Subtropical city charging station layout optimization method and system based on climate adaptive clustering

Abstract

The invention provides a subtropical city charging station layout optimization method and system based on climate adaptive clustering, and belongs to the technical field of city charging infrastructure planning. The problems that in existing planning, climate adaptability is insufficient, analysis of a driving mechanism is not accurate, and strategy pertinence is not high are solved. The method comprises the following steps: collecting data of a city-surrounding high-speed enclosed area of a subtropical city, and constructing an equal-area hexagonal grid with the side length of 1000m after data cleaning and coordinate conversion; a kernel density estimation method is adopted to analyze the aggregation degree of the charging stations, and adaptive features are obtained in combination with a climate adaptive DBSCAN clustering algorithm; constructing a multi-scale geographically weighted regression model by taking the density of the charging station as a dependent variable, and quantifying the influence of spatial heterogeneity; and according to clustering and regression results, providing a charging station layout scheme from driving factor guidance, a climate adaptive hierarchical strategy and a climate adaptation dimension. The method is mainly used in the subtropical city charging network construction field.

Description

Technical Field

[0001] This invention belongs to the field of urban charging infrastructure planning technology, and in particular relates to a method for optimizing the layout of charging stations in subtropical cities based on climate adaptive clustering. Background Technology

[0002] With the rapid development of the new energy vehicle industry, the rational layout of urban charging infrastructure has become a key factor affecting user charging experience and the quality of industrial development. Southern my country, with its vast subtropical monsoon climate, is characterized by high temperatures and humidity in summer. This specific environmental condition has a dual and crucial impact on the planning and operation of electric vehicle charging stations: Firstly, empirical studies show that when the ambient temperature reaches 35℃ or above, the battery range of electric vehicles experiences a significant decrease of 10%-20%. This range anxiety leads to more frequent and shorter-radius charging demands, profoundly altering the spatial distribution of demand based on traditional static traffic models. Secondly, persistently high humidity easily increases the risk of charging equipment interface corrosion and internal circuit short circuits, physically limiting the deployment density and long-term operational stability of facilities in some areas.

[0003] Currently, in the field of urban charging infrastructure planning, most existing layout optimization methods still focus on common driving factors such as population density, road network structure, land use type, and regional economic level. For example, a common planning approach is to identify urban activity centers based on population heat map data or POI (point of interest) density, and then conduct balanced layout of charging stations based on this. Another approach relies on the classic Geographically Weighted Regression (GWR) model to analyze the spatial heterogeneity of charging demand caused by variables such as road density and distance from the city center. However, these existing technologies generally lack specific adaptation to the unique climatic characteristics of subtropical cities, resulting in significant shortcomings in planning schemes.

[0004] Specifically, the shortcomings of existing technologies are mainly reflected in the following aspects: First, in the clustering analysis stage, traditional methods such as DBSCAN (density-based spatial clustering with noise) typically set their core parameters (such as the neighborhood radius Eps) to a fixed value based on experience or static geographical distance. This method fails to consider the changes in user behavior caused by the reduction in the driving range of electric vehicles in high-temperature environments, i.e., users tend to find charging stations within a shorter driving distance, causing the effective service radius to "shrink." This neglect of climate constraints directly leads to the clustering results failing to accurately reflect the actual spatial behavior patterns of users during high-temperature seasons, resulting in a spatial mismatch between supply and demand.

[0005] Secondly, at the level of analyzing driving factors, existing models such as Ordinary Least Squares (OLS) or the basic GWR model, while incorporating climate factors as independent variables, often homogenize or simplify their effects. For example, they might only arrive at a global conclusion that temperature has a positive impact on charging demand, but fail to quantify the significant spatial heterogeneity of this impact across different regions (such as urban core areas and suburbs). In reality, high temperatures may strongly stimulate charging demand in densely populated areas, while in sparsely populated areas, the impact may be negligible or even negative. This imprecise analysis of the driving mechanisms of climate factors results in a lack of granular spatial guidance for planning decisions.

[0006] Finally, regarding the formulation of deployment strategies, due to the insufficient depth of the aforementioned analysis, existing technologies struggle to develop differentiated solutions that balance the contradiction between the charging demand driven by high temperatures and the safety constraints imposed by high humidity. The consequence is that some areas may experience an imbalance between charging station supply and demand, leading to excessively long queues, while other areas may suffer from low operational efficiency due to high equipment failure rates and low utilization rates, ultimately resulting in the misallocation and waste of public resources. Summary of the Invention

[0007] In view of this, the present invention aims to propose a method and system for optimizing the layout of charging stations in subtropical cities based on climate adaptive clustering, so as to solve the problems of insufficient climate adaptability, inaccurate analysis of driving mechanisms, and weak strategy targeting in existing planning.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: a method for optimizing the layout of charging stations in subtropical cities based on climate-adaptive clustering, the method comprising: Step S1: Collect charging station POI, population heat data, OSM road network data, nighttime light data, summer average maximum temperature, and summer monthly average humidity data in the area enclosed by the subtropical city's ring expressway. After data cleaning and coordinate transformation, construct an equal-area hexagonal grid with a side length of 1000m. Step S2: Analyze the degree of charging station clustering using the kernel density estimation method, and combine it with the climate-adaptive DBSCAN clustering algorithm to obtain adaptation features; Step S3: Construct a multi-scale geographically weighted regression model with charging station density as the dependent variable to quantify the spatial heterogeneity impact of population activity, road density, nighttime light index, climate factors, and economic indicators. Step S4: Based on the clustering and regression results, propose a charging station layout scheme from the perspectives of driving factors, climate-adaptive hierarchical strategy, and climate adaptation.

