A charging station planning method considering user charging experience and distribution network operation risks

Abstract

The present invention provides a charging station planning method considering user charging experience and distribution network operation risk, comprising the following steps: step 1, establishing an electric vehicle load model; step 2, establishing a charging experience index, establishing an upper comprehensive target with the charging experience index, the charging station construction cost and the user convenience as the charging station site selection and capacity optimization target, and proposing an upper planning constraint condition; step 3, proposing a meta-point weighted Voronoi diagram taking into account the spatial weight to partition the service range; step 4, establishing an upper planning process based on the upper optimization target by adopting a genetic taboo hybrid algorithm; step 5, establishing a lower comprehensive target with the new feeder cost and the distribution network operation risk as the charging station load access node optimization target; step 6, establishing a lower planning process based on the lower comprehensive target by adopting a cyclic replacement method; the present invention can plan charging stations by comprehensively considering the user charging experience, the power grid operation status and the charging station construction cost.

Description

Technical Field

[0001] The present application relates to the field of charging station planning, and in particular to a charging station planning method that takes into account user charging experience and distribution network operation risks. Background Art

[0002] With the rapid development of electric vehicles, the demand for electric vehicle charging will increase rapidly. Building charging stations that meet their charging needs is the basic guarantee for the rapid popularization of electric vehicles.

[0003] During the implementation process, the inventors found that there are at least the following problems in the planning of traditional electric vehicle charging stations:

[0004] In the Internet era, people pay more and more attention to the satisfaction of life experience; the planning and construction of charging stations are closely related to the charging experience of electric vehicle users; current studies have considered the user charging experience from the aspects of the time spent on the charging road, the waiting time before charging, and the electric energy consumed by users when driving to the charging station; in addition, the impact of charging station planning on the user charging experience is also reflected in whether users need to queue before charging, whether they can reach the charging station before the battery is deeply discharged, etc.; on the other hand, people in different functional areas have different requirements for the time spent on the charging process due to different time urgency, and the density of electric vehicles is also different, so the above factors should be fully considered when evaluating the user charging experience;

[0005] In terms of the division of charging station service range, many studies have adopted weighted Voronoi diagrams. When using weighted Voronoi diagrams, only the weight of charging stations is considered, without considering the differences in demand for charging stations in different functional areas. On the other hand, building charging stations with larger capacity in places with low land prices will reduce construction costs, so the weight of charging stations should take land prices into consideration. Furthermore, previous studies have mostly used straight-line distances for zoning, but since most roads are vertical or parallel to each other, using right-angle distances can improve the accuracy of zoning.

[0006] In order to ensure the planning of charging stations under the premise of safe operation of the power grid, most studies take the operating state parameters of the power grid (such as voltage, current and load power, etc.) as constraints; however, when the scale of electric vehicles is large, the problem of over-limit operation is difficult to avoid; at the same time, due to the randomness of electric vehicle charging, the uncertainty of the operating state of the distribution network increases. As a basic characteristic of risk, this increases the possibility of serious over-limit of the distribution network; in addition, current research focuses on the site selection and capacity optimization of charging facilities, and rarely pays attention to the optimization problem of charging station load access nodes, simply assuming that the charging station load is connected to the corresponding load node nearby; but in fact, due to the different vulnerabilities of the nodes, the operating risks brought by the charging station load accessing different nodes are different; therefore, it is of great engineering significance to reduce various operating risks by optimizing the charging load access nodes. Summary of the invention

[0007] Based on this, the present invention proposes to partition the service range by using a meta-point weighted Voronoi diagram taking into account spatial weights, establishes a new indicator of charging experience index to measure the extent to which the user's charging experience requirements are met during the charging process, and optimizes the site selection and sizing of charging stations based on the construction cost of the charging station, user convenience and the charging experience index; optimizes the load access nodes of charging stations based on the cost of new feeders and the operating risk of the distribution network; and on this basis, proposes an interactive planning method for charging stations that takes into account user charging experience and the operating risk of the distribution network.

[0008] The present invention is achieved by at least one of the following technical solutions.

[0009] The charging station planning method considering the user charging experience and the distribution network operation risk includes the following steps:

[0010] Step 1: Establish an electric vehicle load model based on the electric vehicle historical data;

[0011] Step 2: Establish a charging experience index, and use the charging experience index, charging station construction cost, and user convenience as the upper-level comprehensive goal for the charging station site selection and capacity optimization goals, and propose upper-level planning constraints;

[0012] Step 3: Propose a meta-point weighted Voronoi diagram taking into account spatial weights to partition the service area;

[0013] Step 4: Based on the upper-level optimization goal, the upper-level planning process is established by using the genetic taboo hybrid algorithm;

[0014] Step 5: Establish the lower-level comprehensive target based on the cost of new feeder and the operation risk of distribution network as the optimization target of charging station load access node, and propose the lower-level constraint conditions;

[0015] Step 6: Establish the lower-level planning process using the cyclic replacement method based on the lower-level comprehensive objectives.

[0016] Furthermore, the step 1 of establishing the electric vehicle load model according to the electric vehicle historical data includes the following steps:

[0017] Step 1-1, using electric vehicle capacity to describe the scale of electric vehicles in the region, where the electric vehicle capacity is the sum of the rated charging power of all electric vehicles in the region, including all electric vehicles in a charging state and a non-charging state; the electric vehicle capacity per unit area is defined as the electric vehicle capacity density;

[0018] Step 1-2: According to the actual charging process of electric vehicle batteries, a two-stage variable power charging model of electric vehicles with constant current and constant voltage is constructed:

[0019] During the charging process, electric vehicles use a constant current-constant voltage two-stage variable power charging mode. When the battery terminal voltage is less than the maximum battery terminal voltage U bmax The battery is charged in constant current mode; when the battery terminal voltage reaches the maximum battery terminal voltage U bmax When the battery enters the constant voltage stage, the current rapidly decays to zero, and the charging power is zero; the equation describing its charging process is:

[0020]

[0021] Where U b (t) is the voltage across the battery at time t, i c is the current during charging, C0 is the nominal capacity of the battery, R is the internal resistance of the battery, K0 is the polarization constant of the battery, E0 is the constant potential of the battery, A0 and B0 are the product coefficient and exponential coefficient of the battery charging exponential segment respectively, SOC(t0) represents the remaining capacity of the battery at the beginning of charging, h(t) is the rechargeable capacity of the battery at time t, that is, the discharged capacity;

[0022] During constant current charging, the current i is a constant, and the charging power of the electric vehicle is expressed as:

[0023] P(t)=U b (t)I n (2)

[0024] In the formula, I n is the charging current in constant current charging mode, P(t) is the charging power at time t;

[0025] Step 1-3: Based on the historical travel data of electric vehicles, construct a time series probability model of electric vehicle charging load:

[0026] The charging load simulation of electric vehicles must select the probability model of each sampling link according to the type of electric vehicle, extract its driving distance, arrival time, initial state of charge and departure time, so as to determine the behavior state of the electric vehicle at each moment, and then calculate the SOC value of the electric vehicle at the corresponding moment to obtain its SOC curve for the day. The calculation formula is:

[0027] (1) When charging

[0028]

[0029] Where SOC(t) is the initial state of charge of the electric vehicle;

[0030] (2) While driving

[0031] SOC(t+Δt)=SOC(t)-d r / Dr ·Δt / T r (4)

[0032] Where, d r is the driving distance of the electric vehicle, D r is the maximum range of an electric vehicle, T r is the driving time of the electric vehicle, Δt is the calculation time interval;

[0033] Then, the charging power for one day is calculated with the help of formula (1) and formula (2); the loads of various electric vehicles are then added one by one to obtain the total charging load of electric vehicles; the calculation formula for the active power of the total charging load of electric vehicles at time t is:

[0034]

[0035] Where: N1, N2 and N3 represent the number of electric private cars, buses and taxis in the charging state in the area at time t; P 1nt , P 2nt and P 3nt They represent the real-time charging power of the nth electric private car, bus and taxi in the charging state at time t respectively;

[0036] Steps 1-4: Establish the maximum simultaneous charging rate:

[0037] In order to clearly understand the maximum proportion of electric vehicles charged simultaneously in a certain area in a day, the maximum simultaneous charging rate of electric vehicles is defined as the ratio of the maximum load of electric vehicles in a day to the capacity of electric vehicles, that is:

[0038]

[0039] Where P t is the total charging load of electric vehicles in the region at time t, K z is the electric vehicle capacity in the area.

[0040] Furthermore, the establishment of the charging experience index in step 2, establishing the upper comprehensive goal with the charging experience index, the construction cost of the charging station and the convenience of the user as the optimization goal of the charging station site selection and capacity determination, and proposing the upper planning constraint conditions include the following steps:

[0041] Step 2-1: Establish a charging experience index based on the charging journey time, SOC value before charging, and whether queuing is required:

[0042] The charging experience index is defined as the degree to which the user's charging experience requirements are met during the charging process, and is reflected by the satisfaction of the charging journey time, pre-charging SOC value, and whether queuing is required.

