Electric vehicle clustering method and system considering driving characteristics and response intentions

Abstract

The invention discloses an electric vehicle clustering method and system considering driving characteristics and response intentions, and mainly relates to the technical field of electric vehicle data analysis. The method comprises the following steps: collecting electric vehicle operation data and charging and discharging data, and constructing an electric vehicle charging and discharging model; according to the collected electric vehicle operation data and charging and discharging data, an AP model considering the response will and driving characteristics of an electric vehicle owner is constructed; based on a gradient descent method, performing algorithm correction on the AP model in the step S2; clustering the cluster centers obtained in the corrected AP algorithm based on a K-means algorithm; and performing effect evaluation on the clustering in the step S4 based on a Silhouette index. The method has the advantages that the utilization efficiency of charging infrastructures can be improved, impact on a power grid is reduced, and efficient management of an electric vehicle group is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle data analysis, and in particular to an electric vehicle clustering method and system that considers the driving characteristics and response willingness of large-scale electric vehicles. Background Art

[0002] With the popularity of electric vehicles and the growth of charging demand, the impact of electric vehicles on the power grid is becoming increasingly significant. In the operation scenario of large-scale electric vehicles, how to effectively manage charging load and optimize power resource scheduling has become an urgent problem to be solved.

[0003] Traditional charging scheduling strategies are mostly based on static models, which are difficult to adapt to the uncertainty of electric vehicle driving characteristics and the differences in individual owners' response willingness. Electric vehicles have a high degree of driving flexibility and uncertainty, and their charging behavior is affected by multiple factors such as the owners' travel patterns, charging costs, and willingness to respond to changes in electricity prices.

[0004] In order to solve this problem, it is particularly important to propose a clustering method that considers the driving characteristics and response willingness of electric vehicles. Summary of the invention

[0005] The purpose of the present invention is to provide an electric vehicle clustering method and system that takes into account driving characteristics and response willingness, which can improve the utilization efficiency of charging infrastructure, while reducing the impact on the power grid and achieving efficient management of electric vehicle groups.

[0006] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0007] On the one hand, a clustering method of electric vehicles considering driving characteristics and response willingness is provided, comprising the following steps:

[0008] S1: Collect electric vehicle operation data and charging and discharging data and build an electric vehicle charging and discharging model;

[0009] S2: Based on the collected electric vehicle operation data and charging and discharging data, an AP model is constructed that considers the response willingness and driving characteristics of electric vehicle owners;

[0010] S3: Based on the gradient descent method, the AP model in step S2 is modified by algorithm;

[0011] S4: clustering the cluster centers obtained in the modified AP algorithm based on the K-means algorithm;

[0012] S5: Evaluate the clustering effect in step S4 based on the Silhouette indicator.

[0013] Preferably, the step S1 is specifically as follows: the electric vehicle charging and discharging model is represented by the remaining power of the battery of the electric vehicle SoC, and the remaining power of the mth electric vehicle at the starting time t in a certain period of time is SoC m，t , then the remaining power at the start of the next period is SoC m，t+1 , the calculation formula is:

[0014]

[0015] Among them, η m，ch and η m，dch are the charging and discharging efficiencies of the mth electric vehicle, P m，ch and P m，dch are the charging and discharging power of the mth electric vehicle, E C is the average capacity of electric vehicle batteries;

[0016] The charging power and remaining power of electric vehicles SoC meet the following conditions:

[0017]

[0018] in, and are the maximum and minimum charging power of electric vehicles, and They are respectively the upper and lower limits of the SoC of the remaining power of electric vehicles.

[0019] Preferably, the step S2 is specifically:

[0020] The cruising range of electric vehicles that need to be aggregated, the owner's response willingness, and the real-time remaining power of all electric vehicles SoC are collected. The cruising range is recorded as {L1, L2, ..., L m}, the timing SoC of the i-th electric vehicle is recorded as {SoC i，1 , SoC i，2 ,…,SoC i，m}, the calculation formula for the electric vehicle owner's response charging / discharging willingness index is:

[0021] w i，t =w i，or +SoC i，t

[0022] w i , or ∈[0, 1]

[0023] w i，t ={w i，1 , w i，2 ,…,w i，N}

[0024] Set L i is the cruising range of the i-th electric car, and L is set max The maximum range of electric vehicles participating in the statistics;

[0025] Among them, w i，t represents the real-time response charging / discharging willingness index of the i-th electric vehicle, w or represents the charging / discharging willingness of electric vehicle owners, w i Represents the response charge / discharge willingness index of the i-th electric vehicle during the collection time.