[0009] Furthermore, a preferred method is proposed, wherein the data cleaning includes: Acquire POI data for charging stations including supercharging, fast charging, slow charging, and battery swapping stations. Use address matching and spatial distance ≤50m as dual criteria to remove duplicate POI records. Obtain road network data including urban expressways, arterial roads, and secondary arterial roads, and delete dead-end roads and duplicate parallel roads in the OSM road network; Climate data were downscaled to a 1000m grid using Kriging interpolation, with interpolation errors controlled within ±0.5℃. Z-score standardization is used to eliminate the dimensions of continuous independent variables.

[0010] Furthermore, a preferred embodiment is proposed, wherein the implementation of the climate-adaptive DBSCAN clustering algorithm includes: For high, medium, and low density areas, the neighborhood radius parameters are traversed in a step size of 100m within a range of 500-1500m, and the optimal benchmark neighborhood radius is determined based on the lowest DB index. Eps base The minimum number of neighboring points, MinPts, is set to 3. Introducing the climate correction factor λ, the calculation formula is as follows: λ=1 / (1-δ) Where δ represents the range reduction rate of electric vehicles under high temperature, and is set to δ=15%; The neighborhood radius is adjusted adaptively based on the climate correction factor. The adjusted neighborhood radius is:

[0011] in, λ hot This is a high-temperature correction factor, keeping MinPts=3 unchanged, used for cluster analysis to match the high-temperature short-distance energy replenishment needs; Calculate the silhouette coefficient of the clustering results for each region to ensure the effectiveness of clustering after climate adaptation.

[0012] Furthermore, a preferred method is proposed, wherein the construction of the multi-scale geographically weighted regression model includes: Based on a hexagonal grid, a unique UID is assigned to each grid, and all variables are associated through the UID, which is then aggregated into a spatial-attribute integrated dataset. Establish model relationships:

[0013] in, Let the charging station density be the density of the i-th grid. For the spatial intercept term of the i-th grid, For grid coordinates, Let b be the spatial regression coefficient of the j-th influencing factor at location i. wjThe value x represents the bandwidth parameter used in the calculation of this coefficient. ij For the j-th influencing factor of the i-th grid, Let be the error term for the i-th grid; Collinearity interference was eliminated by variance inflation factor test, and all VIF values ​​were <7.5.

[0014] Furthermore, an optimal approach is proposed, in which the multi-scale geographically weighted regression model undergoes robustness verification by adjusting the hexagonal grid side length to 800m and 1200m for repeated fitting, and comparing the R-values. 2 Adjust R 2 And the AICc index ensures model scale invariance.

[0015] Furthermore, a preferred approach is proposed, wherein the driving factor orientation includes: Based on the clustering results of the multi-scale geographical weighted regression model, global and local influencing factors are distinguished, and the global influencing factors are used as the basic framework for the layout of charging stations. The layout density of charging stations is optimized based on the spatial distribution of the coefficients of the factors with the largest absolute value of the regression coefficients among the local influencing factors.

[0016] Furthermore, a preferred approach is proposed, wherein the climate adaptive stratification strategy includes: The value of λ is calculated based on δ, and the radius of the DBSCAN clustering benchmark neighborhood is determined. Eps base Increase by 10%-20% to accommodate the reduced battery life in summer; In high-density areas, the focus is on upgrading facilities; in medium-density areas, the focus is on filling coverage gaps; and in low-density areas, small and medium-sized charging stations are deployed. A dynamic analysis model is established to update the clustering results quarterly.

[0017] Based on the same inventive concept, this invention also proposes a subtropical city charging station layout optimization system based on climate adaptive clustering, the system comprising: Data acquisition unit: used to collect charging station POI, population heat data, OSM road network data, nighttime light data, summer average maximum temperature, and summer monthly average humidity data in the area enclosed by the ring expressway of subtropical cities. After data cleaning and coordinate transformation, a hexagonal grid with a side length of 1000m is constructed. Analysis Unit: Used to analyze the degree of charging station clustering using the kernel density estimation method, and combined with the climate-adaptive DBSCAN clustering algorithm to obtain adaptive features; Quantification Unit: Used to construct a multi-scale geographically weighted regression model with charging station density as the dependent variable, quantifying the spatial heterogeneity of population activity, road density, nighttime light index, climate factors, and economic indicators; Layout optimization unit: Based on clustering and regression results, optimize the layout scheme of charging stations from the perspectives of driving factors, climate-adaptive hierarchical strategy, and climate adaptation.

[0018] Based on the same inventive concept, the present invention also proposes a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes a subtropical city charging station layout optimization method based on climate adaptive clustering as described in any of the preceding claims.

[0019] Based on the same inventive concept, the present invention also proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of a climate-adaptive clustering-based subtropical city charging station layout optimization method as described above.

