Spatial effect evaluation method, system and equipment for charging demand of electric vehicle

Abstract

The invention relates to the technical field of electric vehicle charging demand analysis, and discloses an electric vehicle charging demand spatial effect evaluation method, system and device. The system obtains electric vehicle track data, social and economic data and built environment data; according to a charging demand extracted from electric vehicle track data, a Mean-Shift clustering method based on a kernel function is utilized to calculate a charging site of an electric vehicle, a Voronoi diagram is introduced to divide a charging service range into irregular polygons, and a principle of nearby charging is met; a Voronoi polygon is used as a research unit, and the time-space characteristics of charging requirements and influence factors are deeply analyzed; multi-collinearity test is adopted to eliminate redundant factors, global and local space self-correlation test is adopted to explore space clustering and correlation of charging demands and influence factors, charging demand space effect evaluation comprises space heterogeneity and space dependence, influence factors of different scale effects are considered, and the charging demand space effect evaluation method based on the space heterogeneity and the space dependence is realized. The method has the advantages of dynamically updating spatial effect evaluation, being more comprehensive and more accurate and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle charging demand analysis, and in particular to a method, system and device for evaluating the spatial effect of electric vehicle charging demand. Background Art

[0002] With growing awareness of environmental protection and energy conservation, electric vehicles are rapidly growing in China. This, coupled with a growing mismatch between charging demand and infrastructure, and the resulting imbalance in charging infrastructure development, is becoming increasingly prominent. Unlike traditional fuel vehicles, electric vehicle drivers must consider the battery's state of charge, charging time, and the availability of charging stations while driving. Therefore, rationally planning charging infrastructure and accurately estimating charging demand are crucial for the healthy development of electric vehicles and the sustainable development of cities. This requires a deep understanding of electric vehicle charging patterns and demand, as well as their relationship with socioeconomic factors, the urban built environment, and other influencing factors, particularly spatial effects.

[0003] The development of geographic big data and geospatial artificial intelligence has brought new opportunities and challenges to the classic topic of spatial relationship modeling. However, existing methods for assessing the spatial effects of charging demand suffer from incomplete consideration of multi-source heterogeneous data (both static and dynamic) and a lack of ability to simultaneously address spatial dependencies and heterogeneity. Furthermore, further optimization and a comprehensive framework are needed to estimate electric vehicle charging locations and determine the appropriate scope of charging services. Therefore, a dynamically updated, more comprehensive, and more accurate charging demand spatial effect assessment solution is urgently needed to meet the needs of urban and transportation planners, infrastructure companies, and charging station operators in dynamically assessing the spatial effects of charging demand and influencing factors. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a method, system and equipment for evaluating the spatial effect of electric vehicle charging demand, which has the advantages of being able to dynamically and accurately evaluate the spatial effect of electric vehicle charging demand, and solves the above technical problems.

[0005] To achieve the above object, the present invention provides the following technical solution: a method for evaluating the spatial effect of electric vehicle charging demand, comprising the following steps:

[0006] S1: Acquire electric vehicle trajectory data, socioeconomic data, and built environment data;

[0007] S2: Initial charging demand is obtained based on the trajectory data of electric vehicles. The clustering is performed using the Mean-Shift clustering algorithm based on the Gaussian kernel function. Combined with the optimal bandwidth obtained by the silhouette coefficient method, the optimal number of cluster centers is determined and the selected centers are set as the charging locations of the electric vehicles.

[0008] The initial charging demand is a charging order point;

[0009] S3: With the electric vehicle charging location as the center, the Voronoi diagram is introduced to divide the charging service range into multiple irregular polygons, and the charging demand on the boundary of adjacent irregular polygons is at the same distance from the center point of the two irregular polygons;

[0010] S4: Matching electric vehicle trajectory data, socioeconomic data, and built environment data to Voronoi polygons;

[0011] S5: Multicollinearity test was used to eliminate redundant factors, and global and local spatial autocorrelation tests were used to explore the spatial clustering and correlation between charging demand and influencing factors;

[0012] S6: Assess the spatial effects of charging demand, including spatial heterogeneity and spatial dependence, and the spatial impact of different scale effects, based on a hybrid MGWR-SAR model consisting of spatial autoregressive modules stacked into a multiscale geographically weighted model.

