Multi-modal transport single logistics traceability method and system

Abstract

The application relates to the field of logistics information technology and discloses a multimodal transport single-system logistics tracking method and system. The method comprises the following steps: acquiring multimodal real-time transportation data such as highway, railway, waterway and airway associated with a cargo object identifier, and integrating the initial transportation network data set; then, extracting a network topology feature vector to quantify the route delay distribution, and identifying and classifying key transfer nodes according to the network topology feature vector; subsequently, constructing a dynamic tracking graph based on the classified node groups, analyzing data processing delay based on the graph edge weight, triggering a synchronization mechanism to optimize and update the cargo state, and then fusing the node change attribute to obtain the associated path state data set; finally, updating the delay weight and event trigger judgment of the graph according to the path state data set, and obtaining the accurate logistics tracking result. The method can realize dynamic cooperation and efficient tracking of cross-mode transportation.

Description

Technical Field

[0001] This invention relates to the field of logistics information technology, and in particular to a multimodal transport single-document logistics traceability method and system. Background Technology

[0002] Currently, logistics traceability technology is the core of modern supply chain management. By integrating information flow with the dynamics of physical goods, it achieves transparent monitoring of goods from origin to destination, and is widely used in complex scenarios such as cross-border e-commerce, bulk commodity trade, and cold chain transportation. The core of this technology lies in assigning a unique object identifier to each transportation unit and building a complete data link around it to ensure cargo safety and improve circulation efficiency. However, in multimodal transport scenarios, due to the alternating collaboration of multiple modes of transportation such as road, rail, water, and air, the heterogeneous data sources and dynamically changing network topologies place higher demands on the real-time performance and accuracy of traceability technology.

[0003] In one existing technology, logistics traceability primarily relies on piecing together information from static transport records provided by carriers at each transport segment. For example, when a batch of goods is unloaded at a port and transferred to a railway, the traceability system first queries the port authority's outbound records, and then queries the railway transport system's on-the-go information based on the waybill number. This segmented, static node-based traceability method operates under conditions where each transport entity's information system is independent and data exchange is delayed. This method struggles to adapt to temporary changes in transport routes and nodes when dealing with the dynamic and complex nature of multimodal transport. For instance, if transport is temporarily switched from rail to road due to weather or capacity issues, the system cannot quickly detect this dynamic adjustment, leading to information silos between data nodes and interrupting the traceability link. Furthermore, due to the lack of extraction and utilization of transport network topology features, the system struggles to predict and address potential delay risks and cannot achieve dynamic optimization of the global path.

[0004] In summary, existing technologies suffer from delayed and incomplete traceability results in multimodal transport networks. Summary of the Invention

[0005] This invention provides a multimodal transport single-document logistics traceability method and system to solve the problem of delayed and incomplete traceability results in multimodal transport networks.

[0006] Firstly, in order to solve the above-mentioned technical problems, the present invention provides a multimodal transport single-document logistics traceability method, comprising: Acquire real-time route information, railway node details, waterway connectivity, and air transport records, and integrate the data to obtain an initial transportation network dataset; Based on the initial transportation network dataset, network topology features are extracted to obtain a topology feature vector; Based on the topological feature vector, the transit nodes are identified and classified to obtain the classified node groups; Based on the classified node groups and the topological feature vectors, a dynamic tracing graph is constructed to obtain a complete tracing relationship graph; Based on the complete tracing relationship graph, path status association and optimization are performed to obtain the associated path status dataset. Based on the associated path status dataset, the traceability results are generated and the complete traceability relationship graph is updated to obtain the final logistics traceability result.

[0007] Preferably, the step of acquiring real-time path information, railway node details, waterway connectivity, and air transport records, and integrating the data to obtain an initial transportation network dataset, includes: Real-time path information and railway node details are obtained from sensors and railway node monitoring devices to obtain multi-source data; Retrieve waterway connectivity and air transport records from a pre-established database; Based on the multi-source data, data alignment and correction are performed to obtain consistent transportation network data; Based on the consistent transportation network data, the waterway connectivity, and the air transport records, initial route planning is performed to obtain an initial transportation network dataset containing route optimization schemes.

[0008] Preferably, the step of extracting network topology features based on the initial transportation network dataset to obtain a topology feature vector includes: Based on the initial transportation network dataset, the route delay distribution is calculated to obtain the delay distribution feature vector; Based on the delay distribution feature vector, network topology is determined and route priority is optimized to obtain an optimized cargo route allocation scheme. Based on the optimized cargo route allocation scheme, feature matching degree analysis is performed to obtain the topological feature vector.

[0009] Preferably, the step of identifying and classifying transit nodes based on the topological feature vector to obtain classified node groups includes: Based on the topological feature vector, path adjustment event analysis is performed to obtain a set of path adjustment events; The number of events in the path adjustment event set is counted. If the number of events exceeds a preset threshold, the transfer node to be optimized is identified to obtain the set of transfer nodes to be optimized. Based on the set of transit nodes to be optimized, node feature clustering and group merging are performed to obtain classified node groups.

[0010] Preferably, the step of constructing a dynamic tracing graph based on the classified node groups and the topological feature vectors to obtain a complete tracing relationship graph includes: Based on the classified node groups and the topological feature vectors, a preliminary route query is performed to obtain a preliminary set of cargo routes; Based on the initial set of cargo routes, the transportation relationships are dynamically updated to obtain a dynamic set of transportation relationships; Based on the dynamic transportation relationship set, route clustering and optimization are performed to obtain the optimized route set; Based on the optimized path set, the final graph is constructed to obtain a complete traceability graph.

[0011] Preferably, the step of performing path state association and optimization based on the complete tracing relationship graph to obtain the associated path state dataset includes: Based on the complete tracing relationship graph, edge weight analysis and synchronization frequency optimization are performed to obtain the optimized synchronization mechanism. Based on the optimized synchronization mechanism, cargo status is updated and delayed paths are identified to obtain a set of transportation routes that need to be adjusted. Based on the set of transportation routes that need to be adjusted, the tracking path to be optimized is determined, and the set of tracking paths to be optimized is obtained. Based on the set of tracking paths to be optimized, node change attributes are fused to obtain the associated path status dataset.

