Power distribution network power supply grid division method based on improved robustness algorithm

Abstract

The invention discloses a power distribution network power supply grid division method based on an improved robust algorithm, and the method comprises the following steps: 1, extracting a typical daily time sequence scene through employing an optimization extraction model according to the wind power, photovoltaic and load historical data in a power distribution network power supply grid; step 2, on the basis of the typical time sequence scene, taking the optimal comprehensive index as a target, and obtaining an optimal power supply grid division scheme corresponding to each typical day by using an improved robustness algorithm in which a tree structure pruning link is introduced; and step 3, establishing a grid division scheme evaluation index system comprising a division result universality index, a simulation operation rationality index and a clustering extraction representative index, and selecting an optimal grid division scheme in a radar map form. The method can provide guidance for division from the power supply grid of the power distribution network to the power supply units of the power distribution network, and promote standardized grid connection of distributed source, load and storage resources and efficient cooperation among the power supply units.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system planning and operation, and relates to a reasonable division method for dividing a power supply grid into a plurality of power supply units under the background of a new distribution system, and specifically relates to a new distribution network power supply grid division method which is solved by taking into account comprehensive division indicators and by introducing an improved Louvain algorithm that introduces a tree structure pruning link. Background Art

[0002] With the rapid development of renewable energy and the continuous growth of electricity demand, traditional distribution networks are facing unprecedented challenges. In order to improve the flexibility and reliability of the power system, promoting the construction of a new distribution network has become an important task at present. Dividing the new distribution network power supply grid into new distribution network power supply units aims to achieve more efficient and intelligent power distribution and lay the foundation for the sustainable development of future power systems. The distribution network power supply unit refers to the 10kV distribution network power supply range obtained by dividing the power supply grid according to the function of the plot, development status, geographical conditions, load distribution, current power grid, etc.

[0003] With the widespread application of distributed generation and energy storage technology, the structure of the power system has become increasingly complex, and the traditional centralized power supply mode can no longer meet the diversity and flexibility of modern power demand. Therefore, a reasonable division of the power supply grid can help optimize resource allocation and improve the overall efficiency of the system. This division can not only effectively integrate a variety of energy resources, but also realize dynamic load allocation between different power supply units, thereby improving the responsiveness of the power grid. Moreover, by subdividing the power supply grid into multiple power supply units, more refined management and control can be achieved, and the power grid can be enhanced to respond quickly to load fluctuations and faults. Each power supply unit can operate independently and work with other units at the same time, thereby improving the flexibility and stability of the system and ensuring that the continuity of power supply can be maintained in emergencies. In addition, this division can also promote the collaborative work between the power supply units and realize multi-source complementarity. Through intelligent scheduling and control technology, each power supply unit can be optimally configured according to the real-time load and power generation conditions to achieve optimal utilization of resources. The power supply units of the new distribution network can not only cope with changes in local loads, but also improve efficiency at the overall power grid level, further enhancing the resilience of the power system.

[0004] In summary, dividing the new distribution network power supply grid into new distribution network power supply units is not only an inevitable choice to meet the development needs of the times, but also a key step to realize the intelligence and flexibility of the power system. This innovative measure will provide strong support for the sustainable development of the future power system, promote energy transformation, and help achieve the goal of carbon neutrality. However, the existing power supply grid division methods often have problems such as single division indicators, the inability to balance the optimization ability and computational efficiency of the division algorithm, and the difficulty in objectively comparing the effects of the division schemes. It is difficult to substantially guide the grid and unit operation of the distribution network. Summary of the invention

[0005] In order to overcome the above technical defects of the existing power grid division method, the present invention provides a distribution network power grid division method based on the improved Louvain algorithm. The method can provide guidance for the division of the distribution network power grid to the distribution network power supply unit, promote the standardized grid connection of distributed source, load and storage resources and the efficient coordination between power supply units.

[0006] The objective of the present invention is achieved through the following technical solutions:

[0007] A method for dividing a power distribution network into power grids based on an improved Louvain algorithm comprises the following steps:

[0008] Step 1: Based on the historical data of wind power, photovoltaic power generation and load in the power grid of the distribution network, the optimization extraction model is used to extract the typical daily time series scene. The specific steps are as follows:

[0009] Step 11: According to the principal component analysis method (PCA), the multivariate time series data of 365 days a year and including the "wind-solar-load" coupling is reduced to single variable time series data;

[0010] Step 12: Determine the number of clusters according to the elbow rule, use the K-means algorithm to classify the principal component sequence, and divide the annual univariate time series data into several time series data clusters;

[0011] Step 13: Establish a multi-objective mixed integer linear programming model as a typical day optimization extraction model, with the goals of minimizing the number of typical days, minimizing the deviation of typical scenes, maximizing the surrounding number density, and maximizing the radiation radius, and extract the most representative and differentiated typical day time series scenes from each cluster time series data, where:

[0012] The objective function of the multi-objective mixed integer linear programming model is:

[0013]

[0014] in:

[0015]

[0016] In the formula, u=[u1,…,u N ] T ,ω=[ω1,…,ω N ] T are the binary selection decision variables and corresponding weight coefficient variables for each typical day; A ξ 、b ξ are the original data matrix of each data type and the total amount of data in each period; f1 is the number of days selected as typical days in the current sample set; f2 is the sum of the relative deviations between the typical day and the original data, that is, the degree of deviation; f3 is the number density around the typical day; f4 is the total radiation radius of the typical day;

[0017] The constraints of the typical day optimization extraction model are as follows:

[0018]

[0019] In the formula, constraint (a) is the consistency constraint between the typical day weight coefficient variable ω and the selected decision variable u; constraint (b) aims to ensure that the sum of all typical day weight coefficients is consistent with the sample size; constraint (c) is the total relative deviation range constraint, α is the range relaxation coefficient; constraint (d) is the type constraint of the two types of decision variables;

[0020] Step 2: Based on typical time series scenarios and with the goal of optimizing comprehensive indicators, the improved Louvain algorithm with the introduction of tree structure pruning is used to obtain the optimal power grid division scheme corresponding to each typical day, where:

[0021] The comprehensive index includes four aspects: source-load-storage active power matching index O1, modularity index based on electrical distance O2, reactive power support capacity index O3, and power supply unit scale balance index O4. The specific calculation method is as follows:

[0022] (1) Source-load-storage active power matching index O1

[0023] The distributed generation power and generalized load power are optimized with the goal of minimizing the net load, thereby obtaining the distributed generation power and generalized load power corresponding to the optimal solution. The goals of the optimization model are as follows:

[0024]

[0025] Where H is the number of scheduling periods; Ω k is the node set of the kth power supply unit; is the net load value after optimal scheduling of source, load and storage of the kth power supply unit at time h; They are the original load power of the kth power supply unit at time h, the generalized load power after considering the energy storage charging and discharging power and the demand side response, and the DG power; are the transferred load power, reduced load power, energy storage charging power, energy storage discharging power, WT output, PV output, and MT output of the i-th node in the power supply unit;

[0026] Calculate the source-load-storage matching index O1 based on the optimization results:

[0027]

[0028]

[0029] Where N Z The number of power supply units in the current power supply grid division scheme; is the source-load-storage matching degree of the kth power supply unit in the hth period, and

[0030] (2) Modularity index O2 based on electrical distance

[0031] The electrical distance based on the active / reactive power-voltage amplitude sensitivity matrix is ​​used to characterize the coupling relationship between nodes. The sensitivity matrix and electrical distance are calculated as follows:

[0032]

[0033] Where n is the number of distribution network nodes; ΔP i , ΔQ i , ΔV i are the changes of injected active power, injected reactive power and voltage amplitude of each node respectively; S PV , S QV They are active power-voltage amplitude sensitivity matrix and reactive power-voltage amplitude sensitivity matrix respectively; are the ratios of the voltage changes on node j (node ​​j) and node i when the active / reactive power of node j changes, indicating the degree of influence of the active / reactive power changes on the two nodes; D ij They are respectively the active electrical distance, reactive electrical distance and comprehensive electrical distance between node i and node j;

[0034] By using the electrical distance and distribution network connection relationship, the distribution network is transformed into a relationship network that can directly calculate the modularity by calculating the edge weight. The edge weight calculation method is as follows:

[0035]

[0036] Where D is the comprehensive electrical distance matrix; A is the distribution network connection relationship matrix. When there is a direct connection line between node i and node j, A(i, j) = 1, otherwise A(i, j) = 0; e ij is the edge weight between node i and node j after being transformed into a relational network;

[0037] After the distribution network is converted into a relational network, the modularity index O2 based on electrical distance is calculated according to the modularity definition. The calculation method is as follows:

[0038]

[0039] In the formula, k i , k j are the sum of the weights of all the connecting edges of node i and node j respectively; m is the sum of the weights of all the edges in the entire relationship network; δ(i,j) indicates the same belonging status of node i and node j. If the two nodes are divided into the same power supply unit, δ(i,j)=1, otherwise δ(i,j)=0;

[0040] (3) Reactive power support capability index O3

[0041] The reactive power demand is evaluated, the reactive power supply capacity is evaluated, and the evaluation index with a design range of [0,1] is carried out in three steps.

[0042] In terms of reactive power demand assessment of each power supply unit, the voltage of overvoltage nodes at the peak of new energy penetration rate and the voltage of undervoltage nodes at the underestimated penetration rate in a typical day will be calculated under completely uncontrolled conditions. The reactive power demand of each power supply unit will be estimated according to the reactive power-voltage amplitude sensitivity matrix. The calculation method is as follows:

[0043]

[0044] In the formula, Ω k1 ,Ω k2 are the overvoltage node set and undervoltage node set of the kth power supply unit respectively; ΔV i is the absolute value of the voltage amplitude of the overvoltage / undervoltage node i from the normal voltage range; is the minimum capacitive reactive power required by the power supply side for the overvoltage node in the kth power supply unit; is the minimum inductive reactive power required by the power supply side for the undervoltage node in the kth power supply unit;

[0045] In terms of reactive power supply capacity evaluation of each power supply unit, according to the set DG power factor range and SVC reactive capacity, the reactive power output of DG and SVC under the constant power factor on the source side are considered to describe the reactive power supply capacity of each power supply unit. The specific calculation method is as follows:

[0046]

[0047] In the formula, are the source-side capacitive reactive power and inductive reactive power supply capabilities of the kth power supply unit respectively; They are the maximum capacitive reactive power that PV, WT and SVC can provide respectively; They are the maximum inductive reactive power that PV, WT, MT and SVC can provide respectively;

[0048] Based on the estimated reactive power demand and reactive power supply capacity of each power supply unit, a reactive power support capacity index O3 is formed, which is in the following form:

[0049]

[0050] In the formula, They are the capacitive reactive power balance, inductive reactive power balance and comprehensive reactive power balance of power supply unit k, and their values ​​range from 0 to 1;

[0051] (4) Power supply unit scale balance index O4

[0052] The power supply unit scale balance index O4 is introduced to limit the scale of the power supply unit, and its form is as follows:

[0053] O4=1 / [ασ(n Z )+1];

[0054] In the formula, σ(n Z ) is the standard deviation of the number of nodes contained in each power supply unit under the current power supply grid division scheme; α is the adjustment coefficient;

[0055] The power grid division includes four stages: initialization stage, tree structure pruning stage, node merging stage, and network collapse stage. The specific steps are as follows:

[0056] Step 21, initialization phase: read the original load of each node on each typical day, wind and solar output data, network initial weight matrix, node voltage information at peak and valley times of new energy penetration, treat each node as a power supply unit, and initialize the partitioning algorithm process parameters;

[0057] Step 22, tree structure pruning stage: find the leaf nodes in the current distribution network, merge them with the upper root node to form a new network, and repeat this process until the upper limit of the tree structure pruning layer is reached;

[0058] Step 23, node merging stage: traverse each node, add each node to the unit where the directly connected node is located in turn, calculate the increment of the comprehensive index after the node merging, select the node merging scheme with the largest increment, and merge the two nodes; repeat the node merging stage until the comprehensive index cannot increase, where:

[0059] For the active matching index O1 and reactive support capacity index O3, when calculating the index value of the node merging stage, only the index of the part where the composition of the power supply unit has changed is additionally calculated, that is:

[0060]