[0020] Compared with the prior art, the beneficial effects of the present invention are: This invention constructs a climate-endogenous charging station layout analysis and decision-making framework, fundamentally changing the traditional approach of treating climate factors as external environmental constants or simple boundary conditions. Existing DBSCAN clustering algorithms typically use a neighborhood radius based on fixed geographical distances or empirical values, failing to reflect the dynamic impact of subtropical high-temperature climates on user behavior. This invention introduces a climate-adaptive clustering machine for the first time. By quantifying the range reduction rate of electric vehicles caused by high temperatures, it transforms this into a calculable climate correction coefficient and uses the formula λ=1 / (1-δ) to dynamically calibrate the baseline neighborhood radius. This directly maps the physical world's climate constraints (range reduction) to algorithm parameters (service radius adjustment), enabling the clustering results to accurately reflect users' spatial behavior patterns of seeking short-distance, high-frequency charging during high-temperature seasons, thus fundamentally solving the problem of mismatch between supply and demand in spatial space. Examples show that the method effectively improves the clustering quality (profile coefficient) in medium and low density areas. For example, in low density areas, the profile coefficient increases from 0.2845 to 0.3339, proving that the method can more scientifically divide service areas under different climate needs and avoid resource mismatch problems such as having demand but no facilities or having facilities but no demand from the planning source.

[0021] Existing technologies, such as ordinary least squares (OLS) or basic geographic weighted regression (GWR) models, while incorporating climate factors, struggle to reveal the spatial variability and scale of their influence. This invention designs a multi-scale geographic weighted regression (MGWR) model as the core analytical tool. The MGWR model allows each independent variable to have different spatial bandwidths, thus enabling precise quantification of the local impacts of factors such as population activity and the global impacts of factors such as road density. Particularly for climate factors like summer average maximum temperature and humidity, the MGWR model accurately characterizes the spatial distribution differences of their regression coefficients; for example, they may exhibit positive promotion in some regions and negative inhibition in others, thereby revealing the complex spatial heterogeneity of climate-driving mechanisms.

[0022] Existing strategies often rely on unidirectional, isolated analytical results. This invention deeply couples the spatial patterns identified by climate-adaptive clustering with the driving mechanism quantified by the MGWR model. For example, it not only considers the correction of demand radius by climate during clustering, but also explicitly requires the deployment of different types of charging equipment (such as liquid-cooled supercharging piles) and protection standards (such as IP67 moisture resistance) based on the high-temperature positive impact zone and high-humidity negative impact zone revealed by MGWR. This model-based, concrete strategy feedback ensures the integrity and synergy of the technical solution, keeping planning decisions synchronized with dynamic climate impacts and spatial patterns.

[0023] This invention is applicable to the planning and construction of urban charging networks in subtropical high-temperature and high-humidity climate regions. Attached Figure Description

[0024] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart of a subtropical city charging station layout optimization method based on climate adaptive clustering, as described in this invention. Figure 2The diagram illustrates the various variables in the grid as described in this invention. In the diagram, the charging station density data from top to bottom are: 0.00-0.80, 0.81-1.89, 1.90-3.06, 3.07-4.50, 4.51-7.43; the average population activity data from top to bottom are: 2.79-106.33, 106.34-230.76, 230.77-365.31, 365.32-653.91, 653.92-1451.90; and the road density data from top to bottom are: 0.36-5.07, 5.08-7.83, 7.84-10.21, 10.22-13.09, 13 0.10-18.49; Nighttime light data from top to bottom are: 3.75-17.01, 17.02-27.54, 27.55-37.56, 37.57-50.65, 50.66-91.94; Maximum temperature data from top to bottom are: 35.15-35.42, 35.43-35.54, 35.55-35.64, 35.65-35.77, 35.78-35.98; Humidity data from top to bottom are: 66.57-66.76, 66.77-67.72, 67.73-68.57, 68.58-69.46, 69.47-70.73; Figure 3This is a comparison chart of the DB index, ordinary clustering, and climate-adaptive DBSCAN clustering results described in this invention. In the chart, the x-axis values ​​of the DB index in the high-density area are 500, 600, 700, 800, 900, 1000, and 1100, and the y-axis values ​​are 0.50, 0.75, 1.00, 1.25, 1.50, 1.75, 2.00, 2.25, and 2.50. The x-axis values ​​of both ordinary clustering and climate-adaptive clustering in the high-density area are 685000 and 68, respectively. 7500, 690000, 692500, 695000, 697500, 700000, 702500; the ordinate (y) values ​​for both ordinary clustering and climate-adaptive clustering in high-density areas are 3.1075, 3.1100, 3.1125, 3.1150, 3.1175, 3.1200, 3.1225, 3.1250, 3.1275; the abscissa (x) values ​​for both ordinary clustering and climate-adaptive clustering in medium-density areas are 680000, 685000, 690000. The values ​​of the DB index x-axis in the medium-density region are 600, 800, 1000, 1200, and 1400, respectively; the values ​​of the DB index y-axis in the medium-density region are 0.7, 0.8, 0.9, 1.0, 1.1, 1.2, and 1.3; the values ​​of the y-axis for both ordinary clustering and climate-adaptive clustering in the medium-density region are 3.110, 3.115, 3.120, 3.125, and 3.130; the values ​​of the DB index x-axis in the low-density region are... The values ​​are 600, 800, 1000, 1200, and 1400. The values ​​of the y-axis of the DB index in the low-density area are 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, and 1.9. The values ​​of the x-axis of the ordinary clustering and climate-adaptive clustering in the low-density area are 680000, 685000, 690000, 695000, 700000, 705000, and 710000. The value of the y-axis of the ordinary clustering and climate-adaptive clustering in the low-density area is 3.110 × 10⁻⁶. 6 3.115×10 6 3.120×10 6 3.125×10 6 3.130×10 6 3.135×10 6 ClusterPoints represents cluster points, Iterations represents the number of iterations, and davies_bouldin_score represents the db index. Figure 4The graph shows the regression coefficient distribution of the influencing factors described in this invention. In the graph, the average population activity data from top to bottom are: -0.12 -0.09, 0.10 -0.29, 0.30 -0.45, 0.46 -0.63, 0.64 -0.86; the road density data from top to bottom are: 0.129 -0.134, 0.135 -0.138, 0.139 -0.143, 0.144 -0.148, 0.149 -0.154; the nighttime light data from top to bottom are: 0.260 -0.269, 0.270 -0.278, 0.279 -0.286, 0.287 -0.293, 0.294 -0.298; and the highest temperature data from top to bottom are: -0.43 -0.22, -0.21 -0.22. -0.10, -0.09 -0.02, 0.03 -0.15, 0.16 -0.30, and the humidity data from top to bottom are: -0.21 -0.12, -0.11 -0.01, 0.02 -0.16, 0.17 -0.28, 0.29 -0.40. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of the present invention can be combined with each other, and the described embodiments are only some embodiments of the present invention, not all embodiments.