[0012] Preferably, the step of generating and updating the complete traceability relationship graph based on the associated path status dataset to obtain the final logistics traceability result includes: Based on the associated path state dataset, the path weights are updated to obtain the optimized path weight dataset. Based on the optimized path weight dataset, state and weight clustering is performed to obtain path state clustering. Based on the path status clustering and grouping, event-triggered judgment and update are performed to obtain the final logistics traceability result.

[0013] Secondly, the present invention provides a multimodal transport single-document logistics traceability system, comprising: The data integration module is used to acquire real-time route information, railway node details, waterway connection relationships and air transport records, and integrate the data to obtain an initial transportation network dataset; The feature extraction module is used to extract network topology features based on the initial transportation network dataset to obtain a topology feature vector; The node classification module is used to identify and classify transit nodes based on the topological feature vector to obtain classified node groups. The graph construction module is used to dynamically construct a traceability graph based on the classified node groups and the topological feature vectors to obtain a complete traceability relationship graph. The path optimization module is used to perform path status association and optimization based on the complete tracing relationship graph to obtain the associated path status dataset. The result generation module is used to generate and update the complete traceability relationship graph based on the associated path status dataset to obtain the final logistics traceability result.

[0014] Thirdly, the present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the multimodal transport single-document logistics traceability method described in any one of the above.

[0015] Fourthly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the multimodal transport single-document logistics traceability method described in any one of the above.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention collects multi-source heterogeneous data from highways, railways, waterways, and aviation in real time, uses timestamp synchronization and time series interpolation techniques for data alignment and correction, and stores the data in a distributed database. In this way, isolated data points with different formats and time deviations under different transportation modes are integrated into a unified initial transportation network dataset that is consistent in time and format. This solves the problems of data silos and information delays caused by existing technologies that rely on segmented static records, ensuring the integrity and real-time nature of traceability data from the source, and providing a highly reliable data foundation for accurate traceability.

[0017] (2) This invention extracts the topological feature vectors of the transportation network and constructs a dynamic traceability map based on them. This method transforms the original transportation data into structured features containing information such as path delay distribution, node connectivity, and transfer frequency. When path adjustment events exceed a threshold, it can proactively identify and classify the transfer nodes that need optimization. This changes the existing technology's passive information splicing traceability mode, realizes dynamic analysis and proactive optimization of the transportation network status, can promptly discover and manage transportation bottlenecks, and improve the collaborative efficiency of the multimodal transport network.

[0018] (3) This invention identifies data processing delays by monitoring the edge weights of the dynamic traceability graph and performs closed-loop optimization based on feedback from cargo tracking status. When the system detects large-scale data processing delays, it adaptively adjusts the data synchronization frequency to balance real-time performance and system load. Simultaneously, it integrates event data such as multi-mode switching and path adjustments discovered during actual tracking with the node change detection attributes to update the delay weights and event trigger judgments of the graph in reverse. This resolves the contradiction between real-time synchronization and massive data processing in existing technologies, forming an adaptive and self-optimizing traceability mechanism that ensures the traceability results remain accurate even in complex and ever-changing transportation environments. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the multimodal transport single-document logistics traceability method provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the multimodal transport single-document logistics traceability system provided in the second embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Reference Figure 1 The first embodiment of the present invention provides a multimodal transport single-document logistics traceability method, including the following steps: S11: Obtain real-time route information, railway node details, waterway connection relationships, and air transport transfer records, and integrate the data to obtain the initial transportation network dataset; S12, Based on the initial transportation network dataset, perform network topology feature extraction to obtain a topology feature vector; S13, Based on the topological feature vector, identify and classify the transit nodes to obtain the classified node groups; S14. Based on the classified node groups and the topological feature vector, a dynamic tracing graph is constructed to obtain a complete tracing relationship graph. S15, Based on the complete traceability relationship graph, perform path status association and optimization to obtain the associated path status dataset; S16. Based on the associated path status dataset, generate and update the complete traceability relationship graph to obtain the final logistics traceability result.

[0022] In step S11, real-time route information, railway node details, waterway connectivity, and air transport records are acquired and integrated to obtain an initial transportation network dataset, including: Real-time path information and railway node details are obtained from sensors and railway node monitoring devices to obtain multi-source data; Retrieve waterway connectivity and air transport records from a pre-established database; Based on the multi-source data, data alignment and correction are performed to obtain consistent transportation network data; Based on the consistent transportation network data, the waterway connectivity, and the air transport records, initial route planning is performed to obtain an initial transportation network dataset containing route optimization schemes.

[0023] In one implementation, the process begins by acquiring multi-source data from sensors and railway node monitoring devices. Here, multi-source data refers to a collection of raw, unprocessed data streams from different modes of transport. In one implementation, real-time route information is collected in real-time by a Global Positioning System (GPS) module deployed on road transport vehicles, specifically including dynamic parameters such as timestamps, vehicle unique identifiers, latitude and longitude, and current speed. Railway node details are obtained through monitoring devices deployed at railway stations or hubs, specifically including timestamps, train unique identifiers, station location, and operational status.

[0024] The waterway connectivity and air transport records are the basic transportation network data pre-stored in the graph database. This graph database is constructed by abstracting hubs such as ports and airports as nodes, and connecting relationships such as routes and flights as edges, storing attributes such as capacity and timetables. Its updates are obtained periodically from the official shipping schedules of shipping companies, port information systems of port authorities, cargo flight schedules of airlines, and databases of authoritative organizations such as the International Air Transport Association through application programming interfaces (APIs) or standardized data exchange protocols.

[0025] The rationale for pre-storing this data is as follows: First, the update frequency of this type of data is much lower than that of real-time location information, belonging to low-frequency change data. Pre-loading can avoid repeated, high-latency external data requests during each route planning. Second, the formats and structures of various data sources differ. The pre-storage process includes necessary extraction, transformation, and loading (ETL) steps, which can transform heterogeneous data into a unified, standardized graph data model, laying the foundation for efficient graph construction and querying in the future. Third, by maintaining a copy of the data locally or in an internal system, the stability and response speed of the route planning service can be improved, reducing the dependence on the availability of external data sources.