[0043] (1) Charging distance time satisfaction rate

[0044] People in different regions have different requirements for the time spent on charging due to different levels of time urgency, that is, different charging journey time requirements; the charging journey time limit is defined as the user's acceptable limit for the time spent on charging; when the time spent on charging is greater than the charging journey time limit, the charging experience requirement is not met; based on this, in order to examine the satisfaction degree of the charging journey time requirements of all users, the charging journey time satisfaction rate can be defined as:

[0045]

[0046] Where η t_sat is the charging distance time satisfaction rate; d (i) is the electric vehicle capacity density at the i-th point; v a is the average speed of the electric vehicle during the charging process; T d (i) is the charging time limit of the i-th point, that is, the charging time limit of the area to which it belongs; M t_sat is the set of electric vehicle capacity density points that meet the charging journey time requirements; N0 is the number of points; L d (i) is the right angle distance from the i-th point in the planning area to the corresponding charging station;

[0047] (2) SOC value satisfaction rate before charging

[0048] The SOC value of an electric vehicle before charging satisfies a certain probability distribution. When the SOC value is lower than a certain value, the life of the electric vehicle battery will be damaged. Therefore, the pre-charging SOC value satisfaction rate can be defined as the probability that the SOC value meets the minimum SOC value that does not damage the battery life of the electric vehicle when the user arrives at the charging station. M SOC values ​​are obtained by M Monte Carlo samplings, and the algorithm for the pre-charging SOC value satisfaction rate is:

[0049]

[0050] Where η s_sat is the SOC value satisfaction rate before charging; p d (i) is the probability that the i-th point meets the SOC value requirement before charging; ε d (i) is the electric vehicle capacity density at the i-th point; S range is the maximum cruising range of the corresponding electric vehicle; SOC(j) is the jth SOC value of Monte Carlo sampling; Ω d (i) is the set of SOC values ​​that meet the pre-charging SOC value requirements among the M SOC values ​​at the i-th point; card() is the number of elements in the set, SOC minis the lowest SOC value that does not damage the battery life; M is the number of Monte Carlo sampling;

[0051] (3) No-queuing satisfaction rate

[0052] If the demand for electric vehicle charging is greater than the configured capacity of the charging station, users need to queue before charging, which does not meet the no-queue requirement. The no-queue satisfaction rate is defined as the probability that users do not need to queue. The charging station load data of each charging station at 24 times a day is obtained through Monte Carlo sampling M times, and the algorithm for the no-queue satisfaction rate is:

[0053]

[0054] Where η p_sat is the satisfaction rate without queuing; p CD (i) is the average probability that charging station i meets the requirement of no queuing; П(j) is the set of times when the charging station load meets the requirement of no queuing in the jth Monte Carlo sampling of charging station i; J(i) is the number of chargers configured at charging station i; P CD (j,k) is the load of charging station i at the kth moment of the jth Monte Carlo sampling; P c is the rated power of a single charger; α c For charging efficiency;

[0055] (4) Calculation of user charging experience index

[0056] The charging experience index is reflected by comprehensively considering the charging distance time satisfaction rate, the pre-charging SOC value satisfaction rate and the no-queue satisfaction rate. The algorithm is:

[0057] η g_sat =η t_sat η s_sat η p_sat (10)

[0058] Where η g_sat It is the charging experience index;

[0059] Step 2-2: Establish the cost of charging station construction:

[0060] The construction cost of a charging station includes the construction cost of transformers, chargers and other equipment as well as land costs; therefore, the construction cost of charging station i is calculated as:

[0061] C z_t (i) = C s (i)+C d (i) (11)

[0062] In the formula, C s(i) is the construction cost of the transformer, charger and other equipment of the i-th charging station, which is determined by the number of chargers in the charging station; C d (i) is the land cost of the i-th charging station, and its calculation method is:

[0063] C d (i) = S d (i)D d (12)

[0064] In the formula, S d (i) is the land area occupied by the i-th charging station, which is determined by the number of chargers in the charging station, D d is the land cost per unit area;

[0065] The total cost of building charging stations in the planning area is:

[0066] C g_t =∑C z_t (i) (13)

[0067] Step 2-3: Establish user convenience indicators:

[0068] User convenience is expressed by the average charging distance. Points are averaged in the planning area. The electric vehicle capacity density of each point is the electric vehicle capacity density of the area to which it belongs. The charging distance is weighted averaged with the electric vehicle capacity density of each point as the weight to obtain the average charging distance. The algorithm for the average charging distance in the planning area is:

[0069]

[0070] In the formula, ε d (i) and L d (i) are the electric vehicle capacity density of the i-th point in the planning area and its right-angle distance to the corresponding charging station;

[0071] Step 2-4: Establish upper-level comprehensive goals:

[0072] The upper-level planning of charging stations takes the construction cost of charging stations, user convenience and charging experience index as the goals, and optimizes the location and capacity of charging stations; therefore, the algorithm for the upper-level comprehensive goal is:

[0073]

[0074] In the formula, α1, α2, and α3 are weight coefficients, which are used to make the order of magnitude of each indicator the same and to eliminate the units of each indicator, so that the upper-level comprehensive target is dimensionless. B1 is the upper-level comprehensive target;

[0075] Step 2-5: Establish upper-level planning constraints:

[0076] (1) Distance constraints between adjacent stations

[0077] To ensure the charging needs of users, the service range of the charging station should not be too large. At the same time, in order to avoid wasting resources, the charging stations should not be too concentrated. Therefore, the right-angle distance L between adjacent charging stations is cc To satisfy the constraints:

[0078] L min <L cc <L max (16)

[0079] Where, L min and L max are the upper and lower limits of the distance constraints between adjacent stations respectively;

[0080] (2) Constraints on the number of chargers

[0081] In order to meet the charging needs of users and make full use of social resources to ensure a high utilization rate of chargers, the number of chargers configured in the charging station should meet the charger number constraint:

[0082] J min ≤J CD ≤J max (17)

[0083] In the formula, J min and J max are the minimum and maximum number of charger configurations, respectively, CD is the number of chargers configured in the charging station, and the algorithm is:

[0084]

[0085] In the formula, Z CD is the maximum charging demand within the service range of the charging station; K CD The capacity of electric vehicles within the service area of ​​the charging station.

[0086] Furthermore, step 3 proposes that the service range partitioning using the element-point weighted Voronoi diagram taking into account the spatial weight includes the following steps:

[0087] Step 3-1, propose the element-point weighted Voronoi diagram:

[0088] In view of the defect that the traditional weighted Voronoi diagram cannot reflect the difference in weights of different regions, the present invention proposes a meta-point weighted Voronoi diagram to divide the service range, and describes the size of the weights of different regions with spatial weights; at this time, the speed at which the generator generates the Voronoi diagram is not only related to the weight of the generator, but also to the spatial weight of each region; since the spatial weight is taken into account, the weighted distance between two points on different driving paths may be different, so the present invention defines the directed path from point A to point B as a path where A first drives longitudinally to B and then drives transversely, and defines the weighted directed distance as the weighted distance of the directed path;

[0089] Based on this, the meta-point weighted Voronoi diagram is defined as:

[0090] V(R i )={P∈V(R i )|D(R i ,P)≤D(R j ,P),j=1,2,…,m1;j≠i} (19)

[0091] In the formula, D(R i , P) is point R i The weighted directed distance to point P, m1 is the number of generators, D(R j ,P) is point R j The weighted directed distance to point P, if R i The weight is v i , R i The directed path to point P passes through regions Q1, Q2, ..., Q n , n is R i The number of regions that the directed path to point P passes through, and the corresponding spatial weights are θ1, θ2, …, θ n , R i The directed path to point P passes through the boundary points P1, P2, ..., P n-1 , d(P n-1 ,P) is point P n-1 The directed distance to point P, then R i The weighted directed distance to point P is:

[0092]

[0093] Step 3-2: Establish charging station weights:

[0094] Building charging stations with a larger service range and larger capacity in places with low land prices can reduce the total cost of charging station construction. Therefore, when dividing the service range, the square of the charging station weight can be set to be inversely proportional to the inverse of the unit price of the land it occupies. The weight of charging station i is:

[0095]

[0096] Where D i is the unit price of land occupied by charging station i;

[0097] Step 3-3, establish spatial weights:

[0098] The present invention establishes the spatial weight of the region by comprehensively considering the charging journey time requirements and electric vehicle capacity density of different types of regions. The algorithm for the spatial weight of region type i is:

[0099]

[0100] In the formula, θ q (i), ε q (i) T q (i) are the spatial weight, electric vehicle capacity density and charging time limit of regional type i, β1 and β2 are weight coefficients;

[0101] Step 3-4: Division of charging station service area

[0102] For the convenience of calculation, when using the element-point weighted Voronoi diagram for charging station planning, points are generated uniformly in the planning area, and the moving step length of adjacent points is Δx = Δy = Δl. The coordinates of each point, electric vehicle capacity density, land unit price, charging journey time limit and other information are recorded in a matrix to establish a point database; when drawing the Voronoi diagram, each point is represented by a square, the side length of the square is the moving step length, and the point is located at the center of the square; the electric vehicle capacity of the square is the product of the area of ​​the square and the electric vehicle capacity density of the point;

[0103] If the weight is v i Charging station R i The directed path to point O passes through O1, O2, ..., O n-1 , the corresponding spatial weight of each point is θ q1 ,θ q2 ,…,θ qn , then R i The weighted directed distance to point O is:

[0104]

[0105] Compare the weighted directed distances from each generator to a point, and the generator corresponding to the smallest weighted directed distance will include the point in its service range;

[0106] The electric vehicle capacity of a charging station is the sum of the electric vehicle capacities of all points within its service range. The algorithm is:

[0107]

[0108] In the formula, ε d (i) is the electric vehicle capacity density at the i-th point within the service range of the charging station, and ΔS is the area of ​​a single square.

