Global freight link planning method and system

Abstract

The application discloses a global freight link planning method and system, comprising: acquiring multi-source data of a freight network, constructing a directed weighted graph and a composite edge weight, executing a reverse Dijkstra algorithm based on the composite edge weight, constructing a shortest path tree SPT with a logistics terminal as a root, and calculating a bypass cost delta of each bypass edge relative to the SPT; constructing a path graph, the nodes in the path graph are extended by replacing pointers and appending pointers to generate candidate paths, in the extension process, a preset business rule check is performed on each generated candidate path, if the check passes, the candidate path is put into a priority queue, and if the check fails, the candidate path is discarded directly and the generation of subsequent branches is prevented; taking the shortest path corresponding to the SPT as the first result, and circularly executing priority queue popping and path graph extension until Top-K legal paths are obtained for a user to select and recommend according to actual conditions. The application realizes overall optimization of route planning.

Description

Technical Field

[0001] This application relates to the field of freight link planning technology, and in particular to a method and system for planning freight links across the entire domain. Background Technology

[0002] As competition in the logistics industry intensifies, customers' demands for delivery timeliness, service quality, and logistics visibility are constantly increasing. The rationality of the routing network has become a key factor determining the core competitiveness of logistics companies. As the "central nervous system" of logistics transportation, the routing network directly determines the efficiency, cost, and service quality of the entire delivery chain from dispatch to receipt. However, most current route planning relies heavily on manual, extensive management based on experience, resulting in prominent problems such as suboptimal route selection, redundant transit levels, large fluctuations in delivery time, and slow response to anomalies. Furthermore, even formulating a single route plan often requires dispatchers to spend more than half an hour reviewing data from multiple systems and making manual judgments. The differences in experience among dispatchers lead to poor route quality, and traditional manual route planning often only focuses on the rationality of a single route, ignoring the coupling effect of the entire network topology. Summary of the Invention

[0003] This application provides a method and system for planning freight links across the entire network, which solves problems such as inconsistent routing quality, large fluctuations in timeliness, delayed response to anomalies, serious data silos, and lack of global optimization. It realizes the automated and intelligent generation and dynamic adjustment of routing schemes across the entire network, and achieves overall optimal routing planning.

[0004] The technical solution is as follows: Firstly, embodiments of this application provide a method for planning the entire freight transport chain, including: Acquire multi-source data from the freight network, including OMS waybill data, TMS transport segment execution data, GPS vehicle trajectory data, logistics transshipment center operation data, meteorological data, and road construction and control data; Based on the aforementioned multi-source data, a directed weighted graph is constructed with logistics network points as vertices and transportation routes as directed edges. The edge weights of the directed edges are composite weights determined according to evaluation criteria based on route mileage, route type, and the site association relationships between the network points at both ends of the transportation route. The formula for calculating the composite edge weights is as follows: Where k is a natural number 10, and is determined based on whether the logistics service nodes at both ends of the logistics transportation route share the same physical site: if they share the same physical site, then n takes the first preset value, otherwise it takes the second preset value, and the second preset value is greater than the first preset value; P is the priority value corresponding to the route type of the logistics transportation route; m is the mileage of the logistics transportation route in kilometers; and M is a preset amplification factor and M≥10000. Based on the composite edge weights, the reverse Dijkstra algorithm is executed to construct the shortest path tree (SPT) rooted at the logistics endpoint, and the bypass cost δ of each bypass edge relative to the SPT is calculated. A path graph is constructed based on each side path. Nodes in the path graph are expanded by replacing pointers and appending pointers to generate candidate paths. During the expansion process, a preset business rule is executed to verify each generated candidate path. If the verification passes, it is added to the priority queue. If the verification fails, it is discarded and its subsequent branches are prevented from being generated. Using the shortest path corresponding to SPT as the first result, the priority queue popping and path graph expansion are performed repeatedly until the top-K valid paths matching the OD pair consisting of the logistics origin and logistics destination are obtained, which are then recommended to the user based on the actual situation.

[0005] Furthermore, the P-value gradually increases from the original line, secondary branch line, intercity line, interprovincial line, point branch line, air branch line, customer branch line to split line. The original line has the smallest P-value, and the split line has the largest P-value. The split line is a derivative line obtained by splitting the original line according to business rules.

[0006] Furthermore, the bypass cost ,in Let be the weight of the side edge of each directed edge, where the starting node of the side edge is u and the ending node is v. It is represented as the weight of the shortest edge from the end point v of each directed edge to the starting point u. It represents the weight of the shortest edge from the starting point u of the side path to the destination of the logistics.

[0007] Furthermore, the method also includes: caching the results of the Top-K shortest paths, and within the daily scheduled route planning time, detecting whether there are any newly added special time-sensitive items or abnormal events that have already occurred; when there are no newly added special time-sensitive items or abnormal events that have already occurred, performing route planning on the cached Top-K shortest paths for all waybills based on their respective preset business rules; and when a newly added special time-sensitive item or an abnormal event is detected, performing real-time rerouting planning on waybills affected by the newly added special time-sensitive item or the abnormal event.

