's parent nodes, ensuring the bidirectionality of the branch ij power flow; the second item restricts the nodes other than the substation node to have only one parent node; the third item indicates that the substation node has no parent node.

[0103] ③Node voltage constraint is:

[0104] V i min ≤V i,t ≤V i max

[0105] Where V i max 、V i min represent the maximum and minimum voltages at node i, respectively.

[0106] ④ Branch current constraint is

[0107]

[0108] Where, is the maximum current allowed to flow through branch ij.

[0109] ⑤ For active distribution networks, the integration of wind power and photovoltaic power improves the system's redundancy and the ability to continuously supply power to critical loads. At the same time, the upper and lower limits of distributed power output will also be affected by typhoon disasters. With the continuous development of related control technologies, solar photovoltaic power stations and wind power stations have corresponding adjustment capabilities. This invention simplifies the distributed power supply into an adjustable PQ model for calculation and analysis:

[0110]

[0111] Where, and Respectively represent the i DG The active and reactive output of the DG, and They represent the upper and lower limits of the active power output of the i-th DG, and They represent the upper and lower limits of reactive power output of the i-th DG respectively.

[0112] ⑥ During the maintenance process, the amount of materials that the maintenance team can carry in a single trip must comply with the total material capacity limit. The maintenance constraints are:

[0113]

[0114] Where g k is the maintenance material required for the kth faulty equipment, g max The maximum amount of supplies a maintenance team can carry.

[0115] ⑦ It is generally assumed that the fault point will not be damaged again after repair, that is, a single fault point can only be repaired once, and the single repair constraint of the fault point is:

[0116]

[0117] Where r f,i is a 0-1 variable, r f,i 1 means the i-th r The repair point is the fault point f, otherwise it is 0.

[0118] In step C, the active distribution network fault repair model is solved using the adaptive ant colony algorithm.

[0119] During the search process of the traditional ant colony algorithm, the concentration of pheromones will change due to factors such as the time the ants spend traversing the path and the quality of the path. Ants use the roulette method to decide the direction of transfer based on the pheromone concentration on different paths. It represents the selection probability of ant k transferring from fault point i to fault point j at time t:

[0120]

[0121] In the formula, allowed k Represents the set of unrepaired fault points, is the pheromone concentration distributed between fault point i and fault point j, is the heuristic factor, which indicates the expected degree of ants moving from fault site i to fault site j; α and β are the relative importance strengths of pheromone and heuristic factor, respectively. The larger the value, the higher the probability that the ants will move to the roadblock point j in the next stage. When all ants complete a round trip, the pheromone remaining on the path needs to be updated. The pheromone concentration on the path (i, j) is updated by the following formula:

[0122] τ ij (t+n)=(1-ρ)τ ij (t)+Δτ ij (t)

[0123]

[0124]

[0125] Where ρ (0<ρ<1) represents the pheromone evaporation coefficient, and 1-ρ represents the persistence of the pheromone; It represents the amount of pheromone on the path (i, j) after ant k completes one iteration, Q is a constant, L k is the total length of the path traveled by the kth ant in this cycle.

[0126] The present invention improves the state transition rule of ants and uses an adaptive method to adjust the state transition probability to meet the needs of different search stages. Define the new state transition probability after each cycle as follows

[0127]

[0128] Where q is a random number between 0 and 1. The value of q0 determines the relative importance of utilizing prior information versus exploring new paths. When q ≤ q0, the path is selected based on the prior information. Otherwise, a random proportional selection method is used to select a path.

[0129] This section simulates the evaporation status of ant pheromones in the real world. When the optimal value has not been significantly improved after continuous iteration, the pheromone evaporation coefficient ρ is adaptively adjusted to reduce ρ to

[0130] ρ(t)=max[0.95ρ0(t-1),ρ min ]

[0131] Where ρ0 is the initial value of ρ, ρ minis the minimum value of ρ, which can prevent ρ from being too small and reducing the algorithm convergence speed.

[0132] The specific steps of the adaptive ant colony algorithm to solve the active distribution network fault repair model in step C are as follows:

[0133] Step 1: Initialize the parameters of the ant colony algorithm and the taboo table, and select n ant ants and place them on m faulty lines. Initialize the amount of pheromone τ ij (0) = c, and at the initial moment, Δτ ij (0) = 0; set the initial value ρ0 of the pheromone evaporation coefficient, the importance degree α of the pheromone, the importance degree β of the heuristic factor, and the maximum number of iterations G c,max .