[0109] Furthermore, the upper-level planning process of using the genetic taboo hybrid algorithm based on the upper-level optimization target described in step 4 includes the following contents:

[0110] The present invention adopts the Genetic Taboo Hybrid Algorithm (GATS) for upper-level planning of charging stations; GATS combines the characteristics of GA (genetic algorithm) with multiple starting points and TS (tabu search algorithm) with memory function and strong climbing ability, introduces the unique memory function of TS into the GA evolutionary search process, and constructs a new recombination operator (TSR); TS is used to improve the climbing ability of GA, that is, TS is used as the mutation operator (TSM) of GA; the main steps of upper-level planning of charging stations are as follows:

[0111] Ⅰ. Collect point database information as basic data for optimization;

[0112] Ⅱ. Using the electric vehicle capacity density information of the point database, calculate the electric vehicle capacity in the planning area, that is, K z , and then calculate the maximum charging demand Z in the planning area z , the algorithm is:

[0113]

[0114] III. Based on the maximum charging demand Z in the planning area z , with the minimum number of chargers configured in the charging station J min and the maximum number of chargers J max As a constraint, calculate the minimum number of charging stations N min and the maximum number of seats N max ,Right now:

[0115]

[0116] IV. For different charging station number schemes, GATS and element-node weighted Voronoi diagram are used to optimize the location and capacity based on the upper comprehensive objectives, including the following steps:

[0117] (1) Let the number of charging stations be g and set the maximum number of iterations of GA to N. gen , length of taboo table T length1 , group size N pop , recombination probability p ch and mutation probability p m ; The length of the taboo table in TSM is T length2 、Number of neighborhood solutions h and the maximum number of iterations N tAnd other parameters;

[0118] (2)t r represents the number of iterations, let t r = 0, divide the planned area into g areas equally, randomly select an address from each area, and obtain the chromosome address sequence [x1, y1; ...; x i ,y i ;…;x g ,y g ],(x i ,y i ) is the i-th address in the address sequence of the chromosome, and the addresses do not satisfy the inter-station distance constraint for regeneration;

[0119] (3) Calculate the upper-level comprehensive goals of the contemporary group;

[0120] (4) Select N by scrolling pop chromosomes;

[0121] (5) Generate a random number r between 0 and 1 i , i=1,2,…,N pop , if r i <p ch , then the i-th chromosome is the parent; perform a crossover operation on each pair of chromosomes, randomly swap two genes, and determine whether the offspring meets the inter-station distance constraint. If not, select the optimal parent to move to the next generation; if it does, compare the upper comprehensive goal of the offspring with the desired level. If it is better, break the taboo and move to the next generation; if it is worse but not taboo, accept the offspring. If it is taboo, select the optimal parent to move to the next generation;

[0122] (6) Generate a random number q between 0 and 1 i , i=1,2,…,N pop If q i <p m , then the mutation operation is performed on the i-th chromosome:

[0123] a. Take the current chromosome as the optimal chromosome, take its upper comprehensive target as the optimal upper comprehensive target, and set the taboo table of TSM to empty;

[0124] b. Generate neighboring address genes (x+Δx, y), (x-Δx, y), (x, y+Δy) and (x, y-Δy) with the current address gene as the center, add the current address gene to form the allele sequence, perform full permutation to obtain 5^g chromosomes and randomly extract N from them h indivual;

[0125] c. This N hThe chromosomes that do not meet the inter-station distance constraint are removed; the remaining chromosomes that are not in the taboo table are candidate chromosomes, and the upper-level comprehensive goal that meets the charger number constraint after partitioning is calculated;

[0126] d. Select the best candidate chromosome based on the minimum comprehensive objective of the above layer, and use it to replace the current chromosome and the earliest unsatisfied chromosome;

[0127] e. Determine whether the upper comprehensive target of the optimal candidate chromosome is less than the optimal upper comprehensive target. If yes, replace the optimal chromosome with it, and replace the optimal upper comprehensive target with its upper comprehensive target, and then go to step f; otherwise, go directly to step f;

[0128] f. Determine whether the maximum number of iterations N has been reached t , or the optimal upper comprehensive target has remained unchanged for 30 consecutive times? If yes, output the optimal chromosome; otherwise, return to step b;

[0129] (7)t r =t r +1, if t r <N gen , go to step (3); otherwise, output the optimal solution and terminate the algorithm.

[0130] Furthermore, the establishment of a lower-level comprehensive target based on the cost of the newly built feeder and the operating risk of the distribution network as the optimization target of the charging station load access node in step 5, and the proposal of the lower-level constraint conditions include the following steps:

[0131] Step 5-1: Calculate the cost of new feeder:

[0132] The charging station load needs to be connected to the distribution network by building a new feeder to connect to the node of the distribution network; assuming that the new feeder is laid along a straight line, the cost algorithm of the new feeder of the charging station is:

[0133] C k =D k ∑L k (i) (27)

[0134] Where D k is the cost per unit length of the feeder; L k (i) is the length of the newly built feeder of the i-th charging station, that is, the right-angle distance from the i-th charging station to the access node;

[0135] Step 5-2: Calculate the distribution network operation risk:

[0136] (1) Component risk indicators

[0137] Based on the weighted entropy principle, the weighted voltage distribution entropy is used to characterize the voltage risk caused by voltage over-limit and voltage distribution uncertainty, and the weighted power flow distribution entropy is used to characterize the power flow risk caused by power flow over-limit and power flow distribution uncertainty. The algorithm is:

[0138]

[0139] In the formula, H v_i (t) and H l_i (t) are the weighted voltage distribution entropy of node i and the weighted power flow distribution entropy of branch i at time t; n v_i (t) and n l_i (t) are the voltage state number of node i and the power flow state number of branch i at time t; S v_j (t) and P v_j (t) are the voltage loss severity and probability of the j-th voltage state of node i at time t; S l_j (t) and P l_j (t) are the severity and probability of power loss of the jth power state of branch i at time t;

[0140] The algorithm for loss severity S is:

[0141]

[0142] Where, μ is the loss severity amplification factor to adjust the sensitivity of the loss severity; w is the loss amount;

[0143] (2) Network risk indicators

[0144] The present invention comprehensively evaluates the operation risk of the distribution network containing large-scale electric vehicles by combining voltage risk and power flow risk. In order to examine the change of the distribution network operation risk in one day, the network operation risk at time t can be defined as:

[0145] H w (t) = γ1H v (t)+γ2H l (t) (30)

[0146] In the formula, H w (t) is the network operation risk at time t; γ1 and γ2 are weight coefficients; H v (t) and H l (t) are the network voltage risk and network flow risk at time t, respectively. The algorithm is:

[0147]

[0148] In the formula, z1 and z2 are the number of nodes and branches of the distribution network respectively;

[0149] In order to study the most serious risk faced by the safe operation of the distribution network, the maximum network operation risk is defined as the maximum value of the network operation risk in one day, that is:

[0150] H wmax =max(H w (t)) t=1,2,…,24 (32)

[0151] Step 5-3: Propose the lower-level comprehensive goals:

[0152] The lower-level planning of charging stations comprehensively considers the cost of new feeder lines and the operating risks of the distribution network, and optimizes the load access nodes of charging stations. The lower-level comprehensive objectives are defined as:

[0153] B2=ρ1C k +ρ2H wmax (33)

[0154] In the formula, C k and H wmax are the cost of new feeder and the maximum operation risk of the network respectively; ρ1 and ρ2 are weight coefficients;

[0155] Step 5-4: Propose lower-level planning constraints:

[0156] (1) Feeder length constraints

[0157] When the newly built feeder is too long, not only the feeder cost is too high, but also the feeder impedance is too large, which will increase power loss and voltage loss. Therefore, the newly built feeder should meet the feeder length constraint:

[0158] L k (i)≤L kmax (34)

[0159] Where, L kmax The length limit of the newly built feeder;

[0160] (2) Charging station load constraints

[0161] Since the charging load of a charging station cannot be greater than its installed capacity, the following constraints must be met:

[0162] P CD (i)≤α c J(i)P c (35)

[0163] Where P CD (i) is the charging load of charging station i.

[0164] Furthermore, the step 6 of establishing the lower-level planning process based on the lower-level comprehensive target by using the cyclic replacement method includes the following contents:

[0165] (1) Input the address of the charging station, the capacity of electric vehicles in each zone, the parameters of the charging process, and the data of the distribution network parameters;

[0166] (2) Calculate the N m next-day charging load data for each zone of the charging station through Monte Carlo sampling;

[0167] (3) Connect the load of the charging station to the nearest node of the distribution network as the current plan, and calculate the lower-level comprehensive objective;

[0168] (4) Denote the number of charging stations as g, determine the optimization order of each zone, b m = 0;

[0169] (5) b m = b m + 1, calculate the distances from charging station b m to each node, and take the nodes that meet the constraints as candidate nodes;

[0170] (6) Keep other access nodes unchanged, connect the loads of charging station b m to each candidate node respectively, calculate the lower-level comprehensive objective, and select the optimal candidate plan with the minimum lower-level comprehensive objective;

[0171] (7) Judge the advantages and disadvantages of the optimal candidate plan and the current plan. If the optimal candidate plan is better than the current plan, replace the current plan with the optimal candidate plan; otherwise, keep the current plan unchanged;

[0172] (8) Judge whether there is b m < g. If so, return to step (5); otherwise, go to step (9);

[0173] (9) Continuously loop and optimize until the charging station load access plan no longer changes.

[0174] Compared with the prior art, the charging station interactive planning method of the present invention taking into account the user charging experience and the distribution network operation risk has the following significant effects: the element-point weighted Voronoi diagram taking into account the spatial weight proposed in the present invention can comprehensively consider the occupied land price, the charging journey time requirement and the electric vehicle capacity density, so that the charging station can try to choose a low-price and low-cost location as the site, and at the same time can reflect the difference in demand for charging stations in different regions, so that the charging station is closer to commercial areas and office areas and other areas with higher charging requirements, and can better reflect the actual requirements of users for charging stations; considering the user's charging experience, the charging experience index is used as the optimization target for the site selection and capacity determination of the charging station, which effectively improves the user's charging experience, thereby better meeting the user's charging experience needs; taking the operation risk of the distribution network as the optimization target of the charging station load access node, the operation risk caused by the large-scale charging station load access to the distribution network is effectively reduced at a relatively small cost of newly built feeders, and the distribution network can be guaranteed to operate more safely and stably, which is of practical significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0175] Figure 1 is a flowchart of lower-layer planning based on a cyclic replacement method in an embodiment of the present invention;

[0176] Figure 2 It is a regional classification and distribution network diagram of the planning area in the embodiment of the present invention;

[0177] Figure 3 is a curve diagram of various types of conventional loads in an embodiment of the present invention;

[0178] Figure 4 This is a charging station service range zoning diagram of the optimal planning solution when the number of charging stations is 6 in an embodiment of the present invention. DETAILED DESCRIPTION

[0179] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below through embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0180] The embodiment of the present invention will provide a specific application example of a charging station interactive planning method that takes into account user charging experience and distribution network operation risks, including the following steps:

[0181] Step 1: Based on the historical data of electric vehicles, the establishment of electric vehicle load model includes the following steps:

[0182] Step 1-1, using electric vehicle capacity to describe the scale of electric vehicles in the region, where the electric vehicle capacity is the sum of the rated charging power of all electric vehicles in the region, including all electric vehicles in a charging state and a non-charging state; the electric vehicle capacity per unit area is defined as the electric vehicle capacity density;

