Rotary power flow controller planning method considering reliability of power distribution network

Abstract

The invention provides a rotary power flow controller planning method considering the reliability of a power distribution network, and belongs to the technical field of rotary power flow controller application. According to the method, the economical efficiency and the reliability of a power grid are improved by matching the bypass switch with the RPFC for locating, sizing and operation optimization of the RPFC. In order to ensure that the planning result is suitable for the annual operation state of a power grid while improving the solving efficiency, the fluctuation of new energy output and load is considered, and scene analysis is carried out by using a Monte Carlo method and a K-means clustering method. The rotation power flow controller is planned and configured by taking the minimum annual total cost of the whole scheme as an objective function, solving is carried out by adopting an improved ant lion optimization algorithm mixed second-order cone programming method, and finally a rotation power flow controller locating and sizing scheme and an operation strategy with the lowest annual total cost, the best economical efficiency and the highest reliability are obtained.

Description

Technical Field

[0001] The present invention relates to the field of rotational power flow controller planning, and particularly relates to a rotational power flow controller planning method considering the reliability of a distribution network. Background Art

[0002] Connecting new energy generating units to the distribution network in the form of distributed generation (DG) is an important way to consume new energy, and is of great significance for achieving sustainable development. However, the increase in its integration has brought problems such as bidirectional power flow, voltage rise, and power fluctuation to the active distribution network, resulting in a substantial increase in the demand for flexible resources. How to achieve efficient and high-quality consumption of new energy has become a challenge. Traditional distribution network planning and control methods are gradually showing their inadequacy in dealing with these new challenges, and new technical means and planning methods are urgently needed to ensure the stable and reliable operation of the distribution network.

[0003] As a flexible AC transmission system device, the Rotating Power Flow Controller (RPFC) has powerful power flow control capabilities. It can effectively optimize the power flow distribution of the distribution network by adjusting parameters such as voltage amplitude and phase. Under normal operating conditions, the RPFC can reduce network losses, improve power quality and transmission efficiency, and enhance the economic efficiency of the distribution network operation. When a fault occurs in the distribution network, the RPFC can also undertake the load transfer task to a certain extent, reduce the power outage scope and outage time, and play an important role in ensuring the reliability of the distribution network. However, if the RPFC is not reasonably planned, its advantages will be difficult to fully exert. The reliability of the distribution network is directly related to the user's power consumption experience and the stable development of the social economy. A single power outage accident may cause huge economic losses to industrial production, affect the normal living order of residents, and even cause serious consequences in some key fields such as medical treatment and transportation. Studying the RPFC planning method considering the reliability of the distribution network can ensure the reasonable layout and parameter optimization of the RPFC in the distribution network. On the one hand, during normal operation, by optimizing the power flow, each component of the distribution network operates in a more reasonable state, reducing the probability of faults; on the other hand, in the event of a fault, make full use of the collaborative effect of the RPFC and its bypass switch to efficiently achieve load transfer, minimize the load power loss, and significantly enhance the distribution network's ability to respond to faults, thereby comprehensively improving the reliability of the distribution network. In the operation and management of the distribution network, economy and reliability are not independent of each other, but are interrelated and interact with each other. Simply pursuing reliability and over-investing may lead to excessive costs, affecting the economic benefits and sustainable development of power enterprises; on the contrary, one-sided emphasis on economy while ignoring reliability will increase the risk of faults and power outage losses, which is also not conducive to the overall benefits in the long run. The RPFC planning method considering the reliability of the distribution network can find a balance between the two. By reasonably planning the RPFC, while improving reliability, optimize the operation cost of the distribution network, achieve the coordinated improvement of economy and reliability, and meet the long-term needs of the efficient and stable operation of the power system.

[0004] Therefore, the present invention proposes a method for planning a rotating power flow controller considering the reliability of the distribution network, which coordinates the economic efficiency and reliability of network operation through the RPFC, optimizes network operation, reduces network operation costs, and improves power supply reliability. Summary of the Invention

[0005] In order to reasonably plan the rotating power flow controller in the distribution network, the present invention proposes a method for planning a rotating power flow controller considering the reliability of the distribution network, which coordinates the economic efficiency and reliability of network operation, and includes the following steps:

[0006] Step 1: Construction of the equivalent model and constraint conditions of the rotating power flow controller: According to the structural characteristics of the rotating power flow controller, build its equivalent model and determine the corresponding constraint conditions;

[0007] Step 2: Data Acquisition and Scenario Processing: Obtain the photovoltaic power generation data, wind turbine power generation data, load data, and distribution network topology information accessed by the active distribution network; process these data using scenario analysis methods. Among them, the Monte Carlo simulation method is selected for the scenario generation link, and the K-means algorithm is used for scenario clustering;

[0008] Step 3: Planning Model Construction: With the minimum annualized comprehensive cost as the objective function, construct a planning model for the rotating power flow controller considering the reliability of the distribution network;

[0009] Step 4: Model Solving and Scheme Determination: Based on the pre-set optimization objective function, use a hybrid algorithm of the improved ant lion optimization algorithm (IALO) and second-order cone programming (SOCP) to solve the model, so as to obtain the optimal configuration scheme of the rotating power flow controller in the active distribution network.

[0010] Specifically, in Step 1, according to the structure of the rotating power flow controller, establish its equivalent model and constraint conditions: The main circuit of the RPFC consists of 2 rotating phase-shifting transformers (RPSTs). By changing the rotor angles of the 2 RPSTs, a voltage with continuously adjustable amplitude and phase can be inserted into the line, and then the line power can be changed. The RPFC can not only achieve precise control of the line power but also achieve power decoupling control. Therefore, when building the RPFC model, the active power and reactive power of the line it controls are set as the decision variables of the RPFC operation strategy; The power transmission of the RPFC needs to meet the following constraints:

[0011]

[0012]

[0013]

[0014]

[0015]

[0016] In the formula: P RPFC,i,t 、P RPFC,j,t and Q RPFC,i,t 、Q RPFC,j,t are the active power and reactive power injected by the RPFC at the nodes i and j of the controlled line at time t, respectively; are the active power loss and reactive power loss of the RPFC on the controlled line ij at time t; A RPFC 、D RPFC are the active power loss coefficient and reactive power loss coefficient of the RPFC, respectively; S RPFC,ij is the capacity of the RPFC connected to the line ij.

[0017] Due to the capacity limitation of the rotating power flow controller, when a fault occurs in the distribution network, if only relying on the RPFC to carry out load transfer, it is very likely that a large number of loads will lose power, thus reducing the reliability of the network. Based on this, after a fault occurs in the distribution network, the bypass switch of the RPFC is selected to perform the load transfer operation. During the normal operation of the distribution network, the bypass switch is in the open state, and at this time, the RPFC is responsible for optimizing the power flow. Once a network fault occurs, the bypass switch closes immediately, removing the RPFC from the network and supplying power to the load through the bypass switch branch. This operation mode can fully exploit the respective advantages of the RPFC and its bypass switch, significantly improving the economy and reliability of network operation.