[0008] Furthermore, the real-time rerouting planning method specifically includes: Obtain information on newly added special time-sensitive shipments or waybills affected by abnormal events, as well as their corresponding new preset business rules; Based on the directed weighted graph and the composite edge weights, the reverse Dijkstra algorithm is executed to obtain a new shortest path tree SPT rooted at the logistics endpoint of the waybill affected by the abnormal event for the newly added special time-sensitive shipment, and the new bypass cost δ of each bypass edge of the new shortest path tree SPT relative to the new shortest path tree SPT is calculated. A new path graph is constructed based on each new bypass edge. Nodes in the new path graph are expanded by replacing pointers and appending pointers to generate new candidate paths. During the expansion process, the new preset business rules are executed to verify each newly generated candidate path. If the verification passes, it is added to the priority queue; if the verification fails, it is discarded and its subsequent branches are prevented from being generated. Using the shortest path corresponding to the new shortest path tree SPT as the first result, repeatedly perform priority queue popping and path graph expansion until a new Top-K valid paths are obtained; For newly added special time-sensitive shipments or shipments affected by abnormal events, the new Top-K valid paths will be recommended, while the remaining shipments will be recommended according to the cached Top-K shortest paths.

[0009] Furthermore, the preset business rules include at least one of the following: embargo zone constraints, capacity limit constraints, time window constraints, inter-provincial over-limit constraints, and contract-specific routing constraints.

[0010] Furthermore, the method further includes: basing the Top-K path on... The genetic algorithm performs multi-objective optimization and outputs a Pareto optimal solution set, which includes at least one of the following: time-optimal solution, cost-optimal solution, balanced solution, and low-carbon solution.

[0011] Furthermore, the method also includes generating an alternative path through a local rerouting algorithm when abnormal events, including capacity fluctuations, sudden weather changes, and road closures, are detected during the waybill.

[0012] Furthermore, the method for generating alternative paths using a local rerouting algorithm includes: The road segment affected by the abnormal event is identified as the section to be adjusted, the upstream and downstream boundary points of the section to be adjusted are locked, and a local dynamic sub-graph is constructed with the upstream and downstream boundary points as the starting and ending points. Based on the composite edge weights, the region in the local dynamic subgraph where the edge weights change due to abnormal events is calculated online using the reverse Dijkstra algorithm to obtain the shortest path between the upstream and downstream boundary nodes; The shortest path is calculated incrementally using the LPA path search algorithm. Each candidate path generated is validated using preset business rules. If the validation passes, the path is added to a priority queue; if the validation fails, the path is discarded. This process continues until a local alternative path is generated to replace the section to be adjusted.

[0013] The local alternative path is concatenated with the unaffected segments in the current routing scheme to form a complete rerouting scheme, which is then output.

[0014] Secondly, this application also provides a comprehensive freight transport link planning system, including: The multi-source data acquisition unit is used to acquire multi-source data from the freight network, including OMS waybill data, TMS transport segment execution data, GPS vehicle trajectory data, logistics transfer station operation data, meteorological data, and road construction and control data. The directed weighted graph construction unit is used to construct a directed weighted graph based on the multi-source data, with logistics network points as vertices and transportation routes as directed edges. The edge weights of the directed edges are composite edge weights determined according to evaluation criteria based on route mileage, route type, and the site association relationships between the network points at both ends of the transportation route. The calculation formula for the composite edge weights is as follows: Where k is a natural number 10, and is determined based on whether the logistics service nodes at both ends of the logistics transportation route share the same physical site: if they share the same physical site, then n takes the first preset value, otherwise it takes the second preset value, and the second preset value is greater than the first preset value; P is the priority value corresponding to the route type of the logistics transportation route; m is the mileage of the logistics transportation route in kilometers; and M is a preset amplification factor and M≥10000. The bypass cost acquisition unit is used to execute the reverse Dijkstra algorithm based on the composite edge weight, construct the shortest path tree SPT with the logistics destination as the root, and calculate the bypass cost δ of each bypass edge relative to the SPT. The path graph construction unit is used to construct a path graph based on each side path. Nodes in the path graph are expanded by replacing pointers and appending pointers to generate candidate paths. During the expansion process, a preset business rule is executed to verify each generated candidate path. If the verification passes, it is added to the priority queue; if the verification fails, it is discarded and its subsequent branches are prevented from being generated. The Top-K valid path acquisition unit is used to take the shortest path corresponding to SPT as the first result, and repeatedly perform priority queue popping and path graph expansion until the Top-K valid paths are obtained for users to choose and recommend according to the actual situation.