[0134] Step 2: For the set of faulty lines to be repaired, ant individuals select faulty lines and move forward according to the probability calculated by the state transition formula of the adaptive ant colony algorithm.

[0135] Step 3: Modify the taboo table pointer and determine whether there exists k < m. If so, jump to step 3; otherwise, continue to complete a round of the artificial ant, and finally form a fault repair strategy to update the amount of information on each path.

[0136] Step 4: Calculate the objective function F in formula (4.1) for the artificial ants that meet the constraint conditions b , and store the minimum value F of the objective function at the current iteration bmin .

[0137] Step 5: If the number of iterations G c > G c,max then end the loop and output the program optimization result; otherwise, clear the taboo table and re-initialize the ant colony.

[0138] The active distribution network optimal reconfiguration model in step D takes minimizing the current moment's distribution network operation cost as the objective function, which can be expressed as:

[0139]

[0140] In the formula, F a is the operation cost of the distribution network, t is the duration of network optimal reconfiguration, and in this paper, t = 1h; the first term is the cost of purchasing electricity from the transmission network, C1 represents the cost coefficient of the distribution network purchasing electricity from the transmission network, and P t represents the active power output of the substation node at time t; the second and third terms are the output costs of distributed power sources, C PV represents the photovoltaic power sales cost, and N PV represents the total number of PVs connected to the distribution network, represents the output power of the i PV th PV at time t, CWT represents the cost of selling wind power, N WT Represents the total number of WTs connected to the distribution network, represents the i-th WT The fourth item represents the economic loss caused by the distribution network loss, C2 is the cost of the distribution network loss, represents the i-th br The fifth item represents the switch operation cost, C3 represents the switch operation cost of the distribution network, k is the switch number, SW is the switch set in the distribution network, sw k,0 and sw k,1 Indicates the switch status before and after network reconstruction.

[0141] In step D, the active distribution network optimization reconstruction model is solved by calling the cplex algorithm package using the YALMIP toolbox in Matlab. Therefore, the power flow constraints and branch current constraints need to be transformed into second-order cone relaxations and defined as follows:

[0142]

[0143] Substituting the above equation into the power flow constraint and current constraint, it can be transformed into

[0144]

[0145] Further relax the last term, that is

[0146]

[0147] It can be further deformed into a standard second-order cone form

[0148]

[0149] Example 2

[0150] To verify the method of the present invention, the present invention adopts Figure 1 The IEEE 33-node distribution system shown in the example is tested. The system base capacity is 10 MVA, the base voltage is 12.66 kV, the system base load is 3715 + j2300 kVA, the active power loss under normal operating conditions is 202.68 kW, and the minimum node voltage is 0.9131 pu. The DG types in this example include photovoltaic and wind turbines. Basic data of the photovoltaic and wind turbines are shown in Table 1.

[0151] Table 1. Distributed power parameters connected to a 33-node system

[0152]

[0153] The time-varying characteristics of the DG output and the load per unit value of each node in the distribution network in a typical day are set as follows: Figure 3 As shown in Figure 2, the actual output of PV panels on cloudy days is calculated using the per-unit PV output and the cloudy weather attenuation coefficient. The actual output of wind turbines is calculated by superimposing the per-unit wind power output and the typhoon wind speed. The actual output of PV and WT on a typical day under the influence of a typhoon can be determined. The load importance and weights, as well as the parameters related to distribution companies and power users, are shown in Tables 2 and 3, respectively.

[0154] Table 2. Load importance and weight

[0155]

[0156] Table 3 Parameters related to distribution companies and electricity users

[0157]

[0158] Assume that lines 8, 11, 16, 24, 25, and 32 have permanent faults at 12:00. First, the contact switches are actuated to quickly recover from the fault. The calculated topology is as follows: Figure 5 As shown in (a), further component repair optimization is performed based on the above results. It is assumed that the maintenance personnel will depart from the maintenance center to the fault point at 12:00. The geographical location of each fault point and the emergency repair time are set as shown in Table 4. In this embodiment, two component repair plans are set for comparative analysis.