[0018] During the period from the occurrence of the fault to the closing of the bypass switch, it is necessary to rely on the main branch of the RPFC to carry out load transfer. To improve the reliability of the system, the RPFC needs to transfer as many loads as possible to the downstream area of the fault without exceeding its own capacity limit. If the capacity of the RPFC cannot meet the load demand in the downstream area of the fault area, some loads must be cut, and the priority of cutting is determined according to the distance from the RPFC access point. The magnitude of the transferred power is closely related to the load amount in the transferable area, the output of distributed generation (DG), and the capacity of the RPFC. The calculation formula is as follows:

[0019]

[0020] In the formula: P RPFC is the load that can be transferred by the RPFC and its connection line when a fault occurs in the distribution network; T S is the time required for the bypass switch to operate; N l is the number of load nodes downstream of the fault point; L i,t is the load value of node i at time t; η i,t is the load reduction state of node i at time t. If it is reduced, the value is 1; if it is not reduced, the value is 0; P DG,t is the output of DG downstream of the fault point at time t.

[0021] Specifically, in step 2, the photovoltaic power generation data, wind turbine power generation data, and load data are used to generate and cluster scenarios by using the scenario analysis method:

[0022] In the field of distribution networks, there are problems of intermittency and randomness in the output of wind and light and the load, while the scenario analysis method can effectively address them. It transforms the problem of the uncertain probability model of wind and light output into a solvable deterministic typical scenario for processing, avoiding the trouble of establishing and solving complex models. The specific steps are as follows: First, scenario generation is carried out. The probability models of wind, light, and load are discretized. A large number of initial scenarios are generated randomly, and each generated scenario corresponds to a probability value. Since the accuracy requirements need to be met, the number of randomly generated scenarios is often large, but this will also lead to a corresponding increase in computational complexity. To solve the problem of excessive computational volume, scenario clustering needs to be carried out. This step is to cluster the set of initial scenarios into representative scenarios with relatively high occurrence probabilities. Its basic principle is to cluster samples into different categories by measuring the similarity between samples. If the similarity between some samples is high, they will be classified into the same category. Scenario generation and scenario clustering together constitute the entire process of scenario analysis. Finally, with the help of the scenario analysis method, a set of typical scenarios can be obtained to describe the uncertainty of wind, light, and load.

[0023] The Monte Carlo method is selected for scenario generation. It is also known as the statistical simulation method. It is a random event simulation technology developed based on probability theory and mathematical statistics. Its core principle is to carry out sampling operations and simulation calculations with the help of random numbers according to the probability distribution of the sample space of random variables. The implementation of this method for random event simulation specifically includes the following steps:

[0024] (1) Establish the probability model of the problem: The primary task is to collect and analyze the dataset of characteristic quantities of the random event to be solved. On this basis, a probability distribution model that can describe this random event is constructed.

[0025] (2) Use statistical methods to calculate parameters and obtain the probability density function: Calculate the relevant parameters of the probability distribution model and use statistical means to obtain the corresponding probability density function expression.

[0026] (3) Conduct random sampling based on the probability density function and generate initial scenarios: Continuously repeat the random sampling action within the known probability distribution model according to the probability density function. According to the internal logic of the event development, datasets of different characteristic quantities are extracted and combined into a random array of event occurrence, thereby generating a large number of initial scenarios.

[0027] The K-means algorithm is selected for scenario clustering. It is a representative of the typical prototype-based objective function clustering method. It uses a certain distance from data points to prototypes as the optimized objective function and obtains the adjustment rules for iterative operations by using the method of finding the extreme value of the function. The K-means clustering algorithm usually only needs to perform iterative calculations 5-10 times, so it has the advantages of fast convergence speed, simple structure, and convenient operation. The K-means algorithm uses the Euclidean distance as the similarity measure. It finds the optimal classification corresponding to a certain initial clustering center vector to minimize the evaluation index. The algorithm uses the sum of squared errors criterion function as the clustering criterion function. The main content of this algorithm is as follows:

[0028] (1) Randomly select K sample scenarios in the original scenario as the initial centroids;

[0029] (2) Assign each sample scenario to the nearest clustering center to form categories centered on different sample scenarios;

[0030] (3) Continue to find new clustering centers in the categories divided in the previous step and iteratively update the existing clustering centers at the same time;

[0031] (4) Repeatedly calculate the difference between the updated clustering center and the existing clustering center. When the difference between the two is within the specified threshold range, the algorithm stops, indicating that the centroid will not move significantly.

[0032] Specifically, in step 3, a planning model for a rotary power flow controller considering the reliability of the distribution network is established with the minimum annualized comprehensive cost as the objective function:

[0033]

[0034] In the formula: f is the annual comprehensive cost; is the annual investment and construction cost of the RPFC after conversion; is the annual operation and maintenance cost of the RPFC; C line is the annual investment and construction cost of the access line of the RPFC after conversion; C opt is the cost transferred from the operation layer to the planning layer;

[0035]

[0036] In the formula: d is the discount rate; v is the service life of the RPFC; c RPFC is the unit capacity investment and construction cost of the RPFC; S RPFC,y is the installed capacity of the y-th RPFC; Y is the total installed number of RPFCs;

[0037]

[0038] In the formula: ε RPFCis the annual operation and maintenance cost coefficient of RPFC;

[0039]

[0040] Where: c m is the investment and construction cost per unit length of the line required for RPFC access; l line is the length of the line to be built

[0041] C opt = C loss + C fault

[0042] Where: C loss represents the line loss cost during network operation; C fault represents the power outage loss under network faults;

[0043]

[0044] Where: N is the number of scenarios obtained by clustering; p s is the probability of the s-th scenario occurring; μ is the loss cost coefficient of the distribution network; n is the total number of branches in the distribution network; P l,t,s is the line loss of branch l under the s-th scenario; is the active power loss of the y-th RPFC under the s-th scenario.

[0045] C fault = αβEENS

[0046] Where, β represents the average electricity price conversion multiple; EENS represents the amount of power outage loss of the system, which can be calculated by the following formula:

[0047] EENS = ∑L i t u,i

[0048]

[0049] Where: L i is the average load of node i; t u,i is the average outage time of users in a year; λ a and t r,a are the failure rate and repair time of the equipment respectively; A represents the set of equipment that causes power outages at load points.

[0050] Specifically, the constraint conditions of the rotational power flow controller planning model considering the reliability of the distribution network built in step 3 include rotational power flow controller constraints, rotational power flow controller access position and capacity constraints, grid power flow balance constraints, node voltage and line safety constraints, distributed power output constraints, network reconfiguration radial constraints, and switch operation times constraints.

[0051] Specifically, for the rotational power flow controller planning model considering the reliability of the distribution network established in step 3, island division needs to be carried out first when conducting reliability assessment:

[0052] The traditional distribution network has a single - source radial structure. When an upstream component fails, all downstream loads will lose power. With the access of distributed generation (DG), the user side also has the attribute of a power source, changing the single - source power supply structure of the distribution network. When the main grid fails, by operating sectionalizing devices, DGs can flexibly form islands within their power supply ranges to restore the power supply to important loads affected by the fault within the islands, thereby reducing the system power shortage and improving system reliability. Therefore, when calculating the reliability of a distribution network with distributed generation, the formation of islands needs to be considered.