[0026] Next, based on the multi-source data, data alignment and correction operations are performed to obtain consistent transportation network data. Consistent transportation network data here refers to a standardized dataset that maintains consistency in the time dimension after time synchronization and deviation correction processing. Specifically, timestamp synchronization technologies such as network time protocols are used to align the timestamps of the collected multi-source data to millisecond-level precision. If time deviations are still found between the data after alignment, correction is performed through time series interpolation. The calculation process involves determining the proportion of the time interval between the target timestamp and the start and end timestamps, applying this proportion to the total displacement between the start and end positions to calculate the interpolated displacement, and finally superimposing this interpolated displacement onto the start position to obtain the corrected position estimate.

[0027] Finally, based on the consistent transportation network data, the waterway connectivity, and the air transport records, initial route planning is performed to obtain an initial transportation network dataset containing route optimization schemes. Here, the route optimization scheme refers to a planning result that defines the specific transportation modes, route nodes, and estimated timelines from the origin to the destination. The specific process of this planning operation is as follows: First, based on pre-stored basic transportation network data, the system constructs a multimodal transport network graph in the graph database, with cities, ports, airports, or railway hubs as nodes and connections between different transportation modes as edges. Then, real-time status data from the consistent transportation network data, such as road congestion and station busyness, are dynamically loaded as weights onto the edges and nodes of the graph.

[0028] Finally, a node scheduling algorithm, such as Dijkstra's algorithm, is executed to calculate a globally optimal transportation path. The iterative calculation process of this algorithm includes: First, initializing the path weights of all nodes to infinity, with the starting node having a path weight of 0; second, starting from the starting node, traversing all its neighboring nodes, calculating and updating the path weights from the starting node to these neighboring nodes; third, selecting the node with the smallest path weight among all currently calculated nodes that has not been visited, setting it as the new current node, and repeating step two until the target node is reached. The path weight between any two nodes is calculated by summing the base transportation time, dynamic delay time, and node transfer time. The node transfer time here is a time penalty term calculated based on the node's historical transfer efficiency and current capacity occupancy. After the algorithm completes, the output sequence of nodes with the smallest path weight is the path optimization scheme. Combining this scheme with consistent transportation network data constitutes the final output initial transportation network dataset.

[0029] In step S12, based on the initial transportation network dataset, network topology features are extracted to obtain a topology feature vector, including: Based on the initial transportation network dataset, the route delay distribution is calculated to obtain the delay distribution feature vector; Based on the delay distribution feature vector, network topology is determined and route priority is optimized to obtain an optimized cargo route allocation scheme. Based on the optimized cargo route allocation scheme, feature matching degree analysis is performed to obtain the topological feature vector.

[0030] In one implementation, firstly, based on the initial transportation network dataset, route delay distribution is calculated to obtain a delay distribution feature vector. This delay distribution feature vector is a multi-dimensional array that quantifies the statistical characteristics of delays along each transportation path, specifically including statistical indicators such as delay mean and variance. The specific process involves extracting highway traffic flow and path length data from the initial transportation network dataset and processing this data through time series analysis.

[0031] The specific analysis process is as follows: First, calculate the actual travel time for each trip record, which is the end timestamp minus the start timestamp. Second, aggregate the actual travel times according to different time periods (e.g., weekday peak hours, off-peak hours, and nighttime hours). Third, calculate the average travel time for each time period. Subsequently, the system subtracts a preset ideal travel time from this average travel time to obtain the average delay time for each path in different time periods. The ideal travel time here is a preset benchmark value, based on the legal maximum speed limit of the path as the ideal speed, calculated by dividing the path length by this ideal speed, or the travel time achievable under ideal operating conditions (e.g., no congestion, no weather impact) based on the design speed or legal speed limit. Finally, the system combines the average delay time of each path and its variance and other statistical indicators into a multidimensional array, forming the delay distribution feature vector.

[0032] Next, based on the delay distribution feature vector, network topology determination and route priority optimization are performed to obtain an optimized freight route allocation scheme. This optimized freight route allocation scheme is a specific multi-modal transportation path and timing arrangement assigned to a particular commodity, meeting its transportation timeliness requirements. Specifically, the system combines the delay distribution feature vector with railway node details (including node capacity and scheduling data) from the initial transportation network dataset, queries the connectivity of each node in the graph database, and determines a multi-modal freight route topology integrating road and rail transport. Here, node connectivity refers to the number of other nodes directly connected to a given node in the network topology.

[0033] Subsequently, the system determines if the estimated transport time in the cargo route topology does not match the cargo transport requirements. Here, the estimated transport time is a dynamically calculated time assessment value, set by summing the base transport time of all transport segments on a given path (derived from flight schedules, railway timetables, etc. in the initial transport network dataset), the average delay time corresponding to each segment (derived from the delay distribution feature vector), and the node transfer time.

[0034] For example, a batch of fresh produce requiring delivery within 12 hours is expected to take 15 hours via the optimal rail route under the current topology. This 15-hour calculation is obtained by adding the base travel time of 13 hours for the rail route to the average delay of 2 hours during the transit period, obtained from the delay distribution feature vector. Since the expected delivery time does not meet the requirements, the system initiates priority optimization. By prioritizing routes of other transportation modes (such as air travel) that can meet the time requirements, the system increases their priority and re-plans the delivery sequence, ultimately generating an optimized cargo route allocation scheme that specifies the transportation mode, transfer nodes, and specific train or departure times.

[0035] Finally, based on the optimized cargo route allocation scheme, feature matching degree analysis is performed to obtain a topology feature vector. This topology feature vector is a comprehensive data structure that not only includes the network's delay distribution characteristics and topology but also incorporates node operational efficiency evaluation metrics. The specific process of this analysis involves the system calculating the matching degree between the connectivity and transfer frequency of key nodes in the network. Transfer frequency refers to the number of cargo transfers a node processes per unit of time. The matching degree measures whether the network importance of a node (reflected by connectivity) matches its actual operational busyness (reflected by transfer frequency). This matching degree is calculated by normalizing the node's transfer frequency and connectivity, and then calculating their correlation coefficient or Euclidean distance. The system integrates the calculated matching degree with previously obtained parameters such as delay distribution characteristics and node connectivity to form the final output topology feature vector, which comprehensively describes the network's static structure, dynamic state, and operational efficiency.