[0026] Preferably, the cruising range L of the electric vehicle and the response charge / discharge willingness index w of the electric vehicle during the collection time i As a key variable for calculating cluster centers, the AP model construction method includes:

[0027] S21: Calculate the similarity between the real-time remaining power SoC of all electric vehicles and the charging / discharging willingness index of electric vehicle owners according to the cruising range of the electric vehicle and the charging / discharging willingness index of electric vehicle owners, and use the result of the similarity measurement as a set of weight matrices. The calculation of the similarity between the real-time remaining power SoC of all electric vehicles and the charging / discharging willingness of electric vehicle owners is specifically:

[0028] s L (i, k) = -||L i -L k || 2

[0029] s w (i, k) = -||w i -w k || 2

[0030] Among them, s L (i, k) represents L i With L k The similarity between w (i, k) represents w i With w k The similarity between

[0031] S22: Set initial responsibilities and availability based on the similarity matrix between electric vehicles:

[0032] Let the initialization responsibility and availability between the i-th and k-th electric vehicles be r and a respectively:

[0033] r x (i, k) = 0

[0034] a x (i, k) = 0

[0035] r w (i, k) = 0

[0036] a w (i, k) = 0;

[0037] S23: r(i, k) represents the fitness of the i-th electric car selecting the k-th electric car as the cluster center:

[0038]

[0039] The above formula represents the fitness of data point i evaluating k relative to all other possible cluster centers, where 0≤λ≤1;

[0040] S24: Availability calculation for cluster center:

[0041]

[0042]

[0043]

[0044]

[0045] S25: According to step S21 and step S22, the responsibility and availability matrices are updated alternately until convergence reaches the maximum number of iterations, and the cluster center is determined:

[0046] k = arg max k {a x (i,k)+r x (i,k)+a w (i,k)+r w (i, k)}.

[0047] Preferably, the step S3 comprises:

[0048] S31: Determine the driving characteristics of the electric vehicle and the evaluation function of the owner's response willingness;

[0049] S32: Use gradient descent to find weights in combination with fuzzy information entropy;

[0050] S33: construct an algorithm for the modified AP model.

[0051] Preferably, the step S31 is specifically:

[0052] The n-dimensional driving characteristics and response willingness of the i-th electric vehicle are represented by w i and L iIndicates that, let its corresponding weight vector be v i , v i =(v i1 , v i2 ), the weighted Euclidean distance is

[0053]

[0054] Select fuzzy information entropy for the driving characteristics of electric vehicles and the evaluation function of the owner's response willingness:

[0055]

[0056]

[0057] Preferably, the step S32 includes:

[0058] Initialize the weight vector:

[0059] right and Δv t To solve;

[0060] like but

[0061] Continue to iterate and calculate w i，or ∈[0, 1] and w i，t ={w i，1 , w i，2 ,…,w i，N}Until the iteration reaches the preset number of times or convergence;

[0062] Continue to iterate and calculate w i，or ∈[0,1], w i，t ={w i，1 , w i，2 ,…,w i，N}、s L (i, k) = -||L i -L k || 2 , find the total weight of all electric vehicles involved in the statistics;

[0063] v after iterative calculation i , v i =(v i1 , v i2 ), v i1 represents the weight of the driving characteristics of the i-th electric vehicle after being corrected by the gradient descent method, v i2 represents the weight of the response willingness of the i-th electric vehicle after being corrected by the gradient descent method, v iAs the weight parameter of the driving characteristics of electric vehicles and their owners' response willingness.