[0053] Taking the unit adjacent relationship and load priority as heuristic rules, starting from the initial island (i.e., the positive unit that meets the local load demand), a combination of search and fusion ideas is adopted to solve the island division problem. Searching is to find the units adjacent to the initial island according to the connection attribute matrix, and then sort them according to the power attribute and the priority of the load points included. During the sorting process, the source - point unit has the highest priority, and the load units form a sequence according to the load level from high to low. Fusion is to select one of all adjacent feeder units to be incorporated into the island. The unit should first meet the basic principle of power balance. At the same time, to improve the power supply reliability of load points, two additional screening principles are determined: the island should give priority to ensuring the power supply of important loads, and under the condition that the output of distributed generation allows, give priority to restoring the power supply of high - level load points (loads are divided into first - level, second - level, and third - level loads according to importance, secondary importance, and general importance); the island should contain as many users as possible. The selected units are incorporated into the island and the search process continues. Assuming that unit f fails, the steps of island division formed according to the above rules are as follows:

[0054] (1) Each source - point unit forms an initial island by itself.

[0055] (2) Perform the island expansion operation on each initial island. For a certain source - point unit i, find the units adjacent to it according to the adjacency matrix. If there is a source - point unit s among the adjacent units, directly execute the island fusion program to incorporate this unit. For the load units among the adjacent units, sort them according to the load level table, then screen out the units that meet the power balance condition, and incorporate the one with the highest level g into the island. The specific fusion process includes power attribute correction and connection attribute correction. After island fusion, units s and g are incorporated into unit i to become one unit, and the power attribute of unit i is updated to P i +P s +P g , and the power attributes of units s and j are set to 0. The units adjacent to the original s and j units in the adjacency matrix are modified to be adjacent to unit i.

[0056] (3) Repeat step 2 in the order of unit numbers until all source point units are traversed and no more island fusions occur.

[0057] (4) Merge the boundaries of source point units. Search for load units adjacent to multiple source point units simultaneously, and determine whether the requirements for island combined power supply are met through the power attribute matrix. If they are met, the unit with the highest load level is included in the island according to the method in step 2 until no more island fusions occur. After executing the above algorithm, the load points divided inside the island can be powered by the DG when a fault occurs in unit f, and the repair time is shortened from the repair time of the faulty component to the maximum value of the operation time of the sectionalizing device and the switching operation time required for island formation.

[0058] Specifically, in the model of the rotating power flow controller considering the reliability of the distribution network established in step 3, the reliability assessment of the active distribution network adopts the minimum path method combined with island search:

[0059] First, find the minimum path of each load point. All equipment is divided into two categories: equipment on the minimum path and equipment not on the minimum path. The processing principle for equipment on the minimum path is:

[0060] If the equipment on the minimum path is within the scope of island division, the failure of this type of equipment on the minimum path will cause the outage of the load point. These equipment participate in the reliability calculation.

[0061] If the equipment on the minimum path is outside the scope of island division and there is a sectionalizing device (such as a disconnector) installed on the main feeder, then the outage time of the load point caused by the failure of this type of equipment is only max{S, T}, where S is the operation time of the sectionalizing device and T is the switching operation time required for island formation. Therefore, the outage time and the outage rate of the corresponding equipment participate in the reliability calculation.

[0062] The processing principle for equipment not on the minimum path is:

[0063] For branch lines, if there is a fuse installed at the head end of the branch line, the failure of the equipment on the branch line and the melting of the fuse do not affect other branch lines.

[0064] For the equipment behind the disconnector on the main feeder, the outage time of the previous section of the load point caused by its failure is the operation time S of the disconnector.

[0065] For the DG, only when both the DG and the main feeder fail does it affect the reliability calculation of the load point. The failure rate and the annual average power outage time are converted according to the second-order failure of the DG and the main feeder, and the conversion formulas are as follows:

[0066]

[0067]

[0068] t U,i = λ i t γ,i

[0069] Wherein, λ D and t γ,D are respectively the failure rate and the fault repair time of the DG; λ k and t γ,k are the failure rate and the fault repair time of the k-th feeder; N D is the number of possible faulty devices before the DG and the load node.

[0070] From the above formula, the equivalent failure rate of the load point and the duration of the fault power outage can be obtained. Furthermore, the power supply reliability index of the distribution network containing DG can be obtained, and then the power outage loss of the network can be obtained.

[0071] Specifically, the rotation power flow controller planning model considering the reliability of the distribution network built in step 4 uses an improved ant lion optimization algorithm hybrid second-order cone programming method to solve the built model. Considering that the IALO algorithm has good convergence in integer programming, and SOCP can quickly solve the optimal solution in continuous variable problems, this paper uses the IALO-SOCP hybrid algorithm to solve the model, which can not only reduce the solution space of the algorithm but also reduce the complexity of SOCP. The improved ant optimization algorithm is as follows:

[0072] Use the IALO algorithm to solve the installation location and capacity of the RPFC. Compared with traditional optimization algorithms such as the particle swarm optimization algorithm and the genetic algorithm, the performance of the ant lion optimization (ALO) algorithm is more superior; the ALO algorithm is inspired by the hunting of ant lions for ants in nature: the ant lion makes a trap in advance to wait for the ant. After the ant is caught by the ant lion, the ant lion will continue to dig the trap and wait for the next ant. The ant lion searches for the optimal solution of the problem by continuously preying on ants; the ALO algorithm has the disadvantage of being easily trapped in local optima. Therefore, this paper proposes an IALO algorithm based on a continuous boundary contraction factor and a position update dynamic weight coefficient:

[0073] Compared with the discontinuous increase in the change trend of the boundary contraction factor of the ALO algorithm, the IALO algorithm uses a continuously increasing boundary contraction factor, which can search the solution space more comprehensively. The calculation formula is as follows:

[0074]

[0075] Where: I is the boundary contraction factor; κ is the current iteration number; c κ , f κ are respectively the upper and lower bounds of the variable at the f κ -th iteration; K *is the maximum number of iterations; γ is the shrinkage adjustment coefficient; ξ is the scaling factor;

[0076] The weight coefficient of the ALO algorithm is fixed. To improve the balance ability of the algorithm between global exploration and local development, the IALO algorithm adopts a position update dynamic weight coefficient. Therefore, the ant position update formula is:

[0077]

[0078] In the formula: ω1 and ω2 are weight coefficients; is the position where the ant randomly walks around the antlion selected by roulette wheel; is the position where the ant randomly walks around the elite antlion. At the initial stage of iteration, ω1 takes a larger value, the weight is larger. As the number of iterations increases, the weight ω2 of the best elite antlion is gradually increased. In this way, it can effectively ensure the global optimization of the algorithm in the early stage and guarantee the local search of the algorithm in the later stage.

[0079] Specifically, the established planning model of the rotating power flow controller considering the reliability of the distribution network in step 4 uses an improved antlion optimization algorithm hybrid second-order cone programming method to solve the established model. Among them, the RPFC operation strategy adopts cone relaxation constraints, converts it into a second-order cone programming model and then solves it.