[0013] As a preferred technical solution of the present invention, the Voronoi diagram in S3 divides the charging service range into multiple irregular polygons, wherein the mth center point C in the Voronoi diagram is m The expression of the Voronoi cell centered at is as follows:

[0014]

[0015] Among them, VR(C m ,S) indicates that C m A single Voronoi cell is centered, S represents the set of charging locations, ∩ represents the intersection operator, C m+1 ∈S represents the center point C m+1 Belongs to the set S, C m ≠C m+1 Indicates the center point C m with C m+1 Not equal, D(C m |(C m ,C m+1 )) contains the m+1th center point C m+1 The split half plane, and the m+1th center point C m+1 and the mth center point C m adjacent;

[0016] The specific expression of the Voronoi diagram is as follows:

[0017]

[0018] Where V(S) represents the overall Voronoi diagram of set S, ∪ is the union operator, and Respectively represent the intersection and C m and C m+1 The Voronoi cell centered at C m+1 ∈S represents the center point C m+1 Belongs to the set S, C m ≠C m+1 Indicates the center point C m with C m+1 Not equal.

[0019] As a preferred technical solution of the present invention, the m+1th center point C m+1 The split half plane D(C m |(C m ,C m+1 )) is expressed as follows:

[0020] D(C m |(C m ,C m+1 ))={x|d(C m ,x)<d(C m+1 ,x)}

[0021] Among them, x represents the m+1th center point C m+1 Compared to the distance from the mth center point C m More recent data points.

[0022] As a preferred technical solution of the present invention, the multicollinearity test in S5 is specifically a multicollinearity test performed by the variance inflation factor, and the specific expression is as follows:

[0023]

[0024] Among them, VIF p and They represent the variance inflation factor and goodness of fit obtained by regressing the p-th independent variable on other variables.

[0025] As a preferred technical solution of the present invention, the specific expression of the MGWR-SAR model is as follows:

[0026]

[0027] Among them, y a 、 and ε a They represent the electric vehicle charging demand, intercept term and random error term of the research unit a, respectively. (u a ,v a ) represents the centroid of the research unit a, ua and v a They represent the latitude and longitude coordinates of the research unit a, B is the number of adjacent research units, P is the total number of variables, represents the estimated coefficient of variable p in research unit a, bw p represents bandwidth, α bw (u a ,v a ) represents the spatial hysteresis coefficient, W ab is the spatial weight matrix, represents the spatial lag term, reflecting the y a Electric vehicle charging demand of neighboring units b The spatial dependency relationship between (b=1,...,B) is investigated, and the research unit a represents the Voronoi polygon.

[0028] The present invention also provides a spatial effect evaluation system for electric vehicle charging demand, which is based on the above-mentioned spatial effect evaluation method for electric vehicle charging demand and includes: a data acquisition module, a charging location identification module, a charging service range determination module, a spatiotemporal feature analysis module, a multicollinearity and spatial autocorrelation test module, and a charging demand spatial effect evaluation module;

[0029] Data acquisition module, used to obtain electric vehicle trajectory data, socioeconomic data and built environment data;

[0030] The charging location identification module obtains the initial charging demand based on the trajectory data of the electric vehicle, clusters it using the Mean-Shift clustering algorithm based on the Gaussian kernel function, and combines it with the optimal bandwidth obtained by the silhouette coefficient method to determine the optimal number of cluster centers. The selected center points are then set as the charging locations of the electric vehicles.

[0031] A charging service range determination module is used to determine the coverage of the charging service based on the charging location of the electric vehicle and the spatial partitioning method of Voronoi polygons;

[0032] The spatiotemporal feature analysis module is used to analyze the spatiotemporal features of charging demand and influencing factors using Voronoi polygons as research units;

[0033] Multicollinearity and spatial autocorrelation test modules are used to eliminate redundant factors using multicollinearity tests, and to analyze the spatial clustering and correlation between charging demand and influencing factors using global and local spatial autocorrelation tests.