[0015] The beneficial effects of the technical solutions provided in some embodiments of this application are as follows: By acquiring multi-source data of the freight network, including OMS waybill data, TMS transport segment execution data, GPS vehicle trajectory data, logistics transshipment center operation data, meteorological data, and road construction and control data; a directed weighted graph is constructed based on the multi-source data, with logistics network points as vertices and transport routes as directed edges, wherein the edge weight of the directed edge is a composite edge weight determined according to the evaluation criteria of route mileage, route type, and site association relationship between the network points at both ends of the transport route, and the calculation formula of the composite edge weight is: Where k is a natural number 10, determined based on whether the logistics service nodes at both ends of the logistics transportation route share the same physical site: if they share the same physical site, n takes a first preset value; otherwise, it takes a second preset value, and the second preset value is greater than the first preset value. P is the priority value corresponding to the route type of the logistics transportation route, m is the mileage of the logistics transportation route in kilometers, and M is a preset amplification factor, where M≥10000. Based on the composite edge weights, the reverse Dijkstra algorithm is executed to construct the shortest path tree SPT rooted at the logistics endpoint, and the bypass cost δ of each bypass edge relative to the SPT is calculated. Based on each bypass edge, a path graph is constructed. Nodes in the path graph are expanded by replacing pointers and appending pointers to generate candidate paths. During the expansion process, preset business rules are executed to verify each generated candidate path. If the verification passes, it is added to the priority queue; if the verification fails, it is discarded and its subsequent branch generation is prevented. Taking the shortest path corresponding to the SPT as the first result, the priority queue popping and path graph expansion are executed cyclically until the Top-K valid paths are obtained for users to choose and recommend according to the actual situation. It achieves automated and intelligent generation and dynamic adjustment of the entire network routing scheme, realizing the overall optimal routing planning. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic flowchart illustrating the steps of the whole-domain freight link planning method provided in the embodiments of this application; Figure 2 This is a schematic block diagram of the global freight link planning system provided in the embodiments of this application. Detailed Implementation

[0018] The technical solutions of 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, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0020] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0021] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0022] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0023] It should be noted that all data involved in this application was obtained with the permission of the relevant users, complies with relevant policies and regulations, and will not infringe on user privacy.

[0024] Please see Figure 1 , Figure 1 This is a schematic flowchart illustrating a comprehensive freight transport link planning method provided in an embodiment of this application. The method can be implemented by a computer device, which can be deployed on a single server or a server cluster. It can also be deployed on a handheld terminal, laptop, wearable device, or robot, etc.

[0025] It should be noted that the acquisition of any information involved in the provided methods is in compliance with relevant regulations and is carried out with the user's consent, and will not infringe on the user's privacy or violate relevant laws and regulations.

[0026] Specifically, such as Figure 1 As shown, this embodiment of a whole-domain freight link planning method includes steps S101 to S105.

[0027] Step S101: Obtain multi-source data of the freight network, including OMS waybill data, TMS transport segment execution data, GPS vehicle trajectory data, logistics transfer station operation data, meteorological data, and road construction and control data. The breadth of data sources, the timeliness of data collection, and the standardization of processing and storage are crucial for the stable support and guarantee of the entire freight transport chain. This embodiment acquires data from OMS waybill data, TMS transport segment execution data, GPS vehicle trajectory data, logistics transshipment center operation data, meteorological data, and road construction and control data to provide a dynamic data support foundation for the specific algorithms of subsequent full-domain freight transport chain planning.

[0028] Specifically, OMS waybill data may include waybill number (unique), originating point code, receiving point code (freight destination), cargo type (e.g., general cargo / fresh goods / valuable goods), weight and volume, service level (e.g., express, standard, or regular), expected delivery time, customer level (VIP / regular), waybill number, and order source. TMS transport segment execution data may include the internal transport segment number, actual departure / arrival time, actual vehicle travel time, planned travel time, actual mileage, vehicle type and tonnage, transport costs (e.g., fuel consumption, tolls, labor), historical transport timeliness achievement rate, driver and vehicle information, all freight outlets, outlet affiliation (primary, secondary, pickup / delivery), route type (e.g., original route, secondary branch, intercity route, interprovincial route, branch line, air branch, customer branch, and split route), whether it is air / rail, and the relationship between the route and outlet (e.g., whether it contains a co-location marker). GPS vehicle trajectory data may include vehicle ID and license plate number, GPS latitude and longitude coordinates, reporting timestamp, instantaneous speed, driving direction (azimuth), cumulative mileage, equipment online status, and affiliated network / region. Logistics transfer center operation data may include transfer center code, current sorting line processing load, number / total number of loading / unloading points occupied, number of vehicles in queue, operator shift schedule and on-duty personnel, average daily processing capacity, current amount of goods held up, and the transfer center's geographical location and affiliated region. Meteorological data may include weather phenomena, affected areas, warning levels, visibility, road conditions (e.g., water accumulation, icing, slippery conditions), and forecast lead time (e.g., 0-72 hours). Road construction and control data may include construction and control section numbers or latitude and longitude ranges, control type (e.g., fully closed, semi-closed, traffic restrictions, speed limits), effective start and end times, reasons for control (e.g., construction, accidents, large-scale events, holiday restrictions), and truck no-entry signs.

