Medium-voltage power distribution network structure form evolution method considering distributed power supply and load growth

Abstract

The invention provides a medium-voltage power distribution network structure form evolution method considering distributed power supply and load growth. The method comprises the following steps: firstly, establishing a model of a target power distribution network, wherein the model comprises a node geographic position, node connection information, a load condition and prediction information of annual load and new energy in an evolution target year; establishing a power distribution network natural evolution model based on a complex network theory by considering load and distributed new energy growth; predicting a key time point of an evolution stage, obtaining a plurality of evolution results based on a Monte Carlo method according to a natural evolution model, and forming an initial population; and establishing a natural evolution optimization model of the power distribution network, and obtaining an optimal network frame through a genetic algorithm. According to the method, a power distribution network evolution model considering the influence of the high-proportion distributed power supply is provided, key nodes of the network transformed from a radiation type are estimated, and optimization solution is carried out on a future power distribution network frame structure.

Description

Technical Field

[0001] The present invention belongs to the field of distribution network planning, and in particular relates to a medium voltage distribution network framework evolution method taking into account distributed power sources and load growth. Background Art

[0002] In the context of the clean, low-carbon and intelligent energy revolution, as the global energy supply is transformed towards electrification, clean and low-carbon, a high proportion of distributed power access has become a prominent feature of the future development of the power system, leading to a huge change in the structure of the power system. The new distribution network has also assumed an increasingly important responsibility, and the distribution network is also rapidly evolving towards a future advanced form.

[0003] As the proportion of distributed power sources increases, the boundaries of the transmission and distribution network will become increasingly unclear. Distributed photovoltaic power generation will change the radial network structure of the traditional distribution network, change the trend, and the economic and safe operation of the distribution network will be affected to varying degrees. The traditional planning method for grid evolution research is difficult to reflect the development process and characteristics of the grid, while the method based on complex networks ignores the physical properties of the grid. Therefore, it is necessary to combine the two methods to establish a model for the evolution of the distribution network under the condition of distributed power generation access and optimize the solution to achieve the optimal distribution network morphological structure evolution under the index evaluation system.

[0004] In response to the above problems, a method for the structural evolution of medium-voltage distribution network considering the growth of distributed generation and load is designed. Summary of the invention

[0005] The purpose of the present invention is to address the defects of the prior art, to reflect the evolution mechanism of the distribution network under the access of a high proportion of distributed power sources, to provide an appropriate basis for the development and planning of the future new distribution network morphology, and to propose a method for the evolution of the structure of a medium-voltage distribution network taking into account the growth of distributed power sources and loads.

[0006] The technical solution of the present invention is as follows: the proposed method for calculating the carrying capacity of distributed photovoltaic power generation in a distribution network is characterized by comprising the following steps:

[0007] Step 1: Establish distribution network topology model, power flow calculation model and reliability calculation model;

[0008] Step 2: Considering the growth of load and distributed renewable energy, on the basis of the distribution network topology model in step 1, a distribution network natural evolution model is established based on complex network theory;

[0009] Step 3: Predict the key nodes of the evolution stage through the evolution model established in the previous step and obtain multiple evolution results according to the evolution model based on the Monte Carlo method to form an initial population;

[0010] Step 4: Establish a natural evolution optimization model for the distribution network, and use the initial population formed in step 3 to obtain the optimal grid through genetic algorithm;

[0011] Preferably, the distribution network topology structure model in step 1 refers to a model obtained by abstracting the geographical distribution network feeder into a graph consisting of points and edges based on graph theory and complex network theory, ignoring the physical properties of the grid components, and projecting it into an M×M grid graph, wherein each node on the feeder falls on the intersection of the grid graph. The distribution network topology structure model thus obtained includes the morphological structure, connection status and node geographical location information of the feeder;

[0012] The power flow calculation model described in step 1 specifically refers to compiling the distribution network feeder related data into a data file used in MATPOWER power flow calculation;

[0013] The reliability calculation model described in step 1 specifically refers to the data file used when calculating the reliability index of the distribution network feeder, including branch data and load data; the branch data must include the head end node n i , terminal node n j , line length l, failure rate λ, repair time t re , head end switch type st i , end switch type st j , switching time t s 、Line Type ty , the load data includes user type u ty , peak load P peak , average load P avg With the number of users N user ;

[0014] As a preferred embodiment, the natural evolution model of the distribution network based on complex network theory considering the growth of load and distributed new energy sources in step 2 divides the evolution of the distribution network into parallel node growth process and edge growth process. The original distribution network topology network is used as the basis, and new nodes are added in each evolution cycle. The position of the new node "growth" is determined by the probability method, and then the existing nodes are selected to establish connection lines based on factors such as the physical distance between nodes and the load density. With the increase of the evolution cycle, new nodes (new load points, distributed power generation devices, etc.) and new edges (new lines) are constantly "born" and connected to the original power grid, simulating the process of continuous evolution and development of the power grid in both time and space dimensions.

[0015] The method described in step 2 is specifically to set the parameters of the distributed generation devices, substations and lines in the power grid, and set the total number of evolution cycles T according to different power grid planning requirements, and use years as the time unit as the evolution cycle to achieve the growth of load points and distributed power sources. The steps of each evolution cycle are as follows;

[0016] First, the load data and energy data of each node are corrected by proportional conversion based on the predicted information of the load and new energy in each evolution cycle of the distribution network. For load point i, the load data of the current evolution cycle is corrected as follows:

[0017] P i,lgoal =α i (P lgoal -P lnow )+P i,lnow

[0018] Among them, P lgoal is the total load predicted by the distribution network in the current evolution period, P lnow is the total load of the distribution network in the last evolution cycle, α i is the load growth coefficient, which is related to the node position, P i,lnow is the load of load node i in the distribution network in the current evolution period;

[0019] For distributed power sources, the data at load point i is corrected as follows:

[0020] P i,egoal =β i (P egoal -P enow )+P i,enow

[0021] Among them, P egoal is the maximum energy amount predicted by the distribution network in the current evolution period, P enow is the maximum energy of the distribution network in the previous cycle, β i is the proportional coefficient of energy growth, which is related to the node location, P i,enow is the maximum energy of load node i in the region at the current stage;

[0022] Furthermore, the new load point location is selected and the new line is established to complete the load node growth process and edge growth process. It is assumed that one load node is added in each evolution cycle, 10 growth points are randomly selected in the coordinate system, and the candidate location is selected as the new node location according to the probability represented by ζ. According to Π(k i ) to establish a new line between the probability selection node and the newly added load point; the coordinate system refers to the coordinate system corresponding to the grid diagram in step 1.

