An end delivery path planning method based on historical trajectory

Abstract

The application discloses an end distribution path planning method based on historical trajectories, comprising: acquiring all express address and distribution trajectory point data in a preset statistical period; constructing a total set of logistics end delivery nodes according to all express addresses; obtaining a closed distribution trajectory based on the distribution trajectory point data, and constructing a residence node set corresponding to each closed distribution trajectory; establishing an association relationship between a delivery node and a residence node for each pickup and delivery operation in the preset statistical period, and giving the delivery node a start time and an end time; acquiring a closed pickup and delivery path of a single pickup and delivery operation, splitting the closed pickup and delivery path into a plurality of delivery path segments and storing the delivery path segments until the storage of the delivery path segments of all pickup and delivery operations is completed, and obtaining a distribution network connectivity table according to the stored data; and planning an end distribution path based on the distribution network connectivity table. The application can use the long-term accumulated distribution knowledge of pickup and delivery personnel for end distribution path planning, and can effectively improve the distribution efficiency and quality.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of logistics management and spatial big data processing, and particularly relates to an end delivery path planning method based on historical trajectories. BACKGROUND

[0002] At present, modern logistics business has become the basis of e-commerce, deeply integrated into modern production and social life, and greatly changed people's shopping way. In order to continuously improve the logistics quality and customer experience, fast and accurate pick-up has become the pursuit goal of logistics enterprises. However, due to the huge express business volume, the large number of customer locations, the complex delivery road conditions, the imperfect road network required by the end delivery, even the lack of the road network, the rapid change of transportation conditions, etc., the end delivery from the logistics service network or station to the customer "last mile" is the most difficult part of the pick-up and delivery task, and directly affects the express service quality. In order to facilitate management and improve the pick-up efficiency, usually a logistics service network divides the service range into multiple fixed delivery sections (or delivery fences), and each section is usually responsible for delivery by a pick-up staff. In the delivery operation, the pick-up staff first plans the delivery order according to the express addresses and their spatial positions to be delivered in the section, i.e. the delivery path, then loads the express according to the reverse order of the delivery, and finally starts the delivery operation according to the planned delivery path.

[0003] Although the spatial information technology (such as geographic information technology) and logistics delivery have been highly integrated, the pick-up staff in the section can plan the delivery path with the help of modern navigation system, but this delivery path planning mode still faces many challenges and difficulties. These difficulties include: (1) when the service network modifies the electronic fence or adjusts the pick-up staff, the pick-up staff needs to explore for a long time to be familiar with the distribution state of the customers in the section, the surrounding traffic environment and other conditions, so as to efficiently arrange the delivery path; (2) when the section lacks traffic network data or the customers are far away from the traffic route or the road changes quickly, the navigation system is difficult to plan a suitable delivery route; (3) the delivery path experience accumulated by the end pick-up staff is difficult to record and inherit. SUMMARY

[0004] To solve the above technical problems, the present application provides an end delivery path planning method based on historical trajectories to solve the problems existing in the prior art.

[0005] To achieve the above purpose, the present application provides an end delivery path planning method based on historical trajectories, comprising:

[0006] obtaining all express addresses and end section pick-up staff delivery trajectory point data of the logistics end delivery section pick-up in a preset statistical period;

[0007] spatially aggregating the all express address after preprocessing, constructing a total set of logistics end delivery nodes based on the aggregation result, obtaining a closed distribution track based on the distribution track point data, and constructing a residence node set corresponding to each closed distribution track;

[0008] establishing an association between the delivery node and the residence node of each pick-up and delivery operation within a preset statistical period, and assigning a start time and an end time to the delivery node;

[0009] obtaining a closed pick-up and delivery path of a single pick-up and delivery operation, splitting the closed pick-up and delivery path into a plurality of delivery path segments and storing the delivery path segments, until the storage of the delivery path segments of all pick-up and delivery operations is completed, obtaining a path segment data table, and obtaining a distribution network connectivity table according to the path segment data table;

[0010] planning an end distribution path based on the distribution network connectivity table.

[0011] Optionally, the process of spatially aggregating the all express address after preprocessing includes:

[0012] converting the express address into geographic coordinates, and constructing an express element set; obtaining an interest surface of a logistics end distribution path segment, merging the express addresses of each interest surface into one delivery node; using a spatial clustering algorithm to fuse the express addresses that are not covered by the interest surface and are spatially adjacent into one delivery node, and calculating the geographic coordinates of each delivery node based on the express element set.

