Optimal community charging pile distribution method under time-sharing mechanism

Abstract

The invention discloses a community charging pile optimal distribution method under a time-sharing mechanism, and belongs to the field of planning. Comprising the following steps: mastering space-time distribution characteristics of community charging demands, obtaining time characteristic quantity distribution through fitting, and predicting the charging load of the electric vehicle by using a Monte Carlo simulation method; comprising the following steps: 1) inputting data including the number of simulated electric vehicles, first vehicle use time distribution, staying duration distribution, mileage distribution, charging power of charging piles, battery capacity of the electric vehicles and power consumption per kilometer; (2) the vehicle number j is made to be equal to j + 1, travelled distance distribution is extracted, and the initial charging battery state of the electric vehicle is calculated; 3) extracting the vehicle using time and the staying time, calculating the charging time, and determining the actual charging time according to the required charging electric quantity; 4) determining the system charging power at each time point, and updating the system load; 5) judging whether all the electric vehicles are calculated or not; and if not, returning to the step 2), and if yes, outputting a total system node load result.

Description

Technical Field

[0001] The present invention belongs to the field of linear programming, and in particular relates to an optimization method for distributing shared charging piles in a community. Background Art

[0002] A charging pile, also known as an electric vehicle charging station or electric vehicle power supply equipment, is a device that provides electrical energy to electric vehicles, enabling them to store enough electricity to support their operation.

[0003] As an important part of the new energy vehicle ecosystem, the demand for charging piles is increasing day by day with the rapid development of the new energy vehicle industry.

[0004] In recent years, power supply companies have implemented some shared charging pile construction projects in large and old residential areas.

[0005] Currently, when considering the layout of shared charging piles, the main approach is to choose locations with heavy traffic based on the direction of the cables. This experience-based method is relatively rough and cannot guarantee maximum sharing among the charging piles and maximize customer satisfaction.

[0006] The invention patent application with the application publication date of 2024.05.31 and the application publication number of CN 118114826 A discloses a "campus two-wheeled electric vehicle charging pile re-layout optimization method and system", including: obtaining the number of two-wheeled electric vehicles entering and leaving the campus every day, the frequency of entering and leaving the charging station, and the user's preference and demand distribution point data for selecting charging piles; pre-processing the acquired data and calibrating the data; analyzing the data to determine the constraints that the site selection plan must meet; formulating a site selection plan to construct an objective function with the minimum total investment cost and the minimum user charging cost; taking full account of the user's preferences and needs, using the walrus optimization algorithm to perform two rounds of optimization on the objective function to obtain the best electric vehicle charging pile layout optimization plan. This technical solution ensures that the total cost of the optimization target is minimized, so that the optimized charging pile layout can meet the needs of users while balancing the interests of investors and users, and achieve reasonable parking and charging. However, this technical solution focuses on planning the distribution of charging stations by taking into account vehicle flow, without considering the problem of increased peak-to-valley differences in load within the community. It fails to fully tap the shared capacity reduction potential, which is not conducive to the safe and stable operation of the power supply system.

[0007] Judging from the research situation at home and abroad, the current relevant research is mostly aimed at the entire urban network or highway. The analysis of the temporal and spatial distribution of charging demand is mainly based on the travel chain theory, and then combined with the Monte Carlo method to complete the simulation of the electric vehicle travel process, and then obtain the time and spatial distribution of charging and swapping demand.

[0008] The existing deployment method has two major limitations when applied to the new operation environment of community shared charging piles:

[0009] (1) In the urban networks and highways targeted by the current methods, vehicles have a strong route orientation when driving in these areas, so they often choose charging stations / charging piles based on the principle of "charging nearby along the way". The charging needs of car owners after returning to their residential communities are different and have higher transferability. Car owners are more likely to consider both distance and waiting time and choose between multiple charging piles within a certain range. In this regard, the current charging demand distribution characteristic analysis method is not very suitable.

[0010] (2) "Time-sharing" is the characteristic and advantage of community charging piles under the new operating environment. However, existing methods mostly optimize the layout of charging stations / piles based on the spatial distribution of charging demand during peak traffic hours, which makes it impossible to fully tap the capacity reduction potential of shared charging piles in the layout and capacity determination stages. Summary of the invention

[0011] The technical problem to be solved by the present invention is to provide an optimal method for distributing community charging piles under a time-sharing sharing mechanism. It fully taps the capacity reduction potential of sharing, comprehensively considers the cost of charging stations and user costs, and helps to improve the technical level of power supply companies in the work of distributing community shared charging piles, and fully taps the capacity reduction benefits of charging pile sharing.

[0012] The technical solution of the present invention is to provide a method for optimally distributing community charging piles under a time-sharing sharing mechanism, including mastering the spatiotemporal distribution characteristics of community charging demand, determining the optimal path for users to reach the charging station from their location, and using the Monte Carlo simulation method to predict the charging load of electric vehicles by fitting the distribution of the time characteristic quantity obtained; the method is characterized in that the prediction of the charging load of electric vehicles specifically includes the following steps:

[0013] 1) Input data including the number of simulated electric vehicles, the distribution of first-time vehicle use, the distribution of dwelling time, the distribution of driving mileage, the charging power of charging piles, the battery capacity of electric vehicles, and the power consumption per kilometer;

[0014] 2) Let vehicle number j = j + 1, extract the mileage distribution, and calculate the battery status of the electric vehicle at the beginning of charging;

[0015] 3) Take the vehicle usage time and the length of stay, calculate the charging time, and then determine the actual charging time based on the required charging power;

[0016] 4) Determine the system charging power at each time point and update the system load;

[0017] 5) Determine whether all electric vehicles have been calculated; if not, return to step 2); if so, output the total system node load result.

[0018] Specifically, the optimal layout method of community charging piles under the time-sharing sharing mechanism takes into account both charging stations and users, optimizes the number of charging piles in the same charging station, optimizes the number and location of charging stations, and comprehensively considers the cost of charging stations and the charging cost of users to obtain the optimal location and number of charging stations.

[0019] Specifically, the optimal distribution method of community charging piles under the time-sharing sharing mechanism incorporates the user transfer probability within the community into the load simulation, so that the prediction is more consistent with the actual charging situation of users in the community.

[0020] Furthermore, the optimal distribution method of community charging piles under the time-sharing sharing mechanism obtains the user's hypothetical remaining power of the electric vehicle by fitting the distribution of time feature quantities; obtains the distribution of the user's mileage by investigating the user's basic situation and charging habits; and assumes that the charging time when the user uses a fast charging pile is: the electric vehicle starts charging until the electric vehicle is fully charged and stops charging; the charging time when using a slow charging pile is: the smaller value between the time from the start of charging to full charging and the stay time.

[0021] Furthermore, the optimal layout of charging piles includes two parts: one is to optimize the number of charging piles in the same charging station so that the charging needs of users can be met in most cases, and at the same time, there will be no large number of idle charging piles; the other is to optimize the number and location of charging stations, considering both the charging station and the user, and comprehensively considering the cost of the charging station and the charging cost of the user to obtain the best location and number of charging stations.