[0034] The charging demand spatial effect evaluation module is used for the MGWR-SAR hybrid model to evaluate the charging demand spatial effect, which includes spatial heterogeneity and spatial dependence.

[0035] A device for evaluating the spatial effect of electric vehicle charging demand includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus, the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the above-mentioned method for evaluating the spatial effect of electric vehicle charging demand.

[0036] Compared with the prior art, the present invention provides a method, system, and device for evaluating the spatial effect of electric vehicle charging demand, which have the following beneficial effects:

[0037] The present invention first obtains electric vehicle trajectory data, socioeconomic data, and built environment data. Secondly, based on the charging demand extracted from the electric vehicle trajectory data, the kernel function-based Mean-Shift clustering method is used to infer the charging locations of electric vehicles. The Voronoi diagram is then introduced to divide the charging service range into irregular polygons based on the charging locations of electric vehicles, in line with the "charging nearby" principle. Then, using Voronoi polygons as research units, the spatiotemporal characteristics and influencing factors of charging demand are deeply analyzed. Subsequently, multicollinearity tests are used to eliminate redundant factors, and global and local spatial autocorrelation tests are used to explore the spatial clustering and correlation between charging demand and influencing factors. Finally, a hybrid model based on MGWR-SAR is proposed to dynamically and accurately evaluate the spatial effect of charging demand, including spatial heterogeneity and spatial dependence, and consider the influencing factors of different scale effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic diagram of the process of the present invention;

[0039] Figure 2 Schematic diagram of the structure of the system of the present invention;

[0040] Figure 3 Schematic diagram of the terminal device of the present invention. DETAILED DESCRIPTION

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0042] See also Figure 1-Figure 3 , a spatial effect evaluation method for electric vehicle charging demand, comprising the following steps:

[0043] S1: Acquire electric vehicle trajectory data, socioeconomic data, and built environment data;

[0044] (1) Electric vehicle trajectory data: charging orders, frequent dwell times, and low state of charge (SOC) trajectory points;

[0045] (2) Socioeconomic data: housing prices, population density;

[0046] (3) Built environment data: parking density, road network density, distance to the CBD, consumption-related points of interest (POI), occupation-related POI, housing-related POI, tourism-related POI, transportation hub POI, and land use mix;

[0047] S2: Initial charging demand is obtained based on the trajectory data of electric vehicles. The clustering is performed using the Mean-Shift clustering algorithm based on the Gaussian kernel function. Combined with the optimal bandwidth obtained by the silhouette coefficient method, the optimal number of cluster centers is determined and the selected centers are set as the charging locations of the electric vehicles.

[0048] The initial charging demand is a charging order point;

[0049] (1) Original Mean-Shift clustering algorithm:

[0050] Given n charging order points T={t1,t2,...,t n},as well as The center position coordinate t at the jth update in the high-dimensional spherical space j and radius Q, defining a spherical area S Q It can be expressed as:

[0051] S Q ={t i |||t i -t j ||<Q,i=1,2,…,n}

[0052] Here, ||·|| represents the norm of the vector, that is, the distance between two points.

[0053] The Mean-Shift drift vector of the center point of the sphere can be expressed as:

[0054]

[0055] Where, t i and g are S Q The charging order points and quantities within the sphere are updated to:

[0056] tj+1 =M(t j )+t j

[0057] Where, t j+1 Indicates the updated center position coordinates;

[0058] The original Mean-Shift treats all points in the hypersphere equally and does not consider the influence of points at different distances from the center (offset vectors). Therefore, a Gaussian kernel function is introduced to obtain the contribution of different samples to the offset vector.