[0036] In step S13, based on the topological feature vector, transit nodes are identified and classified to obtain classified node groups, including: Based on the topological feature vector, path adjustment event analysis is performed to obtain a set of path adjustment events; The number of events in the path adjustment event set is counted. If the number of events exceeds a preset threshold, the transfer node to be optimized is identified to obtain the set of transfer nodes to be optimized. Based on the set of transit nodes to be optimized, node feature clustering and group merging are performed to obtain classified node groups.

[0037] In one implementation, firstly, based on the topological feature vector, path adjustment event analysis is performed to obtain a path adjustment event set. Here, a path adjustment event refers to a record of a transportation route deviating from the predetermined plan due to factors such as weather, port congestion, or temporary capacity adjustments. Specifically, the system extracts the connectivity data of each transportation route from the topological feature vector and queries the graph database for the node connectivity and edge weights of the corresponding routes. When a sudden change in edge weights (such as transportation time) exceeding the normal fluctuation range is detected, the system marks this as a path adjustment event and stores its detailed information, including the affected nodes, event type, and occurrence timestamp, in the path adjustment event set.

[0038] It should be noted that the normal fluctuation range here is dynamically set based on historical data statistics. The system statistically analyzes the historical transportation time data for each route, calculates its mean and standard deviation, and sets upper and lower control limits according to statistical process control theory (e.g., the mean plus or minus three times the standard deviation). These control limits constitute the normal fluctuation range. When the real-time monitored transportation time exceeds this range, it is considered a sudden change.

[0039] Next, the number of events in the route adjustment event set is counted. If the number of events exceeds a preset threshold, the transfer nodes to be optimized are identified, resulting in a set of transfer nodes to be optimized. The threshold is a numerical standard used to judge the frequency of network anomalies. It is set based on statistical analysis of historical transportation data to calculate the distribution of the number of route adjustment events per unit time (e.g., daily), and the high percentile of this distribution (e.g., the 95th percentile) is taken as the threshold. Historical data collection must cover at least one complete business cycle (e.g., 12 months), and outliers (data exceeding Q1-1.5IQR or Q3+1.5IQR) are removed using box plots. This ensures that the system will only initiate optimization when the event frequency is significantly higher than most normal operating conditions, avoiding overreaction to normal fluctuations.

[0040] If the threshold is exceeded, the system initiates the identification of transfer nodes to be optimized, for example, using a breadth-first search (BFS) algorithm. The specific operation of this algorithm is as follows: First, the node with the highest event frequency in the path adjustment event set is placed into a queue as the starting node; second, a node is taken from the queue, and its node capacity and real-time scheduling data are queried. If the node's capacity utilization exceeds a preset capacity threshold (e.g., 90%), the node is added to the set of transfer nodes to be optimized. This capacity threshold is set individually for each node, based on a combination of the node's historical performance data and operational safety margin. Specifically, by analyzing the relationship curve between the node's historical capacity utilization and transfer delay time, the inflection point where the delay time begins to rise sharply is determined, and a certain safety buffer (e.g., 10%) is reserved on this basis, thus setting a capacity threshold (e.g., 90%) that ensures both efficiency and the ability to cope with emergencies; third, all unvisited adjacent nodes of the node are added to the queue; fourth, steps two and three are repeated until the queue is empty.

[0041] Finally, based on the set of transfer nodes to be optimized, node feature clustering and group merging are performed to obtain classified node groups. The specific process is as follows: First, the system extracts the node features of each transfer node to be optimized. Here, the node feature is a multi-dimensional vector containing parameters such as cargo throughput, flight or voyage frequency, and average transfer time. In one implementation, to eliminate the influence of differences in feature data scales (such as "tons" and "hours") on the clustering results, the system first performs Z-score normalization on each dimension of the node feature vector before performing clustering. The specific normalization process is as follows: for any feature dimension in the node feature vector (e.g., "cargo throughput"), the system first calculates the mean of all transfer nodes to be optimized along that feature dimension. with standard deviation Subsequently, for each node's eigenvalue... All are obtained through formulas Convert it to a normalized value This operation transforms all feature dimension data into a standard normal distribution with a mean of 0 and a standard deviation of 1. Then, a clustering algorithm such as K-means is used to group these normalized feature vectors.

[0042] The determination of the number of clusters K in this algorithm, i.e., the preset operational efficiency level, is based on a combination of business needs and data distribution characteristics. On the one hand, the number of levels is initially determined according to the business management requirement of stratifying nodes (such as core hubs, regional centers, and ordinary nodes); on the other hand, it is verified through a data-driven approach, for example, by using the elbow rule to analyze the sum of squares within clusters (WCSS) under different K values, and selecting the K value corresponding to the elbow of the curve in the figure as the optimal number of clusters to ensure that the classification results have natural discriminative power in the data. The specific operation process of the algorithm is as follows: First, determine the number of clusters K; second, randomly select K nodes as initial cluster centers; third, calculate the Euclidean distance from each node to each cluster center and assign it to the nearest cluster center to form K initial node groups; fourth, recalculate the center of each group; fifth, repeat steps three and four until the cluster centers no longer change.

[0043] Subsequently, group merging is performed. The cosine similarity between the center vectors of any two initial node groups is calculated. If the similarity value exceeds a preset similarity threshold, the two groups are merged. The similarity threshold here is set based on multiple clustering experiments on historical data to evaluate the group partitioning effect under different thresholds (e.g., using silhouette coefficients for evaluation), selecting an optimal value that can effectively merge groups with highly overlapping functions while maintaining clear boundaries between different functional groups.

[0044] The specific process of using silhouette coefficient for evaluation is as follows: First, a test interval for similarity thresholds is set (e.g., from 0.80 to 0.99, with a step size of 0.01). Second, for each threshold in the test interval, a complete group merging operation is performed, that is, all preliminary node groups with cosine similarity higher than the threshold are merged to obtain a final group partitioning scheme for the current iteration. Third, the system calculates the global average silhouette coefficient of the final group partitioning scheme. Here, the silhouette coefficient is used to measure the compactness (small intra-cluster distance) and separation (large inter-cluster distance) of the clustering results. Fourth, steps two and three are repeated, traversing all test thresholds. Finally, the system selects the threshold that maximizes the global average silhouette coefficient as the preset similarity threshold. Let the center vectors (based on normalized features) of any two preliminary node groups be A and B, where A = ,..., And B= ,..., Then, the formula for calculating the cosine similarity Sim(A,B) is: If the calculated similarity value Sim(A,B) exceeds the similarity threshold, the two groups are merged into one. This merging process is repeated until the similarity between any two groups is below the threshold. The final set of groups is the classified node group.