[0061] In the formula, k old , k in , k out They are the power supply unit numbers of the unchanged constituent nodes, the power supply unit numbers of the nodes that have migrated in, and the node numbers of the nodes that have migrated out; They are the average active matching degree of the corresponding power supply unit in all periods; are the comprehensive reactive power balance of the corresponding power supply units; N′ Z , O1′, O3′ are the number of power supply units, active matching index and reactive matching index after node merging respectively;

[0062] For the modularity index O2, only the influence of the edge weights of the nodes whose subordinate relationships have changed on the index increment is considered, that is:

[0063]

[0064] In the formula, S in , S out are the sum of the internal edge weights of the power supply unit to be incorporated into node i and the sum of the edge weights at the boundary of the power supply unit to be incorporated into node i; k i , k i,in are the sum of the edge weights of node i and the sum of the edge weights of node i and the power supply unit to be merged; O2′ is the modularity index after node merging;

[0065] Step 24, network collapse phase: collapse the nodes contained in each unit into super nodes, calculate the edge weights between the nodes of each power supply unit and the edge weights between the internal nodes, that is:

[0066]

[0067] In the formula, Ω k ,Ω k′ are the node sets of the kth and k′th network power supply units respectively; E kk′ is the edge weight between network power supply unit k and network power supply unit k′ after network collapse; s i is the sum of the edge weights of all connected edges of node i in the network before collapse; E kk is the self-weight of the network power supply unit k after the network collapses;

[0068] Step 25: Use the node edge weights between power supply units as the edge weights between super nodes, and use the node edge weights within the power supply unit as the self-weights of the super nodes to form a new network. Repeat the node merging stage and the network collapse stage until the comprehensive index no longer increases, that is, the distribution network power supply grid division scheme corresponding to the current typical day is obtained;

[0069] Step 3: Establish a grid partitioning scheme evaluation index system that includes the general applicability index of the partitioning results, the rationality index of the simulation operation, and the representative index of cluster extraction, and select the optimal grid partitioning scheme in the form of a radar chart, where:

[0070] The form of clustering to extract representative indicators is as follows:

[0071]

[0072]

[0073] In the formula, Γ i is the division scheme of the i-th power supply unit; F1 and F2 are the evaluation indexes of the surrounding number density and the radiation radius of the representative day respectively;

[0074] The form of the general applicability index of the division result is as follows:

[0075]

[0076] Where N p The number of power supply unit division schemes; O1(Γ j ,Data i ) is the power supply unit division scheme Γ j Substitute into the power supply unit division scheme Γ i The O1 index value obtained after the corresponding scene data; F3 and F4 are the functional index versatility and structural index versatility respectively;

[0077] The form of the simulation operation rationality index is as follows:

[0078]

[0079] In the formula, Dividing the power supply unit into Γ i Number of power supply units; Ω bus,k ,Ω line,k are the node set and line set in the kth power supply unit, N b 、N l is the corresponding number of nodes and lines; S b,h , are the actual usage value and upper limit of the distributed resource at node b in the hth period; σ(U b) is the standard deviation of the voltage series in the simulated operation cycle of node b; h(P l <0) is the power flow reversal time of line 1; F5, F6, and F7 are resource utilization, voltage stability, and power flow stability, respectively.

[0080] In the present invention, the power supply grid of the distribution network refers to a power supply range consisting of several adjacent plots (or user blocks) with the same power supply area classification level and basically the same power consumption nature or power supply reliability requirements. The power supply area is usually 5 to 12 km 2 .

[0081] Compared with the prior art, the present invention has the following advantages:

[0082] The present invention can obtain a division scheme from the distribution network power supply grid to the distribution network power supply unit according to specific wind, light and load timing scenarios and the actual distribution network grid structure, so as to enhance the system's new energy absorption capacity and improve orderly electricity consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 This is a flow chart of the distribution network power grid division method based on the improved Leuven algorithm;

[0084] Figure 2 This is a schematic diagram of the specific process of the distribution network grid division method based on the improved Leuven algorithm.

[0085] Figure 3 This is the topology diagram of the modified IEEE123 node example.

[0086] Figure 4 Decision diagrams are selected for typical days of various time series samples.

[0087] Figure 5 This is a heat map of the electrical distance calculation results.

[0088] Figure 6 This figure shows the changes of various indicators during the power supply grid division process under the two grid structures.

[0089] Figure 7 The power supply grid division process for two grid structures and different tree structure pruning layers.

[0090] Figure 8 The power supply grid division plan corresponding to each typical day.

[0091] Fig. 9 A radar chart showing the evaluation and selection results of each power supply grid division scheme. DETAILED DESCRIPTION

[0092] The technical solution of the present invention is further described below in conjunction with the accompanying drawings, but is not limited thereto. Any modification or equivalent replacement of the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention should be included in the protection scope of the present invention.

[0093] The present invention provides a distribution network power supply grid division method based on an improved Louvain algorithm. The method starts from the historical operation data of the new distribution network, realizes the division from the distribution network grid to the distribution network unit in a typical scenario, and aims at the binary paradox problem of the optimization ability and calculation efficiency of the division algorithm. An improved Louvain algorithm that considers the comprehensive optimization of functional indicators and structural indicators and introduces a tree structure pruning link is proposed to realize the division of the distribution network power supply grid, and a comparison process of each typical day division scheme is given.

[0094] like Figure 1 As shown, the specific steps include:

[0095] Step 1: Extract typical daily time series scenarios based on historical data under the combination of “wind-solar-load”

[0096] Based on the historical data of wind power, photovoltaic power and load within the distribution network, the optimization extraction model is used to extract typical daily time series scenarios.

[0097] In this step, the annual wind power data, annual photovoltaic data, and annual load data within the distribution network power supply grid are used as input, and the typical daily time series scene extraction is realized according to the following steps:

[0098] (1) According to the principal component analysis (PCA) method, the multivariate time series data of 365 days a year and including the "wind-solar-load" coupling are reduced to univariate time series data.

[0099] (2) The number of clusters is determined according to the elbow rule, and the K-means algorithm is used to classify the principal component sequence, dividing the annual univariate time series data into several time series data clusters.

[0100] (3) A multi-objective mixed integer linear programming model (MOMILP) is established as a typical day optimization extraction model to extract the most representative and differentiated typical day time series scenes from each cluster of time series data. The extraction model aims to minimize the number of typical days, minimize the deviation of typical scenes, maximize the surrounding number density, and maximize the radiation radius.