[0159] Solution 1: Using the method of the present invention, taking into account the time-varying nature of load demand and DG output, the distribution network optimization and reconstruction and fault repair rolling optimization are carried out. The optimization and reconstruction plan and the repair plan are adjusted every time a fault line is repaired.

[0160] Option 2: Single-stage emergency repair, that is, without considering distribution network optimization and reconstruction and fault repair rolling optimization, the fault is directly repaired based on the initial repair order of the first solution after rapid fault recovery.

[0161] Table 4 Location of each fault point and repair time requirements

[0162]

[0163] The single-stage emergency repair without considering the optimization reconstruction and the reconstruction and fault repair rolling optimization method proposed in this paper are used to repair the fault components. The distribution network topology structures obtained in different time periods are as follows: Figure 5 and Figure 6 shown.

[0164] There are 6 fault lines in the fault loop. Substituting them into the fault repair model, we can directly solve the optimal repair sequence of the fault components: 25-24-32-16-11-8. Option 1: First, repair the fault line 25. By closing the section switches 24-25 and the tie switches 25-29 on both sides of the fault, the load at node 25 is restored in advance. The optimal grid structure is obtained as follows: Figure 5 (b) shows. Secondly, Figure 5 Based on (b), the repair team's repair time and the economic losses from the power outage are taken into account. The repair order for all faults except line 25 is redefined. This problem is solved repeatedly. After each fault is repaired, the maintenance personnel check whether the distribution network has been effectively reconfigured and optimize it. This process continues until all six faults are repaired, resulting in the final fault recovery topology.

[0165] Ultimately, the fault repair sequence obtained using option 1 for component repair was 25-32-16-11-8-24. After repairing faulty line 32, a valid reconstruction was detected. Closing the section switches 32-33 on both sides of the line restored power to the lost loads at nodes 17, 18, and 33, effectively restoring power to all loads. Maintenance personnel then repaired faulty lines 16, 11, 8, and 24 in sequence. After each repair, they optimized the grid structure and DG output with the goal of minimizing the operating cost of the active distribution network. They opened tie switches 18-33 and 9-15 in sequence. After line 8 was repaired, tie switch 9-15 was closed. Section switches 9-10 and 14-15 were opened. After line 24 was repaired, tie switch 18-33 was closed, and section switches 17-18 and 28-29 were opened. The optimal repair sequence obtained using option 1 for component repair was 25-24-32-16-11-8.

[0166] Table 5 Component repair technical indicators for scenario 2 (Scheme 1)

[0167]

[0168]

[0169] Table 6 Component repair technical indicators for scenario 2 (Scheme 2)

[0170]

[0171] Comparing the parameters in Tables 5 and 6, it can be seen that the component repair plan formulated by the method in this paper restored power to all loads after repairing the fault line 32, which took 5 hours and 10 minutes, and the load power supply was 41622.15kW·h; while the single-stage emergency repair restored all loads after repairing the fault line 16, which took 15 hours and 43 minutes, and the load power supply was 38883.59kW·h; in contrast, the method proposed in this paper took a longer driving distance, and the total component repair time was slightly higher than the single-stage emergency repair. This is because the reconstruction of the method in this paper after repairing individual faulty components can quickly restore the load in the non-fault area, so that some of the power-lost loads do not need to wait until all faulty components are repaired before restoring power, thereby changing the economic loss of the power outage of the corresponding fault line. Since the fault repair model C in the simulation experiment in this section t The experience value setting is smaller, so the priority of line repair is changed.

[0172] Figure 7 This reflects the load level of the active distribution network during the component repair phase. As shown in the figure, the load levels of the active distribution network corresponding to the two schemes are equal during the periods from 12:00 PM to 2:42 PM and from 3:43 AM to 9:00 AM the next day. This is because both component repair strategies prioritized restoring the islands with the largest loads (26, 27, 28, 29, 30, and 32) at 2:42 PM. Scheme 2 restored all de-energized loads at 3:43 AM, equalizing the system power supply levels. For the remaining periods, F1(t) > F2(t). Combining the analysis of Scenario 1 and Scenario 2, we can see that the proposed optimized reconstruction and fault repair rolling optimization method is more suitable for scenarios with a high fault multiplicity and a large number of fault-generated islands. In particular, for loads that can be restored by tie switches during the repair phase, performing restoration and reconstruction after repairing a single fault can significantly improve the load recovery rate during the component repair phase.