[0013] Optionally, the process of constructing a residence node set corresponding to each closed distribution track includes:

[0014] spatially clustering all track points of each closed distribution track, grouping track points that meet a density requirement into a category to form a residence track subsegment, fusing the track points of each residence track subsegment into one residence node, calculating the track point center coordinates of each residence track subsegment as the coordinates of the residence node, and obtaining the residence node set.

[0015] Optionally, the process of establishing an association between the delivery node and the residence node of each pick-up and delivery operation within a preset statistical period includes:

[0016] reading a set of express addresses of a pick-up and delivery operation, obtaining a subset of delivery nodes of the pick-up and delivery operation based on the set of express addresses and the total set of logistics end delivery nodes, and assigning a start time and an end time to each delivery node in the subset of delivery nodes according to the residence node set corresponding to the pick-up and delivery operation.

[0017] Optionally, the process of assigning a start time and an end time to each delivery node in the subset of delivery nodes includes:

[0018] Extracting the delivery node subset and the delivery node and the resident node belonging to the same interest surface in the resident node set, assigning the start time and the end time of the resident node to the corresponding delivery node; if there is no resident node in the interest surface where the single delivery node is located, finding the nearest resident node, assigning the start time and the end time of the resident node to the corresponding delivery node; if there is no matching resident node, assigning the start time and the end time of the nearest trajectory point in the delivery trajectory to the corresponding delivery node.

[0019] Optionally, the process of obtaining the closed pick-up and delivery path comprises:

[0020] Taking the station yard of the pick-up and delivery operation as the starting point, the delivery nodes in the delivery node subset are sequentially connected according to the order of the start time, and the last delivery node is connected to the station yard to obtain the closed pick-up and delivery path.

[0021] Optionally, the process of obtaining the path segment data table comprises:

[0022] Taking the station yard as the starting point, the continuous two delivery nodes of the closed pick-up and delivery path of the single pick-up and delivery operation are sequentially extracted, the time difference between the end time of the previous delivery node and the start time of the next delivery node is calculated to obtain the path segment time consumption, and the delivery path segment data is composed and stored based on the path segment start node number, the next node number and the path segment time consumption, until the delivery path segment storage of all pick-up and delivery operations is completed, to obtain the path segment data table.

[0023] Optionally, the process of obtaining the distribution network connection table comprises: traversing the delivery path segment data in the path segment data table, calculating the average time consumption value of the delivery path segment with the same start node number and next node number, judging whether the delivery path segment exists in the distribution network connection table, if it exists, updating the path segment time consumption in the distribution network connection table through the average time consumption value; if it does not exist, directly storing the delivery path segment data into the distribution network connection table, wherein the structure of the distribution network connection table is completely same as that of the path segment data table.

[0024] Compared with the prior art, the present application has the following advantages and technical effects:

[0025] The present application mines the distribution path of the terminal pick-up and delivery personnel on the basis of sorting the terminal pick-up and delivery personnel distribution historical trajectory data, and constructs a distribution network; and then uses the constructed distribution network to plan a suitable path for the distribution of the terminal logistics express. BRIEF DESCRIPTION OF DRAWINGS

[0026] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application. The application illustratively described in the detailed description will be apparent to those skilled in the art from this

[0027] Figure 1 A flow chart of the method according to an embodiment of the present application;

[0028] Figure 2 An example of express address and AOI according to an embodiment of the present application;

[0029] Figure 3 An example of track points of a road section according to an embodiment of the present application;

[0030] Figure 4 An example of a closed delivery track and its track points according to an embodiment of the present application;

[0031] Figure 5 An example of a stay node extracted from a delivery track and its start and end time according to an embodiment of the present application;

[0032] Figure 6 An example of a delivery node and track stay node of a pick-up and delivery operation according to an embodiment of the present application;

[0033] Figure 7 An example of a delivery path according to an embodiment of the present application;

[0034] Figure 8 A network connectivity graph according to an embodiment of the present application;

[0035] Figure 9 An example of a delivery route according to an embodiment of the present application. DETAILED DESCRIPTION

[0036] It should be noted that the embodiments and features of the embodiments in the present application can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0037] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described herein can be executed in an order different from that shown herein.