[0026] Implementation Method 1, see Figure 1 This implementation method addresses the problems of insufficient climate adaptability, inaccurate analysis of driving mechanisms, and weak strategy targeting in existing planning. It proposes a subtropical city charging station layout optimization method based on climate adaptive clustering. The method includes: Step S1: Collect charging station POI, population heat data, OSM road network data, nighttime light data, summer average maximum temperature, and summer monthly average humidity data in the area enclosed by the subtropical city's ring expressway. After data cleaning and coordinate transformation, construct an equal-area hexagonal grid with a side length of 1000m. Step S2: Analyze the degree of charging station clustering using the kernel density estimation method, and combine it with the climate-adaptive DBSCAN clustering algorithm to obtain adaptation features; Step S3: Construct a multi-scale geographically weighted regression model with charging station density as the dependent variable to quantify the spatial heterogeneity impact of population activity, road density, nighttime light index, climate factors, and economic indicators. Step S4: Based on the clustering and regression results, propose a charging station layout scheme from the perspectives of driving factors, climate-adaptive hierarchical strategy, and climate adaptation.

[0027] Traditional methods treat climate as a homogeneous or secondary factor. This implementation, however, introduces a quantifiable climate correction mechanism to dynamically incorporate the actual impact of high temperature and humidity on electric vehicle range and equipment operation into the deployment decision. The direct effect is the ability to accurately identify and respond to short-distance, high-frequency charging needs caused by reduced range in summer. Simultaneously, it proactively avoids operational risks in high-humidity areas during equipment selection and site selection, thereby reducing resource mismatch and equipment failure rates at the source and improving the overall robustness of the charging network.

[0028] Existing technologies (such as OLS or basic GWR models) often struggle to capture the complex spatial patterns of various driving factors, especially climate factors. This invention, through the design of a multi-scale geographically weighted regression (MGWR) model, provides the first clear quantification of the spatial heterogeneity and scale of influence of various factors (such as population activity, road density, temperature, and humidity) on charging station layout. For example, the model results reveal which areas are primarily driven by high temperatures and which are primarily constrained by humidity, thus providing unprecedented scientific evidence for developing differentiated layout plans tailored to specific regions and overcoming the inaccuracy of existing technologies in driving analysis.

[0029] The method proposed in this implementation does not simply add climate variables to existing models, but rather constructs a completely new set of technical principles, realizing a paradigm shift from static geographic planning to climate-spatial dynamic collaborative planning. Traditional DBSCAN clustering or other spatial clustering methods rely on geometric distance or empirical values ​​for core parameters such as the neighborhood radius Eps, representing a static geospatial concept. It assumes that user behavior patterns (such as charging search radius) are constant. This implementation proposes the concept of climate-adaptive clustering. Its technical principle is to transform the climate physical constraint (range reduction caused by high temperatures) into a calculable algorithm correction parameter. Through formula calculation, the concept of range reduction, a behavioral economics and vehicle engineering concept, is quantified as a dynamic adjustment coefficient for the cluster neighborhood radius. This allows the clustering algorithm to simulate the behavioral changes of users seeking shorter charging distances in high-temperature environments, making the clustering results more realistically reflect the spatial distribution of demand under climate influence. This is a fundamental innovation that deeply integrates external environmental stimuli with internal algorithmic logic.

[0030] Existing spatial regression models (such as GWR) typically assume that all variables have the same scale of influence, i.e., they are fitted with a uniform bandwidth. This fails to distinguish which factors are globally stable (e.g., main road networks) and which are locally variable (e.g., population micro-distribution). The core principle of the MGWR model adopted in this implementation is to allow each independent variable to have its own independent bandwidth. This principle enables the model to automatically identify and separate the scales of influence of different driving factors. For example, the results of this embodiment show that road density and nighttime light are global scale factors with relatively stable effects; while population activity, temperature, and humidity are local scale factors with drastic spatial variations in their influence coefficients. This multi-scale separation capability allows us to accurately understand that climate factors (temperature, humidity) are not simply single positive or negative influences, but that their intensity and direction vary with spatial location, thus providing a fundamental theoretical basis for precise interventions based on spatial differentiation.