[0064] On the other hand, a system is provided based on the above-mentioned electric vehicle clustering method considering driving characteristics and response willingness, comprising:

[0065] Data collection and automobile charging and discharging model building module, used to: collect electric vehicle operation data and charging and discharging data and build an electric vehicle charging and discharging model;

[0066] The clustering model building module is used to: build an AP model that takes into account the response willingness and driving characteristics of electric vehicle owners based on the collected electric vehicle operation data and charging and discharging data; perform algorithm correction on the AP model in step S2 based on the gradient descent method; and cluster the cluster centers obtained in the modified AP algorithm based on the K-means algorithm;

[0067] Clustering effect evaluation module, used to evaluate the effect of clustering in step S4 based on the Silhouette index

[0068] Compared with the prior art, the beneficial effects of the present invention are:

[0069] 1. Taking full advantage of the differences in car owners’ willingness to charge and their responses to dynamic electricity prices, large-scale electric vehicles will be aggregated to interact with the power grid;

[0070] 2. By building multi-dimensional data models such as driving mode, dwelling time, and charging behavior, electric vehicles with similar characteristics can be classified to achieve refined charging scheduling strategies;

[0071] 3. On the basis of fully considering the differences among different types of large-scale electric vehicles, it can improve the accuracy of grid load management, improve the utilization efficiency of charging infrastructure, optimize the charging experience of car owners, and promote efficient coordination between smart grids and electric vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 is a flow chart of the method of the present invention;

[0073] Figure 2 It is a schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0074] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the application equally.

[0075] In the present invention, terms such as "upper", "lower", "left", "right", "front", "back", "vertical", "horizontal", "side", "bottom", etc. indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings. They are relational words determined only for the convenience of describing the structural relationships of the various parts or elements of the present invention, and do not specifically refer to any part or element in the present invention and should not be understood as limitations on the present invention.

[0076] In the present invention, terms such as "fixed connection", "connected", "connection", etc. should be understood in a broad sense, indicating that it can be fixedly connected, integrally connected or detachably connected; it can be directly connected or indirectly connected through an intermediate medium. Relevant scientific research or technical personnel in this field can determine the specific meanings of the above terms in the present invention according to specific circumstances, and they should not be understood as limiting the present invention.

[0077] Example:

[0078] like Figure 1 As shown, this embodiment provides an electric vehicle clustering method considering driving characteristics and response willingness, comprising the following steps:

[0079] S1: Collect electric vehicle operation data and charging and discharging data and build an electric vehicle charging and discharging model;

[0080] S2: Based on the collected electric vehicle operation data and charging and discharging data, an AP model is constructed that considers the response willingness and driving characteristics of electric vehicle owners;

[0081] S3: Based on the gradient descent method, the AP model in step S2 is modified by algorithm;

[0082] S4: clustering the cluster centers obtained in the modified AP algorithm based on the K-means algorithm;

[0083] S5: Evaluate the clustering effect in step S4 based on the Silhouette indicator.

[0084] Among them, the electric vehicle charging and discharging model is constructed as follows:

[0085] In step S1, the battery of the electric vehicle provides power for it. The electric vehicle can be regarded as a mobile energy storage device. The remaining power of the battery is usually represented by SoC. The remaining power of the mth electric vehicle at the start time t in a certain period of time is SoC m，t , then the remaining power at the start of the next period is SoC m，t+1 :

[0086]

[0087] Among them, η m，ch and ηm，dch are the charging and discharging efficiencies of the mth electric vehicle, P m，ch and P m，dch are the charging and discharging power of the mth electric vehicle, E C is the average capacity of electric vehicle batteries. In order to ensure the service life of the battery, the charging power and SoC of the electric vehicle must meet the following constraints:

[0088]

[0089] in, and are the maximum and minimum charging power of electric vehicles, and They are respectively the upper and lower limits of the SoC of the remaining power of electric vehicles.

[0090] The specific implementation of step S2 includes:

[0091] Collect the mileage of electric vehicles that need to be aggregated {L1, L2, ..., L m}, responsiveness and real-time SoC for all electric vehicles, {soC i，1 , SoC i，2 ,…,SoC i，m},{SoC i，1 , SoC i，2 ,…,SoC i，m} represents the timing SoC of the i-th electric vehicle. Assuming that there are N electric vehicles participating in the aggregation of electric vehicles, the owner's response to charging and discharging willingness index is:

[0092] w i，t =w i，or +SoC i，t

[0093] w i，or ∈[0, 1]