[0059] (2) Mean-Shift clustering algorithm based on Gaussian kernel function:

[0060] The Gaussian kernel function can map data points that are originally linearly inseparable in low-dimensional space to high-dimensional space, making them linearly separable without constructing complex mapping functions; as a distance mapping, the Gaussian kernel function can ensure that closed points in low-dimensional space are still close in the mapped high-dimensional space; therefore, the Gaussian kernel function is used to calculate the center and S Q The distance between inliers gives data points closer to the center a larger weight, thus avoiding the problem that all samples contribute equally to the offset vector. For a given Gaussian kernel function k and bandwidth h, the translation vector and translation distance marker can be expressed as:

[0061]

[0062] dis(M)=||M'(t j )||

[0063] Then, the center of the sphere is improved to:

[0064] t j+1 =M'(t j )+t j

[0065] In the formula, M'(t j ) and dis(M) are the direction and distance of the hypersphere respectively; therefore, the center point will continue to move toward the high-density area until dis(M) no longer changes significantly or reaches the predetermined number of iterations; then, a fixed value of t is determined. l The cluster C is the center l ; For another f The cluster C is the center f , if t l and t f The distance between them is less than a predefined threshold δ (δ>0), then the two clusters are merged into one category, namely C f All points in will be labeled as cluster C l :

[0066] ||t l -t f ||<δ

[0067] If a cluster converges, a new center point will be randomly selected from the unmarked points to continue the above process to generate a new cluster; the process of determining the cluster center is a process of moving to a new high-density area, so different clusters reflect different charging demand densities;

[0068] (3) Determine the optimal bandwidth h:

[0069] In order to obtain the best bandwidth and evaluate the clustering results, the silhouette coefficient is used to describe the similarity between samples in the clustering data; sample t i and all other points n in the cluster m The average distance of -1 can be expressed as:

[0070]

[0071] Similarly, in C m Point t in i and all other points in the sample n m+1 The next nearest neighbor cluster C m+1 The average distance in can be expressed as:

[0072]

[0073] Then the silhouette coefficient SC(t i ) can be expressed as:

[0074]

[0075] C m The silhouette coefficient SC(C m ) is the mean of the silhouette coefficient of each sample:

[0076]

[0077] SC(C m ) has a value range of [-1, +1];

[0078] SC(C m ) is closer to 1, indicating that C m The clusters are more compact and well separated from other clusters;

[0079] In this way, determine C m The optimal bandwidth h;

[0080] Therefore, we first cluster the charging demand (i.e., charging order points) using the Mean-Shift clustering algorithm with a Gaussian kernel function. Secondly, we combine the optimal bandwidth obtained using the silhouette coefficient method to determine the optimal number of cluster centers. Finally, we set these centers as the charging locations for electric vehicles, i.e., the locations with the highest charging demand density.

[0081] S3: With the electric vehicle charging location as the center, the Voronoi diagram is introduced to divide the charging service range into multiple irregular polygons, and the charging demand on the boundary of adjacent irregular polygons is at the same distance from the center point of the two irregular polygons;

[0082] According to the charging location of electric vehicles, the Voronoi diagram is introduced to divide the charging service range into irregular polygons, which conforms to the principle of "charging nearby";

[0083] The Mean-Shift clustering algorithm with a Gaussian kernel function is used to determine the best EV charging locations. The Voronoi diagram (Thyssen polygon) is introduced to divide the urban area into irregular polygons of different sizes. The center of the Voronoi polygon corresponds to each EV charging location, and the Voronoi polygon is used to divide the reasonable charging service range.

[0084] For A charging locations C on a two-dimensional plane i ∈S(i=1,2,...,A), S represents the set of charging locations, any two adjacent center points C m and C m+1 , the longitude and latitude are (u m ,v m ) and (u m+1 ,v m+1 ), C m and C m+1 The distance between can be expressed as:

[0085]

[0086] For any two adjacent center points C m and C m+1 ,B(C m ,C m+1 ) is the perpendicular bisector of the line segment passing through these two points; then, B(C m ,C m+1 ) and C m and C m+1 The distance d(C m ,y) and d(C m+1 ,y) are equal and contain C m The split half plane is expressed as:

[0087] B(C m ,C m+1 )={y|d(C m ,y)=d(C m+1 ,y)}

[0088] Then, for the distance from the center point C m A closer data point x if:

[0089] D(C m |(C m ,C m+1 ))={x|d(C m ,x)<d(C m+1 ,x)};

[0090] Where, d(C m ,x) represents the center point C m The distance from the data point x, d(C m+1 ,x) represents the center point C m+1 The distance from the data point x.

