Intelligent site selection method for urban and rural parcel loading and unloading sites based on radiation range

Abstract

The invention relates to the technical field of logistics management, in particular to an intelligent site selection method for urban and rural parcel loading and unloading sites based on a radiation range, and the method comprises the steps: obtaining geographical coordinates, business attributes and passing time sequence data of routes of demand points and candidate sites; calculating a reference transit time, a peak transit time, and a route instability index based on the time series data; calculating a demand priority index based on the business attribute and matching a dynamic service timeliness threshold; constructing a traffic cost function after asymmetric risk adjustment, and applying a penalty related to priority and instability to a service failure condition in which peak time exceeds an aging threshold; urban and rural demand points are separated based on priorities, and a station with the lowest total risk cost is searched by applying a K-center point algorithm and using a traffic cost function. Through the technical scheme of the invention, the service toughness of the selected site can be high, and the practical stability of the radiation range under various time-varying conditions is ensured.

Description

Technical Field

[0001] This application relates to the field of logistics management technology, and in particular to an intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range. Background Technology

[0002] In the integrated urban and rural logistics network, parcel loading and unloading stations are the core hubs connecting the main lines and the last mile. The location of the station directly determines its coverage area, that is, the geographical area that can provide efficient and reliable services. A reasonable location plan is the cornerstone of controlling operating costs and ensuring service quality.

[0003] Currently, existing technologies often employ clustering or heuristic optimization methods for site selection. The core of these methods is to construct a cost matrix to evaluate the delivery cost from each candidate site to each demand point. This cost is usually defined as the geographic straight-line distance between two points or the average travel time calculated based on Geographic Information System (GIS) road network data.

[0004] However, this site selection method, which relies on static averages, suffers from serious flaws and technical problems in complex urban and rural environments. Urban and rural transportation networks exhibit high heterogeneity, with highways, national roads, and rural dirt roads coexisting, resulting in vastly different traffic efficiencies and significant time-varying characteristics. For example, market days in towns and villages can cause periodic road congestion at specific times; the large number of slow-moving agricultural machines during busy farming seasons severely impacts the efficiency of rural roads; and severe weather can even lead to the closure of some low-grade roads. Existing technologies use static average travel time as a fixed value, completely ignoring this dynamic fluctuation. An optimal site selected based on this average might happen to depend on a route with a short average travel time but extremely high peak risk, posing a significant service resilience risk to logistics operations. Summary of the Invention

[0005] To address the technical problem that existing site selection methods rely on static average travel time, resulting in poor service resilience, this application provides an intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range.

[0006] This application provides an intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range, comprising: acquiring the geographical coordinates, local coverage population, and average daily parcel volume of multiple demand points; acquiring the geographical coordinates of multiple candidate stations; and acquiring time-series data of travel time for multiple routes between the demand points and the candidate stations; traversing multiple routes and calculating the baseline travel time, peak travel time, and route instability index for each route based on the time-series data of travel time; calculating the demand priority index for each demand point based on the local coverage population and the average daily parcel volume, and dynamically matching a dedicated service timeliness threshold for each demand point based on the demand priority index; constructing an asymmetric risk-adjusted travel cost function, wherein the travel cost function uses the peak travel time as the base cost and applies a penalty related to both the demand priority index and the route instability index to cases where the peak travel time exceeds the service timeliness threshold; separating the demand points into a high-priority demand point set and a low-priority demand point set based on the demand priority index, and applying the K-centroid algorithm within the respective candidate station ranges and using the travel cost function as the cost to find the parcel loading and unloading station with the lowest total risk cost.

[0007] This application can proactively avoid high-risk routes with drastic fluctuations in travel time, and the site selection result changes from average optimal to stable optimal, which greatly improves the service resilience and the actual stability of the coverage area under various time-varying conditions.

[0008] In one embodiment, the baseline passage time is the median of the passage time time series data; the peak passage time is the 95th percentile of the passage time time series data.