[0031] Implementation Method Two: This implementation method provides a complete implementation process for the climate-adaptive clustering-based subtropical city charging station layout optimization method described in Implementation Method One. The method includes: Step S1: Collect charging station POI, population heat data, OSM road network data, nighttime light data, summer average maximum temperature, and summer monthly average humidity data in the area enclosed by the subtropical city's ring expressway. After data cleaning and coordinate transformation, construct an equal-area hexagonal grid with a side length of 1000m. The data acquisition methods described in this embodiment include: Using Python and map APIs, we collected POI data for charging stations, including those categorized as supercharging, fast charging, slow charging, and battery swapping stations. Average population activity data for weekdays and weekends were obtained through the Baidu Smart Eye Population Heat Map Platform. Obtain road network data including urban expressways, arterial roads, and secondary arterial roads through OpenStreetMap (OSM); Data on average summer maximum temperature at 1km resolution and average summer humidity at 0.1° resolution were obtained from the National Tibetan Plateau Scientific Data Center; data on nighttime light intensity with annual average radiance were obtained from the National Oceanic and Atmospheric Administration (NOAA).

[0032] Data cleaning includes: Using Python and map APIs, we collected POI data for charging stations, including supercharging, fast charging, slow charging, and battery swapping stations. We then used address matching and spatial distance ≤50m as dual criteria to remove duplicate POI records. Obtain road network data including urban expressways, arterial roads, and secondary arterial roads, and delete dead-end roads and duplicate parallel roads in the OSM road network; The coordinate system was converted to WGS-1984. An equal-area hexagonal grid with a side length of 1000m was constructed using ArcGIS. Spatial overlay analysis was used to eliminate grids outside the study area, ultimately retaining 200-300 effective grids. Kernel density analysis was performed on charging station POI and road data, and then the density within the grid was statistically calculated using ArcGIS spatial connectivity tools. Population activity and nighttime light intensity were calculated as average values ​​within the grid using a tabular, zonal statistical method. Kriging interpolation was used to downscale the 1km resolution summer average maximum temperature to a 1000m grid, with interpolation errors controlled within ±0.5℃. Bilinear interpolation was used to resample the 0.1° resolution humidity data to a 0.008333° resolution, corresponding to approximately a 1km spatial scale, ensuring spatial matching with the grid. Finally, Z-score standardization was performed on continuous variables such as population activity and road density using SPSS to eliminate dimensional differences.

[0033] Step S2: Analyze the degree of charging station clustering using the kernel density estimation method, and combine it with the climate-adaptive DBSCAN clustering algorithm to obtain adaptation features; In this embodiment, the kernel density estimation method divides the kernel density results into high-density, medium-density, and low-density regions using the natural breakpoint method, which intuitively reflects the spatial agglomeration of charging stations and the distribution of hotspot areas. The implementation of the climate-adaptive DBSCAN clustering algorithm described in this embodiment includes: Using DMSAS software, for high, medium, and low density areas, the neighborhood radius parameters were traversed in a step size of 100m within a range of 500-1500m, and the optimal benchmark neighborhood radius was determined based on the lowest DB index. Eps base Find three optimal neighborhood radii respectively, and set the minimum number of neighborhood points MinPts to 3; Introducing the climate correction factor λ, the calculation formula is as follows: λ=1 / (1-δ) Where δ represents the range reduction rate of electric vehicles at high temperatures, set to δ=15%; according to domestic and international test reports, when the temperature reaches 35℃ or above, the range of electric vehicles decreases by an average of 10-20%. Finally, the high-temperature correction factor λ is calculated. hot =1 / (1-0.15)≈1.18; The neighborhood radius is adjusted adaptively based on the climate correction factor. The adjusted neighborhood radius is:

[0034] in, λ hot This is a high-temperature correction factor, keeping MinPts=3 constant, used for cluster analysis to match the high-temperature short-distance energy replenishment needs; Calculate the silhouette coefficient of the clustering results for each region to ensure the effectiveness of clustering after climate adaptation.

[0035] Step S3: Construct a multi-scale geographically weighted regression model with charging station density as the dependent variable to quantify the spatial heterogeneity impact of population activity, road density, nighttime light index, climate factors, and economic indicators. The construction of the multi-scale geographically weighted regression model described in this embodiment includes: Based on a hexagonal grid, a unique UID is assigned to each grid, and all variables are associated through the UID, which is then aggregated into a spatial-attribute integrated dataset. Establish model relationships:

[0036] in, Let the charging station density be the density of the i-th grid. For the spatial intercept term of the i-th grid, For grid coordinates, Let b be the spatial regression coefficient of the j-th influencing factor at location i. wj The value x represents the bandwidth parameter used in the calculation of this coefficient. ij For the j-th influencing factor of the i-th grid, Let be the error term for the i-th grid; Collinearity interference was eliminated by variance inflation factor test, and all VIF values ​​were <7.5.

[0037] The multi-scale geographic weighted regression model described in this embodiment also undergoes robustness verification. The hexagonal grid side lengths are adjusted to 800m and 1200m for repeated fitting, and the R-value is compared. 2 Adjust R 2 And the AICc index ensures model scale invariance.

[0038] Step S4: Based on the clustering and regression results, propose a charging station layout scheme from the perspectives of driving factors, climate-adaptive hierarchical strategy, and climate adaptation.

[0039] The driving factor orientation described in this embodiment includes: Based on the clustering results of the multi-scale geographical weighted regression model, global and local influencing factors are distinguished, and the global influencing factors are used as the basic framework for the layout of charging stations. The layout density of charging stations is optimized based on the spatial distribution of the coefficients of the factors with the largest absolute value of the regression coefficients among the local influencing factors.