[0023] The new node is set to be "born" in the "empty position" near the existing node, that is, there is no grid intersection around the grid intersection where the existing node is located. The position where a new node can be generated near the grid node is called the growth point. One node is added in each evolution cycle, and n growth points are randomly selected in the coordinate system as candidate growth points. Considering the economic and power supply costs, the load density of each candidate growth point is used as the probability of site selection to determine the location of the new node. In the formula, P a , P c They represent the loads of nodes a and c adjacent to node d, respectively, and P is the sum of the loads of all nodes;

[0024]

[0025] The newly added load point is connected to an existing node with a certain probability through a specific preferential principle. The probability of a new node connecting to an existing node is determined by the physical distance and the degree of the node. The new node tends to connect to nodes with large node degrees and short distances from itself. The probability of a new node connecting to node i is:

[0026]

[0027] Where: k i is the degree of node i; It is the preferred factor for connecting new nodes to existing nodes. The specific expression is as follows:

[0028]

[0029] Where: l i is the physical distance from node i to the new node. The exponents α and β can be used to adjust the influence of node degree and physical distance;

[0030] Furthermore, the new distributed generation location is selected according to the load concentration degree ζ, and connected to the nearest load node, completing the distributed power node growth process and edge growth process. Unlike the load point, the distributed power location selection can be selected not only at the growth point location, but also at the original load node location;

[0031] Furthermore, the distribution network topology model, power flow model and reliability calculation model are updated. The distribution network topology model updates the node location information and connection information, the power flow model updates the node load information and connection line information, the reliability calculation model updates the node load information, connection line information and system fault information, and the power flow is calculated to check whether the voltage exceeds the limit. If it exceeds the limit, the evolution is stopped;

[0032] Repeat the above steps and finally reach the evolution target time to end the evolution.

[0033] The key time nodes of the predicted evolution stage described in step 3 are the time nodes predicted by summarizing the evolution results of multiple distribution network natural evolution models. In the evolution model, voltage over-limit will cause the evolution to stop, and the average value of the evolution stop time is determined as the key time node of the distribution network evolution;

[0034] In step 3, multiple evolution results are obtained according to the natural evolution model based on the Monte Carlo method, and the natural evolution model of the distribution network is simulated multiple times according to the key time nodes as the evolution target years to obtain the results;

[0035] The forming of the initial population in step 3 refers to simulating the natural evolution of the distribution network based on the Monte Carlo method, repeating it N times, calculating the index and sorting the index calculation results from high to low, and selecting the evolution results with the top N / 2 index rankings as the initial population for genetic algorithm optimization;

[0036] The index system includes reliability index, status index and economic index. The reliability index specifically includes the system average power outage frequency coefficient K SAIFI , System average power outage time coefficient K SAIDI 、Expected value of insufficient power K EENS and power supply reliability K ASAI , the state indicators include improved power flow distribution entropy H n The voltage Theil entropy T iu , economic indicators include network loss rate C loss and construction cost C price .

[0037] The reliability index calculation formula is:

[0038] K SAIFI =Σλ i N i / ΣN i

[0039] K SAIDI =Σ(N i U i ) / U i

[0040]

[0041] K ASAI =(8760ΣN i -∑N i U i ) / (8760∑N i )

[0042] In the formula, λ i is the failure rate at load point i, N i is the number of users at load point i, U i is the average annual power outage time of the load point, Pijk (X) is the probability that the outage capacity is greater than X at the kth hour on the jth day of the i-th time period; m is the number of time periods in a year; n i is the number of days in the i-th time period; L ijk is the hourly load at the kth hour of the jth day in the i-th time period.

[0043] Assume that after node i stops operating due to a fault, the real power flow of line l is P l 0 (i) Define the load rate of line l as:

[0044]

[0045] Where P l max (i) is the maximum power flow limit of the line.

[0046] Divide the line load rate into n reasonable sections and set the constant sequence X = [0, e, 2e, ..., ne], satisfying ne = 1. Place each line in the appropriate load rate interval [ke, (k + 1) e) and the overloaded line in the interval [1, ∞), then the improved power flow distribution entropy H n The indicator expression is:

[0047]

[0048] In the formula, is the probability that the line is in the interval [ke, (k+1)e); l k is the number of lines whose load factor is in the interval [ke,(k+1)e); is the average load rate of the line in the section [ke, (k+1)e); δ is the overload coefficient, when the line exceeds the limit, δ>1, otherwise δ=1.

[0049] Suppose that a certain type of object h in a system is partitioned into M zones with a total of N objects. It is defined that a certain link z in the system changes, and a certain indicator y of the object h in the system changes accordingly. The Theil entropy T in the zone g is g for:

[0050]

[0051] In the formula, h is the classification object in the system; N is the total number of classification objects; N g is the total number of objects in region g; Y h(h∈g) Indicates the change of a certain indicator of object h in region g; Y g Represents the sum of the index changes of all objects h in the region g.