[0009] By using the median as the baseline time, the impact of a single extreme outlier can be effectively resisted, and the normal performance can be represented more robustly. By using the 95th percentile as the peak time, the service ceiling and worst-case scenario of the route can be accurately grasped, providing a realistic basis for service level agreement commitments.

[0010] In one embodiment, the route instability index satisfies the following relationship: ;in, To be from candidate stations To the point of demand The route instability index, To be from candidate stations To the point of demand Peak traffic time To be from candidate stations To the point of demand The baseline travel time.

[0011] This index assesses the relative volatility risk of route travel time; a higher index indicates a more unstable route and a higher risk.

[0012] In one embodiment, the demand priority index satisfies the following relationship: ;in, For demand points The demand priority index For demand points The comprehensive density index is calculated based on the local coverage population and the average daily parcel volume. and These are the mean and standard deviation of the overall density of all demand points, respectively.

[0013] This index assesses the dual characteristics of high-density urban areas and low-density rural areas as a smooth index between [0,1], providing accurate input for subsequent matching of differentiated service standards.

[0014] In one embodiment, the service timeliness threshold satisfies the following relationship: ;in, For demand points Service timeliness threshold For the longest time limit in the entire rural area, The strictest time limit for all towns and cities in the region. For demand points The demand priority index.

[0015] Through linear interpolation, the radiation range was successfully transformed from a fixed time value into a refined timeliness indicator that dynamically changes with local demand density and is point-to-point.

[0016] In one embodiment, the travel cost function satisfies the following relationship: in, To be from candidate stations To the point of demand The cost of passage To be from candidate stations To the point of demand Peak traffic time For demand points The demand priority index To be from candidate stations To the point of demand The route instability index, For demand points Service timeliness threshold This is a preset penalty constant.

[0017] By combining reliable operating costs with service failure penalties, and linking the severity of the penalties to priority and instability, the model is forced to avoid the trap route with short average time but high peak risk.

[0018] In one embodiment, when the demand priority index approaches 0, the penalty approaches 0, and the passage cost approaches the peak passage time.

[0019] In one embodiment, the high-priority demand point set is the demand points whose demand priority index is greater than a preset high threshold; the low-priority demand point set is the demand points whose demand priority index is not greater than the preset high threshold.

[0020] In one embodiment, the comprehensive density index satisfies the following relationship: ; in, For demand points Local coverage population, For demand points The average daily parcel volume Preset weights.

[0021] The technical solution of this application has the following beneficial technical effects: This application can automatically apply strict timeliness standards to high-priority urban areas while applying lenient coverage standards to rural areas. This solves the resource misallocation problem caused by the traditional one-size-fits-all site selection and actively avoids high-risk routes with drastic fluctuations in travel time. The site selection result is no longer average optimal but stable optimal, ensuring the reliability of the station's coverage area under various complex time-varying conditions.

[0022] Furthermore, by linking penalties to both priority and instability, this invention achieves asymmetric penalties, imposing significant costs only when high-priority towns experience service failures on highly unstable routes. This shifts site selection decisions from cost-oriented to risk-averse and service quality-oriented, greatly enhancing the practical rationality of site selection. Attached Figure Description

[0023] Figure 1 This is a flowchart of an intelligent site selection method for urban and rural parcel loading and unloading stations based on the radiation range, according to an embodiment of this application.

[0024] Figure 2 This is a schematic diagram illustrating the comparison between the site selection method of the present invention and the traditional site selection method. Detailed Implementation

[0025] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0026] Figure 1 This is a flowchart illustrating an intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range, according to an embodiment of this application. Figure 1 As shown, the intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range includes steps S101 to S105, which are described in detail below.

[0027] S101: Obtain the geographical coordinates, local coverage population, and average daily parcel volume of multiple demand points; obtain the geographical coordinates of multiple candidate sites; and obtain the time series data of travel time for multiple routes between demand points and candidate sites.