[0183] Step 1-2: According to the actual charging process of electric vehicle batteries, a two-stage variable power charging model of electric vehicles with constant current and constant voltage is constructed:

[0184] During the charging process, electric vehicles use a constant current-constant voltage two-stage variable power charging mode. When the battery terminal voltage is less than the maximum battery terminal voltage U bmax The battery is charged in constant current mode; when the battery terminal voltage reaches the maximum battery terminal voltage U bmax When the battery enters the constant voltage stage, the current rapidly decays to zero, and the charging power is zero; the equation describing its charging process is:

[0185]

[0186] Where U b (t) is the voltage across the battery at time t, i c is the current during charging, C0 is the nominal capacity of the battery, R is the internal resistance of the battery, K0 is the polarization constant of the battery, E0 is the constant potential of the battery, A0 and B0 are the product coefficient and exponential coefficient of the battery charging exponential segment respectively, SOC(t0) represents the remaining capacity of the battery at the beginning of charging, h(t) is the rechargeable capacity of the battery at time t, that is, the discharged capacity;

[0187] During constant current charging, the current i is a constant, and the charging power of the electric vehicle is expressed as:

[0188] P(t)=U b (t)I n (2)

[0189] In the formula, I n is the charging current in constant current charging mode, P(t) is the charging power at time t;

[0190] Step 1-3: Based on the historical travel data of electric vehicles, construct a time series probability model of electric vehicle charging load:

[0191] The charging load simulation of electric vehicles must select the probability model of each sampling link according to the type of electric vehicle, extract its driving distance, arrival time, initial state of charge and departure time, so as to determine the behavior state of the electric vehicle at each moment, and then calculate the SOC value of the electric vehicle at the corresponding moment to obtain its SOC curve for the day. The calculation formula is:

[0192] (1) When charging

[0193]

[0194] Where SOC(t) is the initial state of charge of the electric vehicle;

[0195] (2) While driving

[0196] SOC(t+Δt)=SOC(t)-d r / D r ·Δt / T r (4)

[0197] Where, d r is the driving distance of the electric vehicle, D r is the maximum range of an electric vehicle, T r is the driving time of the electric vehicle, Δt is the time interval;

[0198] Then, the charging power for one day is calculated with the help of formula (1) and formula (2); the loads of various electric vehicles are then added one by one to obtain the total charging load of electric vehicles; the calculation formula for the active power of the total charging load of electric vehicles at time t is:

[0199]

[0200] Where: N1, N2 and N3 represent the number of electric private cars, buses and taxis in the charging state in the area at time t; P 1nt , P 2nt and P 3nt They represent the real-time charging power of the nth electric private car, bus and taxi in the charging state at time t respectively;

[0201] Steps 1-4: Establish the maximum simultaneous charging rate:

[0202] In order to clearly understand the maximum proportion of electric vehicles charged simultaneously in a certain area in a day, the maximum simultaneous charging rate of electric vehicles is defined as the ratio of the maximum load of electric vehicles in a day to the capacity of electric vehicles, that is:

[0203]

[0204] Where P t is the total charging load of electric vehicles in the region at time t, K z is the electric vehicle capacity in the area.

[0205] Step 2: Establish a charging experience index. Use the charging experience index, charging station construction cost, and user convenience as the charging station site selection and capacity optimization targets to establish the upper-level comprehensive target, and propose the upper-level planning constraints, including the following steps:

[0206] Step 2-1: Establish a charging experience index based on the charging journey time, SOC value before charging, and whether queuing is required:

[0207] The charging experience index is defined as the degree to which the user's charging experience requirements are met during the charging process, and is reflected by the satisfaction of the charging journey time, pre-charging SOC value, and whether queuing is required.

[0208] (1) Charging distance time satisfaction rate

[0209] People in different regions have different requirements for the time spent on charging due to different levels of time urgency, that is, different charging journey time requirements; the charging journey time limit is defined as the user's acceptable limit for the time spent on charging; when the time spent on charging is greater than the charging journey time limit, the charging experience requirement is not met; based on this, in order to examine the satisfaction degree of the charging journey time requirements of all users, the charging journey time satisfaction rate can be defined as:

[0210]

[0211] Where η t_sat is the charging distance time satisfaction rate; d (i) is the electric vehicle capacity density at the i-th point; v a is the average speed of the electric vehicle during the charging process; T d (i) is the charging time limit of the i-th point, that is, the charging time limit of the area to which it belongs; M t_sat is the set of electric vehicle capacity density points that meet the charging journey time requirements; N0 is the number of points; L d (i) is the right angle distance from the i-th point in the planning area to the corresponding charging station;

[0212] (2) SOC value satisfaction rate before charging

[0213] The SOC value of an electric vehicle before charging satisfies a certain probability distribution. When the SOC value is lower than a certain value, the life of the electric vehicle battery will be damaged. Therefore, the pre-charging SOC value satisfaction rate can be defined as the probability that the SOC value meets the minimum SOC value that does not damage the battery life of the electric vehicle when the user arrives at the charging station. M SOC values ​​are obtained by M Monte Carlo samplings, and the algorithm for the pre-charging SOC value satisfaction rate is:

[0214]

[0215] Where η s_sat is the SOC value satisfaction rate before charging; p d (i) is the probability that the i-th point meets the SOC value requirement before charging; ε d (i) is the electric vehicle capacity density at the i-th point; S range is the maximum cruising range of the corresponding electric vehicle; SOC(j) is the jth SOC value of Monte Carlo sampling; Ω d(i) is the set of SOC values ​​that meet the pre-charging SOC value requirements among the M SOC values ​​at the i-th point; card() is the number of elements in the set, SOC min is the lowest SOC value that does not damage the battery life; M is the number of Monte Carlo sampling;

[0216] (3) No-queuing satisfaction rate

[0217] If the demand for electric vehicle charging is greater than the configured capacity of the charging station, users need to queue before charging, which does not meet the no-queue requirement. The no-queue satisfaction rate is defined as the probability that users do not need to queue. The charging station load data of each charging station at 24 times a day is obtained through Monte Carlo sampling M times, and the algorithm for the no-queue satisfaction rate is:

[0218]

[0219] Where η p_sat is the satisfaction rate without queuing; p CD (i) is the average probability that charging station i meets the requirement of no queuing; П(j) is the set of times when the charging station load meets the requirement of no queuing in the jth Monte Carlo sampling of charging station i; J(i) is the number of chargers configured at charging station i; P CD (j,k) is the load of charging station i at the kth moment of the jth Monte Carlo sampling; P c is the rated power of a single charger; α c For charging efficiency;

[0220] (4) Calculation of user charging experience index

[0221] The charging experience index is reflected by comprehensively considering the charging distance time satisfaction rate, the pre-charging SOC value satisfaction rate and the no-queue satisfaction rate. The algorithm is:

[0222] η g_sat =η t_sat η s_sat η p_sat (10)

[0223] Where η g_sat It is the charging experience index;

[0224] Step 2-2: Establish the cost of charging station construction:

[0225] The construction cost of a charging station includes the construction cost of transformers, chargers and other equipment as well as land costs; therefore, the construction cost of charging station i is calculated as:

[0226] C z_t (i) = C s (i)+C d (i) (11)

[0227] In the formula, C s (i) is the construction cost of the transformer, charger and other equipment of the i-th charging station, which is determined by the number of chargers in the charging station; C d (i) is the land cost of the i-th charging station, and its calculation method is:

[0228] C d (i) = S d (i)D d (12)

[0229] In the formula, S d (i) is the land area occupied by the i-th charging station, which is determined by the number of chargers in the charging station, D d is the land cost per unit area;

[0230] The total cost of building charging stations in the planning area is:

[0231] C g_t =∑C z_t (i) (13)

[0232] Step 2-3: Establish user convenience indicators:

[0233] User convenience is expressed by the average charging distance. Points are averaged in the planning area. The electric vehicle capacity density of each point is the electric vehicle capacity density of the area to which it belongs. The charging distance is weighted averaged with the electric vehicle capacity density of each point as the weight to obtain the average charging distance. The algorithm for the average charging distance in the planning area is:

[0234]

[0235] In the formula, ε d (i) and L d (i) are the electric vehicle capacity density of the i-th point in the planning area and its right-angle distance to the corresponding charging station;

[0236] Step 2-4: Establish upper-level comprehensive goals:

[0237] The upper-level planning of charging stations takes the construction cost of charging stations, user convenience and charging experience index as the goals, and optimizes the location and capacity of charging stations; therefore, the algorithm for the upper-level comprehensive goal is:

[0238]

[0239] In the formula, α1, α2, and α3 are weight coefficients, which are used to make the orders of magnitude of each indicator the same and to eliminate the units of each indicator, so that the upper-level comprehensive target is dimensionless. B1 is the upper-level comprehensive target.

[0240] Step 2-5: Establish upper-level planning constraints:

[0241] (1) Distance constraints between adjacent stations

[0242] To ensure the charging needs of users, the service range of the charging station should not be too large. At the same time, in order to avoid wasting resources, the charging stations should not be too concentrated. Therefore, the right-angle distance L between adjacent charging stations is cc To satisfy the constraints:

[0243] L min <L cc <L max (16)

[0244] Where, L min and L max are the upper and lower limits of the distance constraints between adjacent stations respectively;

[0245] (2) Constraints on the number of chargers

[0246] In order to meet the charging needs of users and make full use of social resources to ensure a high utilization rate of chargers, the number of chargers configured in the charging station should meet the charger number constraint:

[0247] J min ≤J CD ≤J max (17)

[0248] In the formula, J min and J max are the minimum and maximum number of charger configurations, respectively, CD is the number of chargers configured in the charging station, and the algorithm is:

[0249]

[0250] In the formula, Z CD is the maximum charging demand within the service range of the charging station; K CD The capacity of electric vehicles within the service area of ​​the charging station.