[0052] Considering global changes, the improved Theil entropy index is:

[0053]

[0054] Where, T zy T is the amount of change in index y caused by the change in index z in the region; l is the unevenness of the change in y in each region; T b is the unevenness of the y variation between regions; z is the sum of the changes in indicator y.

[0055] Based on the definition of Theil entropy, the power network with N nodes is partitioned according to the node type, with PQ nodes in zone 1 and PV nodes and balance nodes in zone 2. When the unit load is increased at node i, the voltage change of node j is obtained, and then the node voltage change of the global network is obtained as follows:

[0056]

[0057] Where U ij is the voltage change caused by node j; U' j is the initial voltage value of node j.

[0058] Voltage Theil entropy T iu for:

[0059]

[0060] The network loss and network loss rate are expressed as:

[0061]

[0062] Where P iiloss is the power loss on the ii-th branch. The total network loss P of the system is obtained by summing up the losses of all branches. loss .P load is the total power load in the system.

[0063] Construction cost price It is expressed as:

[0064] C price =C line ×l+C euip ×nn-I

[0065] Where C line is the line cost per unit length, l is the line length, C euip is the equipment cost, nn is the number of equipment, and I is the benefit of policy incentives.

[0066] System average power outage frequency coefficient K SAIFI , System average power outage time coefficient K SAIDI、Expected value of insufficient power K EENS , Improved power flow distribution entropy H n , Voltage Theil entropy T iu , network loss rate C loss and construction cost C price The smaller the index value, the better the power grid evolution result is. ASAI The larger the index value, the better the grid evolution result.

[0067] The indicators are subjectively weighted by the hierarchical analysis method and objectively weighted by the entropy weight method. Finally, the comprehensive weights w1 to w8 are obtained, and the evolution results of each time are evaluated by the following formula;

[0068] f=-w1SAIFI-w2SAIDI-w3EENS+w4ASAI-w5C loss -w6C price -w7H n -w8T iu

[0069] Preferably, the natural evolution optimization model of the distribution network described in step 4 optimizes the distribution network frame of the target year based on the natural evolution model of the distribution network, takes the maximum index evaluation score as the optimization target, takes the location of the load point and the distributed power source and the safe operation of the distribution network as constraints, takes the site selection and access of the load point and the distributed power source as decision variables, and adopts a genetic algorithm to solve;

[0070] The objective function of the optimization model described in step 4 is to maximize the weighted sum of all indicators in the indicator system. Compared with the evaluation of evolution results, C pos With C con To measure the extent to which the model conforms to the two rules that new load points have a greater probability of appearing at locations with higher load density and have a greater probability of connecting to nodes with closer physical distances and larger node degrees.

[0071] max f′=-w′1K SAIFI -w′2K SAIDI -w3′K EENS +w′4K ASAI -w5′C loss -w′6C price -w′7H n -w8′T iu +w9′C pos +w′ 10 C con

[0072] In the formula, w1' to w 10 ' is the weight of each indicator that is comprehensively weighted through the hierarchical analysis method and the entropy weight method.

[0073] The constraints of the optimization model in step 4 include:

[0074] Power flow constraints:

[0075]

[0076] Where P i With Q i are the active and reactive power injected at node i, U i , U j are the voltage amplitudes of nodes i and j respectively, G ij With B ij is the real and imaginary part of the corresponding element in the system node admittance matrix. ij The phase angle difference between the voltages at nodes i and j when the node voltages are expressed in polar coordinates.

[0077] Node voltage constraints:

[0078] U min ≤U i ≤U max

[0079] Where U i is the voltage at node i, U max and U min They are the upper and lower voltage limits specified during system operation.

[0080] Circuit thermal limit constraints:

[0081] I i %≤I max %

[0082] The line thermal limit constraint actually means that the line load rate should not exceed the thermal stability limit of the line. i % is the load factor of the line, then I max % is the thermal stability limit of the line.

[0083] The decision variables of the optimization model in step 4 are the location information and connection information of the newly added load points and the location information of the distributed power sources each year;

[0084] Furthermore, the location information of the newly added load point is represented by the existing node number to which the growth point belongs and its position information relative to the existing node, that is, its location in the grid diagram at the upper left, directly above, upper right, directly left, directly below, and lower right of the node is represented by numbers 1 to 8 respectively.

[0085] Furthermore, the connection information refers to the number of the connected node. The newly added distributed power source location information is also determined by the existing node number and relative position number. The relative position number is 0 to 8, where the number 0 represents that the distributed power source is located above the load point, and the numbers 1 to 8 represent that it is located at the upper left, directly above, upper right, directly left, directly below, and lower right of the node in the grid diagram;

[0086] Furthermore, the decision variables are binary-encoded onto chromosomes, where the information of newly added load points each year includes the 8-bit binary number of the existing node, the 3-bit binary number of the relative position and the 8-bit binary number of the connected node, and the newly added distributed power source includes the 8-bit binary number of the existing node and the 4-bit binary number of the relative position. A total of 31 bits are regarded as a gene, and the evolutionary years are the number of genes, which together constitute a chromosome;

[0087] Further, the formed initial population is encoded according to the above method to obtain the initial population of the genetic algorithm;

[0088] Furthermore, the selection algorithm, crossover algorithm and mutation algorithm in the genetic algorithm are solved by using a non-playback random sampling algorithm, a two-point crossover algorithm and a global mutation respectively; the selection operation in the genetic algorithm adopts a non-playback random remainder selection method, the crossover operation selects a two-point crossover, the mutation operation adopts a global mutation, and the fitness function is the objective function constructed in the above steps; by repeatedly implementing selection, crossover and mutation operations, the fitness value of each individual is calculated, offspring with good adaptability are retained, and offspring that do not meet the conditions are eliminated. After several generations of inheritance, it converges to the chromosome with the best conditions, and then the decoding operation corresponding to the encoding method is used to obtain the final optimization result.