[0091] D(C m |(C m ,C m+1 )) contains C m+1 The split half plane is C m The Voronoi cell centered on can be represented as:

[0092]

[0093] VR(C m ,S) indicates that C m A single Voronoi cell centered at ∩ is the intersection operator, C m+1 ∈S represents the center point C m+1 Belongs to the set S, C m ≠C m+1 Indicates the center point C m with C m+1 Not equal;

[0094] The Voronoi diagram for a set S can be defined as:

[0095]

[0096] V(S) represents the overall Voronoi diagram of set S, ∪ is the union operator, and Respectively represent the intersection and C m and C m+1 The Voronoi cell centered at C m+1 ∈S represents the center point Cm+1 Belongs to the set S, C m ≠C m+1 Indicates the center point C m with C m+1 Not equal.

[0097] The charging demand points within each Voronoi polygon are closer to the current center point and farther from the centers of other polygons. The charging demand points on the boundaries of two polygons are equidistant from the centers of both polygons. Therefore, based on this "charging nearby" principle, the charging service area is divided into irregular polygons of different sizes.

[0098] S4: Matching electric vehicle trajectory data, socioeconomic data, and built environment data to Voronoi polygons;

[0099] Using Voronoi polygons as the research unit, multi-source data such as electric vehicle trajectory data, socioeconomic data, and built environment data are matched to Voronoi polygons;

[0100] From a temporal perspective, using a variety of visualization methods such as line charts and bar charts, the dynamic trends of charging demand in various regions over different time periods, as well as the evolution of various related factors over time, are vividly and intuitively presented. This allows for accurate real-time information, timely understanding of peak and trough periods of charging demand, and the interaction between various factors over time.

[0101] From a spatial perspective, the powerful visualization capabilities of Geographic Information Systems (GIS), such as maps and 3D visualization, are comprehensively and meticulously presented to present the spatial distribution of multiple data sources, including electric vehicle trajectory data, socioeconomic data, and built environment data, throughout the day and at different time periods. This allows for further analysis of existing spatial heterogeneity and clustering, as well as differences in cluster density across different research units, to accurately grasp the spatial pattern of charging demand.

[0102] S5: Multicollinearity test was used to eliminate redundant factors, and global and local spatial autocorrelation tests were used to explore the spatial clustering and correlation between charging demand and influencing factors;

[0103] (1) Multicollinearity test

[0104] Multicollinearity refers to the phenomenon that two or more independent variables in a model are highly correlated. Strong correlation between variables indicates serious multicollinearity, which can lead to unstable coefficient estimates, unreasonable coefficient signs or sizes, weak model interpretability, or even errors. Variance Inflation Factor (VIF) is usually used to test for multicollinearity to eliminate the impact of collinearity on model accuracy and ensure the reliability of the results. VIF can be expressed as:

[0105]

[0106] Where, VIF p and They represent the variance inflation factor (VIF) and goodness of fit obtained by regressing the p-th independent variable on other variables;

[0107] VIF p >10 indicates that there is a strong correlation between the variables and they should be eliminated;

[0108] (2) Global and local spatial autocorrelation test

[0109] Global spatial autocorrelation describes whether an attribute has clustering or different spatial characteristics within a region. The commonly used evaluation index is the Global Moran's Index (GMI). GMI can be expressed as:

[0110]

[0111] Where x a and x b are the values ​​of a variable in research units a and b respectively; is the mean value of the variable; w ab is the spatial weight between study units a and b; A is the number of study units; the GMI value range is between -1 and 1. A positive value indicates a positive correlation, meaning that similar attribute values ​​tend to cluster in space; a negative value indicates a negative correlation, meaning that attribute values ​​tend to be in opposite directions in spatial distribution.