[0080] First, introduce auxiliary variables A i,t 、C ij,t and D ij,t :

[0081]

[0082] Then the above network constraints can be converted into the following formula:

[0083]

[0084]

[0085]

[0086]

[0087]

[0088] Introducing the second-order rotation cone theory can convert the RPFC constraint into a second-order cone constraint:

[0089]

[0090]

[0091] The technical effects demonstrated by the present invention are as follows: With the help of a rotating power flow controller, the flexibility of the distribution network can be deeply explored and fully utilized. During the daily operation of the distribution network, the rotating power flow controller can adjust relevant parameters, thereby effectively reducing the line loss during network operation and playing a good role in smoothing voltage fluctuations, greatly improving the power quality. When the distribution network encounters a fault, the rotating power flow controller can also actively intervene to effectively reduce the losses caused by power outages. The present invention considers the problem of site selection and capacity determination of the rotating power flow controller and optimizes its operation. While ensuring the improvement of the reliability of the distribution network, it also takes into account the economy of network operation. This dual-consideration characteristic makes the present invention have extremely broad application prospects. In addition, the present invention proposes a hybrid algorithm of IALO and SOCP, which greatly simplifies the planning solution process and significantly improves the solution efficiency, providing a solid technical support for the scientific planning and efficient application of the rotating power flow controller in the distribution network. Description of the Drawings

[0092] Appendix Figure 1 is the flowchart of the island search of the present invention;

[0093] Appendix Figure 2 is the flowchart of the reliability assessment of the distribution network based on island search;

[0094] Appendix Figure 3 is the structural diagram of the example network;

[0095] Appendix Figure 4 is the iterative convergence curve. Detailed Embodiment

[0096] The following further describes the present invention with reference to the drawings.

[0097] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present application belongs. It should be noted that the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0098] The present invention proposes a planning method for a rotating power flow controller considering the reliability of the distribution network, which coordinates the economy and reliability of network operation, including the following steps:

[0099] Step 1: Construction of the equivalent model and constraints of the Rotating Power Flow Controller (RPFC): Based on the structural characteristics of the RPFC, build its equivalent model and determine the corresponding constraints;

[0100]

[0101]

[0102]

[0103]

[0104]

[0105] Where: P RPFC,i,t 、P RPFC,j,t and Q RPFC,i,t 、Q RPFC,j,t are the active power and reactive power injected by the RPFC at the nodes i and j of the controlled line at time t, respectively; are the active power loss and reactive power loss of the RPFC on the controlled line ij at time t, respectively; A RPFC 、D RPFC are the active power loss coefficient and reactive power loss coefficient of the RPFC, respectively; S RPFC,ij is the capacity of the RPFC connected to the line ij.

[0106] Due to the capacity limitation of the Rotating Power Flow Controller, when a fault occurs in the distribution network, if relying solely on the RPFC for load transfer, it is very likely that a large number of loads will lose power, thereby reducing the reliability of the network; based on this, after a fault occurs in the distribution network, the bypass switch of the RPFC is selected to perform the load transfer operation; during the normal operation of the distribution network, the bypass switch is in the open state, and at this time, the RPFC is responsible for optimizing the power flow; once a network fault occurs, the bypass switch closes immediately, removing the RPFC from the network, and using the bypass switch branch to supply power to the load; this operation mode can fully exploit the advantages of the RPFC and its bypass switch, significantly improving the economy and reliability of network operation;

[0107] During the period from the occurrence of the fault to the closing of the bypass switch, it is necessary to use the main branch of the RPFC for load transfer; to improve the reliability of the system, the RPFC needs to transfer as many loads as possible to the downstream area of the fault without exceeding its own capacity limit; if the capacity of the RPFC cannot meet the load demand in the downstream area of the fault area, some loads must be reduced, and the reduction priority is determined according to the distance from the RPFC access point; the magnitude of the transferred power is closely related to the load quantity in the transferable area, the output of distributed generation (DG), and the capacity of the RPFC, and its calculation formula is as follows:

[0108]

[0109] In the formula: P RPFC is the load that can be transferred when a fault occurs in the distribution network for the RPFC and its tie lines; T S is the time required for the bypass switch to operate; N l is the number of load nodes downstream of the fault point; L i,t is the load value of node i at time t; η i,t is the load shedding state of node i at time t. If load shedding occurs, it takes the value of 1; if not, it takes the value of 0; P DG,t is the output of the DG downstream of the fault point at time t.

[0120] Step 2: Data acquisition and scenario processing: Obtain the photovoltaic power generation data, wind turbine power generation data, load data, and distribution network topology information accessed by the active distribution network; process these data using scenario analysis methods. Among them, the Monte Carlo simulation method is selected for the scenario generation link, and the K-means algorithm is used for scenario clustering.

[0121] In the field of distribution networks, there are problems of intermittency and randomness in the output of wind and light and the load. The scenario analysis method can effectively address this. It transforms the problem of the uncertain probability model of wind and light output into a solvable deterministic typical scenario for processing, avoiding the trouble of establishing and solving complex models.

[0122] The specific steps are as follows: First, scenario generation is carried out. The probability models of wind, light, and load are discretized. A large number of initial scenarios are generated randomly, and each generated scenario corresponds to a probability value. Due to the requirement of meeting the accuracy, the number of randomly generated scenarios is often large, but this also leads to a corresponding increase in the computational complexity. To solve the problem of excessive computational volume, scenario clustering needs to be carried out. This step clusters the initial scenario set into representative scenarios with relatively high occurrence probabilities. Its basic principle is to cluster samples into different categories by measuring the similarity between samples. If the similarity between some samples is high, they will be classified into the same category.

[0123] Scenario generation and scenario clustering together constitute the entire process of scenario analysis. Finally, with the help of the scenario analysis method, a set of typical scenarios can be obtained to describe the uncertainty of wind, light, and load.

[0124] The Monte Carlo method (Monte Carlo) is selected for scenario generation. It is also called the statistical simulation method. It is a random event simulation technology developed based on probability theory and mathematical statistics. Its core principle is to carry out sampling operations and simulation calculations using random numbers according to the probability distribution of the sample space of random variables. The implementation of random event simulation by this method specifically includes the following steps:

[0125] (1) Establish the probability model of the problem: The primary task is to collect and analyze the dataset of characteristic quantities of the random event to be solved. On this basis, construct a probability distribution model that can describe the random event.

[0126] (2) Use statistical methods to calculate parameters and obtain the probability density function: Calculate the relevant parameters of the probability distribution model, and use statistical means to obtain the corresponding probability density function expression.

[0127] (3) Conduct random sampling based on the probability density function and generate initial scenarios: According to the probability density function, continuously repeat the random sampling action within the known probability distribution model. According to the internal logic of the event development, extract the dataset of different characteristic quantities, combine them into a random array of event occurrence, and thus generate a large number of initial scenarios.

[0128] The K-means algorithm is selected for scenario clustering. It is a representative of the typical prototype-based objective function clustering method. It uses a certain distance from the data point to the prototype as the optimized objective function, and uses the method of finding the extreme value of the function to obtain the adjustment rule of iterative operation.

[0129] The K-means clustering algorithm usually only needs to iterate 5-10 times, so it has the advantages of fast convergence speed, simple structure, and convenient operation; the K-means algorithm uses the Euclidean distance as the similarity measure, and it is to find the optimal classification corresponding to a certain initial clustering center vector to minimize the evaluation index; the algorithm uses the sum of squared errors criterion function as the clustering criterion function; the main content of the algorithm is as follows:

[0130] (1) Randomly select K sample scenarios in the original scenarios as the initial centroids;

[0131] (2) Assign each sample scenario to the nearest clustering center to form categories centered on different sample scenarios;

[0132] (3) Continue to find new clustering centers in the categories divided in the previous step, and at the same time iteratively update the existing clustering centers;

[0133] (4) Repeatedly calculate the difference between the updated clustering center and the existing clustering center. When the gap between the two is within the specified threshold range, the algorithm stops, indicating that the centroid will not move significantly.