[0028] In one embodiment, the geographical coordinates of multiple package delivery points, such as villages and communities, are obtained, i.e., demand points. For example, there are 100 demand points in an area. Each demand point... It must be associated with its business attributes, including: the local coverage population, for example Covering 5,000 people, Covering 200 people, and the average daily parcel volume, for example... An average of 1000 pieces per day An average of 10 cases per day. Simultaneously, a candidate site set is obtained, which consists of the geographic coordinates of multiple potential site construction locations. For example, there are four candidate sites, such as transportation hubs and available storefronts in town centers. It is worth noting that the geographic coordinates are used to determine the route between the demand point and the candidate sites, which is the basis for subsequent calculations of travel time.

[0029] In this optional embodiment, for each possible delivery route, exemplarily, such as from a candidate station To the point of demand of For each route, by batch accessing real-time traffic data from map service providers or analyzing historical GPS trajectory data of the fleet at different times, such as weekday morning rush hour, market mornings, off-peak hours, and nighttime, the time-series data of its travel time over the past quarter can be obtained. For example, the time-series data for route (1, 1) might be [25, 28, 26, 90, 30, 27, ...] (unit: minutes).

[0030] In this way, by collecting basic data on demand points and candidate sites, as well as dynamic travel time series data, unreliable static average values ​​are replaced, providing a realistic and comprehensive data foundation for the subsequent construction of dynamic risk models.

[0031] S102, traverse multiple routes and calculate the baseline travel time, peak travel time and route instability index for each route based on travel time time series data.

[0032] In one embodiment, statistical analysis can be performed on the time-series data of each route to calculate the baseline travel time, peak travel time, and route instability index for each route. The baseline travel time is the median of the travel time time-series data, which more robustly represents the route's typical performance; the peak travel time is the 95th percentile of the travel time time-series data, representing the route's worst-case performance in 95% of cases; and the route instability index is used to assess the relative volatility risk of each route.

[0033] Specifically, the route instability index satisfies the following relationship: in, To be from candidate stations To the point of demand The route instability index, To be from candidate stations To the point of demand Peak traffic time To be from candidate stations To the point of demand The baseline travel time.

[0034] For example, for the treacherous route A, assuming the median of the obtained travel time time series data is 30 minutes and the 95th percentile is 150 minutes, the instability index can be calculated as 4.84; while for the safe route B, the median is 40 minutes and the 95th percentile is 45 minutes, so its instability index can be calculated as 1.098.

[0035] Thus, by calculating the baseline time, peak time, and instability index, key risk indicators for assessing service resilience were successfully extracted from time-series data, laying the foundation for the subsequent construction of a cost model.

[0036] S103 calculates the demand priority index for each demand point based on the local coverage population and average daily parcel volume, and dynamically matches a dedicated service timeliness threshold for each demand point based on the demand priority index.

[0037] In one embodiment, a comprehensive density index corresponding to the demand point can be calculated based on the acquired local coverage population and average daily parcel volume data. The comprehensive density index satisfies the following relationship: in, For demand points The comprehensive density index, For the local population coverage, This represents the average daily parcel volume. For example, the preset weight is set to a value of 1. .

[0038] Furthermore, to transform the comprehensive density index into a smooth demand priority index between [0,1], a hyperbolic tangent function can be used for normalization. The demand priority index satisfies the following relationship: in, For demand points The demand priority index and These are the mean and standard deviation of the overall demand density across all demand points, respectively. It is a Z-Score standardization of the overall density.

[0039] For example, for urban locations, assuming a local coverage population of 5000 and an average daily parcel volume of 1000, the comprehensive density index is 0.5 × 5000 + 0.5 × 1000 = 3000; for rural locations, assuming a local coverage population of 200 and an average daily parcel volume of 10, the comprehensive density index is 0.5 × 200 + 0.5 × 10 = 105; assuming all 100 locations... It is 800. If the value is 600, then the demand priority index for urban areas can be calculated as 0.999, corresponding to high priority, while the demand priority index for rural areas is 0.09, corresponding to low priority.