[0022] Specifically, the optimal distribution method of community charging piles under the time-sharing sharing mechanism adopts queuing theory to determine the number of charging piles; that is, the process of electric vehicles arriving at the system conforms to the Poisson distribution with parameter λ, and the interval time to arrive at the charging station and the charging time of the charging pile conform to the negative exponential distribution with parameter μ, s is the number of charging piles, and k is the maximum number of parking spaces in the charging station.

[0023] Specifically, the optimal layout method of community charging piles under the time-sharing sharing mechanism adopts a genetic algorithm to determine the location of the charging station, and distributes and optimizes the location of the charging station by improving the genetic algorithm and the Voronoi diagram and taking the transfer probability into account. While meeting the charging needs of the community, it reduces the access capacity of the charging piles, reduces the construction cost of the charging station, narrows the daily peak-to-valley difference of the charging load, and shortens the average waiting time of users.

[0024] Furthermore, the optimal distribution method of community charging piles under the time-sharing sharing mechanism uses the hierarchical analysis method to make decisions in order to determine the weights of user cost and charging station cost in the objective function.

[0025] Furthermore, the AHP method decomposes the decision problem into a hierarchical structure, which mainly includes three levels: target level, criterion level, and solution level;

[0026] Among them, the target layer is to select a suitable target charging station;

[0027] The criterion layer is the component of the objective function, including the cost of the charging station, the user's time cost and the power loss cost;

[0028] The solution layer is all charging stations in the target area.

[0029] Furthermore, the elements of the criterion layer are compared in pairs to judge their relative importance, and a judgment matrix is ​​obtained through the comparison results; the elements are scaled according to their importance to the target layer, and the larger the value, the higher the importance;

[0030] Perform hierarchical single sorting and consistency check, perform hierarchical single sorting by square root method, and calculate the weight matrix W of the corresponding elements of the criterion layer;

[0031] Calculate the maximum eigenvalue through the weight matrix and the judgment matrix;

[0032] For the hierarchical analysis method with multiple levels, it is also necessary to perform a total hierarchical ranking and consistency check between the levels and calculate the weight of the relative importance of the solution level to the target level.

[0033] Compared with the prior art, the advantages of the present invention are:

[0034] 1. The technical solution of the present invention integrates road information and user information to plan the optimal route for electric vehicles; a method for optimizing the layout of community electric vehicle shared charging piles that comprehensively considers the cost of charging stations and user costs is proposed;

[0035] 2. The technical solution of the present invention is aimed at the community application environment, combined with the actual situation of the time characteristic quantity and the actual situation of the user's travel area, to make a community travel chain analysis; it can reduce the maximum charging power of a single point and shorten the average waiting time of users;

[0036] 3. The technical solution of the present invention fully taps the capacity reduction potential of sharing and comprehensively considers the cost of charging stations and user costs; it helps to improve the technical level of power supply companies in the deployment of shared charging piles in communities and fully tap the capacity reduction benefits of shared charging piles. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1a It is the frequency percentage histogram and probability density function diagram of the initial travel time of community users on weekdays;

[0038] Figure 1bIt is the frequency percentage histogram and probability density function diagram of the initial travel time of community users on rest days;

[0039] Figure 2 It is a frequency percentage histogram and probability density function diagram of the number of trips per day of community users;

[0040] Figure 3 This is a schematic diagram of the probability distribution analysis results of a single parking time in the HOME area;

[0041] Figure 4 This is a schematic diagram of the probability distribution analysis results of a single parking time in the WORK area;

[0042] Figure 5 This is a schematic diagram of the probability distribution analysis results of a single parking time in the OTHERS area;

[0043] Figure 6 This is a schematic diagram of the probability distribution analysis results of a single parking time in the OUTSIDE area;

[0044] Figure 7 It is a flow chart of the calculation process of the charging load of an electric vehicle based on Monte Carlo simulation of the present invention;

[0045] Figure 8 It is a schematic diagram of a charging load simulation curve of the present invention;

[0046] Fig. 9 It is a flowchart of the analytic hierarchy process of the present invention;

[0047] Fig.10 It is a schematic diagram of the hierarchical structure of the present invention;

[0048] Fig.11 It is a specific flow chart of the genetic algorithm of the present invention;

[0049] Fig.12 is a target area diagram of an embodiment of the present invention;

[0050] Fig.13 is a schematic diagram of the responsible area of ​​each charging station in an embodiment of the present invention;

[0051] Fig.14 It is a schematic diagram of the comparison results of the load conditions in the community under the two methods of the present invention according to the vehicle flow and taking the transfer probability into account. DETAILED DESCRIPTION

[0052] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0053] 1. Analysis of spatiotemporal distribution characteristics of community charging demand:

[0054] Understanding the spatiotemporal distribution characteristics of community charging demand is the basis for implementing the optimal layout of community charging piles.

[0055] By linking the single travel status of individual users through travel chains, the travel activities of a large number of users can be analyzed, focusing on the traffic distribution and travel time during peak periods.

[0056] Among electric vehicle users, the trip chain represents the number of trips, departure point, time spent on the way, stopovers, stay time, destination, etc. It is specifically divided into two types: time feature quantity and spatial feature quantity.

[0057] 1.1. Probability distribution of time feature quantity:

[0058] The temporal feature quantity describes the temporal variation of a user's travel activities in a day, including:

[0059] (1) Daily initial travel time T0;

[0060] (2) The driving time and speed between destinations (denoted as t for the i-th trip) i and v i );

[0061] (3) Time t to reach the destination of the i-th trip x_i ;

[0062] (4) Charging time t after reaching the i-th trip destination c_i ;

[0063] (5) The parking time t after arriving at the i-th trip destination p_i ;

[0064] (6) Time of departure from the i-th destination (i.e., time of the i+1-th trip) t k_i .

[0065] The initial travel time of the day, the driving time between destinations, and the number of daily trips are the fundamental determinants of the above-mentioned time characteristics, and the remaining time characteristics can be obtained by simulation and reasoning based on the above-mentioned fundamental determinants. Therefore, the technical solution of the present invention only provides an analysis method for the probability distribution of the three fundamental determinants.

[0066] 1.2. Gaussian mixture model analysis method for probability distribution of temporal feature quantities:

[0067] The technical solution of the present invention adopts Gaussian Mixture Model (GMM) to fit the probability distribution of three fundamental determining factors related to time feature quantities.

[0068] GMM considers the probability distribution of a random variable as a combination of multiple normal distributions.

[0069] When there is only a single feature or variable, a one-dimensional Gaussian distribution is used. The one-dimensional Gaussian mixture model is a linear combination of multiple normal distributions in the form of

[0070]

[0071] in

[0072]

[0073] Where: K is the number of normal distribution functions considered in the Gaussian mixture model; parameter model set Θ = {Θ1, Θ2, L, Θ i ,L,Θ K}, α i , μ i and are the weight, mean and variance of the i-th normal distribution function respectively. There are K normal distribution functions in total. Formula (3-2) is the i-th normal distribution function.