[0038] Embodiment One

[0039] As shown in Figure 1 , the present embodiment provides a terminal delivery path planning method based on historical track, comprising:

[0040] (1) Set a statistical period, read all the addresses of express parcels collected and delivered on the last route segment of the statistical period, and use a geocoding tool to convert the express parcel addresses into geographic coordinates to obtain the express parcel element set C = {c1, c2, ..., c i ,...}, where c i Let c be the element of the i-th package. i =<id,name,address,lon,lat> id is the customer ID of the package, name is the customer name, address is the package address, and lon and lat are the longitude and latitude coordinates of the package address. Figure 2 The triangles in the diagram represent the geographical locations of all parcel addresses on the route segment, where the triangles represent parcel addresses and the polygons represent AOIs (Areas of Interest).

[0041] (2) Using Area of ​​Interest (AOI) data within the logistics terminal segment, parcel addresses within the same AOI are merged into a single delivery node. If there are parcel addresses not covered by AOIs, the DBSCAN spatial clustering algorithm is used to merge spatially adjacent parcel addresses into a single delivery node.

[0042] (3) Calculate the center coordinates of the parcel addresses contained in each delivery node, and set these center coordinates as the geographical coordinates of that delivery node to obtain the final segment delivery node set D = {d1, d2, ..., d...} i ,...}, where d i Let d be the i-th delivery node. i =<id,lon,lat> , where id is the delivery node number, and lon and lat are the longitude and latitude coordinates of the delivery node.

[0043] Step 3: Construct delivery trajectory nodes, including:

[0044] (1) Collect trajectory point data of delivery personnel on the last section of the road during the statistical period, P = {p1, p2, ..., p...} i ,...}, where the i-th trajectory point p i =<t,lon,lat,h,d,s> t represents the timestamp of the trajectory point, lon represents the longitude, lat represents the latitude, h represents the elevation value, d represents the direction of movement, and s represents the speed of movement. Figure 3 The dataset of delivery trajectory points for each road segment is shown.

[0045] (2) Extract the closed delivery trajectory point set. The last-mile delivery personnel sign for packages at the parcel station and load the parcels to be delivered in reverse order along the delivery route, completing the delivery in sequence before returning to the station, forming a closed delivery trajectory, such as... Figure 4 As shown, a delivery route consists of multiple ordered route points.

[0046] (3) Extract the resident nodes of the delivery trajectory. A delivery trajectory is composed of a pickup and delivery vehicle resident trajectory segment and a transportation trajectory segment, and each trajectory segment is composed of a plurality of ordered trajectory points. The vehicle resident trajectory segment includes trajectory points uploaded when the vehicle is loading parcels at a station, and trajectory points uploaded when the vehicle is parked on the road or delivering parcels. The trajectory points in the resident trajectory segment are close to each other and have a high spatial density. The transportation trajectory segment is composed of trajectory points generated when the pickup and delivery vehicle moves from one resident location to the next resident location. These trajectory points are far from each other and have a low spatial density. In this embodiment, the DBSCAN spatial clustering algorithm is used to perform spatial clustering on all trajectory points of a delivery trajectory, and a plurality of trajectory points with high density are clustered into a class to form a resident trajectory segment. In this embodiment, the distance is 10 m, and the minimum number of points is 3. Then, the trajectory points of each resident segment are fused into a resident node, the center coordinates of the trajectory points of each resident trajectory segment are calculated, and the resident node is assigned. The resident node set of the delivery trajectory is obtained. The resident node set of the kth delivery trajectory is expressed as: S k = {s1, s2,... s i ,...}, where s i is the ith resident node, denoted as s i = <id, lon, lat, start_time, end_time>, where id is the resident node number, lon and lat are the longitude and latitude coordinates of the resident node, start_time and end_time are the start time and end time of the resident node, i.e., the time of the first trajectory point and the time of the last trajectory point of the resident node. Figure 5 The resident nodes extracted from the delivery trajectory are shown.

[0047] Step 4: Construct the delivery network, including:

[0048] (1) Establish the association between the delivery node and the resident node of the delivery trajectory, including:

[0049] Read all parcel addresses of the ith pickup and delivery operation, and obtain all delivery node set D i of the pickup and delivery operation according to the correspondence between the parcel address and the delivery node.