[0134] Step 3: Build the planning model: Taking the minimum annualized comprehensive cost as the objective function, build a planning model of the rotary power flow controller considering the reliability of the distribution network;

[0135] The objective function is:

[0136]

[0137] Where: f is the annual comprehensive cost; is the annual investment and construction cost of the RPFC after conversion; is the annual operation and maintenance cost of the RPFC; C line is the annual investment and construction cost of the RPFC access line after conversion; C opt is the cost transmitted from the operation layer to the planning layer;

[0138]

[0139] Where: d is the discount rate; v is the service life of the RPFC; c RPFC is the investment and construction cost per unit capacity of the RPFC; S RPFC,y is the installed capacity of the y-th RPFC; Y is the total installed number of RPFCs;

[0140]

[0141] Where: ε RPFC is the annual operation and maintenance cost coefficient of the RPFC;

[0142]

[0143] Where: c m is the investment and construction cost per unit length of the construction line required for RPFC access; l line is the length of the construction line required

[0144] C opt = C loss + C fault

[0145] Where: C loss represents the line loss cost during network operation; C fault represents the power outage loss under network faults;

[0146]

[0147] Where: N is the number of scenarios obtained by clustering; p s is the probability of the s-th scenario occurring; μ is the loss cost coefficient of the distribution network; n is the total number of branches in the distribution network; P l,t,s is the line loss of branch l under the s-th scenario; is the active power loss of the y-th RPFC under the s-th scenario.

[0148] C fault = αβEENS

[0149] Where, β represents the average electricity price conversion multiple; EENS represents the amount of power outage loss of the system, which can be calculated by the following formula:

[0150] EENS = ∑L i t u,i

[0151]

[0152] where: L i is the average load of node i; t u,i is the average outage time of users in a year; λ a and t r,a are the failure rate and repair time of the equipment respectively; A represents the set of equipment that causes the power outage at the load point.

[0153] The constraint conditions include the flexible interconnection device constraint, the access location and capacity constraint of the flexible interconnection device, the power flow balance constraint of the power grid, the node voltage and line safety constraint, the distributed power output constraint, the radial constraint of network reconfiguration, and the switch operation times constraint:

[0154] When conducting reliability assessment, island division needs to be carried out first:

[0155] The traditional distribution network has a single - source radial structure. When an upstream component fails, all downstream loads lose power. With the access of distributed generation (DG), the user side also has the power source attribute, and the single - source power supply structure of the distribution network has changed. When the main grid fails, by operating the sectionalizing device, the DG can flexibly form islands within its power supply range to restore the power supply of important loads affected by the fault within the island, thereby reducing the system power shortage and improving the system reliability. Therefore, when calculating the reliability of a distribution network with distributed generation, the formation of islands needs to be considered.

[0156] Taking the unit adjacent relationship and load priority as heuristic rules, starting from the initial island (i.e., the positive unit that meets the local load demand), a combination of search and fusion ideas is adopted to solve the island division problem. Search is to find the units adjacent to the initial island according to the connection attribute matrix, and then sort them according to the power attribute and the priority of the load points included. During the sorting process, the source point unit has the highest priority, and the load units form a sequence according to the load level from high to low. Fusion is to select one of all adjacent feeder units to be included in the island. The unit should first meet the basic principle of power balance. At the same time, to improve the power supply reliability of the load point, two additional screening principles are determined: the island should give priority to ensuring the power supply of important loads, and under the condition that the distributed power output permits, give priority to restoring the power supply of high - level load points (loads are divided into first - level, second - level, and third - level loads according to importance, secondary importance, and general importance); the island should contain as many users as possible. Incorporate the selected units into the island and continue the search process. The flow chart of island search is as shown in the appendix Figure 1 shown. Assuming that unit f fails, the steps of island division formed according to the above rules are as follows:

[0157] (1) Each source point unit forms an initial isolated island respectively.

[0158] (2) Perform the isolated island expansion operation on each initial isolated island. For a certain source point unit i, search for the adjacent units according to the adjacency matrix. If there is a source point unit s among the adjacent units, directly execute the isolated island fusion program to integrate into this unit. For the load units among the adjacent units, sort them according to the load level table, then screen out the units that meet the power balance condition, and integrate the one with the highest level g into the isolated island. The specific fusion process includes power attribute correction and connection attribute correction. After the isolated island fusion, the units s and g are integrated into unit i to become one unit, and the power attribute of unit i is updated to P i +P s +P g , and the power attributes of units s and j are set to 0. The units adjacent to the original s and j units in the adjacency matrix are modified to be adjacent to unit i.

[0159] (3) Repeatedly execute step 2 in the order of unit numbers until all source point units are traversed and no more isolated island fusions occur.

[0160] (4) Merge the boundaries of source point units. Search for the load units adjacent to multiple source point units at the same time, and judge whether they meet the requirements of isolated island combined power supply through the power attribute matrix. If they meet the requirements, classify the unit with the highest load level into the isolated island according to the method in step 2 until no more isolated island fusions occur. After executing the above algorithm, the load points divided inside the isolated island can be powered by DG when a fault occurs at unit f, and the repair time is shortened from the repair time of the faulty component to the maximum value of the operation time of the sectionalizing device and the switching operation time required for the formation of the isolated island.

[0161] The reliability assessment of the active distribution network adopts the minimum path method combined with isolated island search, and the flow chart is as attached Figure 2 as shown:

[0162] First, obtain the minimum path of each load point. All equipment is divided into two categories: equipment on the minimum path and equipment not on the minimum path. The processing principle for the equipment on the minimum path is:

[0163] If the equipment on the minimum path is within the scope of isolated island division, then the failure of this type of equipment on the minimum path will cause the outage of the load point. These equipment participate in the reliability calculation.

[0164] If the equipment on the minimum path is outside the scope of isolated island division and there is a sectionalizing device (such as a disconnector) installed on the main feeder, then the outage time of the load point caused by the failure of this type of equipment is only max{S, T}, where S is the operation time of the sectionalizing device and T is the switching operation time required for the formation of the isolated island. Therefore, the ones participating in the reliability calculation are these outage times and the outage rates of the corresponding equipment.

[0165] The processing principle for the equipment on non-minimal paths is as follows:

[0166] For the branch line, a fuse is installed at its head end. When a fault occurs to the equipment on the branch line and the fuse blows, the fault does not affect other branch lines.

[0167] For the main feeder, when a fault occurs to the equipment after the disconnecting switch, the outage time of the front-segment load point is the operating time S of the disconnecting switch.

[0168] For the DG, only when both the DG and the main feeder fail will it affect the reliability calculation of the load point. The failure rate and the annual average power outage time are calculated according to the second-order fault of the DG and the main feeder. The conversion formulas are as follows:

[0169]

[0170]

[0171] t U,i =λ i t γ,i

[0172] In the formula, λ D and t γ,D are respectively the failure rate and the fault repair time of the DG; λ k and t γ,k are the failure rate and the fault repair time of the k-th feeder; N D is the number of possible faulty equipment before the DG and the load node.

[0173] From the above formula, the equivalent failure rate and the fault power outage duration of the load point can be obtained. Furthermore, the power supply reliability index of the distribution network containing the DG can be obtained, and then the power outage loss of the network can be obtained.