[0074] Determine the parameter α in the GMM model (i.e., formula (1.1)) i , μ i and

[0075] To this end, the expected maximum algorithm (EM algorithm) is used to solve it.

[0076] First, define the log-likelihood function of parameter Θ as

[0077]

[0078] Formula r i (n) is the posterior probability of the variable Θ, and the formula is given in the subsequent parameter iteration step. N is the total number of data points, that is, the number of valid data in the data set.

[0079] The maximum expected value algorithm iterates to obtain the maximum log-likelihood estimate of the parameter Θ, which satisfies:

[0080]

[0081] To this end, the iterative solution of parameters is implemented as follows:

[0082] 1) Choose the initial value of the parameter Θ (0) , set the initial number of iterations k = 0;

[0083] 2) The prior probability is obtained by performing Kmeans clustering on the initial input data;

[0084] 3) Remember Θ(i) is the estimated value of the parameter bΘ of the i-th iteration, and the posterior probability of each variable is calculated using Bayes' theorem, as follows:

[0085]

[0086] where r i (t) (n) represents the posterior probability generated by the i-th sub-Gaussian distribution in the Gaussian mixture model after t iterations;

[0087] 4) Take the partial derivative of the log-likelihood function with respect to each parameter to obtain the iterative update expression of each parameter:

[0088]

[0089] 5) From formula (1.5), we can know that the maximum log-likelihood estimation is to determine whether the updated Θ has reached the convergence accuracy or the maximum number of iterations. If so, proceed to step (1.6), otherwise repeat steps (1.2) to (1.4) to continue iterating;

[0090] 6) Output the parameters of the Gaussian distribution function in the Gaussian mixture model

[0091] 1.3、Probability distribution analysis of time feature quantity:

[0092] The vehicles in the community include public vehicles (such as electric sanitation vehicles, postal vehicles, etc.) and private cars. In view of the fact that the shared charging stations in the community are mainly aimed at the charging needs of private cars, the subsequent research will only be carried out on private car users. For individual users, since there are many residential areas in the jurisdiction, the main consideration is that residents use residential areas as the starting point for different activities. The travel chains on weekdays and weekends are different and need to be considered separately. Generally speaking, the morning peak on weekdays is earlier than the peak on weekends.

[0093] 1.3.1. Initial travel time

[0094] Taking into account the different work and rest time of users on different day types, the probability distribution function of the initial travel time T0 is solved separately for two scenarios: weekdays and weekends.

[0095] The GMM (Gaussian mixture model) method is used to analyze the probability distribution of the initial travel time of community users on weekdays and weekends. Based on the analysis results, a frequency percentage histogram and probability density function are made as shown in the figure below: Figure 1a and Figure 1b As shown, the GMM fitting parameters are shown in Table 1, where j represents the jth normal distribution fitting function.

[0096] Table 1 GMM fitting parameters for initial travel time

[0097]

[0098] It can be seen that:

[0099] (1) During the working day, the initial travel time is mainly concentrated between 7:00 and 8:00 in the morning, which coincides with the 9:00 start time for enterprises and institutions; while the initial travel time on weekends is later, concentrated between 9:00 and 10:00 in the morning. This means that a large proportion of users’ travel destinations on weekdays are workplaces, while the weekends have longer rest time and they go out relatively late.

[0100] (2) The variance of the GMM fitting parameters on weekdays is slightly smaller. This shows that travel on weekdays is more concentrated than on weekends, mainly for work; on weekends, the work and rest patterns are more scattered, and the purpose of the first trip is also more scattered. Some users go to work, while others go for entertainment, shopping, etc., so the choice of the first trip time is more scattered and flexible.

[0101] 1.3.2 Number of trips per day

[0102] The following definition is given for a single trip: If you drive from home to work, then drive from work to go shopping, and then drive home after shopping in one day, it counts as three trips, each of which is a single trip.

[0103] The number of trips per day can only be fitted uniformly. Using the same basic data, the GMM method is used to analyze the probability distribution of the number of trips per day. The obtained frequency histogram and GMM probability density fitting results are as follows: Figure 2 The fitting parameters are shown in Table 2.

[0104] Table 2 GMM fitting parameters for daily travel times

[0105]

[0106] It can be seen that:

[0107] (1) The number of trips per day is mainly concentrated in 2 and 4 times;

[0108] (2) the number of trips twice coincided with the travel chain of starting from home, working, and returning home;

[0109] (3) the number of trips four times coincided with a travel chain of starting from home-working-returning home for lunch or eating out-working-returning home;

[0110] (4) In addition, on weekdays, people go out for leisure and entertainment after get off work, and on weekends, they go to different areas and then go home, so the frequency of three trips is also relatively high.

[0111] 1.3.3. Single parking duration

[0112] Considering that users have different behavioral purposes when driving and parking their vehicles in different functional areas and need to stay for different lengths of time, the probability distribution of a single parking time is analyzed separately according to the functional areas.

[0113] The functional areas within the community are divided into three categories: residential area (HOME), work area (WORK), and commercial area / leisure area (OTHERS). In addition, users in the community may travel outside the community, so a fourth area is added - outside the community (OUTSIDE), which is a long-distance area.

[0114] When the vehicle's single driving distance exceeds the total perimeter of the community, the vehicle is considered to have left the community and entered the community. The single parking time of a vehicle refers to the parking time of the area after the vehicle enters the area.

[0115] Based on the same basic data, the GMM method is used to analyze the probability distribution of single parking time in four functional areas. The frequency histogram and probability density diagram obtained by analysis are shown in the figure below: Figures 2 to 7 The fitting parameters are shown in Table 3.

[0116] Table 3 GMM fitting parameters for single parking duration

[0117]

[0118] It can be seen that:

[0119] (1) The parking time in different areas varies greatly. The parking time in the work area is concentrated in three parts: less than 100 minutes, 200-300 minutes, and 500-600 minutes. This is consistent with the previous statement that most parking times are 2-4 times. The most common situations are short-term work, half-day parking at work, and driving to work in the morning and leaving in the evening to park for a whole day.

[0120] (2) The parking time in the commercial and leisure area is mainly concentrated around 100 minutes, because users mainly go to this area for leisure, entertainment and social activities, which does not require too much time;

[0121] (3) The parking time outside the jurisdiction is similar to the image of the three superimposed together within the jurisdiction, because users go outside the jurisdiction to go home, work, and business leisure activities, and the proportions are similar;

[0122] (4) The parking time of users in residential areas and commercial / leisure areas is similar, while the parking time in work areas is longer than that in the other two areas.

[0123] 1.3.4. Probability distribution of spatial feature quantities:

[0124] Spatial characteristics are used to describe the changing patterns of a user's travel activities in a spatial area throughout the day.

[0125] To this end, we introduce a set of destination types {S(1), S(2), …, S(k)}, where k is the total number of destinations reached by the trip; D i =S(k) indicates that the destination of the i-th trip is type(k); the distance traveled by the user from the i-th destination to the i+1-th destination is d i The probability distribution of the spatial characteristic quantity of electric vehicles is mainly manifested in the probability of their transfer between destinations within a day.