[0050] Read all resident nodes of the ith delivery trajectory, and obtain the resident node set S i .

[0051] From D i and S iThe system searches for delivery nodes and dwell nodes within the same AOI, assigning the start and end times of dwell nodes within the same AOI to the corresponding delivery nodes. Then, it searches for the nearest dwell node for all unassociated delivery nodes and assigns the start and end times of the dwell node to the corresponding delivery node. Finally, it searches for the nearest trajectory point in the delivery trajectory for each unassociated delivery node and assigns the trajectory point's time to the start and end times of the delivery node. Figure 6 The diagram shows the delivery node (diamond), trajectory dwell node (solid large circle), and trajectory point of a single delivery operation.

[0052] Repeat the above steps until the matching of delivery nodes for all collection and delivery operations during the statistical period is completed.

[0053] (2) Constructing a delivery network, including:

[0054] Read all delivery nodes for the i-th delivery operation.

[0055] Starting from the station yard, the delivery nodes of the i-th delivery operation are connected sequentially according to their start times. Then, the last delivery node is connected to the station yard, thus forming the i-th closed delivery path, represented as: s→d1→d2→...→d k →...d n →e, where s represents the station dwell node for the i-th delivery and collection operation, d k Let represent the k-th delivery node, e represent the station node that returns to the yard after the i-th delivery operation is completed, and n represent the number of delivery nodes in the i-th delivery operation. Figure 7 Example of a delivery route.

[0056] For the i-th delivery route, starting from s, take two consecutive nodes in sequence, calculate the time difference between the end time of the previous node and the start time of the next node, and form a delivery route segment.<id,source,target,cost,reserve_cost> Here, `id` represents the path segment number, `source` represents the starting node number, `target` represents the next node number directly connected to `source`, and `cost` represents the path segment weight. In this embodiment, `cost` is expressed as the time difference between two nodes, representing the transportation time from `source` to `target`. The `reserve_cost` value of a path segment represents the transportation time from `target` to `source`. In this patent, `reserve_cost` is set to "-1", indicating that the path segment is a one-way line segment, meaning that `target` cannot reach `source`. In this embodiment, all path segments of the i-th delivery path are stored in the path segment data table.

[0057] Repeat the above operation until all route segments for all delivery routes are stored in the route segment data table.

[0058] A delivery network connectivity table is created, and the structure of the connectivity table is consistent with the path segment table. All path segments with the same source and target are searched, and the average time of source-target transportation is calculated as the cost weight of the path segment. When storing in the network connectivity table, first check whether the path segment of source-target exists in the network connectivity table, if it exists, update the newly calculated cost to the corresponding path segment, if it does not exist, store the path segment data in the network connectivity table. Figure 8 The network connectivity graph created in this example is shown.

[0059] Step 5: Plan the express delivery route using the constructed network connectivity table, including:

[0060] (1) Read all express addresses to be delivered, map the express to the express node, and get the express node set to be delivered.

[0061] (2) Use Dijkstra's shortest path search algorithm to take the station yard as the starting point and the end point of delivery, take the express node as the path passing point, and use the delivery network connectivity table to plan the shortest delivery path. Figure 9 The delivery path planned in this example is shown using bold lines.

[0062] The present application first aggregates the express addresses of the logistics terminal delivery section within a certain period of time in space, and combines the express addresses of the same interest surface (region) or address proximity into one delivery node; secondly, collects and sorts the track data of the section delivery within the statistical period, analyzes the closed delivery track from the station yard to each express address and back to the station yard, extracts the residence track sub-section from each delivery track, and fuses all track points of the residence track sub-section into a residence node, sets the start time and end time for each residence node; then, read all the express of each pickup and delivery operation, map the express to the delivery node, associate the delivery node to the delivery node of the pickup and delivery, get the delivery node connection path of the pickup and delivery, and use these connection paths to build a delivery network; finally, use Dijkstra's algorithm to plan the delivery path from the constructed delivery network. The present application extracts the terminal delivery semantic path which implies the delivery experience from the spatiotemporal big data of the terminal delivery track and express address, and then constructs the delivery network data for planning the express delivery path in the later period. The present application can use the long-term accumulated delivery knowledge of the pickup and delivery staff for terminal delivery path planning, and can effectively improve the delivery efficiency and quality.

