Logistics goods transportation tracking and scheduling method based on multi-source data fusion

Abstract

The application provides a logistics goods transportation tracking scheduling method based on multi-source data fusion, comprising: acquiring real-time positioning data, planning path data, time efficiency attribute data and road network dynamic event data of each transportation task; predicting the occupation time interval of each transportation task on each road section in a future time window and generating a road section-task occupation matrix; in response to a traffic capacity reduction event, screening a conflict task set from the matrix, and dividing it into a high priority level and a low priority level according to a priority classification rule; the high priority level keeps the original path and allocates the remaining traffic capacity according to the occupation time interval, and takes the occupation time interval and the occupied road section as a space-time obstacle constraint; the low priority level re-plans an alternative path that avoids the constraint in space-time according to the constraint; and the traffic permission and the alternative path are respectively issued. The application can realize the orderly allocation of traffic resources without global unified re-planning, and avoid space-time conflicts between paths.

Description

Technical Field

[0001] This invention relates to the field of logistics information processing technology, and in particular to a logistics cargo transportation tracking and scheduling method based on multi-source data fusion. Background Technology

[0002] In logistics and transportation networks, capacity reduction caused by dynamic events on the road network is the core reason for resource competition among multiple transportation tasks on road segments or nodes. When such events occur, multiple transportation tasks planned to pass through the affected road segment during the affected period simultaneously compete for the limited remaining capacity. How to rationally allocate this capacity to take into account the differentiated timeliness requirements of different transportation tasks is a technical challenge faced in the field of logistics scheduling.

[0003] Existing scheduling methods typically employ two strategies. First, a globally unified route replanning is performed on all affected vehicles. This involves inputting dynamic event information into a route planning algorithm and recalculating the routes of all vehicles en route. Second, passage rights are allocated sequentially according to a first-come, first-served time order. However, globally unified replanning involves the simultaneous recalculation of numerous vehicle routes, resulting in long response times and high computational resource consumption, making it difficult to meet real-time scheduling requirements. The first-come, first-served strategy completely disregards the differences in timeliness requirements among different transportation tasks. Time-sensitive tasks such as cold chain fresh produce and emergency medical services may lose commercial or social value due to waiting in queues. Furthermore, the above strategies do not use the already reserved routes and travel times of some vehicles as hard constraints for other vehicles' route searches after route replanning. This leads to new intersections and competition between replanned alternative routes and unreplanned vehicle routes in subsequent time and space, causing repeated iterative adjustments to the scheduling scheme. Summary of the Invention

[0004] The embodiments of the present invention provide a logistics cargo transportation tracking and scheduling method based on multi-source data fusion, which aims to solve the problem that when dynamic events cause competition for transportation resources, the existing scheduling methods do not establish a differentiated passage arbitration mechanism among multiple tasks, resulting in the inability to guarantee the right of passage for high-time-efficiency tasks and the easy occurrence of spatiotemporal conflicts between replanned paths.

[0005] To achieve the above objectives, this invention provides a logistics cargo transportation tracking and scheduling method based on multi-source data fusion, comprising the following steps: Acquire multi-source data on logistics and cargo transportation, including real-time location data, planned route data, timeliness attribute data, and road network dynamic event data for each of the multiple transportation tasks; Based on the real-time positioning data and the planned route data, the time interval for each of the transportation tasks to occupy each road segment within the future time window is predicted. A road segment-task occupancy matrix is ​​generated based on each of the occupancy time intervals. The road segment-task occupancy matrix uses each road segment in the road network as the row index and each of the transportation tasks as the column index. The matrix elements are the occupancy time intervals of the corresponding transportation tasks on the corresponding road segments. In response to the road network dynamic event data indicating that a traffic capacity reduction event has occurred in a target road segment or target node, all matrix elements with the travel index of the target road segment or target node are filtered from the road segment-task occupancy matrix, and the transportation tasks corresponding to the filtered matrix elements are determined as a set of conflict tasks. Based on the timeliness attribute data of each transportation task in the conflict task set, the conflict task set is divided into a high-priority level and a low-priority level according to a preset priority classification rule, wherein the number of transportation tasks included in the high-priority level is less than or equal to the remaining traffic capacity of the target road segment or target node after the traffic capacity reduction event. For each of the transportation tasks in the high priority level, its planned path remains unchanged, and the remaining traffic capacity is allocated to each of the transportation tasks in the high priority level according to the time interval occupied by each of the transportation tasks. The time interval and route segment occupied by each transportation task in the high priority level are defined as spatiotemporal obstacle constraints; Based on the spatiotemporal obstacle constraints, alternative routes are replanned for each of the transportation tasks in the low priority level, wherein the alternative routes avoid the spatiotemporal region defined by the spatiotemporal obstacle constraints. The passage permits for each transportation task in the high-priority level and the alternative paths for each transportation task in the low-priority level are respectively sent to the corresponding terminal devices.