[0040] The climate adaptive stratification strategy described in this embodiment includes: The value of λ is calculated based on δ, and the radius of the DBSCAN clustering benchmark neighborhood is determined. Eps baseIncrease by 10%-20% to accommodate the reduced battery life in summer; In high-density areas, the focus is on upgrading facilities; in medium-density areas, the focus is on filling coverage gaps; and in low-density areas, small and medium-sized charging stations are deployed. A dynamic analysis model is established to update the clustering results quarterly.

[0041] The preferred layout scheme for charging stations described in this embodiment is as follows: Based on the bandwidth results of the multi-scale weighted geographic regression model, global and local influencing factors are distinguished. Global influencing factors serve as the basic framework for charging station layout, providing fundamental coverage. Then, the factor with the highest influence intensity among the local influencing factors is identified as the core driving force affecting the spatial layout of charging stations. Based on the regression coefficient distribution of this factor, the city's charging station layout is optimized. A dynamic adjustment mechanism is established to adapt to the reduced driving range and equipment operation requirements in summer. The core is to calculate the climate correction coefficient based on the 10%-20% driving range reduction rate δ of electric vehicles in subtropical summers using the formula λ=1 / (1-δ), and expand the DBSCAN clustering baseline neighborhood radius Eps by 10%-20%. At the same time, it is further optimized for the differences between coastal and inland cities—coastal cities are affected by humidity, so Eps is increased by an additional 5%-8%, while inland cities focus on high temperature adaptation; temporary charging points are increased in summer to ensure that the charging needs after the reduction in driving range are covered.

[0042] The impact of high temperature and humidity on charging station layout is being monitored. Based on the results of the implementation examples, high temperatures in some areas promote charging station layout because they reduce the range of electric vehicles, creating a surge in demand for charging. Charging stations equipped with liquid-cooled supercharging piles and intelligent temperature control systems should be deployed in these areas. High-temperature off-peak charging discounts should be offered, and battery cooling and replenishment stations should be established. Simultaneously, newly built charging stations must be uniformly equipped with high-temperature protection and battery safety features. Data from meteorological and charging platforms should be integrated to conduct capacity scheduling and equipment inspections of charging stations in high-impact areas during high-temperature warnings, ensuring stable operation.

[0043] Implementation Method 3, see below Figures 2 to 4 This embodiment describes a specific implementation of the climate-adaptive clustering-based subtropical city charging station layout optimization method described in Embodiment 1. The implementation process includes: Study area delineation. The study area is defined as the area enclosed by the Changsha Ring Expressway, which accounts for more than 85% of Changsha's new energy vehicle ownership and charging demand, conforming to the analysis principle of focusing on core needs.

[0044] Multi-source data collection. Data was collected from multiple sources, including charging station POIs, population activity, road network, nighttime light, average monthly high temperature, and average monthly humidity. An equal-area hexagonal grid with sides of 1000m was constructed using ArcGIS, generating 228 effective grids. Various vector and raster data were spatially connected or statistically analyzed into the hexagonal grid to form an integrated attribute-spatial dataset.

[0045] Table 1: Data Sources

[0046] Data standardization was performed. The average maximum summer temperature at 1km resolution was downscaled to a 1000m grid using Kriging interpolation, with interpolation errors controlled within ±0.5℃. Humidity was resampled from 0.1° resolution to 0.008333° resolution to ensure grid matching. All continuous independent variables were standardized using SPSS Z-scores to eliminate dimensionality effects and ensure the stability of the MGWR model. Charging station POIs were identified as duplicates based on address matching and spatial distance ≤50m, ultimately retaining only valid POIs. Dead-end roads and duplicate parallel roads were removed from the OSM road network.

[0047] Analyze the layout characteristics of charging stations. The specific steps are as follows: First, the kernel density estimation tool of ArcGIS 10.8 platform was used to analyze the results. The results showed that the distribution of charging stations exhibited a clear center-periphery structure. The kernel density results were divided into three levels: high density, medium density and low density areas by the natural breakpoint method.

[0048] Secondly, a clustering algorithm was implemented based on DMSAS software, and climate adaptive parameter optimization was performed. The parameter optimization process is as follows: 1) Baseline parameters are determined by optimizing parameters for high, medium, and low density regions respectively, ultimately determining the optimal neighborhood radius Eps. base ; 2) Based on the characteristics of subtropical cities, the climate correction coefficient λ is calculated using the formula: λ = 1 / (1-δ), where δ is the electric vehicle range reduction rate, δ = 15%. The calculated λ... hot =1.18. The formula for calculating the adaptive neighborhood radius is: Eps adj =λ hot ×Eps base The adjusted neighborhood radius is obtained.

[0049] Table 2: Clustering Results

[0050] 3) Finally, calculate the silhouette coefficient of each clustering result to evaluate the clustering quality.

[0051] A variable system was constructed to analyze the driving factors of charging station distribution. Using the density of charging stations within a grid as the dependent variable, seven core independent variables were selected, taking into account both common driving factors and climate-specific characteristics.

[0052] Table 3: Construction of the variable system

[0053] Model validation and robustness analysis. First, a multicollinearity test was performed on each variable. The variance inflation factor was calculated using SPSS. The VIF values ​​of all independent variables were <7.5, indicating no multicollinearity issues.

[0054] Next, the model fitting was compared. Standardized data was imported into MGWR2.2 software, and the goodness of fit of OLS, GWR, and MGWR models was compared. The results showed that MGWR's goodness of fit was significantly better than OLS and GWR, as follows:

[0055] Grid scale sensitivity analysis was performed, and the hexagonal grid side lengths were adjusted to 800m and 1200m for repeated model fitting. The R-value for a 1000m grid was analyzed. 2 At the highest scale, the selected independent variables have the strongest explanatory power for the distribution of charging stations.