[0089] The advantages of the present invention are:

[0090] Without presetting the classical model, starting from the most intuitive dynamic behavior of the distribution network, the access of distributed power sources and the growth of load, the natural evolution of the distribution network is simulated based on the regularity and randomness of evolution, and a natural growth model of the medium-voltage distribution network is constructed; from the perspective of evolution, different evolutionary driving factors and driving forces are combined to establish a distribution network evolution evaluation index system; based on the index system, the medium-voltage distribution network structural morphology optimization objective function is established, and the genetic algorithm is used to solve it using a coding method suitable for the evolution model. Compared with the traditional planning method, the present invention takes into account the node growth and connection laws of natural evolution, and compared with the complex network method, it takes into account the physical properties of the distribution network, and adopts a more comprehensive index system to evaluate and optimize the evolution results, which is more meaningful for the evolution of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] Figure 1 A flow chart of a method for evolving the structure of a medium voltage distribution network taking into account growth of distributed power sources and loads provided by the present invention;

[0092] Figure 2 Modeling flow chart for natural evolution model of distribution network;

[0093] Figure 3 The feeder topology diagram of a regional distribution network for testing the natural evolution model and optimization method of the distribution network;

[0094] Figure 4 This is the topological structure diagram after the evolution and optimization of the feeder lines of the distribution network in a certain area. DETAILED DESCRIPTION

[0095] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in each embodiment of the present invention described below can be combined with each other as long as they do not conflict with each other. For parameters that require actual situation analysis, we have noted the parameter setting method above and will not repeat it here.

[0096] The steps of the present invention are further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0097] like Figure 1 As shown, it is a flow chart of the grid evolution model and optimization method of high-proportion distributed generation access distribution network based on complex network, including establishing the target distribution network model, establishing the natural evolution model of the distribution network based on complex network theory, predicting the key nodes of evolution and forming the initial population based on the Monte Carlo method, establishing the optimization model and solving it based on the genetic algorithm.

[0098] like Figure 2 As shown in the figure, it is the modeling flow chart of the natural evolution model of distribution network.

[0099] like Figure 3 As shown, it is a feeder topology diagram of an embodiment of the present invention.

[0100] like Figure 4 As shown, it is a diagram of the feeder topology structure after evolution and optimization according to an embodiment of the present invention.

[0101] Combine the following Figures 1 to 4 The specific implementation of the present invention is a method for evolving the structure of a medium voltage distribution network considering the growth of distributed power sources and loads, which specifically includes the following steps:

[0102] Preferably, the distribution network topology model in step 1 refers to a model obtained by abstracting the geographical distribution network feeder into a graph consisting of points and edges based on graph theory and complex network theory, ignoring the physical properties of the grid components, and projecting it into a 100*100 grid graph, wherein each node on the feeder falls on the intersection of the grid graph. The distribution network topology model thus obtained includes the morphological structure, connection status and node geographical location information of the feeder;

[0103] The power flow calculation model described in step 1 specifically refers to compiling the distribution network feeder related data into a data file used in MATPOWER power flow calculation;

[0104] The reliability calculation model described in step 1 specifically refers to the data file used when calculating the reliability index of the distribution network feeder, including branch data and load data; the branch data must include the head end node n i , terminal node n j , line length l, failure rate λ, repair time t re , head end switch type st i , end switch type st j , switching time t s 、Line Type ty , the load data includes user type u ty , peak load P peak , average load P avg With the number of users N user ;

[0105] As a preferred embodiment, the natural evolution model of the distribution network based on complex network theory considering the growth of load and distributed new energy sources in step 2 divides the evolution of the distribution network into parallel node growth process and edge growth process. The original distribution network topology network is used as the basis, and new nodes are added in each evolution cycle. The position of the new node "growth" is determined by the probability method, and then the existing nodes are selected to establish connection lines based on factors such as the physical distance between nodes and the load density. With the increase of the evolution cycle, new nodes (new load points, distributed power generation devices, etc.) and new edges (new lines) are constantly "born" and connected to the original power grid, simulating the process of continuous evolution and development of the power grid in both time and space dimensions.

[0106] The method described in step 2 is specifically to set the parameters of the distributed generation devices, substations and lines in the power grid according to the original power grid information, and set the evolution cycle according to different power grid planning requirements, and use the year as the time unit as the evolution cycle to achieve the growth of load points and distributed power sources. The steps of each evolution cycle are as follows;

[0107] First, the load data and energy data of each node are corrected by proportional conversion based on the predicted information of the load and new energy in each evolution cycle of the distribution network. For load point i, the load data of the current evolution cycle is corrected as follows:

[0108] P i,lgoal =α i (P lgoal -P lnow )+P i,lnow

[0109] Among them, P lgoal is the total load predicted by the distribution network in the current evolution period, P lnow is the total load of the distribution network in the last evolution cycle, α i is the load growth coefficient, which is related to the node position, P i,lnow is the load of load node i in the distribution network in the current evolution period;

[0110] For distributed power sources, the data at load point i is corrected as follows:

[0111] P i,egoal =β i (P egoal -P enow )+P i,enow

[0112] Among them, P egoal is the maximum energy amount predicted by the distribution network in the current evolution period, P enow is the maximum energy of the distribution network in the previous cycle, β i is the proportional coefficient of energy growth, which is related to the node location, P i,enow is the maximum energy of load node i in the region at the current stage;

[0113] Furthermore, the new load point location is selected and the new line is established to complete the load node growth process and edge growth process. It is assumed that one load node is added in each evolution cycle, 10 growth points are randomly selected in the coordinate system, and the candidate position is selected as the new node position according to the probability represented by ζ. i ) represents the probability of selecting nodes and establishing new lines between newly added load points;