[0126] The Markov Chain is used to analyze the transition probability of spatial feature quantities. To this end, the state sequence of electric vehicle users is defined as {X(t), t∈T}, where the time sequence T = {0, 1, 2, ...}, then the conditional probability of its state at any time depends only on the state at the previous moment, that is,

[0127]

[0128] As long as the transition probability between any two states can be calculated, the corresponding Markov chain model can be determined. i Under the condition of j The conditional probability of is the state transition probability between these two states, that is:

[0129]

[0130] The above formula indicates that at time t, the Markov chain changes from state D i Start, after k steps of transfer, arrive at state D at time t+m j When k = t, the transition probability matrix of the Markov chain model is the t-step transition probability matrix, and the elements in the matrix are Any element satisfies The sum of the elements in any row is equal to 1.

[0131] The model parameters of the Markov chain are (S, A, Π), where S represents the set of destinations, P represents the transition probability matrix between destinations, and Π represents the probability distribution of the departure point. The specific steps are as follows:

[0132] (1) Suppose that the set of all destinations that the user can reach is S = {s1, s2, ..., s N};

[0133] (2) Calculate the transition probability matrix P, P = {a ij} indicates that the user at time t, from the destination D i Transfer to D j The probability is calculated as

[0134] a ij =P(q t+1 =s j |q t =s i ),1≤i,j≤N(1.12)

[0135] (3) Calculate the probability distribution of the departure point Π, Π = {π1, π2, ..., π N}, π i It represents the probability that the user departs from the departure place i, and the calculation formula is as follows:

[0136] π i =P(q t =s i )(1.13)

[0137] The frequency of actual data of local users can be approximated into probability to calculate the parameters of the Markov chain. The probability of the user's departure point and the probability of transferring from one destination to another can be obtained by detecting the spatial feature quantity.

[0138] The data used for spatial feature quantities comes from the National Household Travel Survey NHTS2021 in the United States. The number of car trips from various cities in the United States to other cities in 2021 is sorted out. Since all nodes in the target area are reachable to each other, the number of trips for the departure-destination routes with a zero number in the NHTS data is set to 1. Since the highest proportion of all cities is the departure and destination being the same city, the frequency of the same departure and destination is set to 0. Then 68 cities are selected from the data as the destination set S.

[0139] Number the road nodes in the target area, add up all the times for the same departure point, divide one of the departure-destination times by the total to get the transition probability, add up all the departure-destination times, divide the times for the same departure point by the total times to get the probability distribution of the departure point.

[0140] The probability distribution of the departure point and the transition probability matrix are shown in the following two tables.

[0141] Table 4 Probability distribution of departure place

[0142] Departure number Point 1 Point 2 Point 3 … Point 67 Point 68 Frequency 166505 23314781 765885 … 2192604 1048053 frequency 0.000544 0.076223 0.002504 … 0.007168 0.003426

[0143] Table 5 Transition probability matrix

[0144] Node Number Point 1 Point 2 Point 3 … Point 67 Point 68 Point 1 0 9.91E-4 1.86E-4 … 6.01E-6 1.32E-3 Point 2 3.86E-6 0 3.35E-6 … 4.57E-4 2.39E-4 Point 3 5.88E-5 1.38E-4 0 … 1.31E-6 4.58E-3 … … … … … … … Point 67 4.560E-7 424E-3 4.56E-7 0 1.78E-3 Point 68 2.06E-4 5.13E-3 5.17E-3 … 4.74E-3 0

[0145] 1.3.5 Optimal Path Planning

[0146] Optimal path planning is a method for finding the best path from a starting point to a target point in a graph or network, and has wide applications in computer science and engineering.

[0147] The two most common algorithms are Dijkstra's algorithm and the Bellman-Ford algorithm, both of which can be used to find the shortest path from a starting point to all other nodes in a graph.

[0148] The Dijkstra algorithm uses a greedy idea, starting from the initial node, and each time selects the node with the shortest distance from the unvisited nodes adjacent to the node until all nodes are visited.

[0149] The Bellman-Ford algorithm creates a distance array, sets the distance of the starting node to 0 and the distances of other nodes to infinity, and then relaxes all edges, trying to shorten the distance between two nodes through a known path to obtain the optimal path.

[0150] The difference between the two is that the Dijkstra algorithm is applicable to cases without negative weight edges and is more efficient for sparse graphs. The Bellman-Ford algorithm can process graphs with negative weight edges, but its disadvantage is that its time complexity is high.

[0151] In this technical solution, in order to determine the optimal path for the user to reach the charging station from his current location, there will be no negative weight edges in the path, and there are 68 nodes and 106 roads in total, that is, the graph formed is a 68*68 matrix, of which only 212 values ​​are 0 and the rest are 0, which is a sparse matrix. Considering efficiency and accuracy, this technical solution chooses to use the Dijkstra algorithm.

[0152] The traffic road network is abstracted as a weighted directed graph G = ({V}, {E}), where V is a set of n vertices, which are the starting points, end points, stopovers and intersections in the traffic network; E is a set of m arcs, which are the arcs in the traffic network; <v, w> represents the arc from vertex v to vertex w in E; α(v, w) represents the non-negative weight of the arc, which can be expressed as attributes such as the length of the road, driving time, and user familiarity in traffic conditions.

[0153] Use Dijkstra algorithm to solve the shortest path. The specific steps are as follows:

[0154] (1) Initialize the shortest path set dis. Except for the starting point ab, set dis=∞ for other nodes.

[0155] (2) Starting from the starting point ab, if the node v iConnected with ab, the shortest path dis is updated, that is:

[0156] dis(i)=α(ab,v i )(1.14)

[0157] (3) Find the smallest dis(i) among the nodes connected to the starting point ab, and include this node in the shortest path node.

[0158] In the point set T, the shortest path is <ab,v j >, that is:

[0159] dis(j)=min{dis(i)}(1.15)

[0160] (4) Update the starting point, and change v j As a new starting point, find the node v connected to it k , if dis(j)+α(j,k<)di(s), then update dis, that is:

[0161] dis(k)=α(j,k)+dis(k)(1.16)

[0162] (5) If the end point is found or all nodes have been traversed, the operation is terminated; otherwise, the process returns to (3).

[0163] The nodes in the target area are connected by roads, and the length of each road section is measured, considering that the weight is not only related to the road length, but also to the road speed limit (v i ), Lane congestion coefficient (s i ), the user's familiarity with the road (z i ). The road speed limit can be determined by the road type. The urban road grade is divided into four types: expressway, trunk road, secondary trunk road, and branch road. The lane congestion coefficient is also related to the road type. The following table shows the road speed limit and lane congestion coefficient of different road types.

[0164] Table 6 Speed ​​limits and congestion coefficients for different road types

[0165] Road Type Road speed limit (km / h) Lane congestion factor Expressway 80 1.3 Main Road 50 1.7 Secondary trunk road 40 1.5 Branch Road 30 1.6

[0166] Through the survey, it is known that the user's familiarity has little effect on the charging choice, so the familiarity is finely divided according to the different areas where the road is located. The higher the familiarity, the smaller the value. The residential area takes 0.85, the work area takes 0.9, the commercial / leisure area takes 0.95, and the outside of the community takes 1. The mathematical expression of the weight of the directed graph in the technical solution of the present invention is:

[0167]

[0168] Among them l iis the length of each road i. The optimal route for users to travel between destinations can be obtained through optimal path planning.