[0029] In summary, OMS order data allows for the definition of waybill origin, destination, and service level, determining origin, destination pairs, and optimization targets for each route calculation. TMS transport execution data provides actual travel time and costs for each transport segment. GPS vehicle trajectory data provides real-time vehicle location and speed, enabling dynamic updates to road toll costs and triggering abnormal rerouting. Transfer center operation data reflects sorting load and queuing status, allowing for dynamic adjustment of node capacity and achieving adaptive capacity balance. Meteorological data, such as weather warnings and visibility information, helps identify high-risk road sections. Road construction and traffic control data allows for setting hard constraints and adjusting triggers for local rerouting. Information such as the basic transport route table, the relationship between routes and network points, and the network point affiliation table enables more efficient and accurate planning and construction of route maps, considering route weights.

[0030] In some feasible implementations, the acquired multi-source data needs to be connected to the system in real time via standardized APIs or message queues (Kafka). After processing such as format standardization, timestamp alignment, and outlier filtering, the data is distributed and stored as the basis for subsequent data processing in the freight network. Different types of data can have different update frequencies; for example, GPS tracks are streamed in real-time at the second level, transit hub loads at the minute level, meteorological and traffic control data at the hour level, and historical TMS data are synchronized in batches at the daily level. Real-time aggregation of multi-source data ensures the dynamic update capability of the data used in subsequent full-domain freight link planning algorithms.

[0031] Step S102: Based on the multi-source data, construct a directed weighted graph with logistics network points as vertices and transportation routes as directed edges. The edge weights of the directed edges are composite edge weights determined according to the evaluation criteria of route mileage, route type, and the site association relationship between the network points at both ends of the transportation route. The formula for calculating the composite edge weights is as follows: Where k is a natural number 10, and is determined based on whether the logistics service nodes at both ends of the logistics transportation route share the same physical site: if they share the same physical site, then n takes the first preset value, otherwise it takes the second preset value, and the second preset value is greater than the first preset value; P is the priority value corresponding to the route type of the logistics transportation route; m is the mileage of the logistics transportation route in kilometers; and M is a preset amplification factor and M≥10000. In this embodiment, the directed weighted graph can be understood as a mathematical abstraction of a freight network, used to transform real-world logistics infrastructure, such as network points, routes, and transportation costs, into a data structure that can be computed by algorithms. The directed weighted graph includes three main elements: vertices, edges, and edge weights. Vertices represent logistics service points, including distribution centers at various levels (T1 / T2 / SC) and last-mile sales outlets. The vertices in the constructed directed weighted graph include elements such as the vertex's network point ID, name, province, region, and affiliation type. Edges represent transportation routes between network points, with a clear direction, such as the destination. Edges may include factors such as route code, route type, and mileage. Edge weights are composite edge weights, mapped to each edge according to various weight conditions, including composite weight values. Only through the abstract design of the actual physical freight network using mathematical calculations can effective quantitative analysis be performed.

[0032] The composite edge weights can be obtained according to the following formula:

[0033] Where k is a natural number 10, and n is determined based on whether the logistics service nodes at both ends of the logistics transportation route share the same physical site: if they share the same physical site, then n takes a first preset value, otherwise it takes a second preset value, and the second preset value is less than the first preset value. For example, the first preset value is 1 and the second preset value is 0. k can increase the weight of the site in the composite edge weight. m represents the mileage (in kilometers) of the logistics transportation route. Mileage is the most fundamental consideration in logistics transportation, as costs, fuel, tolls, and vehicle wear and tear are all strongly correlated with mileage. The formula uses mileage as the primary determining factor for route ranking, ensuring that the algorithm first approximates the optimal physical location. M is a preset amplification factor, and M≥10000, reducing the importance of mileage in the composite edge weights. P represents the priority value corresponding to the route type of the logistics transportation route. In some embodiments, the route type ranges from the original route, secondary branch line, intercity route, interprovincial route, point branch line, air branch line, customer branch line to split route, with the P value decreasing sequentially. The original route has the largest P value and the highest priority, while the split route has the smallest P value. The split route is a derivative route obtained by splitting the original route according to business rules. This embodiment ensures the business rationality of subsequent full-domain freight link planning by coordinating route mileage, route type, and the site association relationship between the network points at both ends of the transportation route, i.e., the shortest physically and the most reliable operationally.

[0034] Step S103: Execute the reverse Dijkstra algorithm based on the composite edge weights to construct the shortest path tree SPT rooted at the logistics destination. The reverse Dijkstra's algorithm searches from the destination towards the starting point, not from the starting point towards the destination. By executing the reverse Dijkstra's algorithm, the shortest edge weights from the starting point of each waybill, all intermediate points, and their predecessor edges, representing the shortest edge weights, can be obtained in one step, with the destination as the root node. Connecting the predecessor edges of each point forms a shortest path tree (SPT) rooted at the destination. It should be noted that the shortest edge weights here are calculated based on the aforementioned composite edge weights. The significance of obtaining the shortest path tree (SPT) is that, regardless of the starting point, as long as one moves forward along the predecessor edge of this tree, one can reach the destination with the "optimal cost." The shortest path tree (SPT) also serves as the baseline for all subsequent "deviations" in the bypass paths.