[0114] The new node is set to be "born" in the "empty position" near the existing node, that is, there is no grid intersection around the grid intersection where the existing node is located. The position where a new node can be generated near the grid node is called the growth point. One node is added in each evolution cycle, and 10 growth points are randomly selected in the coordinate system as candidate growth points. Considering the economic and power supply costs, the load density of each candidate growth point is used as the probability of site selection to determine the location of the new node. In the formula, P a, P c They represent the loads of nodes a and c adjacent to node d, respectively, and P is the sum of the loads of all nodes;

[0115]

[0116] The newly added load point is connected to an existing node with a certain probability through a specific preferential principle. The probability of a new node connecting to an existing node is determined by the physical distance and the degree of the node. The new node tends to connect to nodes with large node degrees and short distances from itself. The probability of a new node connecting to node i is:

[0117]

[0118] Where: k i is the degree of node i; It is the preferred factor for connecting new nodes to existing nodes. The specific expression is as follows:

[0119]

[0120] Where: l i is the physical distance from node i to the new node. The exponents α and β can be used to adjust the influence of node degree and physical distance;

[0121] Furthermore, the new distributed generation location is selected according to the load concentration degree ζ, and connected to the nearest load node, completing the distributed power node growth process and edge growth process. Unlike the load point, the distributed power location selection can be selected not only at the growth point location, but also at the original load node location;

[0122] Furthermore, the distribution network topology model, power flow model and reliability calculation model are updated. The distribution network topology model updates the node location information and connection information. The power flow model updates the node load information and connection line information. The reliability calculation model considering the distributed power source updates the node load information, connection line information and system fault information. The power flow is calculated to check whether the voltage exceeds the limit. If it exceeds the limit, the evolution is stopped;

[0123] Repeat the above steps and finally reach the evolution target time to end the evolution.

[0124] The key nodes of the predicted evolution stage in step 3 are the time nodes predicted by summarizing the evolution results of multiple distribution network natural evolution models. In the evolution model, voltage over-limit will cause the evolution to stop, and the average value of the evolution stop time is determined as the key node of the distribution network evolution;

[0125] In step 3, a plurality of evolution results are obtained according to the natural evolution model based on the Monte Carlo method, and the natural evolution model of the distribution network is simulated multiple times according to the key time points as the evolution target years to obtain the results;

[0126] The forming of the initial population in step 3 refers to simulating the natural evolution of the distribution network based on the Monte Carlo method, repeating it N times, calculating the index and sorting the index calculation results from high to low, and selecting the evolution results with the top N / 2 index rankings as the initial population for genetic algorithm optimization;

[0127] The index system includes reliability index, status index and economic index. The reliability index includes SAIFI, SAIDI, EENS and ASAI. The status index includes improved power flow distribution entropy H. n The voltage Theil entropy T iu , economic indicators include network loss rate C loss and construction cost C price .

[0128] The indicators are subjectively weighted by the hierarchical analysis method and objectively weighted by the entropy weight method. Finally, the comprehensive weights w1 to w8 are obtained, and the evolution results of each time are evaluated by the following formula;

[0129] f=-w1SAIFI-w2SAIDI-w3EENS+w4ASAI-w5C loss

[0130] -w6C price -w7H n -w8T iu

[0131] Preferably, the natural evolution optimization model of the distribution network described in step 4 optimizes the distribution network frame of the target year based on the natural evolution model of the distribution network, takes the maximum index evaluation score as the optimization target, takes the location of the load point and the distributed power source and the safe operation of the distribution network as constraints, takes the site selection and access of the load point and the distributed power source as decision variables, and adopts a genetic algorithm to solve;

[0132] The objective function of the optimization model described in step 4 is to maximize the weighted sum of all indicators in the indicator system:

[0133] max f′=-w1′SAIFI-w′2SAIDI-w3′EENS+w′4ASAI-w5′C loss -w′6C price -w′7H n -w′8T iu +w′9C pos +w′ 10 C con

[0134] In the formula, w1' to w 10 ' is the weight of each indicator, SAIFI, SAIDI, EENS and ASAI are reliability indicators, Hn To improve the power flow distribution entropy, T iu is the voltage Theil entropy, C loss is the network loss rate, C price In order to reflect the natural evolution model in the algorithm, the newly added load points have a greater probability of appearing in locations with higher load density and have a greater probability of connecting to nodes with shorter physical distances and higher node degrees. pos With C con To measure the degree to which the model complies with the above two laws.

[0135] The constraints of the optimization model in step 4 include:

[0136] Power flow constraints:

[0137]

[0138] Where P i With Q i are the active and reactive power injected at node i, U i , U j are the voltage amplitudes of nodes i and j respectively, G ij With B ij is the real and imaginary part of the corresponding element in the system node admittance matrix. ij The phase angle difference between the voltages at nodes i and j when the node voltages are expressed in polar coordinates.

[0139] Node voltage constraints:

[0140] U min ≤U i ≤U max

[0141] Where U i is the voltage at node i, U max and U min They are the upper and lower voltage limits specified during system operation.

[0142] Circuit thermal limit constraints:

[0143] I i %≤I max %

[0144] The line thermal limit constraint actually means that the line load rate should not exceed the thermal stability limit of the line. i % is the load factor of the line, then I max % is the thermal stability limit of the line.

[0145] The decision variables of the optimization model in step 4 are the location information and connection information of the newly added load points and the location information of the distributed power sources each year;

[0146] Furthermore, the location information of the newly added load point is represented by the existing node number to which the growth point belongs and its position information relative to the existing node, that is, its location in the grid diagram is represented by numbers 1 to 8 at the upper left, directly above, directly left of the upper right, lower left to the right, directly below, and lower right of the node respectively.