[0173] Example 3

[0174] The present invention also provides an active distribution network component repair system that considers network reconstruction and fault repair rolling optimization, including:

[0175] The first module is configured to collect DG, load output data and topology data after the active distribution network fails, and read the geographical location of each node and line of the active distribution network;

[0176] The second module is configured to calculate the required repair time and driving distance between fault nodes after the active distribution network fault based on the data and geographic location obtained by the first module, and perform k-means clustering on the node coordinates to obtain the geographic location of the cluster center as the maintenance center address;

[0177] The third module is configured to use the randomly generated maintenance path of maintenance personnel as a decision variable, establish an active distribution network fault repair model considering the repair team's repair time and the economic loss of power outages, use the adaptive ant colony algorithm to solve the model to obtain an initial fault repair sequence, and repair the fault line that is first in the repair sequence. The maintenance path is a closed-loop path consisting of the maintenance starting point and end point being the maintenance center address obtained by the second module.

[0178] The fourth module is configured to calculate the corresponding grid topology and DG output results based on the network topology updated in the third module, taking the active distribution network grid structure and the output of each distributed generation as optimization targets, and minimizing the current distribution network operating cost as the objective function;

[0179] The fifth module is configured to determine whether all fault points have been repaired. If so, it outputs the final repair sequence and the optimized reconstruction strategy formulated after each repair of a single line. If the repair is not completed, it returns to the third module.

[0180] The above fully verifies that the method of the present invention performs rolling optimization on the two steps of fault repair and optimization reconstruction in the research on active distribution network component repair, which can fully consider the time-varying demand of distribution network load and the output fluctuation of distributed power sources, shorten the power outage time of users while taking into account the economy of system operation, and help improve the fault recovery effect.

[0181] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, it can still be modified within the spirit and principles of the present invention, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for repairing active distribution network components considering network reconstruction and fault repair rolling optimization, characterized in that: The following steps are involved: A: Collect DG, load output data and topology data after the active distribution network fails, and read the geographical location of each node and line in the active distribution network; B: Calculate the required repair time and driving distance between faulty nodes after the active distribution network failure based on the data and geographic location obtained in step A. Perform k-means clustering on the node coordinates to obtain the geographic location of the cluster center as the maintenance center address. C: Using randomly generated maintenance personnel's repair paths as decision variables, an active distribution network fault repair model is established, taking into account the repair team's repair time and the economic losses caused by power outages. The model is solved using an adaptive ant colony algorithm to obtain an initial fault repair sequence. The fault line that comes first in the repair sequence is repaired. The maintenance path is a closed-loop path with both the starting point and the end point being the repair center address obtained in step B. D: Based on the network topology updated in step C, the active distribution network grid structure and the output of each distributed generation are taken as optimization targets, and the minimization of the current distribution network operating cost is used as the objective function to obtain the corresponding grid topology and DG output results; E: Determine whether all fault points have been repaired. If so, output the final repair sequence and the optimized reconstruction strategy formulated after each single line repair. If not, return to step C.

2. The active distribution network component repair method considering network reconstruction and fault repair rolling optimization according to claim 1 is characterized in that: The active distribution network fault repair model in step C includes a fault repair model and constraints. The objective function of the fault repair model uses the repair order of maintenance personnel as a decision variable and considers the repair time of the maintenance team and the economic loss of power outage: Where C p is the electricity price sold by the distribution network unit, f is the fault line number or the switch number that can be operated; m is the total number of fault lines; i rp Number the fault repair sequence; Number The power outage time caused by the fault line before repair; w is the load importance; ω w is the weight coefficient corresponding to the load importance; Number The power value that can be recovered after the fault line is repaired; C t is the repair time cost coefficient, Repair number The time required to connect the fault line; For maintenance personnel from the fault point To the fault point Time spent.