[0094] w i，t ={w i，1 , w i，2 ,…,w i，N}

[0095] L i represents the cruising range of the i-th electric vehicle, L max Represents the maximum range of the electric vehicles involved in the statistics; L i As a key factor in whether electric vehicles respond to dispatch and whether electric vehicle owners respond to dispatch, the longer the mileage, the more willing the owners are to respond to dispatch. Therefore, the mileage can be used as a driving characteristic to participate in the selection of cluster centers. i，trepresents the real-time response charging / discharging willingness index of the i-th electric vehicle, w or represents the charging / discharging willingness of electric vehicle owners, w i Represents the response charging / discharging willingness index of the i-th electric vehicle within the collection time;

[0096] Affinity Propagation (AP) is adopted to consider the response willingness and driving characteristics of electric vehicle owners, where the cruising range L and the response charging / discharging willingness index w i As the key variable for calculating cluster centers, the steps of the AP algorithm are as follows:

[0097] First, when calculating the similarity, we combine the mileage of electric vehicles and the owner's response willingness index, select a suitable distance measurement method, and calculate the real-time status of all electric vehicles (including SoC and charging and discharging willingness). The result of the similarity measurement is a set of weight matrices.

[0098] These similarity matrices are then used in the subsequent clustering process to determine the degree of membership of each EV in different cluster centers;

[0099] Next, the AP algorithm optimizes the selection of clusters by iteratively updating responsibility and availability. In the initialization phase, the initial responsibility and availability are set according to the similarity matrix between electric vehicles. During the responsibility update process, each data point calculates the fitness of selecting a cluster center. Finally, the cluster center is adjusted through availability update to ensure that the algorithm converges to a suitable cluster allocation.

[0100] a) Calculate the similarity between the real-time SoC of all electric vehicles and the charging / discharging intentions of the electric vehicle owners. In this embodiment, negative Euclidean distance is used to represent the similarity (the smaller the distance, the higher the similarity):

[0101] s L (i, k) = -||L i -L k || 2

[0102] s w (i, k) = -||w i -w k || 2

[0103] s L (i, k) represents L i With L k The similarity between w (i, k) represents w i With w k The similarity between

[0104] b) Initialization responsibility and availability: Let r and a represent the initialization responsibility and availability between the i-th and k-th electric vehicles respectively:

[0105] r x (i, k) = 0

[0106] a x (i, k) = 0

[0107] r w (i, k) = 0

[0108] a w (i, k) = 0

[0109] c) Responsibility update: r(i, k) represents the fitness of the i-th electric car selecting the k-th electric car as the cluster center:

[0110] r(i, k) =

[0111] λ{s x (i, k)-max j≠k {a x (i,j)+s x (i, j)}}+

[0112] (1-λ){s w (i, k)-max j≠k {a w (i,j)+s w (i, j)}}

[0113] The above formula represents the fitness of data point i evaluating k relative to all other possible cluster centers, where 0≤λ≤1;

[0114] d) Availability Updates:

[0115]

[0116]

[0117]

[0118]

[0119] The availability of the cluster center itself is calculated using the above formula;

[0120] e) Iterative information transfer:

[0121] According to a) and b) , the responsibility and availability matrices are updated alternately until convergence or the maximum number of iterations is reached;

[0122] f) Determine the cluster center:

[0123] k = arg max k {a x (i,k)+r x (i,k)+a w (i,k)+r w (i, k)}.

[0124] By combining the charging needs of electric vehicles with the response willingness of car owners, the AP algorithm can select cluster centers more accurately, ensuring that the algorithm can adapt to different electric vehicle characteristics and adaptively adjust cluster centers when faced with complex electric vehicle behavior data.

[0125] In the above clustering process, there is a problem that it is difficult to determine the weights of the two parameters, namely, the driving characteristics of electric vehicles and the response willingness of car owners. If the traditional Euclidean distance is used, it means that the contribution of these two parameters in the process of large-scale electric vehicle clustering is the same. However, in reality, due to the differences in the driving characteristics of each electric vehicle and the response willingness of the owners, the two types of parameters will inevitably have different degrees of influence on the aggregation of electric vehicles.