[0147] Furthermore, the connection information refers to the number of the connected node. The newly added distributed power location information is also determined by the existing node number and relative position number, but the difference is that the distributed power location can be the original load point location, so the relative position number is 0 to 8, where 0 represents that the distributed power location is above the load point;

[0148] Furthermore, the decision variables are binary-encoded onto chromosomes, where the information of newly added load points each year includes the 8-bit binary number of the existing node, the 3-bit binary number of the relative position and the 8-bit binary number of the connected node, and the newly added distributed power source includes the 8-bit binary number of the existing node and the 4-bit binary number of the relative position. A total of 31 bits are regarded as a gene, and the evolutionary years are the number of genes, which together constitute a chromosome;

[0149] Further, the formed initial population is encoded in this way;

[0150] Furthermore, the selection algorithm, crossover algorithm and mutation algorithm in the genetic algorithm are solved by using the non-replay random sampling algorithm, the two-point crossover algorithm and the global mutation respectively;

[0151] The following embodiments use a 10kV medium voltage distribution network feeder and its distribution area in a certain area as the research object of the evolution process. The topology is as follows: Figure 3 As shown, the feeder line includes 64 load nodes with a total length of 11.446 km. The main line adopts JKLYJ-240 conductor.

[0152] The total load is 8.64MW, the annual load growth rate is 5%, and the distributed photovoltaic penetration rate increases by 2% each year. The total load, distributed new energy penetration rate, and installed capacity examples every 5 years are shown in Table 1;

[0153] Table 1

[0154]

[0155] After simulating the natural evolution model of the distribution network 50 times, the key evolution node was found to be 15 years;

[0156] The target year of evolution is set to 15 years, and the Monte Carlo method is used to simulate 100 times. Finally, 50 evolution results are retained as the initial population. The final optimization result is obtained by genetic algorithm. Figure 4 As shown;

[0157] The comparison between a random evolution result and the optimization result is shown in Table 2;

[0158] Table 2

[0159]

[0160] It should be understood that the parts not elaborated in detail in this application belong to the prior art.

[0161] It should be understood that the above description of the preferred embodiment is relatively detailed and cannot be regarded as limiting the scope of protection of the patent of this application. Under the guidance of the present invention, ordinary technicians in this field can also make substitutions or modifications without departing from the scope of protection of the claims of this application, all of which fall within the scope of protection of this application. The scope of protection requested for this application shall be based on the attached claims.

Claims

1. A method for the structural evolution of a medium voltage distribution network considering the growth of distributed power sources and loads, characterized in that: The following steps are involved: Step 1: Establish distribution network topology model, power flow calculation model and reliability calculation model; Step 2: Considering the growth of load and distributed renewable energy, on the basis of the distribution network topology model in step 1, a distribution network natural evolution model is established based on complex network theory; Step 3: Predict the key time nodes of the evolution stage through the natural evolution model of the distribution network established in the previous step and obtain multiple evolution results according to the natural evolution model of the distribution network based on the Monte Carlo method to form an initial population; Step 4: Establish a natural evolution optimization model for the distribution network, and use the initial population formed in step 3 to obtain the optimal grid through genetic algorithm.

2. According to claim 1, a method for the structural evolution of a medium voltage distribution network considering the growth of distributed power sources and loads is characterized in that: In step 1, the specific operation method of establishing the distribution network topology model, the power flow calculation model and the reliability calculation model is as follows: Step 1.1, abstract the geographical distribution network feeder into a graph consisting of points and edges, and project it into an N×N grid graph to obtain a model, in which each node on the feeder falls on the intersection of the grid graph, and establish a distribution network topology model; the distribution network topology model thus obtained includes the morphological structure, connection status and node geographical location information of the feeder; Step 1.2, compile the distribution network feeder related data into the data file used in MATPOWER power flow calculation, and establish the power flow calculation model; Step 1.3, create a data file containing reliability calculations for nodes, line length, failure rate, repair time, switch type, switching time, line type, user type, peak load, average load, and number of users.

3. According to claim 1, a method for the structural evolution of a medium voltage distribution network considering the growth of distributed power sources and loads is characterized in that: In step 2, the specific operation method of constructing the natural evolution model of the distribution network is as follows: Step 2.1, setting the parameters of substations, distributed generation, and lines in the distribution network, the number of evolution cycles, and the load and distributed generation data that need to be added in each evolution cycle; Step 2.2: In each evolution cycle, if it is necessary to add a load point or distributed generation, the location of the new load point or distributed generation is selected according to the load density ζ, and the connection probability Π(k i ) Select a node to create a new line; Step 2.3, check the parameters of distribution network evolution in turn, including whether the load growth reaches the limit of the region, whether the voltage and current of the distribution network are over-limit, whether the reliability parameters do not meet the requirements, and whether the evolution time reaches the set number of cycles. If any of the above is found, the evolution simulation ends. Step 2.4: Repeat steps 2.2 and 2.3 until the target evolution time is reached, and then end the evolution.