3. The active distribution network component repair method considering network reconstruction and fault repair rolling optimization according to claim 1 is characterized in that: The active distribution network fault repair model in step C includes a fault repair model and constraint conditions, wherein the constraint conditions include The node power balance constraint is: The spanning tree model is used to describe the radial topology of the distribution network, and the radial structure constraints of the distribution network are established as follows: The node voltage constraints are: In i min ≤V i,t ≤V i max Where V i max 、V i min represent the maximum and minimum voltages of node i respectively; The branch current constraint is Where, is the maximum current allowed to flow through branch ij; The distributed power source is simplified into an adjustable PQ model for calculation and analysis: The maintenance constraints are: It is assumed that the fault point will not be damaged again after repair, that is, a single fault point can only be repaired once. The single repair constraint of the fault point is: Where, P i,i,t , Q i,i,t are the active power and reactive power injected into node i respectively; P i,i,t , Q i,i,t is the voltage amplitude at nodes i and j at time t; C(i) is the set of nodes connected to node i; G ij 、B ij represents the conductance and susceptance of branch ij; δ ij,t is the voltage phase angle difference between node i and node j at time t; E ij,t and Z ij,t Both are Boolean variables; Z ij,t Indicates whether branch ij is connected at time t, E ij,t and E ji,t It is an auxiliary variable related to branch ij, indicating the parent-child relationship. Assuming that the flow direction of branch ij at time t is from node j to node i, then E ij,t =1; the first item restricts the nodes at both ends of the branch from being each other's parent nodes, ensuring the bidirectionality of the branch ij power flow; the second item restricts the nodes other than the substation node to have only one parent node; the third item indicates that the substation node has no parent node; and Respectively represent the i DG The active and reactive output of the DG, and They represent the upper and lower limits of the active power output of the i-th DG, and They represent the upper and lower limits of reactive power output of the i-th DG respectively; g k is the maintenance material required for the kth faulty equipment, g max The maximum amount of supplies that the maintenance team can carry; r f,i is a 0-1 variable, r f,i 1 means the i-th r The repair point is the fault point f, otherwise it is 0.

4. The active distribution network component repair method considering network reconstruction and fault repair rolling optimization according to claim 1 is characterized in that: The active distribution network optimization and reconstruction model in step D takes minimizing the current distribution network operation cost as the objective function, which is expressed as: Where, F a is the operating cost of the distribution network, t is the duration of network optimization and reconstruction, and in this paper, t = 1h; the first item is the cost of purchasing electricity from the transmission network, C1 represents the cost coefficient of the distribution network purchasing electricity from the transmission network, P t represents the active power output of the substation node at time t; the second and third items are the output costs of distributed generation, C PV Represents the photovoltaic electricity sales cost, N PV Represents the total number of PVs connected to the distribution network, represents the i-th PV PV output power, C WT represents the cost of selling wind power, N WT Represents the total number of WTs connected to the distribution network, represents the i-th WT The fourth item represents the economic loss caused by the distribution network loss, C2 is the cost of the distribution network loss, represents the i-th br The fifth item represents the switch operation cost, C3 represents the switch operation cost of the distribution network, k is the switch number, SW is the switch set in the distribution network, sw k,0 and sw k,1 Indicates the switch status before and after network reconstruction.

5. An active distribution network component repair system that considers network reconstruction and fault repair rolling optimization, characterized in that: include: The first module is configured to collect DG, load output data and topology data after the active distribution network fails, and read the geographical location of each node and line of the active distribution network; The second module is configured to calculate the required repair time and driving distance between fault nodes after the active distribution network fault based on the data and geographic location obtained by the first module, and perform k-means clustering on the node coordinates to obtain the geographic location of the cluster center as the maintenance center address; The third module is configured to establish an active distribution network fault repair model using the randomly generated maintenance personnel's repair path as a decision variable, taking into account the repair team's repair time and the economic losses caused by the power outage. The model is solved using an adaptive ant colony algorithm to obtain an initial fault repair sequence, and the fault line that comes first in the obtained repair sequence is repaired. The maintenance path is a closed-loop path with both the repair starting point and the end point being the maintenance center address obtained by the second module. The fourth module is configured to calculate the corresponding grid topology and DG output results based on the network topology updated in the third module, taking the active distribution network grid structure and the output of each distributed generation as optimization targets, and minimizing the current distribution network operating cost as the objective function; The fifth module is configured to determine whether all fault points have been repaired. If so, it outputs the final repair sequence and the optimized reconstruction strategy formulated after each repair of a single line. If the repair is not completed, it returns to the third module.