[0045] In step S14, based on the classified node groups and the topological feature vectors, a dynamic tracing graph is constructed to obtain a complete tracing relationship graph, including: Based on the classified node groups and the topological feature vectors, a preliminary route query is performed to obtain a preliminary set of cargo routes; Based on the initial set of cargo routes, the transportation relationships are dynamically updated to obtain a dynamic set of transportation relationships; Based on the dynamic transportation relationship set, route clustering and optimization are performed to obtain the optimized route set; Based on the optimized path set, the final graph is constructed to obtain a complete traceability graph.

[0046] In one implementation, firstly, a preliminary route query is performed based on the categorized node groups and the topological feature vectors to obtain a preliminary set of freight routes. This preliminary set of freight routes is a collection of initial transport paths between node groups. This query operation is performed on a graph database that has been built and loaded with basic transport network data in previous steps. The graph database is structured with cities, ports, airports, or railway hubs as nodes and connections between different modes of transport as edges, and is regularly retrieved and updated via API calls to authoritative data sources. This graph database, as a persistent data structure, provides the basic multimodal transport network graph for this step. The specific query process involves using the categorized node groups as the start and end points of the query, and employing the query language of the graph database to find all paths connecting these groups. The topological feature vectors serve as constraints for the query; for example, only paths with historical freight volumes exceeding a specific historical freight volume threshold are queried. This specific threshold is set based on business rules; for example, based on traffic volume ranking, only the top 80% of main transport routes are selected to filter out unimportant connections, thereby generating the preliminary set of freight routes.

[0047] Next, based on the initial set of cargo routes, the transportation relationships are dynamically updated to obtain a dynamic set of transportation relationships. The core of this operation is to verify whether each route in the initial set matches the real-time data and adjust the edge weights accordingly. The specific verification process is as follows: for each route in the initial set of cargo routes, the system extracts its baseline performance parameters, such as the average transportation time derived from the topological feature vector. The specific calculation process is as follows: in step S12, the system has extracted the historical travel records of the route from the initial transportation network dataset. The actual travel time is obtained by calculating the difference between the end timestamp and the start timestamp of each record, and these actual travel times are aggregated and averaged according to time periods (such as peak and off-peak periods) to calculate the baseline value.

[0048] Simultaneously, the system calls an external interface to obtain real-time logistics status data related to the route. The verification operation compares these two data points. If the difference between the real-time transportation time and the baseline transportation time is within a preset normal fluctuation range, the verification passes. The definition of the normal fluctuation range here is consistent with that in step S13, and is also dynamically set based on the statistical process control theory of historical transportation time data (e.g., mean plus or minus three standard deviations). If it exceeds this range, the verification fails, and a weight adjustment is triggered. The initial setting of the edge weight is based on historical data, for example, set to the reciprocal of the average transportation time. The weight adjustment process is based on the quantitative calculation of the verification results. Specifically, the new weight is equal to the initial weight minus a delay penalty term, which is the difference between the real-time transportation time and the baseline transportation time multiplied by a preset weight adjustment coefficient. This coefficient is set based on the sensitivity of different transportation modes to time delay; for example, the adjustment coefficient for air transportation is higher than that for road transportation.

[0049] Subsequently, route clustering and optimization are performed on the dynamic transportation relationship set to obtain an optimized path set. This operation uses the K-means clustering algorithm, and its specific process is as follows: First, a feature vector is created for each route in the dynamic transportation relationship set. This vector contains data from multiple dimensions, such as its finally updated edge weights, transportation frequency, on-time rate, and node capacity utilization. Second, the elbow rule is used to determine the optimal number of clusters K to divide the routes into different operational efficiency levels. Third, K cluster centers are randomly initialized. Fourth, the Euclidean distance from the feature vector of each route to these K cluster centers is calculated, and the route is assigned to the cluster represented by the nearest center. Fifth, the centroid of each cluster is recalculated, which is the mean of the feature vectors of all routes within the cluster. Sixth, steps four and five are repeated until the centroids of all clusters no longer change. After clustering is completed, the clusters are labeled according to the feature values ​​of the centroids of each cluster (e.g., high efficiency, medium efficiency), and routes belonging to the high efficiency and medium efficiency clusters are selected to form the optimized path set.

[0050] Finally, based on the optimized path set, a final graph is constructed to obtain a complete traceability graph. Formal modeling here refers to instantiating the logical path set into a physical data structure within a graph database. The specific process is as follows: First, the categorized node groups are created as node objects in the graph database, and each node object is assigned attributes such as group ID, geographical range, and main transportation function. Second, the optimized path set is traversed, creating a directed edge object between the corresponding node objects for each path. Third, the key attributes of the path, including its final dynamically adjusted weight, transportation mode, and real-time on-time rate, are attached as attributes to the corresponding edge objects. Through these operations, a complete traceability graph with node groups as vertices and optimized, dynamically weighted transportation relationships as edges is constructed.

[0051] In step S15, based on the complete tracing relationship graph, path status association and optimization are performed to obtain the associated path status dataset, including: Based on the complete tracing relationship graph, edge weight analysis and synchronization frequency optimization are performed to obtain the optimized synchronization mechanism. Based on the optimized synchronization mechanism, cargo status is updated and delayed paths are identified to obtain a set of transportation routes that need to be adjusted. Based on the set of transportation routes that need to be adjusted, the tracking path to be optimized is determined, and the set of tracking paths to be optimized is obtained. Based on the set of tracking paths to be optimized, node change attributes are fused to obtain the associated path status dataset.

[0052] In one implementation, firstly, based on the complete traceability graph, edge weight analysis and synchronization frequency optimization are performed to obtain an optimized synchronization mechanism. Specifically, the system extracts the weight data of all edges from the complete traceability graph and compares it with a preset edge weight delay threshold. This threshold is set based on statistical process control analysis of the historical edge weight data of the path. Specifically, the system calculates the mean and standard deviation of the edge weights of the path over the past statistical period and uses the mean minus three times the standard deviation as a dynamic delay threshold to identify statistically abnormally low weights.