[0101] The surrounding number density refers to the number of samples within a data sample cutoff distance, and the calculation formula is as follows:

[0102]

[0103] d ij =DTW d (X i ,X j ) (i,j∈[1,N].i≠j) (2);

[0104] Where N is the number of typical day samples, i and j are the numbers of two typical day samples; ρ i is the surrounding number density of the i-th sample; ε(·) is the positive and negative judgment function, which takes 1 when the independent variable is negative and takes 0 when it is non-negative; d ij For sample X i With sample X j The matrix dynamic time warping distance between c is the sample set cutoff distance, which is taken as 1 / 3 of the maximum matrix dynamic time warping distance.

[0105] The radiation radius mainly reflects the differences among samples, and the calculation formula is as follows:

[0106]

[0107] Where j is the sample number whose surrounding number density is greater than the surrounding number density of the current sample; φ is an empty set.

[0108] According to equations (1) to (3), if the surrounding number density of a sample is larger and the radiation radius is larger, the sample is more representative and different, and is more likely to be selected as a typical day; if the radiation radius of a sample is larger but the surrounding number density is smaller, the sample can be regarded as an extreme sample and should also be included.

[0109] In order to reduce the difference between the total amount and distribution information of typical days and original data and limit the number of typical days, the statistical indicators mainly consider the deviation degree Δe of wind, light and load and the total number of typical days. Among them, the deviation degree is the relative deviation between the total amount obtained by weighted summation of the selected typical days and the total amount of original data, and the calculation formula is as follows:

[0110]

[0111] In the formula, ω i is the decision weight for selecting typical day i; ξ is the data type, S ξ,i is the total amount of data on a typical day i, S ξ,all is the total amount of data in the current sample set.

[0112] Based on the above indicators, the present invention establishes a multi-objective mixed integer linear programming model (MOMILP) to achieve the optimal extraction of typical days. The objective function is:

[0113]

[0114] in:

[0115]

[0116] In the formula, u=[u1,…,u N ] T ,ω=[ω1,…,ω N ] T are the binary selection decision variables and corresponding weight coefficient variables for each typical day; A ξ 、b ξ are the original data matrix of each data type and the total amount of data in each time period; f1 is the number of typical days selected in the current sample set; f2 is the sum of the relative deviations between the typical day and the original data, corresponding to formula (4); f3 is the number density around the typical day, corresponding to formula (1); f4 is the total radiation radius of the typical day, corresponding to formula (3).

[0117] The constraints of the typical day optimization extraction model are as follows:

[0118]

[0119] Wherein, constraint (a) is the consistency constraint between the typical day weight coefficient variable ω and the selected decision variable u; constraint (b) aims to ensure that the sum of all typical day weight coefficients is consistent with the sample size; constraint (c) is the total relative deviation range constraint, α is the range relaxation coefficient; constraint (d) is the type constraint of the two types of decision variables.

[0120] The above model is a mixed integer linear programming model, which can be solved directly by calling the solver.

[0121] Step 2: Based on the improved Leuven algorithm, power grid division is achieved with the goal of optimizing comprehensive indicators:

[0122] Based on typical time series scenarios and aiming at optimizing comprehensive indicators, the improved Louvain algorithm with the introduction of tree structure pruning is used to obtain the optimal power supply grid division scheme corresponding to each typical day.

[0123] In this step, the goal of power grid division of the distribution network is to divide the larger distribution network grid into several smaller distribution network units based on typical scenarios, grid topology and other information. From the perspective of supply and demand balance, each power supply unit needs to be as balanced and autonomous as possible, and the power exchange between each power supply unit needs to be as small as possible, so as to reduce the difficulty of power coordination between the power supply units; from the perspective of network structure, each power supply unit needs to have the characteristics of strong coupling between internal nodes and weak coupling between nodes of different power supply units, so as to reduce the impact of the action of a certain power supply unit on other power supply units. Therefore, the present invention starts from the functional and structural requirements, and establishes a comprehensive index for evaluating the effect of power grid division, covering four aspects: source-load-storage active matching index O1, modularity index based on electrical distance O2, reactive power support capacity index O3, and power supply unit scale balance index O4:

[0124] (1) Source-load-storage active power matching index O1

[0125] The net load sequence is a time-varying sequence of the difference between the load in the system and the output of the distributed generator (DG) after energy storage coordination, reflecting the dynamic change process of the supply and demand relationship in the power supply unit. Therefore, the optimized active net load sequence can be used to characterize the external characteristics of the power demand of the power supply unit, and the source-load-storage active matching index can be established based on this.

[0126] When optimizing the net load sequence, the present invention mainly considers uncontrollable resources such as wind turbine (WT), photovoltaic (PV), and controllable resources such as microturbine (MT), energy storage system (ESS), curtailable load, and transferable load. The specific optimization model objectives are as follows:

[0127]

[0128] Where H is the number of scheduling periods; Ω k is the node set of the kth power supply unit; is the net load value after optimal scheduling of source, load and storage of the kth power supply unit at time h; They are the original load power of the kth power supply unit at time h, the generalized load power after considering the energy storage charging and discharging power and the demand side response, and the DG power; They are respectively the transferred load power, reduced load power, energy storage charging power, energy storage discharging power, WT output, PV output, and MT output of the i-th node in the power supply unit.

[0129] In terms of constraints on net load optimization, the source side mainly considers the output range constraints, climbing constraints, and head-to-tail consistency constraints of MT; the load side mainly considers the range constraints and increase / decrease matching constraints of transferable loads, and the range constraints of reducible loads; the energy storage side mainly considers the range constraints and state constraints of ESS charging and discharging power, the range constraints, head-to-tail consistency constraints, and continuity constraints of battery state of charge (SOC).

[0130] According to the net load power of each power supply unit, the average value under a certain time scale is taken to construct the source-load-storage matching index O1:

[0131]

[0132] Where N Z The number of power supply units in the current power supply grid division scheme; is the source-load-storage matching degree of the kth power supply unit in the hth period, and

[0133] According to the indicator definition, the source-load-storage active power matching index O1 takes into account the power regulation capability of various resources, and its value range is between 0 and 1. The larger the value of O1, the closer the net load after the load, ESS and DG in each power supply unit are to 0, and the better the supply and demand balance of the current power supply grid division scheme is.