[0112] GMI can use the standardized statistic z-value to detect spatial autocorrelation in order to better observe the statistical significance of spatial correlation. The z-value can be expressed as:

[0113]

[0114] Where E(GMI) and VAR(GMI) represent the expectation and variance of GMI;

[0115] z-value>1.96 indicates strong spatial correlation;

[0116] As a global indicator, GMI cannot reflect specific spatial variability and local spatial correlation. The Local Moran's Indicator (LMI), also known as the Local Indicators of Spatial Association (LISA), is further applied to measure the variable spatial relationship between the observed research unit and its adjacent units. LMI can be expressed as:

[0117]

[0118] Combined with the significance test, LMI or LISA generates five types of spatial clustering for each spatial charging demand or influencing factor observation, namely, not significant, high-high, low-low, low-high, and high-low;

[0119] S6: Assess the spatial effects of charging demand based on a hybrid MGWR-SAR model, including spatial heterogeneity and spatial dependence, as well as the spatial impact of different scale effects. The MGWR-SAR hybrid model consists of a spatial autoregressive module stacked into a multi-scale geographically weighted model.

[0120] The specific expression of the MGWR-SAR model is as follows:

[0121]

[0122] Among them, y a 、 and ε a They represent the electric vehicle charging demand, intercept term and random error term of the research unit a, respectively. (u a ,v a ) represents the centroid of the research unit a, u a and v a The latitude and longitude coordinates of the research unit a are respectively represented, B is the number of adjacent research units, and P is the total number of variables. represents the estimated coefficient of variable p in research unit a, bw p represents bandwidth, α bw (u a ,v a ) represents the spatial hysteresis coefficient, W ab is the spatial weight matrix, represents the spatial lag term, reflecting the y a Electric vehicle charging demand of neighboring units bThe spatial dependency between (b=1,...,B), and the research unit a represents the Voronoi polygon;

[0123] In this embodiment, the spatial effect assessment method of electric vehicle charging demand is modeled using actual trajectory data of a certain urban area in China from all day, daytime charging peak period (10:00-16:00), nighttime charging peak period (20:00-2:00), and non-charging peak period (16:00-20:00 and 2:00-10:00);

[0124] Goodness-of-fitting (R 2 The model performance was evaluated using five indicators: Root Mean Square Error (RMSE), Corrected Akaike Information Criterion (AICc), Log-likelihood, and Residual Sum of Squares (RSS); R 2 The larger the and Log-likelihood, the smaller the RMSE, AICc, and RSS, indicating that the model has a better fitting effect, stronger interpretability, and smaller fitting error;

[0125] The performance index comparison between the MGWR-SAR model and the comparison model is shown in Table 1;

[0126] Table 1 Performance indicators of local regression models

[0127]

[0128]

[0129] MGWR-SAR can provide more generalizable and robust estimation results (R 2 The highest values ​​are 0.849, 0.855, 0.860 and 0.841, and the RMSE, AICc and RSS are the smallest). Its performance is better than that of the GWR, MGWR and GWR-SAR models. The hybrid model of the variable spatial impact scale and SAR module in MGWR-SAR enables it to simultaneously capture the spatial dependence and heterogeneity of charging demand and external factors, thereby effectively improving the interpretability and goodness of fit of the model.

[0130] Taking the whole day as an example, the spatial influence scale and regression coefficient of the MGWR-SAR model are shown in Table 2:

[0131] Table 2 Spatial influence scale and regression coefficient of MGWR-SAR model

[0132] Factors / Variables Spatial impact scale Minimum Maximum mean variance Low SOC 247 0.557 0.578 0.567 0.006 Housing prices 253 -0.010 0.024 0.006 0.010 population density 54 0.053 0.751 0.346 0.180 Parking density 97 0.214 0.668 0.451 0.099 Road network density 43 -0.120 0.943 0.150 0.225 Land use mix 253 0.014 0.041 0.030 0.007 Distance to CBD 123 0.168 0.432 0.305 0.084 Consumption-related POI 166 -0.558 -0.446 -0.510 0.031 Tourism-related POIs 242 -0.140 -0.108 -0.127 0.009 Transportation hub POI 139 -0.283 0.047 -0.167 0.097 Spatial lag 152 -0.315 -0.219 -0.278 0.028

[0133] Based on the ratio of the spatial impact scale to the total number of study units, the spatial impact scale (bandwidth) was divided into three categories: local, regional, and global. If the ratio was less than 30%, the spatial impact scale was defined as local; if the ratio exceeded 80%, the spatial impact scale was defined as global; the rest corresponded to regional impacts.