[0053] If the weight of an edge is below this threshold, the system determines that the path has a large-scale data processing delay. For all paths identified as delayed, the system uses a time window analysis method to statistically analyze their data synchronization frequency. This method mainly defines a time window size parameter, such as 24 hours. The parameter is set based on selecting a standard duration that matches the complete operational cycle of the logistics business, ensuring that the statistical results can reflect a complete daytime operational cycle, thereby balancing the representativeness and timeliness of the statistics. The specific operation for calculating the data synchronization frequency is as follows: the system queries the data synchronization log, retrieves the timestamp records of all successful synchronization events of the target path within the time window (e.g., the past 24 hours), counts the total number of these records, and then divides the total number by the number of hours in the time window (e.g., 24 hours) to calculate the statistical synchronization frequency of the path, in units of "times / hour".

[0054] Subsequently, the statistical frequency is compared with a preset synchronization frequency threshold. This threshold is determined by comprehensively considering the value of the goods, timeliness requirements, and characteristics of the transportation mode. A common method for setting this threshold is based on the maximum tolerable information delay for the services carried by that route. The specific setting process is as follows: First, based on the value of the goods, timeliness requirements, and characteristics of the transportation mode, a service level is defined for each route or type of service, and a maximum tolerable information delay is pre-defined for each service level. (For example, time-sensitive air transport) (10 minutes). This delay represents the longest acceptable time from the occurrence of an event to its detection by the system. The second step is for the system to... To set the synchronization frequency threshold, the system requires a minimum synchronization interval to ensure timely information acquisition. Must be smaller than .

[0055] In one implementation, it can be set Therefore, the synchronization frequency threshold The formula for calculating (times / hour) is: (in and The unit is minutes. If the statistical frequency is lower than this calculated value... The system then dynamically increases the data acquisition and synchronization frequency for that path, and finally outputs a set containing updated synchronization frequency parameters for that specific path. This set is the optimized synchronization mechanism.

[0056] For example, the system has a built-in rule matrix that sets the synchronization frequency threshold for high-value or cold chain goods in air transport mode to at least 12 times per hour, while setting the threshold for ordinary goods in water transport mode to at least once every 4 hours. If the statistical frequency falls below this threshold, the system dynamically increases the data collection and synchronization frequency for that route, ultimately outputting a set containing updated synchronization frequency parameters for that specific route. This set is the optimized synchronization mechanism.

[0057] Next, based on the optimized synchronization mechanism, cargo status updates and delayed path identification are performed to obtain the set of transportation routes that need adjustment. Specifically, the system applies the optimized synchronization mechanism to obtain the latest highway route information and railway node details from the initial transportation network dataset, and updates the status timestamps of the corresponding cargo in the transportation network.

[0058] The system then calculates the time interval between two consecutive status updates and compares it with a preset status update timestamp interval threshold. This threshold is set based on the physical and technical limitations of each mode of transport under normal operating conditions. Specifically, it is determined based on the average speed of the transport mode and the deployment density of data collection points along the route to ensure that there are no excessively large tracking blind spots between two collection points. For example, for a railway with an average speed of 100 km / h and monitoring equipment every 100 km along the route, the threshold can be set to 1 hour. If the calculated time interval exceeds the threshold for the corresponding transport mode, the cargo is determined to have a status update delay. The system summarizes and deduplicates the transport routes of all cargoes with status update delays, forming a set of transport routes that need adjustment.

[0059] Then, based on the set of transportation routes requiring adjustment, the system determines the tracking paths to be optimized, thus obtaining a set of tracking paths requiring optimization. Specifically, the system analyzes the frequency of transportation mode switching on each route in the set of routes requiring adjustment and compares it with a preset multi-mode switching frequency threshold. This threshold is determined by combining standard operating procedures (SOPs) and historical efficiency data analysis. Specifically, the system first obtains the predetermined number of switching operations for the route in the standard plan and uses this as a benchmark; then, by analyzing historical data, it identifies the inflection point in the number of switching operations that leads to a significant increase in the total transportation delay rate after exceeding the predetermined number of switching operations. This threshold is set as a safe value between the standard number of switching operations and this inflection point to identify unplanned abnormal scheduling that may lead to reduced efficiency. If the switching frequency of a route exceeds this threshold, the system determines that there is a problem with the scheduling and coordination of that route, and further determines it from the unfiltered "needs adjustment" state to "needs optimization" requiring in-depth analysis, ultimately forming a set of tracking paths requiring optimization. Finally, based on the set of tracking paths to be optimized, node change attributes are fused to obtain the associated path status dataset. Specifically, the system extracts change detection attributes of the nodes involved in the set of tracking paths to be optimized from the initial transportation network dataset. These attributes include the node status change timestamp and transportation mode identifier.

[0060] The core of the fusion operation is to perform correlation calculation. The specific process of this calculation is as follows: for a certain cargo delay that occurs on the path, let its occurrence timestamp be... For a node on this path, let its state change timestamp be... The system sets a causal relationship time window. Its length is set based on the typical propagation time of the node type's impact on downstream applications. A high correlation is defined as follows: the cargo delay timestamp follows the node state change timestamp, and the time difference between the two falls within the causal association time window. The mathematical expression for this condition is: If this condition is met, the system determines the change event of that node as the root cause of the delay; the causal relationship time window The length is set according to the node type. For international hub ports, the typical transmission time is longer, and the window can be set to 24 hours; for regional railway hubs, the window can be set to 6 hours; and for urban distribution stations, the window can be set to 2 hours.

[0061] Subsequently, a status association operation is performed. This operation involves adding a set of predefined association fields to the data structure recording the delay path and assigning values ​​to these fields. For example, a boolean field "Has the root cause been identified?" is added and set to true; a string field "Root cause node ID" is added and set to its unique identifier; and a timestamp field "Root cause event time" is added and set to true. Through this operation, the system structurally binds high-level latency phenomena with low-level node events, forming the final output associative path state dataset.