[0134] (2) Modularity index O2 based on electrical distance

[0135] The electrical distance based on the active / reactive power-voltage amplitude sensitivity matrix is ​​used to characterize the coupling relationship between nodes. The sensitivity matrix and electrical distance are calculated as follows:

[0136]

[0137] Where n is the number of distribution network nodes; ΔP i , ΔQ i , ΔV i are the changes of injected active power, injected reactive power and voltage amplitude of each node respectively; S PV , S QV They are active power-voltage amplitude sensitivity matrix and reactive power-voltage amplitude sensitivity matrix respectively; are the ratios of the voltage changes on node j (node ​​j) and node i when the active / reactive power of node j changes, indicating the degree of influence of the active / reactive power changes on the two nodes; D ij They are the active electrical distance, reactive electrical distance and comprehensive electrical distance between node i and node j respectively.

[0138] First, using the electrical distance and distribution network connection relationship, the distribution network is transformed into a relationship network that can directly calculate the modularity by calculating the edge weight. The edge weight calculation method is as follows:

[0139]

[0140] Where D is the comprehensive electrical distance matrix; A is the distribution network connection relationship matrix. When there is a direct connection line between node i and node j, A(i, j) = 1, otherwise A(i, j) = 0; e ij is the edge weight between node i and node j after conversion into a relational network.

[0141] After the distribution network is converted into a relational network, the modularity index O2 based on electrical distance can be directly calculated according to the modularity definition. The calculation method is as follows:

[0142]

[0143]

[0144] In the formula, k i , k j are the sum of the weights of all connecting edges of node i and node j respectively; m is the sum of the weights of all edges in the entire relationship network; δ(i,j) represents the same status of node i and node j. If the two nodes are divided into the same power supply unit, δ(i,j)=1, otherwise δ(i,j)=0.

[0145] (3) Reactive power support capability index O3

[0146] In terms of functionality, in addition to considering the source-load-storage active power matching of each power supply unit, a reactive power support capability index O3 describing the internal voltage regulation capability of the power supply unit should be established for the voltage control problem. The establishment of this index will be carried out through three steps: evaluating reactive power demand, evaluating reactive power supply capacity, and designing an evaluation index in the range of [0,1].

[0147] In terms of reactive power demand assessment of each power supply unit, the present invention will calculate the overvoltage node voltage at the peak of new energy penetration rate and the undervoltage node voltage at the underestimated penetration rate in a typical day under the condition of no control, and estimate the reactive power demand of each power supply unit according to the reactive power-voltage amplitude sensitivity matrix. The calculation method is as follows:

[0148]

[0149] In the formula, Ω k1 ,Ω k2 are the overvoltage node set and undervoltage node set of the kth power supply unit respectively; ΔV iis the absolute value of the voltage amplitude of the overvoltage / undervoltage node i from the normal voltage range; is the minimum capacitive reactive power required by the power supply side for the overvoltage node in the kth power supply unit; It is the minimum inductive reactive power required by the power supply side for the undervoltage node in the kth power supply unit.

[0150] In terms of reactive power supply capacity evaluation of each power supply unit, the present invention can describe the reactive power supply capacity of each power supply unit based on the set DG power factor range and SVC reactive capacity, taking into account the DG reactive output and SVC reactive output under the constant power factor on the source side. The specific calculation method is as follows:

[0151]

[0152] In the formula, are the source-side capacitive reactive power and inductive reactive power supply capabilities of the kth power supply unit respectively; They are the maximum capacitive reactive power that PV, WT and SVC can provide respectively; They are the maximum inductive reactive power that PV, WT, MT and SVC can provide respectively.

[0153] Based on the estimated reactive power demand and reactive power supply capacity of each power supply unit, a reactive power support capacity index O3 can be formed, which is in the following form:

[0154]

[0155] In the formula, They are capacitive reactive power balance, inductive reactive power balance and comprehensive reactive power balance of power supply unit k, and their values ​​range from 0 to 1. The closer to 1, the better the match between the estimated reactive power demand and the estimated reactive power supply capacity.

[0156] (4) Power supply unit scale balance index O4

[0157] Considering that the large difference in the scale of power supply units will have unnecessary impact on the difficulty of subsequent planning and the rationality of base station configuration results, the power supply unit scale balance index O4 is introduced in the structural index to limit the scale of power supply units. Its form is as follows:

[0158] O4=1 / [ασ(n Z )+1] (26);

[0159] In the formula, σ(n Z ) is the standard deviation of the number of nodes contained in each power supply unit under the current power supply grid division scheme; α is the adjustment coefficient. The greater the difference in the size of each power supply unit, the closer the index O4 is to 0, and vice versa.

[0160] In summary, the present invention performs weighted summation of the above four indicators to obtain a comprehensive evaluation index of the power supply grid division scheme.

[0161] Figure 2 The figure is a flow chart of the distribution network grid partitioning algorithm based on the improved Louvain algorithm. The power grid partitioning method proposed in the present invention includes four stages: initialization stage, tree structure pruning stage, node merging stage, and network collapse stage. The specific steps are as follows:

[0162] Step 21, initialization phase: read the original load of each node on each typical day, wind and solar output data, network initial weight matrix, node voltage information at peak and valley times of new energy penetration, and other initial data. Treat each node as a power supply unit and initialize the partitioning algorithm process parameters.

[0163] Step 22, tree structure pruning phase: find the leaf nodes in the current distribution network, merge them with the upper root node, and form a new network. Repeat this process until the upper limit of the tree structure pruning layer is reached.

[0164] Regardless of whether the main grid structure of the distribution network is a ring network or a radial network, there are often many small-scale tree-like lines at its ends. On the one hand, the characteristics of these terminal tree-like lines are often directly related to their root nodes, so their integrity should not be destroyed; on the other hand, if the tree-like lines are divided into different power supply units, power supply units with extremely small scales will inevitably be generated, which is contrary to the actual needs of the distribution network power supply unit division. In summary, the present invention collapses the small-scale tree structure at the end of the network into a super node in advance, which can avoid the frequent calculation and comparison processes during the first few rounds of node merging. In addition, early collapse can also reduce the problem of isolation of small-scale power supply units that may occur when nodes are traversed and merged.