[0035] Step S104: Obtain the side path cost δ relative to each predecessor edge in the shortest path tree SPT, construct a path graph based on each side path edge, and expand the nodes in the path graph by replacing pointers and appending pointers to generate candidate paths. During the expansion process, perform preset business rule verification on each generated candidate path. If the verification passes, it is added to the priority queue; if the verification fails, it is discarded directly and its subsequent branches are prevented from being generated. In some embodiments, the Eppstein algorithm is used to obtain the bypass edges of the predecessor edge and the bypass costs of each bypass edge. The aim is to optimize routing from a network-wide perspective, incorporating thousands of service nodes and tens of thousands of potential paths into a unified mathematical model and solving for the globally optimal routing network structure. A seemingly circuitous route might, from a network-wide perspective, precisely balance the capacity pressure on another route, thus achieving overall optimization. Therefore, for each shipment, it is necessary to find alternative preferred routes between the logistics origin and destination based on the shortest path. This necessitates obtaining the bypass edges of the predecessor edge and the bypass costs of each bypass edge, using these as a basis for ranking and recommending optimal distances.

[0036] In some embodiments, bypass cost ,in Let be the weight of the bypass edge for each predecessor edge, where the starting endpoint of the bypass edge is u and the ending endpoint is v. Let be the weight of the shortest edge from the end point v of each predecessor edge to the starting end point u. This represents the shortest edge weight from the starting endpoint u of the bypass to the freight destination. The edge weight here is determined using the same criteria as the composite edge weight obtained from the Shortest Path Tree (SPT) mentioned above, to ensure that all planned paths meet actual business requirements.

[0037] It should be noted that in this embodiment, the replacement pointer is used to replace the forward edge and each side edge, while the append pointer is used to append from one network node to another along the planned path direction.

[0038] In the path graph, nodes are expanded by replacing and appending pointers to generate candidate paths. During the expansion process, each generated candidate path undergoes a preset business rule verification. Paths that pass the verification are added to a priority queue, while those that fail are discarded and their subsequent branches are prevented from being generated. In some embodiments, the preset business rules include at least one of the following: restricted area constraints, capacity limit constraints, time-sensitive window constraints, inter-provincial over-limit constraints, and contract-specific routing constraints. The preset business rule verification is attached to the hot paths during path expansion. Each candidate path generated by replacement and appending undergoes a legality rule verification first; illegal paths are eliminated as early as possible, and invalid paths are not added to the queue, thus preventing queue expansion.

[0039] Step S105: Using the shortest path corresponding to SPT as the first result, repeatedly perform priority queue popping and path graph expansion until the Top-K valid paths are obtained for the user to choose and recommend according to the actual situation.

[0040] In this embodiment, the path graph pops up in ascending order of cumulative cost (δ). Therefore, the first path to pop up is the shortest, the second is the shortest, and so on. This ensures that the user receives K paths strictly ranked from lowest to highest bypass cost (Top-K). Furthermore, there are no preset business rules in the path graph. Each path undergoes a final verification against these preset business rules after popping up and before being output. This "final check" ensures that the K paths output to the user 100% meet hard constraints such as embargoes, timeliness, and capacity.

[0041] In practice, users may face multiple routing options based on factors such as timeliness, cost, number of transfers, and carbon emissions. Operations personnel can flexibly choose according to business scenarios and customer needs. Therefore, in some embodiments, the method of this application further includes basing the Top-K paths on... The genetic algorithm performs multi-objective optimization and outputs a Pareto optimal solution set, which includes at least one of the following: time-optimal solution, cost-optimal solution, balanced solution, and low-carbon solution. For example, the time-optimal solution may involve more transfers; the cost-optimal solution involves fewer transfers and the lowest cost; the balanced solution compromises timeliness and cost; and the low-carbon solution prioritizes rail transport and avoids air and high-speed rail.