[0251] Step 3: Propose a meta-point weighted Voronoi diagram taking into account spatial weights to partition the service area, including the following steps:

[0252] Step 3-1, propose the element-point weighted Voronoi diagram:

[0253] In view of the defect that the traditional weighted Voronoi diagram cannot reflect the difference in weights of different regions, the present invention proposes a meta-point weighted Voronoi diagram to divide the service range, and describes the size of the weights of different regions with spatial weights; at this time, the speed at which the generator generates the Voronoi diagram is not only related to the weight of the generator, but also to the spatial weight of each region; since the spatial weight is taken into account, the weighted distance between two points on different driving paths may be different, so the present invention defines the directed path from point A to point B as a path where A first drives longitudinally to B and then drives transversely, and defines the weighted directed distance as the weighted distance of the directed path;

[0254] Based on this, the meta-point weighted Voronoi diagram is defined as:

[0255] V(R i )={P∈V(R i )|D(R i ,P)≤D(R j ,P),j=1,2,…,m1;j≠i} (19)

[0256] In the formula, D(R i , P) is point R i The weighted directed distance to point P, m1 is the number of generators, D(R j ,P) is point R j The weighted directed distance to point P, if R i The weight is v i , R i The directed path to point P passes through regions Q1, Q2, ..., Q n , n is R i The number of regions that the directed path to point P passes through, and the corresponding spatial weights are θ1, θ2, …, θ n , R i The directed path to point P passes through the boundary points P1, P2, ..., P n-1 , d(P n-1 ,P) is point P n-1 The directed distance to point P, then R i The weighted directed distance to point P is:

[0257]

[0258] Step 3-2: Establish charging station weights:

[0259] Building charging stations with a larger service range and larger capacity in places with low land prices can reduce the total cost of charging station construction. Therefore, when dividing the service range, the square of the charging station weight can be set to be inversely proportional to the inverse of the unit price of the land it occupies. The weight of charging station i is:

[0260]

[0261] Where D i is the unit price of land occupied by charging station i;

[0262] Step 3-3, establish spatial weights:

[0263] The present invention establishes the spatial weight of the region by comprehensively considering the charging journey time requirements and electric vehicle capacity density of different types of regions. The algorithm for the spatial weight of region type i is:

[0264]

[0265] In the formula, θ q (i), ε q (i) T q (i) are the spatial weight, electric vehicle capacity density and charging time limit of regional type i, β1 and β2 are weight coefficients;

[0266] Step 3-4: Division of charging station service area

[0267] For the convenience of calculation, when using the element-point weighted Voronoi diagram for charging station planning, points are generated uniformly in the planning area, and the moving step length of adjacent points is Δx = Δy = Δl. The coordinates of each point, electric vehicle capacity density, land unit price, charging journey time limit and other information are recorded in a matrix to establish a point database; when drawing the Voronoi diagram, each point is represented by a square, the side length of the square is the moving step length, and the point is located at the center of the square; the electric vehicle capacity of the square is the product of the area of ​​the square and the electric vehicle capacity density of the point;

[0268] If the weight is v i Charging station R i The directed path to point O passes through O1, O2, ..., O n-1 , the corresponding spatial weight of each point is θ q1 ,θ q2 ,…,θ qn , then R i The weighted directed distance to point O is:

[0269]

[0270] Compare the weighted directed distances from each generator to a point, and the generator corresponding to the smallest weighted directed distance will include the point in its service range;

[0271] The electric vehicle capacity of a charging station is the sum of the electric vehicle capacities of all points within its service range. The algorithm is:

[0272]

[0273] In the formula, ε d (i) is the electric vehicle capacity density at the i-th point within the service range of the charging station, and ΔS is the area of ​​a single square.

[0274] Step 4: Based on the upper-level optimization goal, the upper-level planning process is established using the genetic taboo hybrid algorithm, including the following:

[0275] The present invention adopts the Genetic Taboo Hybrid Algorithm (GATS) for upper-level planning of charging stations; GATS combines the characteristics of GA with multiple starting points and TS with memory function and strong climbing ability, introduces the unique memory function of TS into the GA evolutionary search process, and constructs a new recombination operator (TSR); TS is used to improve the climbing ability of GA, that is, TS is used as the mutation operator (TSM) of GA; the main steps of upper-level planning of charging stations are as follows:

[0276] Ⅰ. Collect point database information as basic data for optimization;

[0277] Ⅱ. Using the electric vehicle capacity density information of the point database, calculate the electric vehicle capacity in the planning area, that is, K z , and then calculate the maximum charging demand Z in the planning area z , the algorithm is:

[0278]

[0279] III. Based on the maximum charging demand Z in the planning area z , with the minimum number of chargers configured in the charging station J min and the maximum number of chargers J max As a constraint, calculate the minimum number of charging stations N min and the maximum number of seats N max ,Right now:

[0280]

[0281] IV. For different charging station number schemes, GATS and element-node weighted Voronoi diagram are used to optimize the location and capacity based on the upper comprehensive objectives, including the following steps:

[0282] (1) Let the number of charging stations be g and set the maximum number of iterations of GA to N. gen , length of taboo table T length1 , group size N pop , recombination probability p ch and mutation probability p m ; The length of the taboo table in TSM is T length2 、Number of neighborhood solutions h and the maximum number of iterations N t And other parameters;

[0283] (2)t r represents the number of iterations, let t r = 0, divide the planned area into g areas equally, randomly select an address from each area, and obtain the chromosome address sequence [x1, y1; ...; x i ,y i ;…;x g ,y g ],(x i ,y i ) is the i-th address in the address sequence of the chromosome, and the addresses do not satisfy the inter-station distance constraint for regeneration;

[0284] (3) Calculate the upper-level comprehensive goals of the contemporary group;

[0285] (4) Select N by scrolling pop chromosomes;

[0286] (5) Generate a random number r between 0 and 1 i , i=1,2,…,N pop , if r i <p ch , then the i-th chromosome is the parent; perform a crossover operation on each pair of chromosomes, randomly swap two genes, and determine whether the offspring meets the inter-station distance constraint. If not, select the optimal parent to move to the next generation; if it does, compare the upper comprehensive goal of the offspring with the desired level. If it is better, break the taboo and move to the next generation; if it is worse but not taboo, accept the offspring. If it is taboo, select the optimal parent to move to the next generation;

[0287] (6) Generate a random number q between 0 and 1 i , i=1,2,…,N pop If q i <p m , then the mutation operation is performed on the i-th chromosome:

[0288] a. Take the current chromosome as the optimal chromosome, take its upper comprehensive target as the optimal upper comprehensive target, and set the taboo table of TSM to empty;

[0289] b. Generate neighboring address genes (x+Δx, y), (x-Δx, y), (x, y+Δy) and (x, y-Δy) with the current address gene as the center, add the current address gene to form the allele sequence, perform full permutation to obtain 5^g chromosomes and randomly extract N from them h indivual;

[0290] c. This N hThe chromosomes that do not meet the inter-station distance constraint are removed; the remaining chromosomes that are not in the taboo table are candidate chromosomes, and the upper-level comprehensive goal that meets the charger number constraint after partitioning is calculated;

[0291] d. Select the best candidate chromosome based on the minimum comprehensive objective of the above layer, and use it to replace the current chromosome and the earliest unsatisfied chromosome;

[0292] e. Determine whether the upper comprehensive target of the optimal candidate chromosome is less than the optimal upper comprehensive target. If yes, replace the optimal chromosome with it, and replace the optimal upper comprehensive target with its upper comprehensive target, and then go to step f; otherwise, go directly to step f;

[0293] f. Determine whether the maximum number of iterations N has been reached t , or the optimal upper comprehensive target has remained unchanged for 30 consecutive times? If yes, output the optimal chromosome; otherwise, return to step b;

[0294] (7)t r =t r +1, if t r <N gen , go to step (3); otherwise, output the optimal solution and terminate the algorithm.

[0295] Step 5: Establish the lower-level comprehensive target based on the cost of new feeder and the operation risk of distribution network as the optimization target of charging station load access node, and propose the lower-level constraint conditions, including the following steps:

[0296] Step 5-1: Calculate the cost of new feeder:

[0297] The charging station load needs to be connected to the distribution network by building a new feeder to connect to the node of the distribution network; assuming that the new feeder is laid along a straight line, the cost algorithm of the new feeder of the charging station is:

[0298] C k =D k ∑L k (i) (27)

[0299] Where D k is the cost per unit length of the feeder; L k (i) is the length of the newly built feeder of the i-th charging station, that is, the right-angle distance from the i-th charging station to the access node;

[0300] Step 5-2: Calculate the distribution network operation risk:

[0301] (1) Component risk indicators

[0302] Based on the weighted entropy principle, the weighted voltage distribution entropy is used to characterize the voltage risk caused by voltage over-limit and voltage distribution uncertainty, and the weighted power flow distribution entropy is used to characterize the power flow risk caused by power flow over-limit and power flow distribution uncertainty. The algorithm is:

[0303]

[0304] In the formula, H v_i (t) and H l_i (t) are the weighted voltage distribution entropy of node i and the weighted power flow distribution entropy of branch i at time t; n v_i (t) and n l_i (t) are the voltage state number of node i and the power flow state number of branch i at time t; S v_j (t) and P v_j (t) are the voltage loss severity and probability of the j-th voltage state of node i at time t; S l_j (t) and P l_j (t) are the severity and probability of power loss of the jth power state of branch i at time t;

[0305] The algorithm for loss severity S is:

[0306]

[0307] Where, μ is the loss severity amplification factor to adjust the sensitivity of the loss severity; w is the loss amount;

[0308] (2) Network risk indicators

[0309] The present invention comprehensively evaluates the operation risk of the distribution network containing large-scale electric vehicles by combining voltage risk and power flow risk. In order to examine the change of the distribution network operation risk in one day, the network operation risk at time t can be defined as:

[0310] H w (t) = γ1H v (t)+γ2H l (t) (30)

[0311] In the formula, H w (t) is the network operation risk at time t; γ1 and γ2 are weight coefficients; H v (t) and H l (t) are the network voltage risk and network flow risk at time t, respectively. The algorithm is:

[0312]

[0313] In the formula, z1 and z2 are the number of nodes and branches of the distribution network respectively;

[0314] In order to study the most serious risk faced by the safe operation of the distribution network, the maximum network operation risk is defined as the maximum value of the network operation risk in one day, that is:

[0315] H wmax =max(H w (t)) t=1,2,…,24 (32)

[0316] Step 5-3: Propose the lower-level comprehensive goals:

[0317] The lower-level planning of charging stations comprehensively considers the cost of new feeder lines and the operating risks of the distribution network, and optimizes the load access nodes of charging stations. The lower-level comprehensive objectives are defined as:

[0318] B2=ρ1C k +ρ2H wmax (33)

[0319] In the formula, C k and H wmax are the cost of new feeder and the maximum operation risk of the network respectively; ρ1 and ρ2 are weight coefficients;

[0320] Step 5-4: Propose lower-level planning constraints:

[0321] (1) Feeder length constraints

[0322] When the newly built feeder is too long, not only the feeder cost is too high, but also the feeder impedance is too large, which will increase power loss and voltage loss. Therefore, the newly built feeder should meet the feeder length constraint:

[0323] L k (i)≤L kmax (34)

[0324] Where, L kmax The length limit of the newly built feeder;

[0325] (2) Charging station load constraints

[0326] Since the charging load of a charging station cannot be greater than its installed capacity, the following constraints must be met:

[0327] P CD (i)≤α c J(i)P c (35)

[0328] Where P CD (i) is the charging load of charging station i.