[0134] The spatial coefficients and spatial effects between various influencing factors and EV charging demand are analyzed at three scales: local, regional, and global. These coefficients are then integrated into a GIS, displaying the spatial effects of different influencing factors on charging demand in real time through maps and 3D visualizations, including spatial heterogeneity and spatial dependence. This provides a basis for decision-making by urban / transportation planners, EV manufacturers, or policymakers, supporting the forecasting of charging demand, the site selection, and the operation of charging stations.

[0135] The present invention also provides a spatial effect evaluation system for electric vehicle charging demand, which is based on the above-mentioned spatial effect evaluation method for electric vehicle charging demand and includes: a data acquisition module, a charging location identification module, a charging service range determination module, a spatiotemporal feature analysis module, a multicollinearity and spatial autocorrelation test module, and a charging demand spatial effect evaluation module;

[0136] Data acquisition module, used to obtain electric vehicle trajectory data, socioeconomic data and built environment data;

[0137] The charging location identification module obtains the initial charging demand based on the trajectory data of the electric vehicle, clusters it using the Mean-Shift clustering algorithm based on the Gaussian kernel function, and combines it with the optimal bandwidth obtained by the silhouette coefficient method to determine the optimal number of cluster centers. The selected center points are then set as the charging locations of the electric vehicles.

[0138] A charging service range determination module is used to determine the coverage of the charging service based on the charging location of the electric vehicle and the spatial partitioning method of Voronoi polygons;

[0139] The spatiotemporal feature analysis module is used to analyze the spatiotemporal features of charging demand and influencing factors using Voronoi polygons as research units;

[0140] Multicollinearity and spatial autocorrelation test modules are used to eliminate redundant factors using multicollinearity tests, and to analyze the spatial clustering and correlation between charging demand and influencing factors using global and local spatial autocorrelation tests.

[0141] The charging demand spatial effect evaluation module is used for the MGWR-SAR hybrid model to evaluate the charging demand spatial effect, which includes spatial heterogeneity and spatial dependence.

[0142] A device for evaluating the spatial effect of electric vehicle charging demand includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus, the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the above-mentioned method for evaluating the spatial effect of electric vehicle charging demand.

[0143] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for evaluating the spatial effect of electric vehicle charging demand, characterized by: The following steps are involved: S1: Acquire electric vehicle trajectory data, socioeconomic data, and built environment data; S2: Initial charging demand is obtained based on the trajectory data of electric vehicles. The clustering is performed using the Mean-Shift clustering algorithm based on the Gaussian kernel function. Combined with the optimal bandwidth obtained by the silhouette coefficient method, the optimal number of cluster centers is determined and the selected centers are set as the charging locations of the electric vehicles. The initial charging demand is a charging order point; S3: With the electric vehicle charging location as the center, the Voronoi diagram is introduced to divide the charging service range into multiple irregular polygons, and the charging demand on the boundary of adjacent irregular polygons is at the same distance from the center point of the two irregular polygons; S4: Matching electric vehicle trajectory data, socioeconomic data, and built environment data to Voronoi polygons; S5: Multicollinearity test was used to eliminate redundant factors, and global and local spatial autocorrelation tests were used to explore the spatial clustering and correlation between charging demand and influencing factors; S6: Assess the spatial effects of charging demand, including spatial heterogeneity and spatial dependence, and the spatial impact of different scale effects, based on a hybrid MGWR-SAR model consisting of spatial autoregressive modules stacked into a multiscale geographically weighted model.