[0042] In some feasible embodiments, the method further includes caching the results of the Top-K shortest paths matched by the OD pairs consisting of the logistics origin and destination, and detecting whether there are any newly added special time-sensitive items or abnormal events that have occurred within the daily prescribed route planning time. When no new special time-sensitive items or abnormal events have occurred, route planning is performed on all waybills using the cached Top-K shortest paths based on their respective preset business rules. When a newly added special time-sensitive item or an abnormal event is detected, real-time rerouting planning is performed on the waybills affected by the newly added special time-sensitive item or the abnormal event, specifically including: Obtain information on newly added special time-sensitive shipments or waybills affected by abnormal events, as well as their corresponding new preset business rules; Based on the directed weighted graph and the composite edge weights, the reverse Dijkstra algorithm is executed to obtain a new shortest path tree SPT rooted at the logistics endpoint of the waybill affected by the abnormal event for the newly added special time-sensitive shipment, and the new bypass cost δ of each bypass edge of the new shortest path tree SPT relative to the new shortest path tree SPT is calculated. A new path graph is constructed based on each new bypass edge. Nodes in the new path graph are expanded by replacing pointers and appending pointers to generate new candidate paths. During the expansion process, the new preset business rules are executed to verify each newly generated candidate path. If the verification passes, it is added to the priority queue; if the verification fails, it is discarded and its subsequent branches are prevented from being generated. Using the shortest path corresponding to the new shortest path tree SPT as the first result, repeatedly perform priority queue popping and path graph expansion until a new Top-K valid paths are obtained; For newly added special time-sensitive shipments or shipments affected by abnormal events, the new Top-K valid paths will be recommended, while the remaining shipments will be recommended according to the cached Top-K shortest paths.

[0043] This application divides the business model into batch route planning mode and real-time route planning mode. To improve route planning speed, based on the latest network topology and capacity data, batch pre-calculation of the entire network route can be performed at a certain time to generate a basic routing table covering all predicted waybills for the next day. The output scheme includes a primary route and a backup route. When a waybill enters the system, the system queries the basic routing table and, based on whether the waybill has any abnormalities, the actual departure time, and real-time traffic data, determines whether to trigger real-time rerouting to enter the real-time route planning mode. For time-sensitive express shipments or route segments with abnormalities, single-ticket rerouting can be initiated, thus outputting a personalized optimal solution.

[0044] In some feasible embodiments, the method further includes generating an alternative path using a local rerouting algorithm when abnormal events, including capacity fluctuations, sudden weather changes, and road closures, are detected during the waybill's journey. The method for generating the alternative path using the local rerouting algorithm includes: The road segment affected by the abnormal event is identified as the section to be adjusted. The upstream and downstream boundary points of the section to be adjusted are locked, and a local dynamic sub-graph is constructed with the upstream and downstream boundary points as the starting and ending points. At the same time, the information of the waybills expected to pass through the section to be adjusted is obtained, including which waybills' predetermined routes pass through the abnormal area, how many distribution centers are involved in the affected area, the expected number of affected tickets, and the list of affected customers.

[0045] Based on the composite edge weights, the region in the local dynamic subgraph where the edge weights change due to abnormal events is calculated online using the reverse Dijkstra algorithm to obtain the shortest path between the upstream and downstream boundary nodes; The shortest path is calculated incrementally using the LPA path search algorithm. Each candidate path generated is validated using preset business rules. If the validation passes, the path is added to a priority queue; if the validation fails, the path is discarded. This process continues until a local alternative path is generated to replace the section to be adjusted.

[0046] The local alternative path is concatenated with the unaffected segments in the current routing scheme to form a complete rerouting scheme, which is then output.

[0047] The LPA (Path Algorithm for Shortest Path) algorithm is an algorithm for finding the shortest path from a given starting point to a given destination point. Based on the composite edge weights of this application, the LPA algorithm is used to solve for the shortest path. In the event of an abnormal event, such as a transfer station exceeding its safe capacity (e.g., reaching 85% of its capacity), an alternative path can be generated using the LPA algorithm, automatically diverting the affected road segment to other transfer stations to avoid congestion. When the average vehicle speed on a road segment drops beyond a threshold due to traffic accidents, road construction, etc., routes for packages currently en route and those about to be dispatched are automatically adjusted. Seasonal route templates can also be automatically generated based on different seasons and promotional window routing patterns, automatically switching and taking effect within specific time windows.

[0048] Furthermore, the local rerouting algorithm employs the same composite edge weight sorting logic as the batch pre-calculation, ensuring that the "optimal" definition of the rerouting scheme is consistent with the daily calculation methods. The same-site / cross-site factor in the composite edge weights allows rerouting to prioritize adjustments within the same site when capacity is tight, avoiding secondary transfers.

[0049] The local rerouting algorithm ensures that when an anomaly occurs, the global path planning does not need to be completely restarted. Instead, it locks the affected area, incrementally repairs a compliant alternative path on the local subgraph using the LPA path search algorithm, and then seamlessly reassembles the original route. It also ensures compliance through composite edge weight-driven sorting and preset business rule verification, minimizing local splicing disturbances and achieving real-time rerouting capabilities with second-level response, local correction, and full-chain compliance.

[0050] In summary, this application comprehensively considers multiple constraints such as timeliness, cost, mileage, number of transfers, and capacity saturation. It automatically calculates the optimal routing scheme for all waybills within the network from the sender's point to the receiver's point. Statistical results show that this scheme increases the direct delivery rate across the entire network to over 80%, and reduces the average number of transfers from 3.2 to less than 2. It should be noted that, simultaneously, this application automatically triggers rerouting calculations and pushes adjustment schemes when abnormal events such as capacity fluctuations, sudden weather changes, and road closures occur, further increasing the timeliness achievement rate from 88% to over 95%. This application reduces the overall transportation cost of the network by 8%-12% through optimizing the routing structure, reducing detours, and increasing vehicle loading rates.