[0126] This embodiment uses the gradient descent method to solve the two parameters of the electric vehicle driving characteristics and the owner's response willingness, and further combines this weight with the Euclidean distance to construct a weighted Euclidean distance. The weighted Euclidean distance not only takes into account the driving characteristics of the electric vehicle, but also dynamically adjusts the weight of the owner's response willingness to ensure that these two factors can be reasonably weighted in different situations. When facing large-scale electric vehicle data, it can better reflect the differences in characteristics of different owners and vehicles, thereby optimizing the clustering results. The specific content of step S3 is as follows:

[0127] The n-dimensional driving characteristics and response willingness of the i-th electric vehicle are represented by w i and L i Indicates that, let its corresponding weight vector be v i , v i =(v i1 , v i2 ), the weighted Euclidean distance is

[0128]

[0129] Fuzzy information entropy is often used in fuzzy clustering and decision analysis. Fuzzy information entropy is selected for the driving characteristics of electric vehicles and the evaluation function of the owner's response willingness:

[0130]

[0131]

[0132] The specific steps of using gradient descent to find weights in combination with fuzzy information entropy are as follows:

[0133] (1) Initialize the weight vector:

[0134] (2) Yes and Δv t To solve;

[0135] (3) If but

[0136] (4) Continue iterating to calculate w i，or ∈[0, 1] and w i，t ={w i，1 , w i，2 ,…,w i，N}Until the iteration reaches the preset number of times or convergence;

[0137] (5) Continue iterating to calculate w i，or ∈[0,1], w i，t ={w i，1 , w i，2 ,…,w i，N}、s L (i, k) = -||L i -L k || 2 , find the total weight of all electric vehicles involved in the statistics;

[0138] v after iterative calculation i , v i =(v i1 , v i2 ), v i1 represents the weight of the driving characteristics of the i-th electric vehicle after being corrected by the gradient descent method, v i2 represents the weight of the response willingness of the i-th electric vehicle after being corrected by the gradient descent method, v i As the weight parameter of the driving characteristics of electric vehicles and their owners' response willingness.

[0139] Modified AP algorithm:

[0140] Calculate all electric vehicle driving characteristics L and electric vehicle owners' charging / discharging willingness w i Similarity:

[0141] S L (i, k) = -v1||L i -L k || 2

[0142] S w (i, k) = -v2||w i -wk || 2

[0143] Initialization responsibility and availability: Let r′ and a′ represent the initialization responsibility and availability between the i-th and k-th electric vehicles respectively:

[0144] r x (i, k) = 0

[0145] a′ x (i, k) = 0

[0146] r′ w (i, k) = 0

[0147] a′ w (i, k) = 0

[0148] r′(i, k) represents the fitness of the i-th electric car selecting the k-th electric car as the cluster center:

[0149]

[0150] The above formula represents the fitness of data point i evaluating k relative to all other possible cluster centers, where 0≤λ≤1;

[0151] Availability Update:

[0152]

[0153]

[0154] Iterate message passing and determine cluster centers:

[0155] Alternately update the responsibility and availability matrices according to the first two steps until convergence or the maximum number of iterations is reached:

[0156] k=arg max k {a′ x (i,k)+r x ′(i, k)+a′ w (i,k)+r′ w (i, k)}.

[0157] Step S4 is specifically:

[0158] By improving the AP algorithm, a set of initial cluster centers are solved, and K-means is further optimized based on these centers. The AP algorithm determines K cluster centers c1, c2, ..., c k , these cluster centers are geometrically 2-dimensional;

[0159] Next, based on the initial cluster center determined by the AP algorithm, the K-means algorithm is used to iteratively optimize the data: First, the Euclidean distance between each data point and all cluster centers is calculated, and each data point is assigned to the nearest cluster center; then, the center of each cluster is recalculated, that is, the mean of all data points in the cluster, as the new cluster center; this process is iterated continuously until the change in the cluster center is less than the preset threshold, indicating that the algorithm has converged:

[0160] Electric Vehicle Data Point Distribution:

[0161] By calculating the Euclidean distance between each data point and each cluster center, the data point y i =(x i , w i ) and the cluster centers are assigned:

[0162] C i =arg min||y i -c j || 2

[0163] C i Represents data point y i The cluster to which it belongs;

[0164] Update cluster center:

[0165] After completing the data point assignment, we need to update each cluster center. The new cluster center is the mean of all data points assigned to the cluster:

[0166]

[0167] Among them, N j represents the number of data points belonging to cluster j, It means summing all the data points in the cluster and repeating the above steps until the cluster center no longer changes significantly. The change in the cluster center is used to determine whether convergence can be stopped:

[0168]

[0169] Step S5, clustering effect evaluation based on Silhouette index:

[0170] This embodiment uses the Silhouette index to evaluate the clustering effect. It considers measuring the similarity between a data point and its cluster and the similarity with other clusters to judge the clustering effect. The value range of the Silhouette index is (-1, 1). The larger the value, the better the clustering effect.