4. A method for structural evolution of a medium voltage distribution network considering growth of distributed power sources and loads according to claim 3, characterized in that: The method described in step 2 is specifically to set the parameters of the distributed generation devices, substations and lines in the power grid, and set the total number of evolution cycles T according to different power grid planning requirements, and use years as the time unit as the evolution cycle to achieve the growth of load points and distributed power sources. The steps of each evolution cycle are as follows; First, the load data and energy data of each node are corrected by proportional conversion based on the predicted information of the load and new energy in each evolution cycle of the distribution network. For node i, the load data of the current evolution cycle is corrected as follows: P i,lgoal =a i (P lgoal -P lnow )+P i,lnow Among them, P lgoal is the total load predicted by the distribution network in the current evolution period, P lnow is the total load of the distribution network in the last evolution cycle, α i is the load growth coefficient, which is related to the node position, P i,lnow is the load of node i in the distribution network in the current evolution period; For distributed power sources, the data at node i is corrected to: P i,egoal =β i (P egoal -P enow )+P i,enow Among them, P egoal is the maximum energy amount predicted by the distribution network in the current evolution period, P enow is the maximum energy of the distribution network in the previous cycle, β i is the proportional coefficient of energy growth, which is related to the node location, P i,enow is the maximum energy of node i in the region at this stage; The new load point location is selected and the new line is established to complete the load node growth process and edge growth process; one load node is added in each evolution cycle, n growth points are randomly selected in the coordinate system, and the candidate location is selected as the new load node location according to the probability represented by ζ, and the load node location is selected according to Π(k i ) to establish a new line between the probability selection node and the newly added load node; The new load node is set to be "born" in the "empty position" near the existing node, that is, there is no grid intersection around the grid intersection where the existing node is located; the position where a new node can be generated near the grid node is the growth point; one node is added in each evolution cycle, and n growth points are randomly selected in the coordinate system as candidate growth points; considering the economic and power supply costs, the load density of each candidate growth point is used as the probability of site selection to determine the location of the new node, where Pa and Pc represent the loads of nodes a and c adjacent to node d, respectively, and P is the sum of the loads of all nodes; The newly added load node is connected to an existing node with a certain probability through a specific preferential principle; the probability of the newly added load node connecting to the original node is determined by the physical distance and the degree of the node; the newly added load node tends to connect to the node with a large node degree and a short distance from itself; the probability of the newly added load node connecting to node i is Π(k i )for: Where: ki is the degree of node i; It is the preferred factor for connecting new nodes to existing nodes; the specific expression is as follows: Where: li is the physical distance from node i to the new node; exponents α and β can be used to adjust the influence of node degree and physical distance; According to the load density ζ, the new distributed generation location is selected and connected to the nearest node, completing the distributed power generation node growth process and edge growth process; unlike the load point, the distributed power generation location can be selected not only at the growth point location, but also at the original load node location; Update the distribution network topology model, power flow model and reliability calculation model. The distribution network topology model updates node location information and connection information. The power flow model updates node load information and connection line information. The reliability calculation model updates node load information, connection line information and system fault information. Check in sequence whether the load growth reaches the limit of the region, calculate the power flow to check whether the voltage and current exceed the limit, whether the reliability parameters do not meet the requirements, and whether the evolution time reaches the set number of cycles. If any of the above is found, the evolution simulation ends. Repeat the above steps and finally reach the evolution target time to end the evolution.

5. A method for the structural evolution of a medium voltage distribution network considering the growth of distributed power sources and loads according to claim 1, characterized in that In step 3, the specific operation method of predicting the key time nodes of the evolution stage and forming the initial population is as follows: Step 3.1, the key nodes in the predicted evolution stage are the key time nodes predicted by summarizing the evolution results of multiple distribution network natural evolution models; the average value of the evolution stop time caused by voltage over-limit in the distribution network natural evolution model is determined as the key node of the distribution network evolution; Step 3.2, based on the Monte Carlo method, set the key nodes of the distribution network evolution as the evolution target years to simulate the natural evolution of the distribution network, repeat N times, calculate the values ​​of each indicator in the indicator system and sort the indicator calculation results from high to low, and select the evolution results of the top N / 2 indicators as the initial population for genetic algorithm optimization; Improve the power flow distribution entropy H by using the reliability index and state index n The voltage Theil entropy T iu , Economic indicators Network loss rate C loss and construction cost C price As an indicator system, the indicators are calculated and weighted; the reliability indicators include the system average power outage frequency coefficient KSAIFI, the system average power outage time coefficient KSAIDI, the expected value of power shortage KEENS and the power supply reliability KASAI.

6. The method for structural evolution of a medium voltage distribution network considering growth of distributed power sources and loads according to claim 1, characterized in that: The step 3 obtains multiple evolution results based on the distribution network natural evolution model based on the Monte Carlo method, which means that the distribution network natural evolution model obtains the results by multiple simulations according to the key nodes of distribution network evolution as the evolution target years; The forming of the initial population in step 3 refers to simulating the natural evolution of the distribution network based on the Monte Carlo method. After N repetitions, the values ​​of each indicator in the indicator system are calculated and the indicator calculation results are sorted from high to low. The evolution results of the top N / 2 indicators are selected as the initial population for genetic algorithm optimization. The index system includes reliability index, status index and economic index. The reliability index specifically includes the system average power outage frequency coefficient K SAIFI , System average power outage time coefficient K SAIDI 、Expected value of insufficient power K EENS and power supply reliability K ASAI , the state indicators include improved power flow distribution entropy H n The voltage Theil entropy T iu , economic indicators include network loss rate C loss and construction cost C price ; The reliability index calculation formula is: K SAIFI =∑λ i N i / ∑N i K SAIDI =∑(N i U i ) / U i K ASAI =(8760∑N i -∑N i IN i ) / (8760∑N i ) In the formula, λ i is the failure rate at load node i, N i is the number of users at load point i, U i is the average annual outage time of the load point, X is the outage capacity, C i is the system spare capacity in the i-th time period, P ijk (X) is the probability that the outage capacity is greater than or equal to X at the kth hour on the jth day of the i-th time period; m is the number of time periods in a year; n i is the number of days in the i-th time period; L ijk is the hourly load at the kth hour on the jth day in the i-th time period; Assuming that load node i stops operating due to a fault, the real power flow of line l is P l 0 (i) Define the load rate of line l as: Where P l max (i) is the maximum power flow limit of the line; Divide the line load rate into n reasonable sections and set the constant sequence X = [0, e, 2e, …, ne], satisfying ne = 1; place each line in the appropriate load rate interval [ke, (k+1)e), and place the overloaded line in the interval [1, ∞), then the improved power flow distribution entropy H n The indicator expression is: In the formula, is the probability that the line is in the interval [ke, (k+1)e); l k is the number of lines whose load factor is in the interval [ke,(k+1)e); is the mean load factor of the line in the section [ke, (k+1)e); δ is the overload coefficient. When the line exceeds the limit, δ>1, otherwise δ=1; β kl (i) is the load rate of line l in the interval [ke, (k+1)e) after node i is out of service; Suppose that a certain type of object h in a system is partitioned into M zones with a total of N objects. It is defined that a certain link z in the system changes, and a certain indicator y of the object h in the system changes accordingly. Theil entropy T in zone g is g for: In the formula, h is the classification object in the system; N is the total number of classification objects; N g is the total number of objects in region g; Y h(h∈g) Indicates the change of a certain indicator of object h in region g; Y g It represents the sum of the changes in the y index of all objects h in the region g; Considering global changes, the improved Theil entropy index is: Where, T zy Theil entropy is the change in index y caused by the change in system index z; T l is the unevenness of the change in y in each region; T b is the unevenness of the variation among regions; z It is the sum of the changes in index y of all h-type objects in the system caused by link z; Based on the definition of Theil entropy, the power network with N nodes is partitioned according to the node type, with PQ nodes as zone 1 and PV nodes and balance nodes as zone 2. When the unit load is increased at node i, the voltage change of node j is obtained, and then the node voltage change of the global network is obtained as follows: Where U ij is the voltage change caused by node j; U' j is the initial voltage value of node j; Voltage Theil entropy T iu for: The network loss and network loss rate are expressed as: Where P iiloss is the power loss on the ii-th branch. The total network loss P of the system is obtained by summing up the losses of all branches. loss ;P load is the total power supply load in the system; Construction cost price It is expressed as: C price =C line ×l+C euip ×nn-I Where C line is the line cost per unit length, l is the line length, C euip is the equipment cost, nn is the number of equipment, and I is the benefit of policy incentives; The indicators are subjectively weighted by the hierarchical analysis method and objectively weighted by the entropy weight method. Finally, the comprehensive weights w1 to w8 are obtained, and the evolution results of each time are evaluated by the following formula; f=-w1K SAIFI -w2K SAIDI -w3K EENS +w4K ASAI -w5C loss -w6C price -w7H n -w8T iu In the formula, w1 to w8 are the weights of each indicator, K SAIFI , K SAIDI , K EENS and K ASAI is the reliability index, H n To improve the power flow distribution entropy, T iu is the voltage Theil entropy, C loss is the network loss rate, C price For construction costs.