[0006] Furthermore, predicting the time interval for each transportation task to occupy each road segment within a future time window based on the real-time positioning data and the planned route data includes the following steps: Based on the real-time positioning data, the location coordinates and driving speed of each transportation task at the current moment are extracted, and based on the positioning data sequence of each transportation task in the historical time period, the historical travel time sequence of each transportation task is generated. Based on the planned route data, the sequence of road segments to be traversed by each of the transportation tasks within the future time window is determined; The historical travel time series, current location coordinates, current driving speed, and real-time traffic flow parameters of each road segment in the sequence of road segments to be traversed are input into the pre-trained travel time prediction model. The travel time prediction model is a bidirectional long short-term memory network. The bidirectional long short-term memory network extracts the forward and backward temporal features of the transportation task in the historical travel time series, and corrects the forward and backward temporal features based on the current location coordinates, current driving speed and the real-time traffic flow parameters, and outputs the first predicted arrival time and the first predicted departure time of each of the transportation tasks to each segment in the sequence of road segments to be traversed. The first predicted arrival time and the first predicted departure time are defined as the occupied time interval.

[0007] Furthermore, the preset priority ranking rules include: Obtain the timeliness attribute data for each of the aforementioned transportation tasks, wherein the timeliness attribute data includes task type identifier and agreed delivery time limit; When the task type is identified as cold chain transportation, medical transportation, fresh food transportation and / or emergency relief supplies transportation, the corresponding transportation task will be marked as the first priority. When the task type is identified as general cargo transportation and the time difference between the agreed delivery time limit and the current time is less than the first preset time threshold, the corresponding transportation task will be marked as the second priority. When the task type is identified as general cargo transportation and the time difference between the agreed delivery time limit and the current time is greater than or equal to the first preset time threshold, the corresponding transportation task will be marked as the third priority. The high priority level consists of transportation tasks of the first priority and the second priority, and the low priority level consists of transportation tasks of the third priority.

[0008] Furthermore, the road network dynamic event data includes real-time traffic data, weather warning data, and traffic control data; the capacity reduction event is determined through one or more of the following methods: When the real-time traffic data indicates that the average traffic speed of the target road segment is lower than a first preset percentage threshold of the historical average speed of the same period of the road segment, it is determined that a traffic capacity reduction event has occurred in the target road segment. When the meteorological warning data indicates that the meteorological warning level in the area where the target road segment is located exceeds the preset level, it is determined that a traffic capacity reduction event has occurred on the target road segment; When the traffic control data indicates that traffic control should be implemented on the target road segment or target node, it is determined that a traffic capacity reduction event has occurred on the target road segment or target node.

[0009] Furthermore, the method for calculating the remaining traffic capacity includes: The historical number of vehicles passing through the target road segment or target node per unit time before the occurrence of the capacity reduction event is obtained as the baseline capacity. Based on the number of lanes in the target road segment or target node, the event type of the capacity reduction event, and the capacity reduction factor corresponding to the event type, the predicted number of vehicles passing through per unit time after the event occurs is calculated as the event capacity. The difference between the baseline traffic capacity and the event traffic capacity is determined as the traffic capacity loss of the target road segment or target node; The remaining capacity is determined by multiplying the event throughput capacity by a preset security redundancy coefficient.