[0165] Step 23, node merging stage: traverse each node, add each node to the unit where the directly connected node is located, calculate the increment of the comprehensive index after the node merging, select the node merging scheme with the largest increment, and merge the two nodes. Repeat the node merging stage until the comprehensive index cannot increase.

[0166] For the active matching index O1 and reactive support capacity index O3, when calculating the index value of the node merging stage, only the index of the part where the composition of the power supply unit has changed is additionally calculated, that is:

[0167]

[0168] In the formula, k old , k in , k out They are the power supply unit numbers of the unchanged constituent nodes, the power supply unit numbers of the nodes that have migrated in, and the node numbers of the nodes that have migrated out; They are the average active matching degree of the corresponding power supply unit in all periods; are the comprehensive reactive power balance of the corresponding power supply units; N′ Z , O1′, and O3′ are the number of power supply units, active matching index, and reactive matching index after the node is merged.

[0169] For the modularity index O2, only the influence of the edge weights of the nodes whose subordinate relationships have changed on the index increment is considered, that is:

[0170]

[0171] In the formula, S in , S out are the sum of the internal edge weights of node i to be incorporated into the power supply unit (both ends of the connection edge are in the power supply unit) and the sum of the edge weights at the boundary of the power supply unit to be incorporated (only one end of the connection edge is in the power supply unit); k i , k i,in They are respectively the sum of the edge weights connecting node i and the sum of the edge weights connecting node i and the power supply unit to be merged; O2′ is the modularity index after the node is merged.

[0172] Step 24, network collapse phase: collapse the nodes contained in each unit into super nodes, calculate the edge weights between the nodes of each power supply unit and the edge weights between the internal nodes, that is:

[0173]

[0174] In the formula, Ω k ,Ω k′ are the node sets of the kth and k′th network power supply units respectively; E kk′ is the edge weight between network power supply unit k and network power supply unit k′ after network collapse; s i is the sum of the edge weights of all connected edges of node i in the network before collapse; E kk is the self-weight of the network power supply unit k after the network collapses.

[0175] Step 25: Use the node edge weights between power supply units as the edge weights between super nodes, and use the node edge weights within the power supply units as the self-weights of the super nodes to form a new network. Repeat the node merging stage and the network collapse stage until the comprehensive index no longer increases, and you can get the distribution network power supply grid division plan corresponding to the current typical day.

[0176] Step 3: Establish a grid partitioning scheme evaluation index system that includes the versatility index of the partitioning results, the rationality index of the simulation operation, and the representative index of cluster extraction (sorted by importance), and select the optimal grid partitioning scheme in the form of a radar chart.

[0177] Since different typical scenarios will produce different power supply unit division schemes, it is also necessary to evaluate and select each power supply unit division scheme to determine the final power supply unit division result. The present invention establishes three types of evaluation and selection indicators, namely, the generalization index of the division result, the rationality index of the simulation operation, and the representative index of cluster extraction, to achieve the determination of the final power supply unit division scheme, wherein: the generalization index of the division result includes the generalization of the energy index and the generalization of the structural index; the rationality index of the simulation operation includes the utilization rate of distributed resources, the voltage stability, and the flow stability; the representative index of cluster extraction includes the typical day surrounding number density evaluation index and the typical day radiation radius evaluation index.

[0178] The meaning of clustering and extracting representative indicators is to use the complete annual data as the original sample, and use the typical day surrounding number density and radiation radius corresponding to each division scheme as the evaluation index. This indicator is only used to re-evaluate the rationality of the selection of typical days in the full data space, so it is the least important in the evaluation and selection process of the power supply unit division scheme. Its form is as follows:

[0179]

[0180] In the formula, Γ i is the division scheme of the i-th power supply unit; F1 and F2 are the evaluation indicators of the representative daily surrounding density and the representative daily radiation radius, respectively.

[0181] The meaning of the versatility index of the partition result is to substitute each power supply unit partition scheme into the scenarios corresponding to other partition schemes, obtain the values ​​of functional indicators and structural indicators, and use the ratio of the new indicator value to the original indicator value to reflect the versatility of each power supply unit partition scheme. It is the most important indicator in the evaluation and selection indicators. The specific form of the indicator is as follows:

[0182]

[0183] Where N p The number of power supply unit division schemes; O1(Γ j ,Data i ) is the power supply unit division scheme Γ j Substitute into the power supply unit division scheme Γ i The O1 index value is obtained after the corresponding scene data, and the other similar parameters are the same; F3 and F4 are the functional index versatility and structural index versatility respectively.

[0184] The meaning of the simulation operation rationality index is to splice each typical day into a simulation operation period in date order, solve the optimal power flow problem with the goal of minimizing line loss, and calculate the average resource utilization rate, average voltage fluctuation rate, and average power flow reversal rate of each power supply unit corresponding to each division scheme, and use this as a more important power supply unit division evaluation and selection index. Its form is as follows:

[0185]

[0186] In the formula, Dividing the power supply unit into Γ i Number of power supply units; Ω bus,k ,Ω line,k are the node set and line set in the kth power supply unit, N b 、N l is the corresponding number of nodes and lines; S b,h , are the actual usage value and upper limit of the distributed resource at node b in the hth period; σ(U b ) is the standard deviation of the voltage series in the simulated operation cycle of node b; h(P l <0) is the power flow reversal time of line 1; F5, F6, and F7 are resource utilization, voltage stability, and power flow stability, respectively.

[0187] After initially establishing the evaluation and selection indicators for the three types of power supply unit division schemes, each indicator can be mapped to the [0.1,1] interval according to the corresponding relationship between size and quality, and the power supply unit division schemes can be grouped and selected in the form of a radar chart. The reason why the minimum value of the mapped indicator is not set to 0 is to prevent the radar chart from appearing with spikes that interfere with subjective judgment.

[0188] Example:

[0189] This embodiment takes the modified 123-node calculation example as an example and adopts the modified IEEE 123-node system ( Figure 3 ) for simulation analysis. Assume that the maximum transferable load and the maximum load reduction ratio of each node are both 5%, the wind power photovoltaic power factor range is from leading 0.95 to lagging 0.95, the micro gas turbine power factor range is from leading 0.8 to leading 0.85, the energy storage charge state range is from 0.05 to 0.95, the charging and discharging efficiency is 0.9, and the reactive compensation device working conditions include capacitive and inductive working conditions. The specific division process is as follows:

[0190] (1) After clustering the principal component sequence, six typical time series scenes can be obtained. Based on the optimization extraction algorithm, typical days with outstanding typicality and differences are selected as time series scenes. Figure 4 .