[0329] Step 6: Based on the lower-level comprehensive objectives, use the cyclic replacement method to establish the lower-level planning process such as Figure 1 As shown, including the following:

[0330] (1) Input data such as the address of the charging station, the capacity of electric vehicles in each partition, charging process parameters, and distribution network parameters;

[0331] (2) Calculate the N m next-day charging load data for each partition of the charging station through Monte Carlo sampling;

[0332] (3) Connect the charging station load to the nearest distribution network node as the current plan, and calculate the lower-layer comprehensive objective;

[0333] (4) Denote the number of charging stations as g, determine the optimization order of each partition, b m = 0;

[0334] (5) b m = b m + 1, calculate the distance from charging station b m to each node, and take the nodes that meet the constraints as candidate nodes;

[0335] (6) Keep other access nodes unchanged, connect the load of charging station b m to each candidate node respectively, calculate the lower-layer comprehensive objective, and select the optimal candidate plan with the minimum lower-layer comprehensive objective;

[0336] (7) Judge the advantages and disadvantages of the optimal candidate plan and the current plan. If the optimal candidate plan is better than the current plan, replace the current plan with the optimal candidate plan; otherwise, keep the current plan unchanged;

[0337] (8) Judge whether there is b m < g. If yes, return to step (5); otherwise, go to step (9);

[0338] (9) Continuously loop and optimize until the charging station load access plan no longer changes.

[0339] In the embodiment, the planning area is as Figure 2 shown and is powered by a 10 kV distribution system. Node e s is the balancing node, the voltage is set to 1.05 p.u., and the system equivalent reactance X s is 2 Ω. The distribution network line type is LGJ-185, and the line parameters are shown in Table 1. e0 is the liaison node without load access. Figure 2 Nodes 1-30 in it are the load nodes of the distribution network. The load node positions and load peaks are shown in Table 2. The power factor of the electric vehicle load is 0.95. In the present invention, the power of a single charger is set to 120 kW, the maximum charging simultaneity rate λ max is taken as 0.2, SOC min is taken as 0.15, and v a is taken as 20 km / h. The conventional load curves of each type are as Figure 3As shown. The upper and lower limits of the distance between stations L max and L min Take 6km and 3km respectively. The length limit L of the new feeder kmax Take 4km.

[0340] In order to make the loss severity more sensitive, the minimum increment of loss severity is set to 0.05 when the loss amount increases by 0.001, and the value of the magnification coefficient μ is 82.422. The number of Monte Carlo sampling M is set to 500 times. The weight coefficients α1, α2, and α3 are 0.02, 0.49, and 0.49 respectively; β1, β2, γ1, and γ2 are all 0.5. ρ1 and ρ2 are 0.9 and 0.1 respectively.

[0341] GATS parameters are set as: N in GA pop is 200, N gen is 50, p ch is 0.5, T length1 is 15, p m is 0.2; N in TS h is 30, T length2 is 15, N t is 50.

[0342] Other simulation parameter settings are shown in Table 1-Table 6.

[0343] Table 1 Line parameters

[0344]

[0345] Table 2 Load node locations

[0346]

[0347]

[0348] Table 3 Parameter settings for each area

[0349]

[0350] Table 4 Electric vehicle parameters

[0351]

[0352]

[0353] Table 5 Lithium battery charging process parameters

[0354]

[0355] Table 6 Charging station parameters

[0356]

[0357] In order to prove the effectiveness of the charging station planning method proposed in this invention, this paper carried out the following simulation.

[0358] When GATS is used for charging station planning, there are planning schemes that meet the constraints only when the number of charging stations is 4, 5, and 6. The simulation results of the optimal planning schemes under different numbers of charging stations are shown in Table 7.

[0359] Table 7 Optimal solutions for different numbers of charging stations

[0360]

[0361] As shown in Table 7, as the number of charging stations increases, the total cost of charging station construction increases, but the average charging distance decreases and the user charging experience index increases. Therefore, the improvement of user convenience and charging experience index comes at the expense of increased construction costs. When the number of charging stations is 6, the upper comprehensive goal is the smallest, which can better meet the needs of construction cost, user convenience and charging experience at the same time. The corresponding optimal planning scheme is the optimal scheme for optimizing the location and capacity of charging stations.

[0362] The service range of the optimal planning scheme when the number of charging stations is 6 is as follows Figure 4 The data of each indicator are shown in Table 8.

[0363] Table 8 Charging station locations and costs

[0364]

[0365] Depend on Figure 4 As can be seen from Table 8, when selecting the site for charging stations, industrial areas or leisure areas with lower land prices are preferred as the site to reduce construction costs. Charging stations 3 and 6 located in industrial areas have lower land prices and higher charging station weights, so not only are the construction costs lower, but the service range is also larger. Therefore, the rationality of the present invention to measure the weight of charging stations by the price of occupied land when zoning the service range of charging stations is verified.

[0366] When the number of charging stations is 6, in the optimal planning schemes using weighted Voronoi diagram partitioning (scenario 1) and the element-point weighted Voronoi diagram partitioning (scenario 2) proposed by the present invention, the average charging distance from the points in each area to the corresponding charging station is shown in Table 9.

[0367] Table 9 Average charging distance in each region

[0368]

[0369] As can be seen from Table 9, compared with the weighted Voronoi diagram partitioning, after the meta-point weighted Voronoi diagram considering spatial weights is used for partitioning, the average charging distance of the commercial area and office area is significantly reduced, while the average charging distance of the industrial area and leisure area is increased. This is because the charging time requirements of the commercial area and office area are high, the electric vehicle capacity density is large, and their spatial weights are large. The directed weighted distance in the area is relatively small, and the partition speed of the charging station is faster, so the overall distance to the charging station is closer. On the contrary, the industrial area and the leisure area have a small spatial weight, so the overall distance to the charging station is farther. The location of the charging station is closer to the commercial area and other areas with higher charging time requirements and higher electric vehicle capacity density, that is, areas with greater demand for charging stations, which proves the rationality of the meta-point weighted Voronoi diagram proposed in the invention considering spatial weights and the superiority of the meta-point weighted Voronoi diagram over the weighted Voronoi diagram in range partitioning.

[0370] Before and after considering the user charging experience index, the optimal planning schemes have 4 and 6 charging stations respectively. The user charging experience indicators are shown in Table 10.

[0371] Table 10 Charging experience index

[0372]

[0373] As shown in Table 10, after considering the charging experience index, the satisfaction rate of charging distance time, the satisfaction rate of SOC value before charging and the satisfaction rate of no queuing have been significantly improved, and the charging experience index has increased by 23.6%, and the user's charging experience has been more satisfied. Therefore, taking the charging experience index as the optimization target can better meet the user's charging experience requirements.

[0374] Based on the optimal planning scheme when the number of charging stations is 6, the optimization of charging station load access nodes is carried out. The optimization results are shown in Tables 11 and 12. The over-limit probabilities and state numbers of the most serious over-limit components before and after optimization at 20:00 are shown in Tables 13 and 14 respectively.

[0375] Table 11 Access nodes before and after optimization

[0376]

[0377] Table 12 Target values ​​before and after optimization

[0378]

[0379] Table 13 Components with the most serious risks when connected nearby

[0380]

[0381] Table 14 Components with the most serious risks after optimization

[0382]

[0383] It can be seen from Tables 11 and 12 that the cost of newly built feeders after optimization increases to 3.4 times of the original, and the operating risk of the distribution network is reduced to 4.83% of the original. It can be seen from Tables 13 and 14 that compared with the case where the charging station load is connected nearby, the probability of over-limit of each component is greatly reduced after optimization, and the number of voltage and flow states is reduced, that is, the state distribution of the component is more stable. It can be seen that the node optimization effect is very obvious, and it can effectively reduce the operating risk of the distribution network at the cost of a smaller new feeder cost, so that the safe operation of the distribution network can be guaranteed when large-scale charging loads are connected to the distribution network, proving the effectiveness of the lower-level planning model for optimizing charging load access nodes proposed in the present invention.

[0384] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other modifications, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and should be included in the protection scope of the present invention.