[0171]

[0172] a(i) represents the average distance between the ith electric vehicle and other electric vehicles of the same type, and b(i) represents the minimum value of the average distance between the ith electric vehicle and other types of electric vehicles.

[0173] like Figure 2 As shown, this embodiment also provides a system based on the above-mentioned electric vehicle clustering method considering driving characteristics and response willingness, including:

[0174] Data collection and automobile charging and discharging model building module, used to: collect electric vehicle operation data and charging and discharging data and build an electric vehicle charging and discharging model;

[0175] The clustering model building module is used to: build an AP model that takes into account the response willingness and driving characteristics of electric vehicle owners based on the collected electric vehicle operation data and charging and discharging data; perform algorithm correction on the AP model in step S2 based on the gradient descent method; and cluster the cluster centers obtained in the modified AP algorithm based on the K-means algorithm;

[0176] The clustering effect evaluation module is used to evaluate the effect of the clustering in step S4 based on the Silhouette index.

[0177] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the described embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. An electric vehicle clustering method considering driving characteristics and response willingness, characterized in that: The following steps are involved: S1: Collect electric vehicle operation data and charging and discharging data and build an electric vehicle charging and discharging model; S2: Based on the collected electric vehicle operation data and charging and discharging data, an AP model is constructed that considers the response willingness and driving characteristics of electric vehicle owners; S3: Based on the gradient descent method, the AP model in step S2 is modified by algorithm; S4: clustering the cluster centers obtained in the modified AP algorithm based on the K-means algorithm; S5: Evaluate the clustering effect in step S4 based on the Silhouette indicator.

2. The electric vehicle clustering method considering driving characteristics and response willingness according to claim 1, characterized in that: The step S1 is specifically as follows: the electric vehicle charging and discharging model is represented by the remaining power of the battery of the electric vehicle, SoC, and the remaining power of the mth electric vehicle at the start time t in a certain period of time is SoC m，t , then the remaining power at the start of the next period is SOC m，t+1 , the calculation formula is: Among them, η m，ch and η m，dch are the charging and discharging efficiencies of the mth electric vehicle, P m，ch and P m，dch are the charging and discharging power of the mth electric vehicle, E C is the average capacity of electric vehicle batteries; The charging power and remaining power of electric vehicles SoC meet the following conditions: in, and are the maximum and minimum charging power of electric vehicles, and They are respectively the upper and lower limits of the SoC of the remaining power of electric vehicles.

3. The electric vehicle clustering method considering driving characteristics and response willingness according to claim 1, characterized in that: The step S2 is specifically: The cruising range of electric vehicles that need to be aggregated, the owner's response willingness, and the real-time remaining power of all electric vehicles SoC are collected. The cruising range is recorded as {L1, L2, ..., L m }, the timing SoC of the i-th electric vehicle is recorded as {SoC i，1 , SoC i，2 ,…,SoC i，m }, the calculation formula for the electric vehicle owner's response charging / discharging willingness index is: w i，t =w i，or +SoC i，t w i，or ∈[0，1] In i，t ={in i，1 ,In i，2 ,…,In i，N } Set L i is the cruising range of the i-th electric car, and L is set max The maximum range of electric vehicles participating in the statistics; Among them, w i，t represents the real-time response charging / discharging willingness index of the i-th electric vehicle, w or represents the charging / discharging willingness of electric vehicle owners, w i Represents the response charge / discharge willingness index of the i-th electric vehicle during the collection time.