[0010] Furthermore, the step of replanning alternative routes for each of the transportation tasks in the low-priority level based on the spatiotemporal obstacle constraints includes: Starting from the current location of the target transportation task in the low priority level, and ending at the final destination of the transportation task, a candidate path is searched in the road network. The search space for the candidate paths excludes paths that include the spatiotemporal region defined by the spatiotemporal obstacle constraints. The alternative path is selected from the candidate paths, and the path cost is calculated by weighting the total travel distance, the total estimated travel time, and the real-time congestion index of the road segments traversed by the path. The total estimated travel time of the route shall not exceed the second preset percentage threshold of the total travel time of the original planned route for the transportation task.

[0011] Furthermore, after determining the time interval and route segment occupied by each transportation task in the high-priority level as spatiotemporal obstacle constraints, the method further includes: Continuously acquire real-time location data of each of the transportation tasks in the high priority level, and calculate the actual driving speed of each of the transportation tasks based on the real-time location data; When the absolute value of the deviation between the actual driving speed of any of the high-priority transportation tasks and the predicted driving speed of the task exceeds a preset deviation threshold, the occupancy time interval of its subsequent road segments is re-predicted based on the current real-time positioning data of the transportation task, and the updated occupancy time interval is obtained. Update the corresponding occupancy time interval in the spatiotemporal obstacle constraint according to the updated occupancy time interval; Based on the updated spatiotemporal obstacle constraints, alternative routes are replanned for each transportation task in the affected low-priority level.

[0012] Further, the step of issuing the passage permits for each transportation task in the high-priority level and the alternative paths for each transportation task in the low-priority level to the corresponding terminal devices includes: Based on the terminal device identifier of each transportation task in the high priority level, the passage permission of each transportation task is encapsulated into a first data packet and sent to the corresponding terminal device. The first data packet contains the passage time interval of the transportation task on the target road segment or target node. Based on the terminal device identifier of each transportation task in the low priority level, the alternative path of each transportation task is encapsulated into a second data packet and sent to the corresponding terminal device. The second data packet contains the identifier of each route segment of the alternative path and the predicted time sequence of arrival at each route segment. The first data packet and the second data packet are transmitted using encryption keys pre-negotiated with the terminal device.

[0013] Furthermore, the multi-source data also includes: Acquire multi-mode positioning data for each of the aforementioned transportation tasks, wherein the multi-mode positioning data includes at least two of the following: satellite positioning data, RFID gantry reading records, and AIS Automatic Identification System (AIS) ship data; When the signal strength of the satellite positioning data for any transportation task is lower than a preset signal threshold in the target area, the RFID gantry reading record or the AIS data shall be used instead of the satellite positioning data as the positioning data source for the transportation task. When neither the RFID gantry reading record nor the AIS data can be obtained, the trajectory of the transportation task is calculated based on the historical occupancy time interval in the segment-task occupancy matrix to generate the complete trajectory of the transportation task during the signal interruption.

[0014] The above technical solution has the following technical effects: First, it reduces the computational complexity and response latency of scheduling under dynamic events. This invention constructs a road segment-task occupancy matrix, representing the spatiotemporal occupancy demands of multiple transportation tasks on road network resources using a unified data structure. Based on this, when a capacity reduction event occurs, only matrix elements with travel indices matching the target road segment or node need to be selected to quickly locate the set of conflicting tasks, avoiding redundant computation by globally traversing all transportation tasks. Simultaneously, high-priority levels retain their original paths, while only low-priority levels undergo path replanning. Compared to existing methods that perform global unified replanning on all affected vehicles, the path search space is significantly reduced, and the scheduling response speed is significantly improved.