[0191] (2) Check whether the electrical distance input matrix is ​​correct by using a heat map. Figure 5 .

[0192] (3) The distribution network power grid division method considering comprehensive indicators and improved Leuven algorithm proposed in the present invention is used for each time sequence scenario to obtain the division results. The example results show that the method proposed in the present invention can be applied to both open-loop and closed-loop working conditions. The parameter changes during the algorithm process are shown in Figure 6 , the impact of reducing the number of layers in the tree structure on the algorithm process can be seen in Figure 7 The typical day division results are shown in Figure 8 .

[0193] (4) According to the partitioning scheme optimization process proposed by the present invention, each partitioning scheme is grouped and optimized in the form of a radar chart. Fig. 9 After comprehensive consideration, the division scheme corresponding to Scheme 4 is selected to guide the zoning planning and operation of the new distribution network.

Claims

1. A method for dividing power distribution network grid based on improved Leuven algorithm, characterized in that The method comprises the following steps: Step 1: Based on the historical data of wind power, photovoltaic power generation and load in the power grid of the distribution network, the typical daily time series scene is extracted using the optimization extraction model; Step 2: Based on typical time series scenarios and aiming at optimizing comprehensive indicators, the improved Louvain algorithm with tree structure pruning is used to obtain the optimal power grid division scheme for each typical day; Step 3: Establish a grid partitioning scheme evaluation index system that includes the versatility index of the partitioning results, the rationality index of the simulation operation, and the representative index of cluster extraction, and select the optimal grid partitioning scheme in the form of a radar chart.

2. The method for dividing the power distribution network into grids based on the improved Louvain algorithm according to claim 1, characterized in that The specific steps of step 1 are as follows: Step 11: Reduce the dimension of the multivariate time series data of 365 days a year and including the "wind-solar-load" coupling to single variable time series data according to the principal component analysis method; Step 12: Determine the number of clusters according to the elbow rule, use the K-means algorithm to classify the principal component sequence, and divide the annual univariate time series data into several time series data clusters; Step 13: Establish a multi-objective mixed integer linear programming model as a typical day optimization extraction model, with the goals of minimizing the number of typical days, minimizing the deviation of typical scenes, maximizing the surrounding number density, and maximizing the radiation radius, to extract the most representative and differentiated typical day time series scenes from each cluster of time series data.

3. The method for dividing the power distribution network into grids based on the improved Louvain algorithm according to claim 2 is characterized in that In step 13, the objective function of the multi-objective mixed integer linear programming model is: in: In the formula, u=[u1,…,u N ] T ,ω=[ω1,…,ω N ] T are the binary selection decision variables and corresponding weight coefficient variables for each typical day; A ξ , b ξ are the original data matrix of each data type and the total amount of data in each period; f1 is the number of days selected as typical days in the current sample set; f2 is the sum of the relative deviations between the typical day and the original data, that is, the degree of deviation; f3 is the number density around the typical day; f4 is the total radiation radius of the typical day; The constraints for the typical day optimization extraction model are as follows: Wherein, constraint (a) is the consistency constraint between the typical day weight coefficient variable ω and the selected decision variable u; constraint (b) aims to ensure that the sum of all typical day weight coefficients is consistent with the sample size; constraint (c) is the total relative deviation range constraint, α is the range relaxation coefficient; constraint (d) is the type constraint of the two types of decision variables.

4. The method for dividing the power supply grid of the distribution network based on the improved Louvain algorithm according to claim 1 is characterized in that In step 2, the comprehensive index includes four aspects: source-load-storage active power matching index O1, modularity index based on electrical distance O2, reactive power support capacity index O3, and power supply unit scale balance index O4. The specific calculation method is as follows: (1) Source-load-storage active power matching index O1 The distributed generation power and generalized load power are optimized with the goal of minimizing the net load, thereby obtaining the distributed generation power and generalized load power corresponding to the optimal solution. The goals of the optimization model are as follows: Where H is the number of scheduling periods; Ω k is the node set of the kth power supply unit; is the net load value after optimal scheduling of source, load and storage of the kth power supply unit at time h; They are the original load power of the kth power supply unit at time h, the generalized load power after considering the energy storage charging and discharging power and the demand side response, and the DG power; are the transferred load power, reduced load power, energy storage charging power, energy storage discharging power, WT output, PV output, and MT output of the i-th node in the power supply unit; Calculate the source-load-storage matching index O1 based on the optimization results: Where N Z The number of power supply units in the current power supply grid division scheme; is the source-load-storage matching degree of the kth power supply unit in the hth period, and (2) Modularity index O2 based on electrical distance The electrical distance based on the active / reactive power-voltage amplitude sensitivity matrix is ​​used to characterize the coupling relationship between nodes. The sensitivity matrix and electrical distance are calculated as follows: Where n is the number of distribution network nodes; ΔP i , ΔQ i , ΔV i are the changes of injected active power, injected reactive power and voltage amplitude of each node respectively; S PV , S QV They are active power-voltage amplitude sensitivity matrix and reactive power-voltage amplitude sensitivity matrix respectively; are the ratios of the voltage changes between node j and node i when the active / reactive power of node j changes, indicating the impact of the active / reactive power changes on the two nodes; D ij They are respectively the active electrical distance, reactive electrical distance and comprehensive electrical distance between node i and node j; By using the electrical distance and distribution network connection relationship, the distribution network is transformed into a relationship network that can directly calculate the modularity by calculating the edge weight. The edge weight calculation method is as follows: Where D is the comprehensive electrical distance matrix; A is the distribution network connection relationship matrix. When there is a direct connection line between node i and node j, A(i, j) = 1, otherwise A(i, j) = 0; e ij is the edge weight between node i and node j after being transformed into a relational network; After the distribution network is converted into a relational network, the modularity index O2 based on electrical distance is calculated according to the modularity definition. The calculation method is as follows: In the formula, k i , k j are the sum of the weights of all connecting edges of node i and node j respectively; m is the sum of the weights of all edges in the entire relationship network; δ(i,j) represents the same state of node i and node j. If the two nodes are divided into the same power supply unit, δ(i,j) = 1, otherwise δ(i,j) = 0; (3) Reactive power support capability index O3 The reactive power demand is evaluated, the reactive power supply capacity is evaluated, and the evaluation index with a design range of [0,1] is carried out in three steps. In terms of reactive power demand assessment of each power supply unit, the voltage of overvoltage nodes at the peak of new energy penetration rate and the voltage of undervoltage nodes at the underestimated penetration rate in a typical day will be calculated under completely uncontrolled conditions. The reactive power demand of each power supply unit will be estimated according to the reactive power-voltage amplitude sensitivity matrix. The calculation method is as follows: In the formula, Ω k1 ,Ω k2 are the overvoltage node set and undervoltage node set of the kth power supply unit respectively; ΔV i is the absolute value of the voltage amplitude of the overvoltage / undervoltage node i from the normal voltage range; is the minimum capacitive reactive power required by the power supply side for the overvoltage node in the kth power supply unit; is the minimum inductive reactive power required by the power supply side for the undervoltage node in the kth power supply unit; In terms of reactive power supply capacity evaluation of each power supply unit, according to the set DG power factor range and SVC reactive capacity, the reactive power output of DG and SVC under the constant power factor on the source side are considered to describe the reactive power supply capacity of each power supply unit. The specific calculation method is as follows: In the formula, are the source-side capacitive reactive power and inductive reactive power supply capabilities of the kth power supply unit respectively; They are the maximum capacitive reactive power that PV, WT and SVC can provide respectively; They are the maximum inductive reactive power that PV, WT, MT and SVC can provide respectively; Based on the estimated reactive power demand and reactive power supply capacity of each power supply unit, a reactive power support capacity index O3 is formed, which is in the following form: In the formula, They are the capacitive reactive power balance, inductive reactive power balance and comprehensive reactive power balance of power supply unit k, and their values ​​range from 0 to 1; (4) Power supply unit scale balance index O4 The power supply unit scale balance index O4 is introduced to limit the scale of the power supply unit, and its form is as follows: O4=1 / [ασ(n Z )+1]; In the formula, σ(n Z ) is the standard deviation of the number of nodes contained in each power supply unit under the current power supply grid division scheme; α is the adjustment coefficient.