[0063] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A historical trajectory-based end delivery path planning method, characterized by, Includes the following steps: Obtain all parcel addresses and delivery trajectory data of delivery personnel on the last-mile delivery routes within a preset statistical period; After preprocessing all the express delivery addresses, spatial aggregation is performed, and a total set of logistics last-mile delivery nodes is constructed based on the aggregation results; closed delivery trajectories are obtained based on the delivery trajectory point data, and a set of resident nodes corresponding to each closed delivery trajectory is constructed. Establish the relationship between delivery nodes and stationing nodes for each collection and delivery operation within the preset statistical period, and assign start and end times to delivery nodes; The process of establishing the correlation between delivery nodes and dwell nodes for each pickup and delivery operation within a preset statistical period includes: Read the set of parcel addresses for the pickup and delivery operation, and obtain a subset of delivery nodes for the pickup and delivery operation based on the set of parcel addresses and the total set of delivery nodes at the end of the logistics chain; based on the set of dwelling nodes corresponding to the pickup and delivery operation, assign a start time and an end time to each delivery node in the subset of delivery nodes according to the set of dwelling nodes. The process of assigning start and end times to each delivery node in the subset of delivery nodes includes: Extract delivery nodes and dwell nodes belonging to the same interest plane from the subset of delivery nodes and the set of dwell nodes, and assign the start and end times of the dwell nodes to the corresponding delivery nodes; if there are no dwell nodes in the interest plane of a single delivery node, find the nearest dwell node and assign the start and end times of the dwell node to the corresponding delivery node; if there are no matching dwell nodes, assign the start and end times of the trajectory point closest to the delivery node in the delivery trajectory to the corresponding delivery node. Obtain the closed delivery path for a single pickup and delivery operation, break down the closed delivery path into several delivery path segments and store them, until all delivery path segments for all pickup and delivery operations are stored, obtain the path segment data table, and obtain the delivery network connectivity table based on the path segment data table; The process of obtaining the path segment data table includes: Starting from the station, extract two consecutive delivery nodes of the closed delivery path of a single delivery operation in sequence, calculate the time difference between the end time of the previous delivery node and the start time of the next delivery node, obtain the time cost of the path segment, and form delivery path segment data based on the starting node number, the next node number and the time cost of the path segment and store it until the delivery path segments of all delivery operations are stored, and obtain the path segment data table. The process of obtaining the delivery network connectivity table includes: traversing the delivery path segment data in the path segment data table, calculating the average time cost of delivery path segments with the same starting node number and next node number, determining whether the delivery path segment already exists in the delivery network connectivity table, and updating the path segment time cost in the delivery network connectivity table using the average time cost value if it exists; otherwise, directly storing the delivery path segment data into the delivery network connectivity table, wherein the structure of the delivery network connectivity table is exactly the same as that of the path segment data table. The last-mile delivery route is planned based on the aforementioned delivery network connectivity table.

2. The last-mile delivery route planning method based on historical trajectories according to claim 1, characterized in that, The process of spatial aggregation after preprocessing all the express mail addresses includes: The parcel address is converted into geographic coordinates, and a parcel element set is constructed; the interest surface of the last-mile delivery route is obtained, and the parcel address of each interest surface is merged into a delivery node; a spatial clustering algorithm is used to merge the parcel addresses that are not covered by the interest surface but are spatially adjacent into a delivery node, and the geographic coordinates of each delivery node are calculated based on the parcel element set.

3. The last-mile delivery route planning method based on historical trajectories according to claim 1, characterized in that, The process of constructing the set of inhabited nodes corresponding to each closed delivery trajectory includes: Spatial clustering is performed on all trajectory points of each closed delivery route. Trajectory points that meet the density requirements are grouped into one category to form a stationary trajectory segment. The trajectory points of each stationary trajectory segment are merged into a stationary node. The center coordinates of the trajectory points of each stationary trajectory segment are calculated and used as the coordinates of the stationary node to obtain the stationary node set.

4. The last-mile delivery route planning method based on historical trajectories according to claim 1, characterized in that, The process of obtaining a closed-loop delivery path includes: Starting from the station where the delivery operation takes place, the delivery nodes in the subset of delivery nodes are connected sequentially according to the order of their start times, and the last delivery node is connected to the station to obtain a closed delivery path.

Images