[0040] Furthermore, based on the obtained demand priority index, a dedicated service timeliness threshold can be dynamically matched for each demand point, specifically by setting two global parameters. and Exemplary The time limit is 45 minutes, a strict time requirement that towns must meet; 120 minutes is the longest acceptable service time in rural areas. Therefore, the service time threshold can be calculated using linear interpolation as follows: in, For demand points Service timeliness threshold For the longest time limit in the entire rural area, This is the strictest time limit for all towns and cities.

[0041] For example, the demand priority index for urban locations is 0.999, while that for rural locations is 0.09. Therefore, the service timeliness threshold for urban locations can be calculated to be 120. (120 45) × 1 = 45 minutes; the service timeliness threshold for rural service points is 120. (120 45) × 0.09 = 113.25 minutes.

[0042] In this way, by calculating the demand priority index and dynamic service timeliness threshold, differentiated service standards are established for the urban-rural dual demand.

[0043] S104. Construct an asymmetric risk-adjusted toll cost function. The toll cost function uses peak toll time as the base cost and imposes a penalty on cases where the peak toll time exceeds the service timeliness threshold, which is related to both the demand priority index and the route instability index.

[0044] In one embodiment, an asymmetric risk-adjusted passage cost function can be constructed based on the obtained peak passage time, instability index, priority index, and service timeliness threshold. The passage cost function satisfies the following relationship: in, To be from candidate stations To the point of demand The cost of passage To be from candidate stations To the point of demand Peak traffic time For demand points The demand priority index To be from candidate stations To the point of demand The route instability index, For demand points Service timeliness threshold This is a preset penalty constant, which takes a large value to represent the huge losses caused by timeouts. For example, the value is 1000. This is used to calculate the absolute timeout. If there is no timeout, this item is 0, and the entire penalty item is 0.

[0045] For example, for town points The value is 1. With a peak travel time of 45 minutes, when encountering the treacherous route A, the corresponding peak travel time is 150 minutes, the route instability index is 4.84, and the corresponding absolute timeout is 105 minutes, with a timeout ratio of 105 / 150 = 0.7. The final calculated travel cost is 2586.4. However, if the safe route B is taken, the corresponding peak travel time is 45 minutes, the route instability index is 1.098, and the corresponding absolute timeout is 0 minutes, with a timeout ratio of 0. The final calculated travel cost is 45. Therefore, for towns, route B with a cost of 45 will be chosen to avoid the more expensive route A.

[0046] Thus, by constructing an asymmetric risk cost function, a precise basis for risk avoidance decision-making is provided for site selection.

[0047] S105 separates demand points into a high-priority demand point set and a low-priority demand point set based on the demand priority index, and applies the K-centroid algorithm within their respective candidate site ranges, using the passage cost function as the cost, to find the parcel loading and unloading site with the lowest total risk cost.

[0048] In one embodiment, the K-Medoids algorithm can be used, with its cost function replaced by an asymmetric risk-adjusted passage cost function, employing a split clustering strategy. The split site selection is used to prevent high-density urban demand points from draining sites that should be serving rural areas.

[0049] Specifically, set a target; for example, set the number of target sites in urban areas to 1 and the number of target sites in rural areas to 1. Then perform separate clustering to... Demand points exceeding a preset high threshold are designated as high-priority demand points, while those not exceeding the preset high threshold are designated as low-priority demand points. For example, the preset high threshold is 0.6. For urban site selection: candidate stations in towns are extracted based on their geographical location, and within each candidate station's range, the K-centroid algorithm is run using the travel cost function as the cost to find the parcel loading and unloading station with the lowest total risk cost. One station is selected from the town candidate stations. For rural site selection: candidate stations in rural areas are extracted, and within each candidate station's range, the K-centroid algorithm is run using the travel cost function as the cost to find the parcel loading and unloading station with the lowest total risk cost. One station is selected from the rural candidate stations. Finally, the urban and rural site selection results are merged to obtain the final set of stations.