6. The active distribution network component repair system considering network reconstruction and fault repair rolling optimization according to claim 5 is characterized in that: The active distribution network fault repair model in the third module includes a fault repair model and constraints. The objective function of the fault repair model uses the repair order of maintenance personnel as a decision variable and considers the repair time of the maintenance team and the economic loss of power outage: Where C p is the electricity price sold by the distribution network unit, f is the fault line number or the switch number that can be operated; m is the total number of fault lines; i rp Number the fault repair sequence; Number The power outage time caused by the fault line before repair; w is the load importance; ω w is the weight coefficient corresponding to the load importance; Number The power value that can be recovered after the fault line is repaired; C t is the repair time cost coefficient, Repair number The time required to connect the fault line; For maintenance personnel from the fault point To the fault point Time spent.

7. The active distribution network component repair system considering network reconstruction and fault repair rolling optimization according to claim 5 is characterized in that: The active distribution network fault repair model in the third module includes a fault repair model and constraint conditions, wherein the constraint conditions include The node power balance constraint is: The spanning tree model is used to describe the radial topology of the distribution network, and the radial structure constraints of the distribution network are established as follows: The node voltage constraints are: In i min ≤V i,t ≤V i max Where V i max 、V i min represent the maximum and minimum voltages of node i respectively; The branch current constraint is Where, is the maximum current allowed to flow through branch ij; The distributed power source is simplified into an adjustable PQ model for calculation and analysis: The maintenance constraints are: It is assumed that the fault point will not be damaged again after repair, that is, a single fault point can only be repaired once. The single repair constraint of the fault point is: Where, P i,i,t , Q i,i,t are the active power and reactive power injected into node i respectively; P i,i,t , Q i,i,t is the voltage amplitude at nodes i and j at time t; C(i) is the set of nodes connected to node i; G ij 、B ij represents the conductance and susceptance of branch ij; δ ij,t is the voltage phase angle difference between node i and node j at time t; E ij,t and Z ij,t Both are Boolean variables; Z ij,t Indicates whether branch ij is connected at time t, E ij,t and E ji,t It is an auxiliary variable related to branch ij, indicating the parent-child relationship. Assuming that the flow direction of branch ij at time t is from node j to node i, then E ij,t =1; the first item restricts the nodes at both ends of the branch from being each other's parent nodes, ensuring the bidirectionality of the branch ij power flow; the second item restricts the nodes other than the substation node to have only one parent node; the third item indicates that the substation node has no parent node; and Respectively represent the i DG The active and reactive output of the DG, and They represent the upper and lower limits of the active power output of the i-th DG, and They represent the upper and lower limits of reactive power output of the i-th DG respectively; g k is the maintenance material required for the kth faulty equipment, g max The maximum amount of supplies that the maintenance team can carry; r f,i is a 0-1 variable, r f,i 1 means the i-th r The repair point is the fault point f, otherwise it is 0.

8. The active distribution network component repair system considering network reconstruction and fault repair rolling optimization according to claim 5 is characterized in that: The active distribution network optimization and reconstruction model in the fourth module takes minimizing the current distribution network operation cost as the objective function, which is expressed as: Where, F a is the operating cost of the distribution network, t is the duration of network optimization and reconstruction, and in this paper, t = 1h; the first item is the cost of purchasing electricity from the transmission network, C1 represents the cost coefficient of the distribution network purchasing electricity from the transmission network, P t represents the active power output of the substation node at time t; the second and third items are the output costs of distributed generation, C PV Represents the photovoltaic electricity sales cost, N PV Represents the total number of PVs connected to the distribution network, represents the i-th PV PV output power, C WT represents the cost of selling wind power, N WT Represents the total number of WTs connected to the distribution network, represents the i-th WT The fourth item represents the economic loss caused by the distribution network loss, C2 is the cost of the distribution network loss, represents the i-th br The fifth item represents the switch operation cost, C3 represents the switch operation cost of the distribution network, k is the switch number, SW is the switch set in the distribution network, sw k,0 and sw k,1 Indicates the switch status before and after network reconstruction.

Images