[0169] Since two charging stations have been built in the target area, located at node 1 and node 23 respectively, the optimal weights from these two nodes to other nodes simulated using the above method are shown in the following table. The optimal weights of newly added charging stations can also be simulated using the above method.

[0170] Table 7 Weights from target node to other nodes

[0171] Node Number Point 1 Point 2 Point 3 … Point 67 Point 68 Point 1 0 660 1089 … 3023 2959 Point 23 2122 1635 1278 … 2877 2999

[0172] 1.4 Charging load prediction model

[0173] The technical solution of the present invention uses the Monte Carlo simulation method to predict the charging load of electric vehicles.

[0174] The Monte Carlo method uses a large number of random sampling techniques to simulate the behavior of complex systems and obtain statistical results to solve problems.

[0175] The difficulty in predicting the charging load of electric vehicles lies in the fact that the initial charging time and the initial battery state of charge are difficult to predict, so the Monte Carlo method can be used for simulation. At the same time, this technical solution incorporates the user transfer probability within the community into the load simulation, making the prediction more consistent with the actual charging situation of users in the community.

[0176] Through the distribution of time feature quantities obtained by the previous fitting, we can get the user's assumption of the remaining power of the electric vehicle (battery state of charge). Through the investigation of the basic situation and charging habits of users, we can get the distribution of users' mileage. It is assumed that most users start charging after returning home, and a small number of users also charge during work or leisure and entertainment. Through the investigation of users' charging time, the full charging time of fast charging piles is basically less than two hours, so it is assumed that the charging time when the user uses a fast charging pile is from the start of charging of the electric vehicle to the full charging of the electric vehicle. The charging time of a slow charging pile is the smaller value between the time from the start of charging to the full charging and the stay time.

[0177] The specific charging load prediction calculation process is as follows: Figure 8 As shown:

[0178] 1) Input the simulated number of electric vehicles, the distribution of first-time vehicle use, the distribution of dwelling time, the distribution of driving mileage, the charging power of charging piles, the battery capacity of electric vehicles, the power consumption per kilometer and other data;

[0179] 2) Let vehicle number j = j + 1, extract the mileage distribution, and calculate the battery status of the electric vehicle at the beginning of charging;

[0180] 3) Take the vehicle usage time and the length of stay, calculate the charging time, and then determine the actual charging time based on the required charging power;

[0181] 4) Determine the system charging power at each time point and update the system load;

[0182] 5) Determine whether all electric vehicles have been calculated. If not, return to step 2). Otherwise, output the total system node load result.

[0183] The load simulation curve obtained by averaging 100 simulations is as follows: Figure 8 As shown, it can be seen from the simulation results:

[0184] (1) The charging load has two peaks at 10 a.m. and 10 p.m., and two troughs at 4 p.m. and 5 a.m.;

[0185] (2) The first increase in charging load starts at 8:30 a.m. This is because users enter the work area and start working. At this time, electric vehicles also gradually start to charge, and the load increases. After 10 a.m., electric vehicles finish charging and exit charging;

[0186] (3) The second increase in charging load begins after 6 p.m. This is because most people start to go home from get off work to charge their cars after 6 p.m., and after 10 p.m., more and more electric vehicles finish charging and stop charging, so the load gradually decreases.

[0187] 2. Optimal Distribution Method of Community Charging Pile under Time Sharing Mechanism

[0188] Based on the spatiotemporal distribution characteristics of charging demand obtained above, the optimal layout method of community charging piles is studied.

[0189] The optimization of charging piles mainly includes two parts. The first is to optimize the number of charging piles in the same charging station so that the charging needs of users can be met in most cases, and there will be no large number of charging piles left idle. The second is to optimize the number and location of charging stations. It is necessary to consider both the charging station and the user, and comprehensively consider the cost of the charging station and the user's charging cost to obtain the best location and number of charging stations.

[0190] 2.1 Objective Function and Basic Constraints

[0191] In order to take into account the costs of both the charging station and the user, the overall objective function F includes: the construction cost of the charging station (C c ), charging station operation and maintenance cost (C w), distribution network loss cost (C loss ), user’s annual charging time cost (C2), user’s annual charging electricity cost (C3).

[0192] F = α1·(C c +C w +C loss )+α2·C2+α3·C3(2.1)

[0193] Among them, the construction cost of the charging station mainly includes the fixed investment cost of the charging station and the construction cost of the charging pile, which is calculated by the following formula:

[0194]

[0195] Where N is the number of charging stations, C fix is the fixed investment cost of the charging station, c b,f and c b,s is the purchase cost of fast and slow charging piles, s i,f s i,s is the number of fast-charging and slow-charging charging piles at the i-th charging station, r0 is the discount rate, and z is the operating life.

[0196] The operation and maintenance costs of charging stations mainly include the maintenance and repair costs of charging station equipment and equipment depreciation costs, which are calculated using the following formula:

[0197]

[0198] Among them, α represents the labor cost ratio coefficient (12%), β represents the grid connection cost ratio coefficient (8%), T dp Indicates the average daily operation time of the charging pile in a day, P i,f P i,s is the fast charging and slow charging charging pile power of the i-th charging station.

[0199] The distribution network loss cost is as follows:

[0200]

[0201] where c e Indicates the unit price of electricity, Ω Represents the distribution network line set I ij Indicates the current flowing through the line, R ij Indicates the line resistance, which is calculated as 365 days * 24 hours in a year.

[0202] The annual charging cost of users mainly includes the time cost (C2) used for charging and the power loss cost (C3) of electric vehicles going to charging stations.

[0203]

[0204] where d j represents the distance from the electric vehicle to the charging station, v j represents the average speed of electric vehicles, W s represents the sum of the queuing time and charging time used by the user for charging, c t is the user's unit time cost. The power loss cost is calculated as follows:

[0205]

[0206] Among them, l represents the mileage of electric vehicles per unit of electricity, c p The unit price for charging an electric vehicle.

[0207] The constraints of the charging station are mainly as follows: The number of charging piles and access power constraints in the charging station are shown in equations (2.6) and (2.7):

[0208] N i,min <N i <N i,max (2.7)

[0209] P i <P max (2.8)

[0210] Among them, N i,min and N i,max are the minimum and maximum number of charging piles in the charging station, P max is the maximum power value that the node can provide. The distance constraint between charging stations is

[0211] D>D min (2.9)

[0212] d<d i,max (2.10)

[0213] Where D min In order to ensure economy, the minimum distance constraint between charging stations is set. i,max The maximum distance that a charging station can serve.

[0214] In order to determine the weights of user cost and charging station cost in the objective function, this technical solution uses the analytic hierarchy process (AHP) for decision making.