[0050] like Figure 2The diagram illustrates a comparison between the site selection method of this invention and the traditional site selection method. The left sub-figure shows the effect of traditional site selection based on average time, while the right sub-figure shows the effect of site selection based on this invention. It can be seen that the traditional site selection model is misled by the low average cost of the "trapped site," selecting a site where service quality would catastrophically collapse during peak hours, making the connection of most high-priority customers high-risk. In contrast, this invention successfully identifies the risk of the "trapped site," penalizes it, and selects a safe site with controllable costs, ensuring the high resilience of the coverage area for high-priority customers.

[0051] Thus, by employing a split K-centroid algorithm and applying asymmetric risk cost, it is possible to select sites that can reliably cover rural areas and have the strongest service resilience while meeting the stringent time requirements of urban areas.

[0052] It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the scope of protection of this application. Therefore, the scope of protection of this patent application shall be determined by the appended claims.

Claims

1. A method for intelligent site selection of urban and rural parcel loading and unloading stations based on radiation range, characterized in that, include: Obtain the geographic coordinates, local coverage population, and average daily parcel volume of multiple demand points; obtain the geographic coordinates of multiple candidate sites; and obtain the travel time time series data of multiple routes between the demand points and the candidate sites. Traverse multiple routes and calculate the baseline travel time, peak travel time, and route instability index for each route based on the travel time time series data; The demand priority index for each demand point is calculated based on the local coverage population and the average daily parcel volume, and a dedicated service timeliness threshold is dynamically matched for each demand point based on the demand priority index. Construct an asymmetric risk-adjusted toll cost function, which uses the peak toll time as the base cost and imposes a penalty on cases where the peak toll time exceeds the service timeliness threshold, which is related to both the demand priority index and the route instability index. Based on the demand priority index, demand points are separated into a high-priority demand point set and a low-priority demand point set. Within their respective candidate site ranges, the K-centroid algorithm is applied and the passage cost function is used as the cost to find the parcel loading and unloading site with the lowest total risk cost.

2. The intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range according to claim 1, characterized in that, The baseline passage time is the median of the passage time time series data; the peak passage time is the 95th percentile of the passage time time series data.

3. The intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range according to claim 1, characterized in that, The route instability index satisfies the following relationship: in, To be from candidate stations To the point of demand The route instability index, To be from candidate stations To the point of demand Peak traffic time To be from candidate stations To the point of demand The baseline travel time.

4. The intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range according to claim 1, characterized in that, The demand priority index satisfies the following relationship: in, For demand points The demand priority index For demand points The comprehensive density index is calculated based on the local coverage population and the average daily parcel volume. and These are the mean and standard deviation of the overall density of all demand points, respectively.

5. The intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range according to claim 1, characterized in that, The service timeliness threshold satisfies the following relationship: in, For demand points Service timeliness threshold For the longest time limit in the entire rural area, The strictest time limit for all towns and cities in the region. For demand points The demand priority index.

6. The intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range according to claim 1, characterized in that, The passage cost function satisfies the following relationship: in, To be from candidate stations To the point of demand The cost of passage To be from candidate stations To the point of demand Peak traffic time For demand points The demand priority index To be from candidate stations To the point of demand The route instability index, For demand points Service timeliness threshold This is a preset penalty constant.

7. The intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range according to claim 6, characterized in that, When the demand priority index approaches 0, the penalty approaches 0, and the passage cost approaches the peak passage time.

8. The intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range according to claim 1, characterized in that, The high-priority demand point set consists of demand points whose demand priority index is greater than a preset high threshold; the low-priority demand point set consists of demand points whose demand priority index is not greater than a preset high threshold.

9. The intelligent site selection method for urban and rural parcel loading and unloading stations based on radiation range according to claim 4, characterized in that, The comprehensive density index satisfies the following relationship: in, For demand points Local coverage population, For demand points The average daily parcel volume Preset weights.

Images