[0015] Secondly, this invention achieves the orderly allocation of limited traffic resources among multiple tasks, ensuring the right-of-way for high-time-sensitive tasks. The invention divides the set of conflicting tasks into different levels according to a preset priority classification rule, and limits the number of high-priority tasks to within the remaining traffic capacity of the target road segment or node after a traffic capacity reduction event. This ensures that high-time-sensitive tasks can obtain traffic guarantees and prevents urgent tasks such as cold chain fresh produce and emergency medical services from losing commercial value or social utility due to a first-come, first-served strategy while queuing. By allocating the remaining traffic capacity to each high-priority task according to the order of its occupied time interval, the invention achieves an orderly arrangement of traffic resources in the time dimension, avoiding the exacerbation of resource conflicts caused by simultaneous competition among multiple tasks.

[0016] Third, it avoids secondary spatiotemporal conflicts and repeated iterative adjustments in the scheduling scheme. This invention defines the occupied time intervals and occupied road segments of each high-priority transportation task as spatiotemporal obstacle constraints. Alternative paths of low-priority levels are forced to avoid the spatiotemporal region defined by these constraints during the search process, thus eliminating the possibility of new spatiotemporal intersections and competition between replanned and unreplanned paths at the source. Compared to existing methods where the scheduling scheme needs repeated corrections due to the lack of consideration for the spatiotemporal coordination of vehicles after path replanning, the scheduling results of this invention are deterministic and consistent, and can be stably executed with a single scheduling operation, avoiding computational duplication and execution chaos caused by iterative adjustments.

[0017] Fourth, a complete closed loop is constructed, from data collection, conflict detection, hierarchical decision-making to instruction issuance. This invention issues access permits and alternative routes to the corresponding terminal devices, and each transportation task can execute subsequent travel according to the clearly issued instructions. A clear data flow and instruction transmission relationship is formed between each link in the entire scheduling chain, solving the problem of the disconnect between scheduling decision results and actual execution in existing methods. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a logistics cargo transportation tracking and scheduling method based on multi-source data fusion, according to an embodiment of the present invention. Detailed Implementation

[0019] To further illustrate the various embodiments, the present invention provides accompanying drawings. These drawings are part of the disclosure of the present invention, primarily used to illustrate the embodiments and to explain the operating principles of the embodiments in conjunction with the relevant descriptions in the specification. With reference to these drawings, those skilled in the art should be able to understand other possible implementations and the advantages of the present invention. Components in the drawings are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0020] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments.

[0021] Figure 1 This is a flowchart illustrating a logistics cargo transportation tracking and scheduling method based on multi-source data fusion according to an embodiment of the present invention. Figure 1 As shown, the method of this embodiment includes the following steps: Acquire multi-source data for logistics and cargo transportation. This multi-source data includes real-time location data, planned route data, timeliness attribute data, and road network dynamic event data for each of the multiple transportation tasks. Real-time location data is obtained through satellite positioning terminals installed on transport vehicles and reported at preset intervals, including the current location coordinates, instantaneous speed, and direction of travel for each task. Planned route data consists of the pre-planned routes for each transportation task at departure, stored as a sequence of road segments, each uniquely identified by a segment identifier. Timeliness attribute data includes the task type identifier and agreed delivery time limit for each transportation task; the task type identifier distinguishes different timeliness requirements for tasks such as cold chain transportation, pharmaceutical transportation, fresh produce transportation, emergency relief supplies transportation, and general cargo transportation. Road network dynamic event data includes real-time road condition data, weather warning data, and traffic control data, which can be obtained in real-time through public data interfaces with third-party traffic information service platforms or meteorological and traffic management departments.

[0022] The acquisition of the aforementioned multi-source data provides a data foundation for subsequent road occupancy prediction and conflict detection. Real-time location data reflects the current location and movement status of each task, planned route data reflects the future driving intentions of each task, timeliness attribute data provides the basis for differentiated scheduling, and road network dynamic event data provides the external conditions that trigger scheduling decisions.

[0023] Based on the acquisition of multi-source data, and according to the real-time positioning data and the planned route data, the time interval for each of the transportation tasks to occupy each road segment within the future time window is predicted.

[0024] The start time of the future time window is the current time, and the end time is the current time plus a first preset duration. The first preset duration can be flexibly set according to the application scenario, and its value should be sufficient to cover the time range from the current time to the completion of the last segment of the planned path for each transportation task.