[0056]

[0057] Import the MGWR results table into ArcGIS, select the Bete coefficient column for each variable, plot the regression coefficients for each variable, and interpret the driving mechanism.

[0058] Table 4: Bandwidth and Mean Regression Coefficients Results

[0059] Based on the model results, the factors affecting the POI density of charging stations can be divided into two categories according to their spatial scale: At the local scale, average population activity, maximum temperature, and humidity are located in a relatively small bandwidth range, indicating that their spatial heterogeneity is relatively large and the spatial variation range of their influence coefficients is relatively large. At the global scale, road density and nighttime light are located in a relatively large bandwidth range, indicating that the spatial heterogeneity of these factors is relatively small and the spatial variation range of their influence coefficients is relatively small.

[0060] The regression coefficients show that the order of influence intensity is: average population activity > nighttime light index > road density > humidity > maximum temperature. Among these, population activity intensity and regional activity level are the core driving forces for charging station layout. Average population activity exhibits significant spatial heterogeneity, requiring close attention to the differences in facility demand based on local population structure. Humidity and maximum temperature also show significant spatial heterogeneity, necessitating layout adaptation to subtropical climate characteristics.

[0061] This paper proposes a charging station layout optimization strategy adapted to Changsha. Specifically, it optimizes the charging station layout from three aspects: driving factor orientation, climate-adaptive hierarchical layout, and climate adaptation, ultimately building an efficient and user-friendly urban charging infrastructure system.

[0062] In terms of driving factors, road density and nighttime lighting are the decisive factors for the basic network framework. Their impact is global in scale, with small spatial variations in regression coefficients, indicating stable influence. Priority should be given to basic coverage based on the city's main road network and built-up areas with high nighttime lighting indices to ensure the accessibility and fairness of charging services. Average population activity is the primary driving factor, and its impact exhibits significant local heterogeneity. Population activity gradually decreases from the city center to the periphery, while the regression coefficient changes from negative to positive and gradually increases. Therefore, in the central area, new construction should be strictly limited, and the focus should be on upgrading existing infrastructure. At core locations, ordinary fast charging stations should be replaced with supercharging piles, and battery swapping station models should be explored. Time-of-use pricing should be used to guide vehicles to charge during off-peak hours, improving the efficiency of the existing network. In emerging areas, urban development can proceed in tandem, with fast charging stations deployed in a grid pattern at 1-2 km intervals around key nodes to quickly capture growing demand. In peripheral potential areas, demand is limited but development potential is huge. Anchor-style deployment should be based at ring road entrances and exits, commercial districts, and key town centers, using asset-light models such as co-construction to build the basic network framework, reserving space for future expansion.

[0063] In terms of climate-adaptive hierarchical layout, climate-corrected clustering analysis results are used to guide the division of service areas in different density zones. In high-density areas, since both the ordinary and climate-corrected clustering profile coefficients are 0.5704, the clustering quality is optimal. The focus should be on upgrading facilities in core business districts and transportation hubs. High-demand nodes should be identified using the climate-corrected clustering neighborhood radius of 1062m, and supercharging stations and smart charging stations should be added. In medium-density areas, the rationality of climate-corrected clustering should be improved. Coverage should be expanded and gaps filled in conjunction with climate considerations, and rainproof and heat dissipation facilities should be configured, with dynamic adjustments to operation and maintenance plans. In low-density areas such as urban suburbs, small and medium-sized charging stations should be precisely deployed in streets, county and township centers, and equipped with battery preheating and insulation facilities, based on the climate-corrected clustering neighborhood radius of 802.4m. Long-term development requires establishing a dynamic analysis model of climate and charging demand, updating clustering results quarterly to dynamically adjust construction and operation strategies, and developing differentiated construction standards for different density zones to achieve efficient resource allocation.

[0064] Regarding the impact of humidity on charging station layout, higher humidity levels have a greater negative impact on the layout. To mitigate this, excessive deployment in areas with strong negative humidity influences should be avoided. Construction should be limited to densely populated core areas, utilizing IP67-rated moisture-proof equipment and regular dehumidification maintenance. Market-based models such as time-of-use pricing and membership priority should be implemented to improve utilization rates. In areas with weak negative humidity influences, denser construction should be prioritized, focusing on integrated fast-charging and supercharging stations. A coordinated monitoring system linking humidity, demand, and facilities should be established to classify the moisture-proof requirements of charging equipment based on the intensity of the negative humidity influences in different areas. In the event of a sudden increase in humidity, emergency moisture-proof plans should be activated for charging stations in high-influence areas to ensure stable facility operation.

[0065] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0066] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1The steps of the function specified in one or more boxes.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure and not to limit its protection scope. Although this disclosure has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading this disclosure, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the protection scope of the published pending claims.

Claims

1. A method for optimizing the layout of charging stations in subtropical cities based on climate-adaptive clustering, characterized in that, The method includes: Step S1: Collect charging station POI, population heat data, OSM road network data, nighttime light data, summer average maximum temperature, and summer monthly average humidity data in the area enclosed by the subtropical city's ring expressway. After data cleaning and coordinate transformation, construct an equal-area hexagonal grid with a side length of 1000m. Step S2: Analyze the degree of charging station clustering using the kernel density estimation method, and combine it with the climate-adaptive DBSCAN clustering algorithm to obtain adaptation features; Step S3: Construct a multi-scale geographically weighted regression model with charging station density as the dependent variable to quantify the spatial heterogeneity impact of population activity, road density, nighttime light index, climate factors, and economic indicators. Step S4: Based on the clustering and regression results, propose a charging station layout scheme from the perspectives of driving factors, climate-adaptive hierarchical strategy, and climate adaptation.