[0215] The analytic hierarchy process is a mathematical model for multi-criteria decision-making problems. It decomposes complex decision-making problems into sub-problems at multiple levels by hierarchical analysis, and then compares the sub-problems pairwise to establish the weight relationship between each level. The judgment matrix is ​​formed through comparison, and the weight of each specific factor is obtained from the judgment matrix.

[0216] The specific process of AHP is as follows: Fig.10 as shown in .

[0217] 1. Decompose the decision problem into a hierarchical structure, which mainly includes three levels: target level, criterion level, and solution level. For this technical solution, the hierarchical structure is as follows: Fig.11 As shown in the figure, the target layer is to select a suitable target charging station, the criterion layer is the components of the objective function, including the cost of the charging station, the user's time cost and the power loss cost, and the solution layer is all the charging stations in the target area;

[0218] 1) For the elements of the criterion layer, compare them two by two to judge their relative importance. The judgment matrix is ​​obtained through the comparison results. The judgment matrix is ​​shown in formula (2.11):

[0219]

[0220] The elements in A satisfy For each element a in the judgment matrix ij We use the method of scaling from 1 to 9 and its reciprocal as the comparison scale, as shown in the following table. The scale is based on its importance to the target layer. The larger the value, the higher the importance.

[0221] Table 8 Meaning of the scale of the judgment matrix

[0222] Scale meaning 1 The two elements are equally important. 3 Compared with the two elements, the former is slightly more important than the latter 5 Compared with the two elements, the former is obviously more important than the latter. 7 Compared with the two elements, the former is more important than the latter. 9 Compared with the two elements, the former is more important than the latter. 2，4，6，8 Indicates the middle value of the above adjacent judgments Countdown from 1 to 9 Indicates the importance of comparing the order of the corresponding two elements

[0223] 2) Perform hierarchical single sorting and consistency check. Use the square root method, equations (2.11) and (2.14) to perform hierarchical single sorting and calculate the weight matrix W of the corresponding elements of the criterion layer.

[0224]

[0225]

[0226] Then the maximum eigenvalue λ is calculated through the weight matrix and judgment matrix max :

[0227]

[0228] The maximum eigenvalue can be used for a one-time test. There are three test indicators: consistency index CI, random consistency index RI and consistency ratio CR. The parameters of RI are shown in the following table:

[0229]

[0230] Table 9 Random consistency index RI parameters

[0231] Matrix order n 1 2 3 4 5 6 7 8 9 10 11 12 RI 0 0 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.49 1.51 1.54

[0232] When CR<0.1, it indicates that the constructed judgment matrix passes the consistency test. If the test fails, it is necessary to return to check whether there are errors in the judgment matrix;

[0233] 3) For AHP with multiple levels, it is also necessary to conduct a total ranking and consistency between the levels.

[0234] Check,the weights of the relative importance of the calculation scheme layer to the,target layer are as follows;

[0235]

[0236] A one-time test is also performed. If CR<0.1, the one-time test passes, and the final weight obtained is the weight relationship between different variables.

[0237] 2.2. Determine the number of charging piles based on queuing theory

[0238] When an electric vehicle enters a charging station, if all charging piles are in use, the electric vehicle needs to queue. Generally, six operating indicators are used to characterize the quality of the queuing system. The average length of the queue L s (Vehicles in the system: vehicles being charged + vehicles waiting in line), average queue length L q (number of vehicles waiting in line), average waiting time W q (the average waiting time from when the vehicles start queuing to when they start charging), the average stay time W s (waiting time in line + charging time), busy period T b (the length of time the system charges continuously), service intensity ρ (the number of vehicles fully charged per unit time, reflecting the busyness of the charging station).

[0239] When the process of users arriving at the charging station is Poisson distributed, the time interval between users arriving at the charging station follows Poisson distribution, that is, the charging time of the charging pile follows negative exponential distribution.

[0240] The queue of electric vehicles arriving at the charging station meets the M / M / s / k model, that is, the process of electric vehicles arriving at the system conforms to the Poisson distribution with parameter λ, and the interval time to arrive at the charging station and the charging time of the charging pile conform to the negative exponential distribution with parameter μ, s is the number of charging piles, and k is the maximum number of parking spaces in the charging station.

[0241] At the same time, the user transfer probability is added to the queuing theory. Using the user's expected estimate of the number of vehicles in the queue, it is assumed that the probability of user transfer, that is, the probability of the user switching to another charging station for charging, has a certain relationship with the expected number of vehicles in the queue. When the team leader s≤n, it is assumed that the user's expected probability of entering the queue and the number of vehicles in the queue satisfy an exponential distribution, that is, the more vehicles in the queue, the more likely it is that the user will switch to another charging station for charging.

[0242] Since the system can only accommodate k electric vehicles in the queue state, the probability distribution of the queue leader N when the queue reaches a stable state is:

[0243]

[0244] Among them, formula (2.18) and formula (2.20) represent the service intensity ρ and service efficiency ρ of the charging pile respectively. s , where p0 represents the idle probability of the charging pile. Then the average queue time L can be calculated q , Average Captain L s Average length of stay s and the average waiting time W q .

[0245]

[0246] λ e =λ(1-P lost )(2.26)

[0247] Among them, λ e is the average number of electric vehicles entering the queuing system per unit time, P lost is the probability that the system will suffer losses due to customer departure. The average number of electric vehicles receiving charging services and the number of charging piles being charged in the system is s, which can be calculated using the following formula:

[0248]

[0249] Assuming that there are 20 charging piles in the charging station, the maximum number of vehicles that the charging station can accommodate is 50, and the average charging time of the vehicles is 2 hours, by inputting the vehicle arrival rate (the number of vehicles arriving at the charging station per hour), the average number of vehicles in the charging station, the average number of vehicles in the queue, and the average queuing time of vehicles under different arrival rates can be obtained as shown in the following table.

[0250] Table 10 Relationship between vehicle arrival rate and vehicle situation in charging station

[0251] Vehicle arrival rate Average number of vehicles in a charging station Average number of vehicles in queue Average queuing time of vehicles / h 5 10 0 0 6 12.04 0.04 0.01 7 14.24 0.24 0.03 8 17.36 1.36 0.17 9 27.16 9.16 1.02 10 50 20 2 11 50 20 2

[0252] From the data in the table above, we can see that when the vehicle arrival rate is less than or equal to 5 vehicles, there will be no vehicles queuing, and all arriving vehicles can be charged. When the number of arriving vehicles is 6 to 10, the vehicles need to queue up for charging, and the average queuing time gradually increases with the increase in the number of vehicles. When the number of vehicles is greater than or equal to 11, the charging station is saturated. Only after the current vehicle is fully charged and leaves, can subsequent vehicles park in the charging station and enter the waiting queue to start queuing for charging.

[0253] 2.3 Genetic Algorithm to Determine Charging Station Location

[0254] The key to optimizing the layout of electric vehicle charging stations is to determine the location of the charging stations. This technical solution optimizes the location of charging stations by improving genetic algorithms and Voronoi Diagram.