[0051] This invention also provides a comprehensive freight transport link planning system, comprising: The multi-source data acquisition unit 201 is used to acquire multi-source data of the freight network, including OMS waybill data, TMS transport segment execution data, GPS vehicle trajectory data, logistics transfer station operation data, meteorological data, and road construction and control data. The directed weighted graph construction unit 202 is used to construct a directed weighted graph based on the multi-source data, with logistics network points as vertices and transportation routes as directed edges. The edge weights of the directed edges are composite edge weights determined according to evaluation criteria based on route mileage, route type, and the site association relationships between network points at both ends of the transportation route. The calculation formula for the composite edge weights is as follows: Where k is a natural number 10, and is determined based on whether the logistics service nodes at both ends of the logistics transportation route share the same physical site: if they share the same physical site, then n takes the first preset value, otherwise it takes the second preset value, and the second preset value is greater than the first preset value; P is the priority value corresponding to the route type of the logistics transportation route; m is the mileage of the logistics transportation route in kilometers; and M is a preset amplification factor and M≥10000. The bypass cost acquisition unit 203 is used to execute the reverse Dijkstra algorithm based on the composite edge weight, construct the shortest path tree SPT with the logistics destination as the root, and calculate the bypass cost δ of each bypass edge relative to the SPT. The path graph construction unit 204 is used to construct a path graph based on each side path. Nodes in the path graph are expanded by replacing pointers and appending pointers to generate candidate paths. During the expansion process, a preset business rule is executed to verify each generated candidate path. If the verification passes, it is added to the priority queue. If the verification fails, it is discarded and its subsequent branches are prevented from being generated. The Top-K valid path acquisition unit 205 is used to take the shortest path corresponding to SPT as the first result, and repeatedly perform priority queue popping and path graph expansion until the Top-K valid paths are obtained for users to choose and recommend according to the actual situation.

[0052] For other details regarding the implementation of the above technical solutions by each module in the above-mentioned whole-domain freight link planning system, please refer to the description in the whole-domain freight link planning method provided in the above-mentioned invention embodiments, which will not be repeated here.

[0053] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for planning freight transport links across the entire region, characterized in that, include: Acquire multi-source data from the freight network, including OMS waybill data, TMS transport segment execution data, GPS vehicle trajectory data, logistics transshipment center operation data, meteorological data, and road construction and control data; Based on the aforementioned multi-source data, a directed weighted graph is constructed with logistics network points as vertices and transportation routes as directed edges. The edge weights of the directed edges are composite weights determined according to evaluation criteria based on route mileage, route type, and the site association relationships between the network points at both ends of the transportation route. The formula for calculating the composite edge weights is as follows: Where k is a natural number 10, and is determined based on whether the logistics service nodes at both ends of the logistics transportation route share the same physical site: if they share the same physical site, then n takes the first preset value, otherwise it takes the second preset value, and the second preset value is greater than the first preset value; P is the priority value corresponding to the route type of the logistics transportation route; m is the mileage of the logistics transportation route in kilometers; and M is a preset amplification factor and M≥10000. Based on the composite edge weights, the reverse Dijkstra algorithm is executed to construct the shortest path tree (SPT) rooted at the logistics endpoint, and the bypass cost δ of each bypass edge relative to the SPT is calculated. A path graph is constructed based on each side path. Nodes in the path graph are expanded by replacing pointers and appending pointers to generate candidate paths. During the expansion process, a preset business rule is executed to verify each generated candidate path. If the verification passes, it is added to the priority queue. If the verification fails, it is discarded and its subsequent branches are prevented from being generated. Using the shortest path corresponding to SPT as the first result, the priority queue popping and path graph expansion are performed repeatedly until the top-K valid paths matching the OD pair consisting of the logistics origin and logistics destination are obtained, which are then recommended to the user based on the actual situation.

2. The whole-domain freight link planning method according to claim 1, characterized in that, The route types range from original routes, secondary branches, intercity routes, interprovincial routes, point-to-point branches, air routes, customer branches to split routes, with the P value gradually increasing. The original routes have the smallest P value, while the split routes have the largest P value. The split routes are derivative routes obtained by splitting the original routes according to business rules.

3. The whole-domain freight link planning method according to claim 1, characterized in that: The bypass cost ,in Let be the weight of the side edge of each directed edge, where the starting node of the side edge is u and the ending node is v. It is represented as the weight of the shortest edge from the end point v of each directed edge to the starting point u. It represents the weight of the shortest edge from the starting point u of the side path to the destination of the logistics.

4. The whole-domain freight link planning method according to claim 1, characterized in that, The method further includes: caching the results of the Top-K shortest paths, and within the daily scheduled routing planning time, detecting whether there are any new special time-sensitive items or abnormal events that have occurred. When there are no new special time-sensitive items or abnormal events that have occurred, performing routing planning on the cached Top-K shortest paths for all waybills based on their respective preset business rules. When a new special time-sensitive item or an abnormal event is detected, performing real-time rerouting planning on waybills affected by the new special time-sensitive item or the abnormal event.