Claims

1. A charging station planning method considering user charging experience and distribution network operation risks, characterized in that: Includes steps: Step 1: Establish an electric vehicle load model based on the electric vehicle historical data, including the following steps: Step 1-1, using electric vehicle capacity to describe the scale of electric vehicles in the region, where the electric vehicle capacity is the sum of the rated charging power of all electric vehicles in the region, including all electric vehicles in a charging state and a non-charging state; the electric vehicle capacity per unit area is defined as the electric vehicle capacity density; Step 1-2: According to the actual charging process of the electric vehicle battery, a two-stage variable power charging model of the electric vehicle constant current-constant voltage is constructed, including the following steps: During the charging process, electric vehicles use a constant current-constant voltage two-stage variable power charging mode. When the battery terminal voltage is less than the maximum battery terminal voltage U bmax The battery is charged in constant current mode; when the battery terminal voltage reaches the maximum battery terminal voltage U bmax When the battery enters the constant voltage stage, the current rapidly decays to zero, and the charging power is zero; the equation describing its charging process is: Where U b (t) is the voltage across the battery at time t, i c is the current during charging, C0 is the nominal capacity of the battery, R is the internal resistance of the battery, K0 is the polarization constant of the battery, E0 is the constant potential of the battery, A0 and B0 are the product coefficient and exponential coefficient of the battery charging exponential segment respectively, SOC(t0) represents the remaining capacity of the battery at the beginning of charging, h(t) is the rechargeable capacity of the battery at time t, that is, the discharged capacity; During constant current charging, the current i is a constant, and the charging power P(t) of the electric vehicle is expressed as: P(t)=U b (t)I n (2); where I n is the charging current in constant current charging mode, P(t) is the charging power at time t; Step 1-3: Based on the historical travel data of electric vehicles, a time series probability model of electric vehicle charging load is constructed, which includes the following steps: The charging load simulation of electric vehicles must select the probability model of each sampling link according to the type of electric vehicle, extract its driving distance, arrival time, initial state of charge and departure time, so as to determine the behavior state of the electric vehicle at each moment, and then calculate the SOC value of the electric vehicle at the corresponding moment to obtain its SOC curve for the day. The calculation formula is: (1) When charging Where SOC(t) is the initial state of charge of the electric vehicle; (2) While driving SOC(t+Δt)=SOC(t)-d r / D r ·Δt / T r (4); Where, d r is the driving distance of the electric vehicle, D r is the maximum range of an electric vehicle, T r is the driving time of the electric vehicle, Δt is the time interval; Then, the charging power for one day is calculated by using formula (1) and formula (2); the loads of various types of electric vehicles are superimposed one by one to obtain the total charging load of electric vehicles; the total charging load of electric vehicles in the area at time t P t for: Where: N1, N2 and N3 represent the number of electric private cars, buses and taxis in the charging state in the area at time t; P 1nt , P 2nt and P 3nt They represent the real-time charging power of the nth electric private car, bus and taxi in the charging state at time t respectively; Steps 1-4: Establish the maximum simultaneous charging rate: In order to clearly understand the maximum proportion of electric vehicles charging simultaneously in a certain area in a day, the maximum simultaneous charging rate of electric vehicles λ is defined max It is the ratio of the maximum load of electric vehicles to the capacity of electric vehicles in a day, that is: Where P t is the total charging load of electric vehicles in the region at time t, K z for the electric vehicle capacity in the area; Step 2: Establish a charging experience index, establish an upper-level comprehensive goal for the optimization of charging station site selection and capacity determination based on the charging experience index, charging station construction cost, and user convenience, and propose upper-level planning constraints; including the following steps: Step 2-1: Establish a charging experience index based on the charging journey time, SOC value before charging, and whether queuing is required: The charging experience index is defined as the degree to which the user's charging experience requirements are met during the charging process, and is reflected by comprehensively considering the charging journey time, the SOC value before charging, and whether the queuing requirements are required; (1) Charging distance time satisfaction rate People in different regions have different requirements for the time spent on charging due to different levels of time urgency, that is, different charging journey time requirements; the charging journey time limit is defined as the user's acceptable limit for the time spent on charging; when the time spent on charging is greater than the charging journey time limit, the charging experience requirement is not met; based on this, in order to examine the satisfaction degree of the charging journey time requirements of all users, the charging journey time satisfaction rate is defined as: Where η t_sat is the charging distance time satisfaction rate; d (i) is the electric vehicle capacity density at the i-th point; v a is the average speed of the electric vehicle during the charging process; T d (i) is the charging time limit of the i-th point, that is, the charging time limit of the area to which it belongs; M t_sat is the set of electric vehicle capacity density points that meet the charging journey time requirements; N0 is the number of points; L d (i) is the right angle distance from the i-th point in the planning area to the corresponding charging station; (2) SOC value satisfaction rate before charging For any electric vehicle at any point, its SOC value before charging satisfies a certain probability distribution. Therefore, whether it meets the SOC value requirement before charging also has a certain probability. Through Monte Carlo sampling, its M SOC values ​​are obtained to obtain the probability of meeting the SOC value requirement before charging. In order to examine the average probability of all users meeting the SOC value requirement before charging, the SOC value satisfaction rate before charging is defined as: Where η s_sat is the SOC value satisfaction rate before charging; p d (i) is the probability that the i-th point meets the SOC value requirement before charging; ε d (i) is the electric vehicle capacity density at the i-th point; S range is the maximum cruising range of the corresponding electric vehicle; SOC(j) is the jth SOC value of Monte Carlo sampling; Ω d (i) is the set of SOC values ​​that meet the pre-charging SOC value requirements among the M SOC values ​​at the i-th point; card() is the number of elements in the set, SOC min is the lowest SOC value that does not damage the battery life; M is the number of Monte Carlo sampling; (3) No-queuing satisfaction rate If the charging demand of electric vehicles is greater than the configured capacity of the charging station, users need to queue during the charging process, which does not meet the requirement of no queuing. The charging station load of each charging station at 24 times a day is obtained through Monte Carlo sampling, and the probability of each charging station meeting the requirement of no queuing in a day is obtained. In order to examine the average probability of all charging stations meeting the requirement of no queuing for users, the no queuing satisfaction rate is defined as: Where η p_sat is the satisfaction rate without queuing; p CD (p) is the average probability that charging station p meets the requirement of no queuing; П(j) is the set of times when the charging station load meets the requirement of no queuing in the jth Monte Carlo sampling of charging station p; J(p) is the number of chargers configured at charging station p; P CD (j,k) is the load of charging station p at the kth moment of the jth Monte Carlo sampling; P c is the rated power of a single charger; α c0 is the charging efficiency; K CD (p) is the capacity of electric vehicles within the service area of ​​charging station p; (4) Calculation method of user charging experience index The charging experience index η is reflected by comprehensively considering the charging journey time satisfaction rate, pre-charging SOC value satisfaction rate and no-queue satisfaction rate. g_sat for: or g_sat =the t_sat or s_sat or p_sat (10); Where η g_sat It is the charging experience index; Step 2-2: Establish the cost of charging station construction: The construction cost of a charging station includes the construction cost of transformers, chargers and other equipment as well as land costs; therefore, the construction cost of a charging station p is C z_t (p) is: C z_t (p) = C s (p)+C d (p)(11); where C s (p) is the construction cost of the transformer, charger and other equipment of charging station p, which is determined by the number of chargers in the charging station; C d (p) is the land cost of charging station p, and its calculation method is: C d (p)=S d (p)D d (12); In the formula, S d (p) is the land area occupied by charging station p, which is determined by the number of chargers in the charging station, D d is the land cost per unit area; The total cost of charging station construction in the planning area is C g_t for: C g_t =∑C z_t (p) (13); Step 2-3: Establish user convenience indicators: User convenience is expressed as the average charging distance. The average points in the planning area are taken, and the electric vehicle capacity density of each point is the electric vehicle capacity density of the area to which it belongs. The charging distance is weighted averaged with the electric vehicle capacity density of each point as the weight to obtain the average charging distance. Then the average charging distance L in the planning area g_p for: In the formula, ε d (i) and L d (i) are the electric vehicle capacity density of the i-th point in the planning area and its right-angle distance to the corresponding charging station; Step 2-4: Establish upper-level comprehensive goals: The upper-level planning of charging stations takes the construction cost of charging stations, user convenience and charging experience index as the goals, and optimizes the location and capacity of charging stations; therefore, the algorithm for the upper-level comprehensive goal is: In the formula, α1, α2 and α3 are weight coefficients, which are used to make the order of magnitude of each indicator the same and to eliminate the units of each indicator, so that the upper comprehensive target is dimensionless. B1 is the upper comprehensive target; Step 2-5: Establish upper-level planning constraints: (1) Distance constraints between stations The right angle distance L between adjacent charging stations cc To satisfy the constraints: L min <L CC <L max (16); where L min and L max are the upper and lower limits of the inter-station distance constraint respectively; (2) Constraints on the number of chargers In order to meet the charging needs of users and make full use of social resources to ensure a high utilization rate of chargers, the number of chargers configured at charging station p should meet the charger number constraint: J min ≤J CD (p)≤J max (17); where J min and J max are the minimum and maximum number of charger configurations, respectively, CD (p) is the number of chargers configured at charging station p, and the algorithm is: In the formula, Z CD (p) is the maximum charging demand within the service area of ​​charging station p; K CD (p) is the capacity of electric vehicles within the service area of ​​charging station p; Step 3: Propose a meta-point weighted Voronoi diagram taking into account spatial weights to partition the service area, including the following steps: Step 3-1, propose the element-point weighted Voronoi diagram: Aiming at the defect that the traditional weighted Voronoi diagram cannot reflect the difference in weights of different regions, a meta-point weighted Voronoi diagram is proposed to divide the service range, and the size of the weights of different regions is described by spatial weights; the speed at which the generator generates the Voronoi diagram is not only related to the weight of the generator, but also to the spatial weight of each region; considering the spatial weight, the weighted distance between two points on different driving paths is different, so the directed path from point A to point B is defined as the path where A first drives longitudinally to B and then drives horizontally, and the weighted directed distance is defined as the weighted distance of the directed path; Based on this, the meta-point weighted Voronoi diagram is defined as: V(R i )={P∈V(R i )|D(R i ,P)≤D(R g ,P),g=1,2,…,m1;g≠i}(19); In the formula, D(R i , P) is point R i The weighted directed distance to point P, m1 is the number of generators, D(R g ,P) is point R g The weighted directed distance to point P, if R i The weight is v i , R i The directed path to point P passes through regions Q1, Q2, ..., Q n , n is point R i The number of regions that the directed path to point P passes through, and the corresponding spatial weights are θ1, θ2, …, θ n , R i The directed path to point P passes through the boundary points P1, P2, ..., P n-1 , d(P n-1 ,P) is point P n-1 The directed distance to point P, then R i The weighted directed distance to point P is: Step 3-2: Establish charging station weights: Building charging stations with large service range and large capacity in places with low land prices can reduce the total cost of charging station construction. Therefore, when dividing the service range, the square of the charging station weight is inversely proportional to the inverse of the unit price of the land it occupies, then the weight of charging station p is ν p for: Where D p is the unit price of the land occupied by the charging station p; Step 3-3, establish spatial weights: The spatial weight of the region is established by combining the charging time requirements and electric vehicle capacity density of different types of regions. The algorithm for the spatial weight of region type c is: In the formula, θ q (c), ε q (c) T q (c) are the spatial weight, electric vehicle capacity density and charging time limit of regional type c; β1 and β2 are weight coefficients; Step 3-4: Division of charging station service area: For the convenience of calculation, when using the element-point weighted Voronoi diagram for charging station planning, points are generated uniformly in the planning area, and the moving step length of adjacent points is Δx = Δy = Δl. The coordinates of each point, electric vehicle capacity density, land unit price, and charging journey time limit information are recorded in a matrix to establish a point database; when drawing the Voronoi diagram, each point is represented by a square, the side length of the square is the moving step length, and the point is located at the center of the square; the electric vehicle capacity of the square is the product of the area of ​​the square and the electric vehicle capacity density of the point; If the weight is v p The directed path from charging station p to point M passes through the following points in sequence, namely O1, O2, ..., O n-1 , the corresponding spatial weight of each point is θ q1 ,θ q2 ,…,θ qn , then R i The weighted directed distance D(R i ,O) is: Compare the weighted directed distances from each generator to a point, and the generator with the smallest weighted directed distance will include the point in its service range; K CD (p) is: In the formula, Planning Area The electric vehicle capacity density of each point, ΔS is the area of ​​a single square; Step 4: Based on the upper-level optimization goal, the genetic taboo hybrid algorithm GATS is used to establish the upper-level planning process, including: The genetic taboo hybrid algorithm GATS is used for upper-level planning of charging stations. The steps are as follows: Ⅰ. Collect point database information as basic data for optimization; Ⅱ. Using the electric vehicle capacity density information of the point database, calculate the electric vehicle capacity in the planning area, that is, K z , and then calculate the maximum charging demand Z in the planning area z , the algorithm is: III. Based on the maximum charging demand Z in the planning area z , with the minimum number of chargers configured in the charging station J min and the maximum number of chargers J max As a constraint, calculate the minimum number of charging stations N min and the maximum number of seats N max ,Right now: IV. Based on the upper comprehensive objectives, GATS and the element-point weighted Voronoi diagram are used to optimize the site selection and capacity, including the following steps: (1) Let the number of charging stations be g, and set the maximum iteration number of the genetic algorithm GA to N. gen , length of taboo table T length1 , group size N pop , recombination probability p ch and mutation probability p m ; Taboo table length T in taboo search algorithm TS length2 、Number of neighborhood solutions h and the maximum number of iterations N t parameter; (2)t r represents the number of iterations, let t r = 0, divide the planning area into g addresses on average, randomly select one address from each address, and obtain the chromosome address sequence [x1, y1; x2, y2; ...; x i ,y i ;…;x g ,y g ],(x i ,y i ) is the i-th address in the address sequence of the chromosome, and the addresses do not satisfy the inter-station distance constraint for regeneration; (3) Calculate the upper-level comprehensive goals of the contemporary group; (4) Select N by scrolling pop chromosomes; (5) Generate a random number r between 0 and 1 i , a=1,2,…,N pop , if r i <p ch , then the ath chromosome is the parent; perform a crossover operation on each pair of chromosomes, randomly swap two genes, and determine whether the offspring meets the inter-station distance constraint. If not, select the optimal parent to move to the next generation; if it does, compare the upper comprehensive goal of the offspring with the desired level. If it is better, break the taboo and move to the next generation; if it is worse but not taboo, accept the offspring. If it is taboo, select the optimal parent to move to the next generation; (6) Generate a random number q between 0 and 1 i , a=1,2,…,N pop If q i <p m , then perform mutation operation on the ath chromosome: a. Take the current chromosome as the optimal chromosome, take its upper comprehensive target as the optimal upper comprehensive target, and set the taboo table of the taboo search algorithm TS to empty; b. Generate the neighboring address gene with the current address base as the center, add the current address gene to form the allele sequence, and perform full permutation to obtain 5 g chromosomes and randomly select N h indivual; c. This N h The chromosomes that do not meet the inter-station distance constraint are removed; the remaining chromosomes that are not in the taboo table are candidate chromosomes, and their partitions are calculated to meet the upper-level comprehensive goal of the charger number constraint; d. Select the best candidate chromosome based on the minimum comprehensive objective of the above layer, and use it to replace the current chromosome and the earliest unsatisfied chromosome; e. Determine whether the upper comprehensive target of the optimal candidate chromosome is less than the optimal upper comprehensive target. If so, replace the optimal chromosome with it, and replace the optimal upper comprehensive target with its upper comprehensive target, and then go to step f; otherwise, go directly to step f; f. Determine whether the maximum number of iterations N has been reached t , or the optimal upper comprehensive target has remained unchanged for 30 consecutive times, then output the optimal chromosome; otherwise, return to step b; (7)t r =t r +1, if t r <N gen , go to step (3); otherwise, output the optimal solution; Step 5: Establish the lower-level comprehensive target based on the cost of new feeder and the operation risk of distribution network as the optimization target of charging station load access node, and propose the lower-level constraint conditions; Step 6: Based on the lower-level comprehensive objectives, the lower-level planning process is established using the cyclic replacement method, including the following steps: (1) Input the address of the charging station and the electric vehicle capacity of each zone, charging process parameters, and distribution network parameter data; (2) Calculate the M-day charging load data of each charging station partition through Monte Carlo sampling; (3) Connecting the charging station load to the nearest distribution network node is taken as the current solution, and the lower-level comprehensive objectives are calculated; (4) Let the number of charging stations be g, and determine the optimization order of each partition, b m =0; (5)b m =b m +1, calculate charging station b m The distance to each node, nodes that meet the constraints are taken as candidate nodes; (6) Keep other access nodes unchanged and connect charging station b m The load is connected to each candidate node respectively, the lower-level comprehensive target is calculated, and the optimal candidate solution is selected with the minimum lower-level comprehensive target; (7) Determine the advantages and disadvantages of the optimal candidate solution and the current solution. If the optimal candidate solution is better than the current solution, replace the current solution with the optimal candidate solution; otherwise, keep the current solution unchanged; (8) Determine whether there is b m <g, if yes, return to step (5), otherwise go to step (9); (9) The optimization cycle is continued until the charging station load access plan no longer changes.