[0255] Genetic algorithm is a heuristic search and optimization technology. It simulates the process of biological evolution, natural selection, crossover and mutation in nature. Through continuous iteration, it enables the group to evolve to better and better areas in the search space. Generation after generation of continuous reproduction and evolution finally converges to a group of optimal individuals, thereby obtaining the optimal solution to the problem. It can provide better solutions to problems with a large number of parameters and more complex mathematical representations.

[0256] The Voronoi diagram is a graphical structure that divides space into different regions. For a given set of points, the Voronoi diagram region of each point contains all the points closest to that point in space. Such a partition describes the positional relationship in space and is widely used in computational graphics and geographic information systems.

[0257] For the optimization problem of electric vehicle charging stations, the specific process of genetic algorithm is as follows: Fig.12 shown.

[0258] (1) Initialize the population and create a variety of initial charging station locations and charging pile numbers. Each plan is an individual, and all individuals together constitute the initial population;

[0259] (2) Calculate the fitness of each individual, that is, calculate the minimum value of the charging station cost and the total cost of the user through the objective function, and use the fitness to characterize the quality of the individual. The individual with higher fitness is more likely to be retained and inherited;

[0260] (3) Selection operation: select individuals according to their fitness. Individuals with higher fitness have a greater probability of being selected.

[0261] (4) Crossover operation: Randomly select two of the selected individuals as parents, select half of the charging station locations from each of the two parents, and combine them into new individuals to simulate gene crossover;

[0262] (5) Mutation operation: Let the location of the charging station in the new individual have a certain random fluctuation, for example, it may be generated on its adjacent node to simulate gene mutation;

[0263] (6) Update the population, replacing individuals with lower fitness in the original population with newly generated individuals to form a new population;

[0264] (7) Return to (2) and repeat steps (2) to (6) until the convergence condition or the number of iterations is met. Finally, output the individual with the highest fitness in the population, output the location and number of charging stations within the individual, and make a Voronoi diagram for the individual to show the area responsible for each charging station.

[0265] Example:

[0266] like Fig.12 As shown in the figure, if the intersection of roads is regarded as a node and the lines between nodes are regarded as branches, there are 68 nodes and 106 branches in the target area, including 12 expressways, 15 trunk roads, 53 secondary trunk roads and 26 branches.

[0267] The boundary of the region is approximately a pentagonal road, and the area is about 19km2. There are residential areas such as residential communities, work areas such as software parks and industrial parks, and commercial and leisure areas such as shopping malls and squares in the region. The area outside the boundary of the region is considered outside the community.

[0268] Assuming that there are 100 electric vehicles in each residential area and office building complex in the region, the number of private electric vehicles in the region is about 5,500, and the average frequency of users using public charging piles is once a week, that is, the number of private vehicles using shared charging stations every day is 800. These vehicles are randomly distributed on the nodes according to the probability distribution results of the spatial characteristic quantities.

[0269] The status of existing charging stations in the area is shown in Table 11.

[0270] Table 11 Existing charging stations in the region

[0271]

[0272]

[0273] Charging station layout optimization results

[0274] Through the hierarchical analysis method, the judgment matrix constructed from the survey results is shown in the following table. The weights α1, α2, and α3 of the charging station cost, user time cost, and power loss cost are calculated to be 0.106, 0.633, and 0.261 respectively.

[0275] Table 12 Judgment matrix of three elements

[0276] element Charging station costs User time cost Power loss cost Charging station costs 1 1 / 5 1 / 3 User time cost 5 1 3 Power loss cost 3 1 / 3 1

[0277] Substitute the obtained weights into formula (2.1) to calculate the objective function. In Table 11, different numbers of charging stations or the same number of charging stations in different locations are randomly selected as the initial population.

[0278] After 20 iterations of the genetic algorithm, the results shown in the following table are obtained. The responsible areas of each charging station are drawn using the Voronoi diagram. Fig.13 shown.

[0279] Table 13 Charging station status after genetic algorithm optimization

[0280] Charging station node location Number of charging piles in the station Number of fast charging piles / unit Number of slow charging piles / unit Node 1 10 5 5 Node 3 17 13 4 Node 8 15 2 13 Node 15 20 8 12 Node 16 3 2 1 Node 18 39 3 36 Node 26 17 13 4 Node 31 14 11 3 Node 45 17 7 10 Node 53 6 5 1 Node 62 10 5 5 Node 63 15 2 13 Node 68 21 18 3

[0281] from Fig.13 It can be seen that:

[0282] (1) Each charging station has its own responsible area, i.e., its corresponding color. The cost for users at the boundary of the area to reach the charging stations in the two areas is basically the same. For the charging station at the boundary of the target area, only the part of the target area that it is responsible for is colored.

[0283] (2) At the same time, the genetic algorithm also optimizes the number of charging stations and the number of charging piles in each charging station, and obtains a more adaptable charging station layout and optimization results. Compared with Table 11, it can be seen that when the number of charging stations in the community is 13, it is the optimal solution. That is, compared with the existing layout, at least one charging station needs to be added at the location of node 45.

[0284] (3) In terms of the number of charging piles in the charging stations, six charging stations need to increase the number of charging piles. These six charging stations are located in areas with large traffic flow. Increasing the number of charging piles can reduce queuing time and improve user satisfaction. Four charging stations need to reduce the number of fast charging piles. The idle rate of charging piles in these charging stations is relatively high, and the price of fast charging piles is relatively expensive, which will cause unnecessary investment and economic losses. Reducing charging piles can reduce the construction and maintenance costs of charging stations. Two charging stations need to reduce the number of slow charging piles and increase the number of fast charging piles in order to balance economic benefits and user costs.

[0285] Comparison with other methods

[0286] The technical solution of the present invention also compares the existing charging station situation, the charging station planning situation that only takes into account the vehicle flow, and the charging station planning situation after the transfer is taken into account in terms of charging station construction cost, average user waiting time, community load peak-to-valley difference, and total charging station power, as shown in the following table.

[0287] Table 14 Comparison of the technical solution method of the present invention with other methods

[0288]

[0289] It can be seen that due to the insufficient number of existing charging stations in the community, new charging stations need to be built, so the total construction cost of charging stations will be higher than the construction cost of existing charging stations. From the existing charging stations, there are problems such as the average waiting time for users is too long, the idle rate of some charging stations is high, and the number of charging piles in some charging stations is insufficient.

[0290] If the distribution of charging stations is planned by taking only vehicle flow into account, the average waiting time for users will be shortened, but the peak-to-valley difference in load within the community will increase, which is not conducive to the safe and stable operation of the system.

[0291] By using the technical solution of the present invention and the charging pile layout method taking transfer probability into account, when the total power of the charging station increases, the total construction cost of the charging piles is 288,500 yuan lower than the cost of only taking into account the traffic flow due to reasonable planning of user charging. At the same time, the average waiting time of users is reduced by 20.0% compared with the existing charging stations, and the peak-to-valley difference of the load of the entire community is reduced by 9.4%.

[0292] Fig.14 The load conditions in the community are shown under two methods: one based on traffic flow and the other taking transfer probability into account. It can be seen that when analyzing only based on traffic flow, the peak load in the community is higher and the valley load is lower, and the peak-to-valley difference is greater than the method taking transfer probability into account.