5. The whole-domain freight link planning method according to claim 4, characterized in that, The real-time rerouting plan specifically includes: Obtain information on newly added special time-sensitive shipments or waybills affected by abnormal events, as well as their corresponding new preset business rules; Based on the directed weighted graph and the composite edge weights, the reverse Dijkstra algorithm is executed to obtain a new shortest path tree SPT rooted at the logistics endpoint of the waybill affected by the abnormal event for the newly added special time-sensitive shipment, and the new bypass cost δ of each bypass edge of the new shortest path tree SPT relative to the new shortest path tree SPT is calculated. A new path graph is constructed based on each new bypass edge. Nodes in the new path graph are expanded by replacing pointers and appending pointers to generate new candidate paths. During the expansion process, the new preset business rules are executed to verify each newly generated candidate path. If the verification passes, it is added to the priority queue; if the verification fails, it is discarded and its subsequent branches are prevented from being generated. Using the shortest path corresponding to the new shortest path tree SPT as the first result, repeatedly perform priority queue popping and path graph expansion until a new Top-K valid paths are obtained; For newly added special time-sensitive shipments or shipments affected by abnormal events, the new Top-K valid paths will be recommended, while the remaining shipments will be recommended according to the cached Top-K shortest paths.

6. The method for planning the entire freight transport chain according to claim 1, characterized in that, The preset business rules include at least one of the following: embargo zone constraints, capacity limit constraints, time window constraints, inter-provincial over-limit constraints, and special contract routing constraints.

7. The whole-domain freight link planning method according to claim 1, characterized in that, The method further includes: basing the Top-K path on... The genetic algorithm performs multi-objective optimization and outputs a Pareto optimal solution set, which includes at least one of the following: time-optimal solution, cost-optimal solution, balanced solution, and low-carbon solution.

8. The whole-domain freight link planning method according to claim 1, characterized in that, The method also includes generating alternative routes using a local rerouting algorithm when abnormal events, including capacity fluctuations, sudden weather changes, and road closures, are detected during the waybill.

9. The whole-domain freight link planning method according to claim 8, characterized in that, The method for generating alternative paths using a local rerouting algorithm includes: The road segment affected by the abnormal event is identified as the section to be adjusted, the upstream and downstream boundary points of the section to be adjusted are locked, and a local dynamic sub-graph is constructed with the upstream and downstream boundary points as the starting and ending points. Based on the composite edge weights, the region in the local dynamic subgraph where the edge weights change due to abnormal events is calculated online using the reverse Dijkstra algorithm to obtain the shortest path between the upstream and downstream boundary nodes; The shortest path is calculated incrementally using the LPA path search algorithm. Each candidate path generated is validated using preset business rules. If the validation passes, the path is added to a priority queue; if the validation fails, the path is discarded. This process continues until a local alternative path is generated to replace the section to be adjusted. The local alternative path is concatenated with the unaffected segments in the current routing scheme to form a complete rerouting scheme, which is then output.

10. A comprehensive freight transport link planning system, characterized in that, include: The multi-source data acquisition unit is used to acquire multi-source data from the freight network, including OMS waybill data, TMS transport segment execution data, GPS vehicle trajectory data, logistics transfer station operation data, meteorological data, and road construction and control data. The directed weighted graph construction unit is used to construct a directed weighted graph based on the multi-source data, with logistics network points as vertices and transportation routes as directed edges. The edge weights of the directed edges are composite edge weights determined according to evaluation criteria based on route mileage, route type, and the site association relationships between the network points at both ends of the transportation route. The calculation formula for the composite edge weights is as follows: Where k is a natural number 10, and is determined based on whether the logistics service nodes at both ends of the logistics transportation route share the same physical site: if they share the same physical site, then n takes the first preset value, otherwise it takes the second preset value, and the second preset value is greater than the first preset value; P is the priority value corresponding to the route type of the logistics transportation route; m is the mileage of the logistics transportation route in kilometers; and M is a preset amplification factor and M≥10000. The bypass cost acquisition unit is used to execute the reverse Dijkstra algorithm based on the composite edge weight, construct the shortest path tree SPT with the logistics destination as the root, and calculate the bypass cost δ of each bypass edge relative to the SPT. The path graph construction unit is used to construct a path graph based on each side path. Nodes in the path graph are expanded by replacing pointers and appending pointers to generate candidate paths. During the expansion process, a preset business rule is executed to verify each generated candidate path. If the verification passes, it is added to the priority queue; if the verification fails, it is discarded and its subsequent branches are prevented from being generated. The Top-K valid path acquisition unit is used to take the shortest path corresponding to SPT as the first result, and repeatedly perform priority queue popping and path graph expansion until the Top-K valid paths are obtained for users to choose and recommend according to the actual situation.

Images