[0062] In step S16, based on the associated path status dataset, the complete traceability relationship graph is generated and updated to obtain the final logistics traceability result, including: Based on the associated path state dataset, the path weights are updated to obtain the optimized path weight dataset. Based on the optimized path weight dataset, state and weight clustering is performed to obtain path state clustering. Based on the path status clustering and grouping, event-triggered judgment and update are performed to obtain the final logistics traceability result.

[0063] In one implementation, firstly, the path weights are updated based on the associated path state dataset to obtain an optimized path weight dataset. Specifically, the system iterates through each path identified as needing optimization in the associated path state dataset. For each path, the system obtains its current edge weight from the complete tracing relationship graph and calculates a delay weight adjustment value based on the associated root cause (i.e., node change event). This adjustment value is calculated by dividing the actual delay time caused by the node change event on the path by the ideal travel time defined in step S12 (this ideal travel time is a benchmark value calculated based on the path length and the legal maximum speed limit), obtaining a delay rate, and then multiplying this delay rate by a preset adjustment coefficient. This adjustment coefficient is a parameter used to quantify the impact of different root causes on the future stability of the path; its setting is based on the nature and expected duration of the root cause event.

[0064] Specifically, the system incorporates a rule lookup table that maps different types of node change events to different coefficient values. For example, for events like "temporary equipment maintenance," which have a short expected duration and controllable impact, the corresponding adjustment coefficient is lower (e.g., 0.5) to apply a moderate weight penalty. Conversely, for events like "long-term port congestion" or "structural bottlenecks in the route," which have a long expected duration and wide impact, the corresponding adjustment coefficient is higher (e.g., 0.9) to apply a more significant weight penalty, thereby significantly reducing their priority in subsequent path planning. The system then updates the edge weights of the path. This update process can be described as subtracting the calculated delay weight adjustment value from 1 and multiplying it by the current edge weight to obtain the new edge weight. To ensure the weights remain positive, the system sets an upper limit of 0.95 for the delay weight adjustment value. This multiplicative model conforms to the convention of edge weights representing costs in graph theory, and the adjustment value acts as a penalty factor, reducing the edge's priority. The system records all updated paths and their new edge weights, forming an optimized path weight dataset.

[0065] It should be noted that the rule lookup table is a pre-configured item in the system. Its core logic is to map the classification of root cause events to adjustment coefficients. The event classification mainly includes three categories: short-term temporary events, medium-term recoverable events, and long-term structural events, which correspond to three adjustment coefficient ranges: low, medium, and high, respectively. The system administrator can maintain and calibrate the specific mapping relationship in the lookup table based on historical operational data and business strategies.

[0066] Next, based on the optimized path weight dataset, state and weight clustering is performed to obtain path state clustering groups. The specific process is as follows: The system uses, for example, a K-means clustering algorithm to group the cargo tracking states. First, a feature vector is created for each in-transit cargo. This vector contains at least two dimensions: the first dimension is the optimized new edge weight of the path the cargo is on; the second dimension is the time interval between the latest state update timestamp and the last update timestamp. Then, the number of clusters K is preset according to business needs (e.g., K=3, corresponding to three state levels: "high efficiency and stability," "slight fluctuations," and "high risk and delay," respectively). The algorithm iteratively calculates and groups cargoes with similar feature vectors (i.e., cargoes with similar path weights and state update frequencies) into the same cluster. After clustering, the system assigns a corresponding state label to each cluster, and the final output is the path state clustering grouping of all in-transit cargoes into different state levels.

[0067] Finally, based on the path status clustering, event-triggered judgment updates are performed to obtain the final accurate logistics traceability result. The specific process involves the system updating its built-in event-triggered judgment logic within the complete traceability graph. Here, event-triggered judgment refers to a series of rules used to automatically monitor and respond to changes in cargo status. Specifically, the update operation upgrades the original judgment rules based on a single indicator (such as only judging whether the timestamp interval has expired) to composite judgment rules based on the path status clustering. For example, a new rule could be set as follows: "If the clustering label of a cargo changes from 'highly efficient and stable' to 'slightly fluctuating,' a warning notification is generated; if it changes to 'high-risk delay,' a high-level alarm is immediately triggered, and alternative routes are automatically suggested to the dispatch center." By executing these updated rules, the system can perform differentiated management and proactive intervention for cargo with different risk levels.

[0068] In this step, the updated complete traceability graph is used as the basis for subsequent iterative calculations, thus reflecting the iterative optimization process. Specifically, when a new transportation task triggers step S11 for initial path planning, the node scheduling algorithm directly reads the updated edge weights (e.g., lower weights penalized for delays), thereby automatically avoiding identified high-delay-risk paths during the path generation phase. Simultaneously, this updated graph also serves as input for the next step S12 to extract network topology features, ensuring that the feature vectors reflect the latest network state and achieving iterative optimization and adaptability of the entire traceability method. The graph weight update adopts an event-driven mechanism, with triggering conditions including node state change timestamp refresh and path delay rate exceeding a dynamic threshold (mean ± 2 standard deviations). After each trigger, the system completes weight recalculation within 500 milliseconds and notifies downstream modules to update synchronously via a message queue (such as RabbitMQ).

[0069] It should be noted that the final accurate logistics traceability result is not a static data file, but rather the dynamic tracking service output by the method itself. It is reflected in the system's ability to provide users with cargo tracking status that has been updated in real time based on clustering results and has a clear risk level (such as "high-risk delay"), and to automatically execute corresponding early warning or alarm operations based on the updated event trigger judgment logic.

[0070] In summary, this invention integrates heterogeneous real-time data from multimodal transport to construct a dynamic traceability map containing topological features, and utilizes path adjustment events and data processing delay analysis to form a closed-loop feedback loop, continuously optimizing the map's synchronization mechanism and weight parameters, thereby achieving accurate and efficient end-to-end traceability of goods in complex dynamic networks.