[0174] Step 4: Model solution and scheme determination: Based on the pre-set optimization objective function, a hybrid algorithm of the improved ant lion optimization algorithm (IALO) and the second-order cone programming (SOCP) is used to solve the model, so as to obtain the optimal configuration scheme of the rotating power flow controller in the active distribution network. Considering that the IALO algorithm has good convergence in integer programming, and the SOCP can quickly solve the optimal solution in continuous variable problems, this paper uses the IALO-SOCP hybrid algorithm to solve the model, which can not only reduce the solution space of the algorithm but also reduce the complexity of the SOCP. The improved ant optimization algorithm is as follows:

[0175] Solve the installation location and capacity of RPFC using the IALO algorithm. Compared with traditional optimization algorithms such as the particle swarm optimization algorithm and the genetic algorithm, the ant lion optimization (ALO) algorithm has better performance; the ALO algorithm is inspired by the ant lion preying on ants in nature: the ant lion makes a trap in advance to wait for the ants. After the ants are caught by the ant lion, the ant lion will continue to dig traps and wait for the next ant. The ant lion finds the optimal solution to the problem by continuously preying on ants; the ALO algorithm has the disadvantage of being easily trapped in local optima. Therefore, this paper proposes an IALO algorithm based on a continuous boundary contraction factor and a dynamic weight coefficient for position update:

[0176] Compared with the discontinuous increase in the change trend of the boundary contraction factor of the ALO algorithm, the IALO algorithm uses a continuously increasing boundary contraction factor, so as to more comprehensively search the solution space. The calculation formula is as follows:

[0177]

[0178] In the formula: I is the boundary contraction factor; κ is the current iteration number; c κ 、f κ are the upper and lower bounds of the variable at the f κ -th iteration respectively; K * is the maximum iteration number; γ is the contraction adjustment coefficient; ξ is the proportionality factor;

[0179] The weight coefficient of the ALO algorithm is fixed. In order to improve the balance ability of the global exploration and local development of the algorithm, the IALO algorithm uses a dynamic weight coefficient for position update. Therefore, the update formula of the ant position is:

[0180]

[0181] In the formula: ω1 and ω2 are weight coefficients; is the position where the ant randomly walks around the ant lion selected by roulette; is the position where the ant randomly walks around the elite ant lion. At the beginning of the iteration, the value of ω1 is larger, the weight is larger. As the number of iterations increases, gradually increase the weight ω2 of the best elite ant lion . In this way, it can effectively ensure the global optimization in the early stage of the algorithm and the local search in the later stage of the algorithm.

[0182] The operation strategy of RPFC adopts cone relaxation constraints, which are converted into a second-order cone programming model for solution.

[0183] First, introduce auxiliary variables A i,t 、C ij,t and D ij,t :

[0184]

[0185] Then the above network constraints can be converted into the following formula:

[0186]

[0187]

[0188]

[0189]

[0190]

[0191] Introducing the second-order rotation cone theory can convert the RPFC constraints into second-order cone constraints:

[0192]

[0193]

[0194] Embodiment

[0195] Taking the improved interconnected network of IEEE33-node and IEEE22-node as the background, its structure is as shown in the appendix Figure 3 shown, where photovoltaic units with capacities of 300 kVA, 300 kVA, 200 kVA, and 200 kVA are respectively connected to nodes 7, 10, 24, and 27, and wind turbines with capacities of 1500 kVA and 1200 kVA are respectively connected to nodes 13 and 30.

[0196] The access position and the number of RPFCs connected will affect the network operation. By changing the number of RPFCs connected to find its optimal access position, the results are shown in Table 1. It can be seen from the table that although increasing the number of RPFCs connected can reduce the line loss cost and the outage loss cost, the degree of reduction is limited. When the number of RPFCs connected is 4, the investment and construction cost and the operation and maintenance cost of RPFC are greater than the reduced line loss cost and outage loss cost, and the annual comprehensive cost increases instead. The optimal number of RPFCs connected in the numerical example of this paper is 2.

[0197] Table 1 Influence of the number of RPFCs connected on the planning results

[0198]

[0199] The new energy penetration rate of the above example is 65%. To further analyze the impact of the new energy penetration rate on the model solution results, the model is solved under different penetration rate conditions, and the results are shown in Table 2. It can be seen from the table that both the annual line loss cost and the annual investment and construction cost of RPFC show a trend of decreasing first and then increasing with the increase of the new energy penetration rate. Moreover, with the change of the penetration rate, the optimal access position of RPFC also changes. If the operation state of the system is to be optimized, DG needs to be reasonably configured.

[0200] Table 2 Solution results under different new energy penetration rates

[0201]

[0202] To verify the effectiveness of the proposed IALO-SOCP hybrid algorithm in this paper, the IALO-SOCP hybrid algorithm, the ALO-SOCP hybrid algorithm, and the SOCP-Gurobi solver are used for solution respectively, and the results are shown in Table 3 and the appendix Figure 4 As shown. It can be seen that when only SOCP is used for solution, the result is not obtained after 8 hours of calculation. This is because the dimension of the decision variable is too high and the number of iterations is large, and it is difficult for SOCP to converge quickly when solving alone. When the hybrid algorithm is used to solve the variables in layers, the solution of the planning problem is easier, and compared with the ALO-SOCP hybrid algorithm, the IALO-SOCP hybrid algorithm has a faster calculation speed. The above results verify the applicability of the proposed hybrid algorithm in power grid planning.

[0203] Table 3 Comparison of solution times of different methods

[0204] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0205] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and 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 application, and should all be included in the protection scope of the present application.

Claims

1. A method for planning a rotating power flow controller considering the reliability of a distribution network, characterized in that: The following steps are involved: Step 1: Construction of equivalent model and constraint conditions of rotating power flow controller: According to the structural characteristics of rotating power flow controller (RPFC), its equivalent model is constructed and the corresponding constraint conditions are determined; Step 2: Data acquisition and scenario processing: Obtain the photovoltaic power generation data, wind turbine power generation data, load data, and distribution network topology information connected to the active distribution network; use scenario analysis methods to process these data; the Monte Carlo simulation method is used in the scenario generation link, and the K-means algorithm is used for scenario clustering; Step 3: Planning model construction: Taking the minimization of annualized comprehensive cost as the objective function, a rotating power flow controller planning model considering the reliability of the distribution network is constructed; Step 4: Model solution and solution determination: Based on the pre-set optimization objective function, the hybrid algorithm of the improved ant lion optimization algorithm (IALO) and the second-order cone programming (SOCP) is used to solve the model, so as to obtain the optimal configuration scheme of the rotating power flow controller in the active distribution network.