2. The charging station planning method considering user charging experience and distribution network operation risk according to claim 1 is characterized in that ,Step 5 uses the cost of new feeder and the operation risk of distribution network as the ,optimization target of charging station load access node to establish the ,lower-level comprehensive target, and proposes the lower-level constraints, ,including the following steps: Step 5-1: Calculate the cost of new feeder: The charging station load needs to be connected to the distribution network by building a new feeder to connect to the node of the distribution network. Assuming that the new feeder is laid along a straight line, the cost of the new feeder of the charging station is C k for: C k =D k ΣL k (p)(27); where D k is the unit distance price of the feeder, L k (p) is the right-angle distance from the pth charging station to the access node, that is, the right-angle distance from charging station p to the access node; Step 5-2: Calculate the distribution network operation risk: (1) Component risk indicators Based on the weighted entropy principle, the weighted voltage distribution entropy is used to characterize the voltage risk caused by voltage over-limit and voltage distribution uncertainty, and the weighted power flow distribution entropy is used to characterize the power flow risk caused by power flow over-limit and power flow distribution uncertainty. The algorithm is: In the formula, H v_η (t) and are the weighted voltage distribution entropy of node η and branch at time t respectively. The weighted power flow distribution entropy of v_η (t) and are the voltage state number of node η and the branch voltage state number of node η at time t. The number of power flow states; S v_ε (t) and P v_ε (t) are the voltage loss severity and probability of the εth voltage state of node η at time t; S l_τ (t) and P l_τ (t) are the branches at time t The severity and probability of power flow loss in the τth power flow state; The algorithm for loss severity S is: Where, μ is the loss severity amplification factor to adjust the sensitivity of the loss severity; w is the loss amount; (2) Network risk indicators The operation risk of the distribution network containing large-scale electric vehicles is evaluated by combining voltage risk and power flow risk. In order to examine the changes in the operation risk of the distribution network in one day, the network operation risk at time t is defined as: H w (t)=γ1H v (t)+γ2H l (t) (30); In the formula, H w (t) is the network operation risk at time t; γ1 and γ2 are weight coefficients; H v (t) and H l (t) are the network voltage risk and network flow risk at time t, respectively. The algorithm is: In the formula, z1 and z2 are the number of nodes and branches of the distribution network respectively; In order to study the most serious risk faced by the safe operation of the distribution network, the maximum network operation risk is defined as the maximum value of the network operation risk in one day H wmax ,Right now: H wmax =max(H w (t)) (32); Step 5-3: Propose the lower-level comprehensive goals: The lower-level planning of charging stations comprehensively considers the cost of new feeder lines and the operating risks of the distribution network, and optimizes the load access nodes of charging stations. The lower-level comprehensive objectives are defined as: B2=ρ1C k +ρ2H wmax (33); where C k and H wmax are the cost of new feeder and the maximum operation risk of the network respectively; ρ1 and ρ2 are weight coefficients; Step 5-4: Propose lower-level planning constraints: (1) Feeder length constraints When the newly built feeder is too long, not only the feeder cost is too high, but also the feeder impedance is too large, which will increase the power loss and voltage loss. Therefore, the newly built feeder should meet the feeder length constraint: Y k (p)≤Y kmax (34); where Y kmax The length limit of the newly built feeder; (2) Charging station load constraints Since the charging load of a charging station cannot be greater than its installed capacity, the following constraints must be met: P CD (p)≤α c0 J CD (p)P c (35); where P CD (p) is the charging load of charging station p.

Images