2. The method for optimizing the layout of charging stations in subtropical cities based on climate adaptive clustering according to claim 1, characterized in that, The data cleaning includes: Acquire POI data for charging stations including supercharging, fast charging, slow charging, and battery swapping stations. Use address matching and spatial distance ≤50m as dual criteria to remove duplicate POI records. Obtain road network data including urban expressways, arterial roads, and secondary arterial roads, and delete dead-end roads and duplicate parallel roads in the OSM road network; Climate data were downscaled to a 1000m grid using Kriging interpolation, with interpolation errors controlled within ±0.5℃. Z-score standardization is used to eliminate the dimensions of continuous independent variables.

3. The method for optimizing the layout of charging stations in subtropical cities based on climate adaptive clustering according to claim 1, characterized in that, The implementation of the climate-adaptive DBSCAN clustering algorithm includes: For high, medium, and low density areas, the neighborhood radius parameters are traversed in a step size of 100m within a range of 500-1500m, and the optimal benchmark neighborhood radius is determined based on the lowest DB index. Eps base The minimum number of neighboring points, MinPts, is set to 3. Introducing the climate correction factor λ, the calculation formula is as follows: λ=1 / (1-δ) Where δ represents the range reduction rate of electric vehicles under high temperature, and is set to δ=15%; The neighborhood radius is adjusted adaptively based on the climate correction factor. The adjusted neighborhood radius is: in, λ hot This is a high-temperature correction factor, keeping MinPts=3 constant, used for cluster analysis to match the high-temperature short-distance energy replenishment needs; Calculate the silhouette coefficient of the clustering results for each region to ensure the effectiveness of clustering after climate adaptation.

4. The method for optimizing the layout of charging stations in subtropical cities based on climate adaptive clustering according to claim 1, characterized in that, The construction of the multi-scale geographically weighted regression model includes: Based on a hexagonal grid, a unique UID is assigned to each grid, and all variables are associated through the UID, which is then aggregated into a spatial-attribute integrated dataset. Establish model relationships: in, Let the charging station density be the density of the i-th grid. For the spatial intercept term of the i-th grid, For grid coordinates, Let x be the spatial regression coefficient of the j-th influencing factor at position i, where bwj represents the bandwidth parameter used in calculating this coefficient, and x is the spatial regression coefficient of the j-th influencing factor at position i. ij For the j-th influencing factor of the i-th grid, Let VIF be the error term for the i-th grid; VIF values ​​< 7.5 are eliminated by variance inflation factor test to remove collinearity interference.

5. The method for optimizing the layout of charging stations in subtropical cities based on climate adaptive clustering according to claim 1, characterized in that, The multi-scale geographically weighted regression model was also robustly validated by adjusting the hexagonal grid side lengths to 800m and 1200m for repeated fitting, and comparing the R-values. 2 Adjust R 2 And the AICc index ensures model scale invariance.

6. The method for optimizing the layout of charging stations in subtropical cities based on climate adaptive clustering according to claim 1, characterized in that, The driving factors include: Based on the clustering results of the multi-scale geographical weighted regression model, global and local influencing factors are distinguished, and the global influencing factors are used as the basic framework for the layout of charging stations. The layout density of charging stations is optimized based on the spatial distribution of the coefficients of the factors with the largest absolute value of the regression coefficients among the local influencing factors.

7. The method for optimizing the layout of charging stations in subtropical cities based on climate adaptive clustering according to claim 3, characterized in that, The climate adaptive stratification strategy includes: The value of λ is calculated based on δ, and the radius of the DBSCAN clustering benchmark neighborhood is determined. Eps base Increase by 10%-20% to accommodate the reduced battery life in summer; In high-density areas, the focus is on upgrading facilities; in medium-density areas, the focus is on filling coverage gaps; and in low-density areas, small and medium-sized charging stations are deployed. A dynamic analysis model is established to update the clustering results quarterly.

8. A subtropical city charging station layout optimization system based on climate adaptive clustering, characterized in that, The system includes: Data acquisition unit: used to collect charging station POI, population heat data, OSM road network data, nighttime light data, summer average maximum temperature, and summer monthly average humidity data in the area enclosed by the ring expressway of subtropical cities. After data cleaning and coordinate transformation, a hexagonal grid with a side length of 1000m is constructed. Analysis Unit: Used to analyze the degree of charging station clustering using the kernel density estimation method, and combined with the climate-adaptive DBSCAN clustering algorithm to obtain adaptive features; Quantification Unit: Used to construct a multi-scale geographically weighted regression model with charging station density as the dependent variable, quantifying the spatial heterogeneity of population activity, road density, nighttime light index, climate factors, and economic indicators; Layout optimization unit: Based on clustering and regression results, optimize the layout scheme of charging stations from the perspectives of driving factors, climate-adaptive hierarchical strategy, and climate adaptation.

9. A computer device, characterized in that: The device includes a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes a method for optimizing the layout of subtropical urban charging stations based on climate adaptive clustering as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of a method for optimizing the layout of charging stations in subtropical cities based on climate adaptive clustering as described in any one of claims 1-7.

Images