2. A method for planning a rotating power flow controller considering the reliability of a distribution network according to claim 1, characterized in that: In step 1, according to the structure of the rotating power flow controller, its equivalent model and constraints are established: the main circuit of the RPFC is composed of two rotating phase-shifting transformers (RPSTs). By changing the rotor angles of the two RPSTs, a voltage with continuously adjustable amplitude and phase can be connected in series in the line, thereby changing the line power; the RPFC can not only realize the precise control of the line power, but also realize power decoupling control. Therefore, when building the RPFC model, the active power and reactive power of the line controlled by it are set as the decision variables of the RPFC operation strategy; the RPFC transmission power needs to meet the following constraints: Where: P RPFC,i,t , P RPFC,j,t and Q RPFC,i,t , Q RPFC,j,t are the active power and reactive power injected by RPFC at the controlled line nodes i and j at time t, respectively; are respectively the active power loss and reactive power loss of RPFC on the controlled line ij at time t; A RPFC , D RPFC are the active power loss coefficient and reactive power loss coefficient of RPFC respectively; S RPFC,ij is the capacity of the RPFC connected to line ij; Since the rotating power flow controller has capacity limitations, if the RPFC is relied upon alone to transfer loads when a distribution network failure occurs, it is very likely that a large number of loads will lose power, thereby reducing the reliability of the network. Based on this, after a fault occurs in the distribution network, the bypass switch of RPFC is selected to perform load transfer operation; during the normal operation of the distribution network, the bypass switch is in the disconnected state, and RPFC is responsible for optimizing the power flow; Once a network failure occurs, the bypass switch is immediately closed to remove the RPFC from the network, and the load is powered by the bypass switch branch. This operation mode can fully tap the advantages of the RPFC and its bypass switch, significantly improving the economy and reliability of network operation. During the period from the occurrence of the fault to the closing of the bypass switch, the load needs to be transferred with the help of the main branch of RPFC. To improve the reliability of the system, RPFC needs to transfer as much load as possible to the downstream area of ​​the fault without exceeding its own capacity limit. If the capacity of RPFC cannot meet the load demand downstream of the fault area, some loads must be reduced, and the priority of reduction is determined by the order of distance from the RPFC access point. The amount of transferred power is closely related to the load in the transferable area, the output of distributed generation (DG) and the capacity of RPFC. The calculation formula is as follows: Where: P RPFC is the load that can be transferred by RPFC and its tie line when a fault occurs in the distribution network; T S The time required for the bypass switch to operate; N l is the number of load nodes downstream of the fault point; L i,t is the load value of node i at time t; η i,t is the load reduction state of node i at time t, if it is reduced, the value is 1, if it is not reduced, the value is 0; P DG,t is the output of the DG downstream of the fault point at time t.

3. A method for planning a rotating power flow controller considering the reliability of a distribution network according to claim 1, characterized in that: In step 2, the photovoltaic power generation data, wind turbine power generation data, and load data are subjected to scenario generation and clustering using a scenario analysis method: In the field of distribution network, there are problems of intermittency and randomness in wind and solar power output and load, and the scenario analysis method can effectively deal with it. It transforms the problem of uncertain wind and solar power output probability model into a solvable deterministic typical scenario to avoid the trouble of establishing and solving complex models. The specific steps are to generate scenarios first, discretize the wind and solar power and load probability models, and generate a large number of initial scenarios in a random way. Each generated scenario corresponds to a probability value. In order to meet the accuracy requirements, the number of randomly generated scenarios is often large, but this will also lead to a corresponding increase in computational complexity. In order to solve the problem of excessive computational complexity, it is necessary to carry out scenario clustering. This step is to cluster the initial scenario set into representative scenarios with a higher probability of occurrence. That The basic principle is to cluster samples into different categories by measuring the similarity between samples. If the similarity between some samples is high, they will be classified into the same category. Scene generation and scene clustering together constitute the whole process of scene analysis. Finally, with the help of scenario analysis, a set of typical scenarios can be obtained to describe the uncertainty of wind and solar loads; The Monte Carlo method, also known as the statistical simulation method, is a random event simulation technology developed based on probability theory and mathematical statistics. Its core principle is to use random numbers to perform sampling operations and simulation calculations based on the probability distribution of the random variable sample space. This method implements random event simulation, which specifically includes the following steps: (1) Establishing the probability model of the problem: The first task is to collect and analyze the characteristic data set of the random event to be solved, and on this basis, to construct a probability distribution model that can describe the random event; (2) Use statistical methods to calculate parameters and obtain probability density functions: Calculate the relevant parameters of the probability distribution model and use statistical methods to obtain the corresponding probability density function expression; (3) Random sampling based on probability density function and generating initial scenarios: Based on the probability density function, the random sampling action is continuously repeated within the known probability distribution model. According to the inherent logic of the event development, data sets with different feature quantities are extracted and combined into a random array of event occurrences, thereby generating a large number of initial scenarios. The K-means algorithm is selected for scene clustering. It is a typical representative of the prototype-based objective function clustering method. It uses a certain distance from the data point to the prototype as the optimized objective function, and uses the method of finding the extreme value of the function to obtain the adjustment rule of iterative operation. The K-means clustering algorithm often only needs 5-10 iterations, so it has the advantages of fast convergence speed, simple structure and convenient operation. The K-means algorithm uses Euclidean distance as the similarity measure. It seeks the optimal classification corresponding to a certain initial cluster center vector to minimize the evaluation index. The algorithm uses the error square sum criterion function as the clustering criterion function. The main contents of the algorithm are: (1) Randomly select K sample scenes from the original scene as the initial centroid; (2) Each sample scene is divided into the cluster center with the shortest distance, forming categories with different sample scenes as the core; (3) Continue to find new cluster centers in the categories classified in the previous step, and iteratively update the existing cluster centers; (4) Repeatedly calculate the difference between the updated cluster center and the existing cluster center. When the difference between the two is within the specified threshold range, the algorithm stops, indicating that the center of mass will not move significantly.

4. A method for planning a rotating power flow controller considering the reliability of a distribution network according to claim 1, characterized in that: In step 3, a rotating power flow controller planning model considering the reliability of the distribution network is constructed with the annualized comprehensive cost minimization as the objective function: Where: f is the annual comprehensive cost; is the converted annual investment and construction cost of RPFC; is the annual operation and maintenance cost of RPFC; C line is the converted annual investment and construction cost of RPFC access line; C opt The cost transmitted from the operation layer to the planning layer; Where: d is the discount rate; v is the useful life of RPFC; c RPFC is the unit capacity investment and construction cost of RPFC; S RPFC,y is the installed capacity of the yth RPFC; Y is the total number of installed RPFCs; Where: ε RPFC is the annual operation and maintenance cost coefficient of RPFC; Where: c m The unit length investment and construction cost of the line required for RPFC access; line The length of the line to be constructed C opt =C loss +C fault Where: C loss Indicates the line loss cost during network operation; C fault Indicates the power outage loss due to network failure; Where: N is the number of scenes obtained by clustering; p s is the probability of the sth scenario occurring; μ is the loss cost coefficient of the distribution network; n is the total number of branches in the distribution network; P l,t,s is the line loss of branch l in the sth scenario; is the active power loss of the yth RPFC in the sth scenario; C fault =αβEENS In the formula, β represents the average electricity price conversion multiple; EENS represents the power loss of the system due to power outage, which can be calculated by the following formula: ONCE=∑L i t u,i Where: L i is the average load of node i; t u,i is the average downtime of users in a year; λ a With t r,a are the failure rate and repair time of the equipment respectively; A represents the set of equipment that causes power outage at the load point.

5. A method for planning a rotating power flow controller considering the reliability of a distribution network according to claim 1, characterized in that: In step 3, the constraints of the rotating power flow controller planning model considering the reliability of the distribution network include rotating power flow controller constraints, rotating power flow controller access location and capacity constraints, power grid power balance constraints, node voltage and line safety constraints, and distributed power generation output constraints.