4. The electric vehicle clustering method considering driving characteristics and response willingness according to claim 3 is characterized in that: The cruising range L of the electric vehicle and the response charge / discharge willingness index w of the electric vehicle during the collection time i As a key variable for calculating cluster centers, the AP model construction method includes: S21: Calculate the similarity between the real-time remaining power SoC of all electric vehicles and the charging / discharging willingness index of electric vehicle owners according to the cruising range of the electric vehicle and the charging / discharging willingness index of the electric vehicle owners, and use the result of the similarity measurement as a set of weight matrices. The calculation of the similarity between the real-time remaining power SoC of all electric vehicles and the charging / discharging willingness of the electric vehicle owners is specifically: s L (i,k)=-‖L i -L k ‖ 2 s w (i,k)=-‖w i -w k ‖ 2 Among them, s L (i, k) represents L i With L k The similarity between w (i, k) represents w i With w k The similarity between S22: Set initial responsibilities and availability based on the similarity matrix between electric vehicles: Let the initialization responsibility and availability between the i-th and k-th electric vehicles be r and a respectively: r x (i,k)=0 a x (i,k)=0 r w (i,k)=0 a w (i,k)=0; S23: r(i, k) represents the fitness of the i-th electric car selecting the k-th electric car as the cluster center: The above formula represents the fitness of data point i evaluating k relative to all other possible cluster centers, where 0≤λ≤1; S24: Availability calculation for cluster center: S25: According to step S21 and step S22, the responsibility and availability matrices are updated alternately until convergence reaches the maximum number of iterations, and the cluster center is determined: k=arg max k {a x (i,k)+r x (i,k)+a w (i,k)+r w (i,k)}。 5. The electric vehicle clustering method considering driving characteristics and response willingness according to claim 4 is characterized in that: The calculation of the availability of the cluster center is specifically as follows: a w (k,k)=∑ l≠k max(0,r w (l,k))。 6. The electric vehicle clustering method considering driving characteristics and response willingness according to claim 4, characterized in that: The step S3 comprises: S31: Determine the driving characteristics of the electric vehicle and the evaluation function of the owner's response willingness; S32: Use gradient descent to find weights in combination with fuzzy information entropy; S33: construct an algorithm for the modified AP model.

7. The electric vehicle clustering method considering driving characteristics and response willingness according to claim 5, characterized in that: The step S31 is specifically as follows: The n-dimensional driving characteristics and response willingness of the i-th electric vehicle are represented by w i and L i Indicates that, let its corresponding weight vector be v i , v i =(v i1 , v i2 ), the weighted Euclidean distance is Select fuzzy information entropy for the driving characteristics of electric vehicles and the evaluation function of the owner's response willingness:

8. The electric vehicle clustering method considering driving characteristics and response willingness according to claim 6, characterized in that: The step S32 comprises: Initialize the weight vector: right and Δv t To solve; like but Continue to iterate and calculate w i，or ∈[0, 1] and w i，t ={w i，1 , w i，2 ,…,w i，N }Until the iteration reaches the preset number of times or convergence; Continue to iterate and calculate w i，or ∈[0,1], w i，t ={w i，1 , w i，2 ,…,w i，N }、s L (i, k) = -‖L i -L k ‖ 2 , find the total weight of all electric vehicles participating in the statistics.

9. The electric vehicle clustering method considering driving characteristics and response willingness according to claim 7, characterized in that: v after iterative calculation i , v i =(v i1 , v i2 ), v i1 represents the weight of the driving characteristics of the i-th electric vehicle after being corrected by the gradient descent method, v i2 represents the weight of the response willingness of the i-th electric vehicle after being corrected by the gradient descent method, v i As the weight parameter of the driving characteristics of electric vehicles and their owners' response willingness.

10. A system based on the electric vehicle clustering method considering driving characteristics and response intention according to claim 1, characterized in that: include: Data collection and automobile charging and discharging model building module, used to: collect electric vehicle operation data and charging and discharging data and build an electric vehicle charging and discharging model; The clustering model building module is used to: build an AP model that takes into account the response willingness and driving characteristics of electric vehicle owners based on the collected electric vehicle operation data and charging and discharging data; perform algorithm correction on the AP model in step S2 based on the gradient descent method; and cluster the cluster centers obtained in the modified AP algorithm based on the K-means algorithm; The clustering effect evaluation module is used to evaluate the effect of the clustering in step S4 based on the Silhouette index.

Images