5. The method for dividing the power supply grid of the distribution network based on the improved Louvain algorithm according to claim 4 is characterized in that In step 2, the power grid division includes four stages: initialization stage, tree structure pruning stage, node merging stage, and network collapse stage. The specific steps are as follows: Step 21, initialization phase: read the original load of each node on each typical day, wind and solar output data, network initial weight matrix, node voltage information at peak and valley times of new energy penetration, treat each node as a power supply unit, and initialize the partitioning algorithm process parameters; Step 22, tree structure pruning stage: find the leaf nodes in the current distribution network, merge them with the upper root node to form a new network, and repeat this process until the upper limit of the tree structure pruning layer is reached; Step 23, node merging stage: traverse each node, add each node to the unit where the directly connected node is located in turn, calculate the increment of the comprehensive index after the node merging, select the node merging scheme with the largest increment, and merge the two nodes; repeat the node merging stage until the comprehensive index cannot increase, where: For the active matching index O1 and reactive support capacity index O3, when calculating the index value of the node merging stage, only the index of the part where the composition of the power supply unit has changed is additionally calculated, that is: In the formula, k old , k in , k out They are the power supply unit numbers of the unchanged constituent nodes, the power supply unit numbers of the nodes that have migrated in, and the node numbers of the nodes that have migrated out; They are the average active matching degree of the corresponding power supply unit in all periods; are the comprehensive reactive power balance of the corresponding power supply units; N′ Z , O1′, O3′ are the number of power supply units, active matching index and reactive matching index after node merging respectively; For the modularity index O2, only the influence of the edge weights of the nodes whose subordinate relationships have changed on the index increment is considered, that is: In the formula, S in , S out are the sum of the internal edge weights of the power supply unit to be incorporated into node i and the sum of the edge weights at the boundary of the power supply unit to be incorporated into node i; k i , k i,in are the sum of the edge weights of node i and the sum of the edge weights of node i and the power supply unit to be merged; O2′ is the modularity index after node merging; Step 24, network collapse phase: collapse the nodes contained in each unit into super nodes, calculate the edge weights between the nodes of each power supply unit and the edge weights between the internal nodes, that is: In the formula, Ω k ,Ω k′ are the node sets of the kth and k′th network power supply units respectively; E kk′ is the edge weight between network power supply unit k and network power supply unit k′ after network collapse; s i is the sum of the edge weights of all connected edges of node i in the network before collapse; E kk is the self-weight of the network power supply unit k after the network collapses; Step 25: Use the node edge weights between power supply units as the edge weights between super nodes, and use the node edge weights within the power supply units as the self-weights of the super nodes to form a new network. Repeat the node merging stage and the network collapse stage until the comprehensive index no longer increases, that is, the distribution network power supply grid division plan corresponding to the current typical day is obtained.

6. The method for dividing the power supply grid of the distribution network based on the improved Louvain algorithm according to claim 1 is characterized in that In step 3, the form of clustering to extract representative indicators is as follows: In the formula, Γ i is the division scheme of the i-th power supply unit; F1 and F2 are the evaluation indexes of the surrounding number density and the radiation radius of the representative day respectively; The form of the general applicability index of the division result is as follows: Where N p The number of power supply unit division schemes; O1(Γ j ,Data i ) is the power supply unit division scheme Γ j Substitute into the power supply unit division scheme Γ i The O1 index value obtained after the corresponding scene data; F3 and F4 are the functional index versatility and structural index versatility respectively; The form of the simulation operation rationality index is as follows: In the formula, Dividing the power supply unit into Γ i Number of power supply units; Ω bus,k ,Ω line,k are the node set and line set in the kth power supply unit, N b 、N l is the corresponding number of nodes and lines; S b,h , are the actual usage value and upper limit of the distributed resource at node b in the hth period; σ(U b ) is the standard deviation of the voltage series in the simulated operation cycle of node b; h(P l <0) is the power flow reversal time of line l; F5, F6 and F7 are resource utilization, voltage stability and power flow stability respectively.

Images