[0071] Reference Figure 2 The second embodiment of the present invention provides a multimodal transport single-document logistics traceability system, comprising: The data integration module is used to acquire real-time route information, railway node details, waterway connection relationships and air transport records, and integrate the data to obtain an initial transportation network dataset; The feature extraction module is used to extract network topology features based on the initial transportation network dataset to obtain a topology feature vector; The node classification module is used to identify and classify transit nodes based on the topological feature vector to obtain classified node groups. The graph construction module is used to dynamically construct a traceability graph based on the classified node groups and the topological feature vectors to obtain a complete traceability relationship graph. The path optimization module is used to perform path status association and optimization based on the complete tracing relationship graph to obtain the associated path status dataset. The result generation module is used to generate and update the complete traceability relationship graph based on the associated path status dataset to obtain the final logistics traceability result.

[0072] It should be noted that the multimodal transport single-document logistics traceability system provided in this embodiment of the invention is used to execute all the process steps of the multimodal transport single-document logistics traceability method in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0073] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a multimodal transport single-document logistics traceability program. When the processor executes the computer program, it implements the steps described in the various multimodal transport single-document logistics traceability method embodiments above, for example... Figure 1 The step S11 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above system embodiments, such as the map construction module.

[0074] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0075] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0076] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0077] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0078] If the modules / units integrated into the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or system capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0079] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0080] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A multimodal transport single-document logistics traceability method, characterized in that, include: Acquire real-time route information, railway node details, waterway connectivity, and air transport records, and integrate the data to obtain an initial transportation network dataset; Based on the initial transportation network dataset, network topology features are extracted to obtain a topology feature vector; Based on the topological feature vector, transit nodes are identified and classified to obtain classified node groups, including: analyzing path adjustment events based on the topological feature vector to obtain a set of path adjustment events; counting the number of events in the set of path adjustment events; if the number of events exceeds a preset threshold, identifying transit nodes to be optimized to obtain a set of transit nodes to be optimized; and clustering and merging node features based on the set of transit nodes to be optimized to obtain classified node groups. Based on the classified node groups and the topological feature vectors, a dynamic tracing graph is constructed to obtain a complete tracing relationship graph; Based on the complete traceability graph, path status association and optimization are performed to obtain an associated path status dataset, including: performing edge weight analysis and synchronization frequency optimization based on the complete traceability graph to obtain an optimized synchronization mechanism; performing cargo status updates and delayed path identification based on the optimized synchronization mechanism to obtain a set of transportation routes that need adjustment; determining the tracking paths to be optimized based on the set of transportation routes to be adjusted to obtain a set of tracking paths to be optimized; and fusing node change attributes based on the set of tracking paths to be optimized to obtain an associated path status dataset. Based on the associated path status dataset, the traceability results are generated and the complete traceability relationship graph is updated to obtain the final logistics traceability result.

2. The multimodal transport single-document logistics traceability method according to claim 1, characterized in that, The process involves acquiring real-time path information, railway node details, waterway connectivity, and air transport records, and integrating this data to obtain an initial transportation network dataset, including: Real-time path information and railway node details are obtained from sensors and railway node monitoring devices to obtain multi-source data; Retrieve waterway connectivity and air transport records from a pre-established database; Based on the multi-source data, data alignment and correction are performed to obtain consistent transportation network data; Based on the consistent transportation network data, the waterway connectivity, and the air transport records, initial route planning is performed to obtain an initial transportation network dataset containing route optimization schemes.

3. The multimodal transport single-document logistics traceability method according to claim 1, characterized in that, The step of extracting network topology features based on the initial transportation network dataset to obtain a topology feature vector includes: Based on the initial transportation network dataset, the route delay distribution is calculated to obtain the delay distribution feature vector; Based on the delay distribution feature vector, network topology is determined and route priority is optimized to obtain an optimized cargo route allocation scheme. Based on the optimized cargo route allocation scheme, feature matching degree analysis is performed to obtain the topological feature vector.

4. The multimodal transport single-document logistics traceability method according to claim 1, characterized in that, The step of constructing a dynamic tracing graph based on the classified node groups and the topological feature vectors to obtain a complete tracing relationship graph includes: Based on the classified node groups and the topological feature vectors, a preliminary route query is performed to obtain a preliminary set of cargo routes; Based on the initial set of cargo routes, the transportation relationships are dynamically updated to obtain a dynamic set of transportation relationships; Based on the dynamic transportation relationship set, route clustering and optimization are performed to obtain the optimized route set; Based on the optimized path set, the final graph is constructed to obtain a complete traceability graph.

5. The multimodal transport single-document logistics traceability method according to claim 1, characterized in that, The step of generating and updating the complete traceability relationship graph based on the associated path status dataset to obtain the final logistics traceability result includes: Based on the associated path state dataset, the path weights are updated to obtain the optimized path weight dataset. Based on the optimized path weight dataset, state and weight clustering is performed to obtain path state clustering. Based on the path status clustering and grouping, event-triggered judgment and update are performed to obtain the final logistics traceability result.

6. A multimodal transport single-document logistics traceability system, characterized in that, include: The data integration module is used to acquire real-time route information, railway node details, waterway connection relationships and air transport records, and integrate the data to obtain an initial transportation network dataset; The feature extraction module is used to extract network topology features based on the initial transportation network dataset to obtain a topology feature vector; The node classification module is used to identify and classify transit nodes based on the topological feature vector to obtain classified node groups. This includes: analyzing path adjustment events based on the topological feature vector to obtain a set of path adjustment events; counting the number of events in the set of path adjustment events; if the number of events exceeds a preset threshold, identifying transit nodes to be optimized to obtain a set of transit nodes to be optimized; and performing node feature clustering and group merging based on the set of transit nodes to be optimized to obtain classified node groups. The graph construction module is used to dynamically construct a traceability graph based on the classified node groups and the topological feature vectors to obtain a complete traceability relationship graph. The path optimization module is used to perform path status association and optimization based on the complete traceability relationship graph to obtain an associated path status dataset. This includes: performing edge weight analysis and synchronization frequency optimization based on the complete traceability relationship graph to obtain an optimized synchronization mechanism; performing cargo status updates and delayed path identification based on the optimized synchronization mechanism to obtain a set of transportation routes that need adjustment; determining the tracking paths to be optimized based on the set of transportation routes to be adjusted to obtain a set of tracking paths to be optimized; and performing node change attribute fusion based on the set of tracking paths to be optimized to obtain an associated path status dataset. The result generation module is used to generate and update the complete traceability relationship graph based on the associated path status dataset to obtain the final logistics traceability result.

Images