[0293] The technical solution of the present invention has practical significance and effect in maintaining the safe and stable operation of the power system, reducing construction costs, and reducing the average queuing time of users:

[0294] 1. Compared with the current method of simply distributing points based on vehicle flow adopted by power supply companies, the method proposed by the technical solution of the present invention, although increasing the access capacity of charging stations and the corresponding investment in charging piles, can reduce the daily peak-to-valley difference of charging load by 9.4% and shorten the average queuing time of users by 4.64 minutes (a decrease of 20.0%). In addition, if the benefits of reduced capacity of distribution transformers and less network loss are taken into account, the method proposed by the technical solution of the present invention can still achieve certain economic benefits while improving customer satisfaction compared with the existing method based on vehicle flow;

[0295] 2. Compared with the optimization method that does not take into account the sharing mechanism and the possibility of vehicle transfer, the optimization method for sharing charging piles proposed in the technical solution of the present invention can reduce the charging pile access capacity by 2.1% and the charging station construction cost by 289,000 yuan (a decrease of 2.4%) while meeting the community charging needs, reduce the peak-to-valley difference of charging load by 17.3%, and shorten the average user queuing time by 1.7 minutes (a decrease of 8.4%).

[0296] 3. The community is part of a certain new village, covering an area of ​​about 1.2 square kilometers. The current total capacity of charging pile access is 2495kW, and the total area of ​​the community within the jurisdiction of a certain power supply company is about 320 square kilometers. If deduced based on the area, the charging pile layout method proposed by the technical solution of the present invention can save 76.93 million yuan in charging pile investment compared with the optimization layout method that does not take into account the sharing mechanism and the possibility of vehicle transfer.

[0297] The present invention can be widely used in the field of site design or layout optimization of community shared smart charging piles.

Claims

1. A method for optimally distributing community charging piles under a time-sharing sharing mechanism, including mastering the spatiotemporal distribution characteristics of community charging demand, determining the optimal path for users to reach the charging station from their location, and using the Monte Carlo simulation method to predict the charging load of electric vehicles by fitting the distribution of time characteristic quantities; its characteristics are The prediction of electric vehicle charging load specifically includes the following steps: 1) Input data including the number of simulated electric vehicles, the distribution of first-time vehicle use, the distribution of dwelling time, the distribution of driving mileage, the charging power of charging piles, the battery capacity of electric vehicles, and the power consumption per kilometer; 2) Let vehicle number j = j + 1, extract the mileage distribution, and calculate the battery status of the electric vehicle at the beginning of charging; 3) Take the vehicle usage time and the length of stay, calculate the charging time, and then determine the actual charging time based on the required charging power; 4) Determine the system charging power at each time point and update the system load; 5) Determine whether all electric vehicles have been calculated; if not, return to step 2); if so, output the total system node load result.

2. The optimal distribution method of community charging piles under the time-sharing sharing mechanism according to claim 1 is characterized by: The optimal distribution method of community charging piles under the time-sharing sharing mechanism takes into account both the charging station and the user, optimizes the number of charging piles in the same charging station, optimizes the number and location of charging stations, and comprehensively considers the cost of the charging station and the charging cost of the user to obtain the optimal location and number of charging stations.

3. The optimal distribution method of community charging piles under the time-sharing sharing mechanism according to claim 1 is characterized in that The optimal distribution method of community charging piles under the time-sharing sharing mechanism incorporates the user transfer probability within the community into the load simulation, making the prediction more consistent with the actual charging situation of users in the community.

4. The optimal distribution method of community charging piles under the time-sharing sharing mechanism according to claim 1 is characterized in that The optimal distribution method of community charging piles under the time-sharing sharing mechanism is to obtain the user's assumed remaining electric power of the electric vehicle by fitting the distribution of time feature quantities; to obtain the distribution of the user's mileage by investigating the user's basic situation and charging habits; and to assume that the charging time when the user uses a fast charging pile is: the electric vehicle starts charging until the electric vehicle is fully charged and stops charging; the charging time when using a slow charging pile is: the smaller value between the time from the start of charging to full charging and the stay time.

5. The optimal distribution method of community charging piles under the time-sharing sharing mechanism according to claim 1 is characterized in that The optimal layout of charging piles includes two parts: one is to optimize the number of charging piles in the same charging station so that the charging needs of users can be met in most cases, and at the same time, there will be no large number of idle charging piles; the second is to optimize the number and location of charging stations, considering both the charging station and the user, and comprehensively considering the cost of the charging station and the charging cost of the user to obtain the best location and number of charging stations.

6. The optimal distribution method of community charging piles under the time-sharing sharing mechanism according to claim 1 is characterized by: The optimal distribution method of community charging piles under the time-sharing sharing mechanism adopts queuing theory to determine the number of charging piles; that is, the process of electric vehicles arriving at the system conforms to the Poisson distribution with parameter λ, and the interval time to arrive at the charging station and the charging time of the charging pile conform to the negative exponential distribution with parameter μ, s is the number of charging piles, and k is the maximum number of parking spaces in the charging station.

7. The optimal distribution method of community charging piles under the time-sharing sharing mechanism according to claim 1 is characterized by: The optimal distribution method of community charging piles under the time-sharing sharing mechanism adopts a genetic algorithm to determine the location of the charging station, and distributes and optimizes the location of the charging station by improving the genetic algorithm and the Voronoi diagram and taking the transfer probability into account. While meeting the charging needs of the community, the access capacity of the charging piles is reduced, the construction cost of the charging station is reduced, the daily peak-to-valley difference of the charging load is reduced, and the average waiting time of users is shortened.

8. The optimal distribution method of community charging piles under the time-sharing sharing mechanism according to claim 1 is characterized by: The optimal distribution method of community charging piles under the time-sharing sharing mechanism uses the hierarchical analysis method to make decisions in order to determine the weights of user cost and charging station cost in the objective function.

9. The optimal distribution method of community charging piles under the time-sharing sharing mechanism according to claim 1 is characterized by: The AHP method decomposes the decision problem into a hierarchical structure, which mainly includes three levels: target level, criterion level, and solution level; Among them, the target layer is to select a suitable target charging station; The criterion layer is the component of the objective function, including the cost of the charging station, the user's time cost and the power loss cost; The solution layer is all charging stations in the target area.

10. The optimal distribution method of community charging piles under the time-sharing sharing mechanism according to claim 9 is characterized in that For the elements of the criterion layer, compare them two by two to judge their relative importance, and obtain a judgment matrix through the comparison results; scale them according to their importance to the target layer, and the larger the value, the higher the importance; Perform hierarchical single sorting and consistency check, perform hierarchical single sorting by square root method, and calculate the weight matrix W of the corresponding elements of the criterion layer; Calculate the maximum eigenvalue through the weight matrix and the judgment matrix; For the hierarchical analysis method with multiple levels, it is also necessary to perform a total hierarchical ranking and consistency check between the levels and calculate the weight of the relative importance of the solution level to the target level.

Images