6. A method for planning a rotating power flow controller considering the reliability of a distribution network according to claim 1, characterized in that: In step 3, the active power distribution network needs to be divided into islands before reliability assessment: The traditional distribution network is a single-power radiation structure. If the upstream component fails, all downstream loads will lose power. With the access of distributed power sources, the user side also has power attributes, and the single-power supply structure of the distribution network has changed. When the main power grid fails, the DG can be flexibly formed into an island within the power supply range by operating the segmentation device, and the power supply of important loads affected by the fault in the island can be restored, thereby reducing the power shortage of the system and improving the system reliability. Therefore, the formation of the island needs to be considered when calculating the reliability of the distribution network containing distributed power sources. Taking the unit adjacent relationship and load priority as heuristic rules, starting from the initial island (i.e., the positive unit that meets the local load demand), the island partition problem is solved by combining search and fusion; the search is to find the units adjacent to the initial island according to the connection attribute matrix, and then sort them according to the power attribute and the priority of the load points contained therein; in the sorting process, the source point unit has the highest priority, and the load units form a sequence from high to low according to the load level; fusion is to select one from all adjacent feeder units to be included in the island; the unit should first meet the basic principle of power balance, and at the same time, in order to improve the power supply reliability of the load point, two additional screening principles are determined: the island should give priority to ensuring the power supply of important loads, and when the output of distributed power sources allows, the power supply of high-level load points should be restored first (the loads are divided into primary, secondary, and tertiary loads according to important, secondary, and general); the island should contain as many users as possible; the screened units are integrated into the island and the search process continues; assuming that unit f fails, the steps of island partition formed according to the above rules are as follows: (1) Each source point unit forms an initial island; (2) Perform an island expansion operation on each initial island; for a source point unit i, search for its adjacent units according to the adjacency matrix; if there is a source point unit s among the adjacent units, directly execute the island fusion procedure to merge the unit; for the load units among the adjacent units, sort them according to the load level table, and then select the units that meet the power balance conditions, and merge the units with the highest level g into the island; The specific fusion process includes power attribute correction and connection attribute correction. After island fusion, units s and g are integrated into unit i to become one unit, and the power attribute of unit i is updated to P. i +P s +P g , the power attributes of cells s and j are set to 0; the cells adjacent to the original cells s and j in the adjacency matrix are modified to be adjacent to cell i; (3) Repeat step 2 in the order of unit numbers until all source point units are traversed and no island fusion occurs; (4) Merge the source point unit boundaries; search for load units that are adjacent to multiple source point units at the same time, and determine whether they meet the island joint power supply requirements through the power attribute matrix. If they meet the requirements, the unit with the highest load level is included in the island according to the method in step 2 until the island fusion no longer occurs; after executing the above algorithm, the load points divided into the island can be restored to power by DG when unit f fails, and the repair time is shortened from the repair time of the faulty component to the maximum value of the operation time of the segmentation device and the switching operation time required for the island formation.

7. A method for planning a rotating power flow controller considering the reliability of a distribution network according to claim 1, characterized in that: In step 3, the reliability assessment of the active distribution network adopts the minimum path method combined with island search; The specific steps are as follows: First, find the minimum path for each load point; all equipment is divided into two categories: equipment on the minimum path and equipment on non-minimum paths; the processing principle for equipment on the minimum path is: 1) If the equipment on the minimum path is within the range of the island division, then the failure of such equipment on the minimum path will cause the load point to shut down; these devices participate in the reliability calculation; 2) If the equipment on the minimum path is outside the range of the island division and a sectionalizing device (such as a disconnector) is installed on the main feeder, then the load point outage time caused by the failure of such equipment is only max{S,T}, where S is the operation time of the sectionalizing device and T is the switching operation time required for the island formation. Therefore, the outage time and the outage rate of the corresponding equipment are involved in the reliability calculation; The principles for handling equipment on non-minimum roads are: 1) For branch lines with fuses installed at the head end, if the equipment on the branch line fails and the fuse blows, it will not affect other branch lines; 2) The downtime of the front load point caused by the failure of the equipment behind the main feeder disconnector is the operation time S of the disconnector; 3) For DG, only when both DG and main feeder fail will the reliability of load point be affected. The failure rate and annual average power outage time conversion formula of the second-order failure of DG and main feeder are as follows: t U,i =λ i t γ,i In the formula, λ D With t γ,D They are the failure rate and failure repair time of DG respectively; λ k With t γ,k is the failure rate and fault repair time of the kth feeder; N D is the number of possible faulty devices located between the DG and the load node; The above formula can be used to obtain the equivalent failure rate of the load point and the duration of the fault power outage, and then the power supply reliability index of the distribution network containing DG can be obtained, and then the power outage loss of the network can be obtained.

8. A method for planning a rotating power flow controller considering the reliability of a distribution network according to claim 1, characterized in that: In step 4, the constructed model is solved by using the improved ant lion optimization algorithm hybrid second-order cone programming method: Considering that the IALO algorithm has good convergence in integer programming, and SOCP can quickly solve the optimal solution in continuous variable problems, this paper adopts the IALO-SOCP hybrid algorithm to solve the model, which can not only reduce the solution space of the algorithm, but also reduce the complexity of SOCP; The IALO algorithm is used to solve the installation location and capacity of RPFC. Compared with traditional optimization algorithms such as particle swarm optimization algorithm and genetic algorithm, the performance of the ant lion optimization (ALO) algorithm is better. The ALO algorithm is inspired by the ant lion hunting ants in nature: the ant lion prepares traps in advance to wait for the ants. After the ants are captured by the ant lion, the ant lion will continue to dig traps and wait for the next ant. The ant lion finds the optimal solution to the problem by constantly preying on ants. The ALO algorithm has the disadvantage of being easily trapped in local optimality. For this reason, this paper proposes an IALO algorithm based on the continuous boundary shrinkage factor and the dynamic weight coefficient of position update: Compared with the ALO algorithm, the boundary shrinkage factor has an intermittent increase trend. The IALO algorithm uses a continuously increasing boundary shrinkage factor, which can search the solution space more comprehensively. The calculation formula is as follows: Where: I is the boundary shrinkage factor; κ is the current iteration number; c κ 、f κ The variables are κ The upper and lower bounds of the iterations; K * is the maximum number of iterations; γ is the shrinkage adjustment coefficient; ξ is the proportional factor; The weight coefficient of the ALO algorithm is fixed. In order to improve the balance between global exploration and local development of the algorithm, the IALO algorithm adopts a dynamic weight coefficient for position update. The update formula is: Where: ω1, ω2 are weight coefficients; The positions of the ants' random walks around the antlion selected by the roulette wheel; is the position of the ants randomly walking around the elite ant lion; in the early stage of iteration, ω1 takes a larger value, The weight is large, and as the number of iterations increases, the best elite ant lion is gradually increased. The weight ω2 can effectively ensure the global optimization in the early stage of the algorithm and the local search in the later stage of the algorithm.

9. A method for planning a rotating power flow controller considering the reliability of a distribution network according to claim 1, characterized in that: In step 4, the constructed model is solved by using an improved ant lion optimization algorithm mixed with a second-order cone programming method: the RPFC operation strategy adopts cone relaxation constraints, which are converted into a second-order cone programming model for solution.

Images