7. The method for structural evolution of a medium voltage distribution network considering growth of distributed power sources and loads according to claim 1, characterized in that: In the step 4, a natural evolution optimization model of the distribution network is established, and the distribution network frame of the target year is optimized based on the natural evolution model of the distribution network. The maximum index evaluation score is taken as the optimization target, the load point and the distributed power source location and the safe operation of the distribution network are taken as constraints, and the load point and distributed power source site selection and access are taken as decision variables. The genetic algorithm is used for solving. The specific operation method is as follows: Step 4.1: Based on the evolution result evaluation formula, introduce the indicator node compliance C pos Compliance with line C con To measure the extent to which the results of the natural evolution model of the distribution network conform to the two rules that the newly added load points have a greater probability of appearing at locations with greater load density and have a greater probability of connecting to nodes with a closer physical distance and a larger node degree, the objective function is constructed in combination with the indicator system of step 3: max f′=-w1′K SAIFI -w′2K SAIDI -w3′K EENS +w′4K ASAI -w5′C loss -w′6C price -w′7H n -w8′T iu +w′9C pos +w′ 10 C con In the formula, w1' to w 10 ' is the weight of each indicator, K SAIFI , K SAIDI , K EENS and K ASAI is the reliability index, H n To improve the power flow distribution entropy, T iu is the voltage Theil entropy, C loss is the network loss rate, C price For construction costs; Step 4.2, taking power flow constraints, node voltage constraints, and line thermal limit constraints as constraints of the optimization model; Step 4.3, the location information and connection information of the newly added load points each year and the location information of the distributed power sources are used as decision variables. The location information of the newly added load points is represented by the existing node number to which the growth point belongs and its position information relative to the existing node. The relative position is represented by the upper left, upper right, left, lower right, lower right and lower right of the node in the grid diagram, which are represented by numbers 1 to 8 respectively. The connection information refers to the number of the connected node. The location information of the newly added distributed power sources is also determined by the existing node number and the relative position number. The relative position number is 0 to 8, where the number 0 represents that the distributed power source is located above the load point, and the meaning of the numbers 1 to 8 is the same as the meaning of the relative position information of the newly added load points; Step 4.4, use genetic algorithm for optimization and solution: encode the initial population generated in step 3 according to the method of step 4.3 to obtain the initial population of the genetic algorithm; the selection operation in the genetic algorithm adopts the non-replay random remainder selection method, the crossover operation selects two-point crossover, the mutation operation adopts global mutation, and the fitness function is the objective function constructed in step 4.1; by repeatedly implementing selection, crossover, and mutation operations, the fitness value of each individual is calculated, the offspring with good adaptability are retained, and the offspring that do not meet the conditions are eliminated. After several generations of inheritance, it converges to the chromosome with the best conditions, and then the decoding operation corresponding to the encoding method is used to obtain the final optimization result.

8. A method for the structural evolution of a medium voltage distribution network considering the growth of distributed power sources and loads according to claim 7, characterized in that: Power flow constraints: Where P i With Q i are the active and reactive power injected at node i, U i , U j are the voltage amplitudes of nodes i and j respectively, G ij With B ij is the real and imaginary part of the corresponding element in the system node admittance matrix; θ ij is the phase angle difference between the voltages at nodes i and j when the node voltages are expressed in polar coordinates; Node voltage constraints: IN min ≤U i ≤U max Where U i is the voltage at node i, U max and U min They are the upper and lower voltage limits specified in the operation of the power distribution system; Circuit thermal limit constraints: I i %≤I max % In the formula, I i % is the load factor of the line, then I max % is the thermal stability limit of the line.

Images