2. The method for evaluating the spatial effect of electric vehicle charging demand according to claim 1, characterized in that: The Voronoi diagram in S3 divides the charging service range into multiple irregular polygons, where the mth center point C m The expression of the Voronoi cell centered at is as follows: Among them, VR(C m ,S) indicates that C m A single Voronoi cell is centered, S represents the set of charging locations, ∩ represents the intersection operator, C m+1 ∈S represents the m+1th center point C m+1 Belongs to the set S, C m ≠C m+1 Represents the mth center point C m and the m+1th C m+1 Not equal, D(C m |(C m ,C m+1 )) contains the m+1th center point C m+1 The split half plane, and the m+1th center point C m+1 and the mth center point C m adjacent; The specific expression of the Voronoi diagram is as follows: Where V(S) represents the overall Voronoi diagram of set S, ∪ is the union operator, and Respectively represent the intersection and C m and C m+1 The Voronoi cell centered at C m+1 ∈S represents the m+1th center point C m+1 Belongs to the set S, C m ≠C m+1 Represents the mth center point C m and the m+1th center point C m+1 Not equal.

3. The method for evaluating the spatial effect of electric vehicle charging demand according to claim 1, characterized in that: The m+1th center point C m+1 The split half plane D(C m |(C m ,C m+1 )) is expressed as follows: D(C m |(C m ,C m+1 ))={x|d(C m ,x)<d(C m+1 ,x)} Among them, x represents the m+1th center point C m+1 Compared to the distance from the mth center point C m Closer data points, d(C m ,x) represents the mth center point C m The distance from the data point x, d(C m+1 ,x) represents the m+1th center point C m+1 The distance from the data point x.

4. The method for evaluating the spatial effect of electric vehicle charging demand according to claim 1, wherein: The multicollinearity test in S5 is specifically a multicollinearity test performed using the variance inflation factor, and the specific expression is as follows: Among them, VIF p and They represent the variance inflation factor and goodness of fit obtained by regressing the p-th independent variable on other variables.

5. The method for evaluating the spatial effect of electric vehicle charging demand according to claim 1, characterized in that: The specific expression of the MGWR-SAR model is as follows: Among them, y a 、 and ε a They represent the electric vehicle charging demand, intercept term and random error term of the research unit a, respectively. (u a ,v a ) represents the centroid of the research unit a, u a and v a denote the latitude and longitude coordinates of the research unit a, B is the number of adjacent research units, P is the total number of variables, represents the estimated coefficient of variable p in research unit a, bw p represents bandwidth, α bw (u a ,v a ) represents the spatial hysteresis coefficient, W ab is the spatial weight matrix, represents the spatial lag term, reflecting the y a Electric vehicle charging demand of neighboring units b The spatial dependency relationship between (b=1,...,B) is investigated, and the research unit a represents the Voronoi polygon.

6. A spatial effect evaluation system for electric vehicle charging demand, based on the spatial effect evaluation method for electric vehicle charging demand according to any one of claims 1 to 5, characterized in that: include: Data acquisition module, charging location identification module, charging service range determination module, spatiotemporal feature analysis module, multicollinearity and spatial autocorrelation test module, and charging demand spatial effect assessment module; Data acquisition module, used to obtain electric vehicle trajectory data, socioeconomic data and built environment data; The charging location identification module obtains the initial charging demand based on the trajectory data of the electric vehicle, clusters it using the Mean-Shift clustering algorithm based on the Gaussian kernel function, and combines it with the optimal bandwidth obtained by the silhouette coefficient method to determine the optimal number of cluster centers. The selected center points are then set as the charging locations of the electric vehicles. A charging service range determination module is used to determine the coverage of the charging service based on the charging location of the electric vehicle and the spatial partitioning method of Voronoi polygons; The spatiotemporal feature analysis module is used to analyze the spatiotemporal features of charging demand and influencing factors using Voronoi polygons as research units; Multicollinearity and spatial autocorrelation test modules are used to eliminate redundant factors using multicollinearity tests, and to analyze the spatial clustering and correlation between charging demand and influencing factors using global and local spatial autocorrelation tests. The charging demand spatial effect evaluation module is used for the MGWR-SAR hybrid model to evaluate the charging demand spatial effect, which includes spatial heterogeneity and spatial dependence.

7. A spatial effect assessment device for electric vehicle charging demand, characterized by: The electronic device comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus, the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute a spatial effect assessment method for electric vehicle charging demand according to any one of claims 1 to 5.

Images