[0025] The process of predicting the time interval is as follows: First, the location coordinates and speed of each transportation task at the current moment are extracted based on the real-time positioning data. Then, a historical travel time series for each transportation task is generated based on the positioning data sequence of each transportation task within a historical time period. The historical travel time series reflects the traffic efficiency characteristics of the transportation task in recent operations, including individual differences such as driving habits and vehicle performance.

[0026] Secondly, based on the planned route data, the sequence of road segments to be traversed by each transportation task within the future time window is determined. The sequence of road segments to be traversed refers to the set of road segments that the transportation task will pass through sequentially along the planned route, starting from its current position, and arranged in the order of travel.

[0027] Then, the historical travel time series, current location coordinates, current driving speed, and real-time traffic flow parameters of each segment in the sequence of road segments to be traversed are input into a pre-trained travel time prediction model. The travel time prediction model is a bidirectional long short-term memory network, which includes a forget gate, an input gate, an output gate, and a hidden state. Using the historical travel time series, the bidirectional long short-term memory network extracts the forward and backward temporal features of the transportation task in the historical travel history; based on the current location coordinates, current driving speed, and the real-time traffic flow parameters, the forward and backward temporal features are corrected, and the first predicted arrival time and first predicted departure time of each transportation task to each segment in the sequence of road segments to be traversed are output.

[0028] The first predicted arrival time and the first predicted departure time are defined as the occupancy time interval. This occupancy time interval represents the time span of the corresponding transportation task on the corresponding road segment from the entry time to the departure time, and is the basic data unit for subsequently constructing the road segment-task occupancy matrix. The bidirectional long short-term memory network can effectively capture long-term dependencies in the travel time series. Through its bidirectional structure, it simultaneously utilizes forward and backward information from historical sequences, combined with real-time traffic flow parameters for correction, enabling the prediction results to possess both temporal pattern learning capabilities and the dynamic changes of current road conditions.

[0029] A road segment-task occupancy matrix is ​​generated based on each of the specified occupancy time intervals.

[0030] The road segment-task occupancy matrix uses road segments in the road network as row indices and each transport task as a column indices. Matrix elements represent the occupancy time interval of the corresponding transport task on the corresponding road segment. In other words, each row of the matrix corresponds to a road segment, each column corresponds to a transport task, and the element value at the intersection of a row and a column represents the occupancy time interval of that task on that road segment. This matrix maps multi-source data into a three-dimensional space of road segment-task-time using a unified data structure. This allows for rapid querying and updating of the occupancy order, occupancy time distribution, and occupancy status of any task on any road segment through matrix operations, providing a computationally achievable data foundation for subsequent conflict detection and capacity allocation.

[0031] After generating the road segment-task occupancy matrix, in response to the road network dynamic event data indicating that a traffic capacity reduction event has occurred in the target road segment or target node, all matrix elements with the travel index of the target road segment or target node are filtered from the road segment-task occupancy matrix, and the transportation tasks corresponding to the filtered matrix elements are determined as the conflict task set.

[0032] The capacity reduction event refers to an event that causes a decrease in the number of vehicles that can pass through a target road segment or target node within a unit of time. In this embodiment, a capacity reduction event is determined by one or more of the following methods: when real-time traffic data indicates that the average speed of the target road segment is lower than a first preset percentage threshold of the historical average speed for the same period of the road segment, a capacity reduction event is determined to have occurred on the target road segment; when meteorological warning data indicates that the meteorological warning level in the area where the target road segment is located exceeds a preset level, a capacity reduction event is determined to have occurred on the target road segment; when traffic control data indicates that traffic control is implemented on the target road segment or target node, a capacity reduction event is determined to have occurred on the target road segment or target node. The above determination conditions cover the main event types that lead to capacity reduction, ensuring that the system can respond to various dynamic events in a timely manner. The capacity reduction event has a predicted duration, which can be provided by an external data source or estimated based on a historical statistical model according to the event type and severity.

[0033] When a capacity reduction event is triggered, all matrix elements corresponding to the target road segment or target node are filtered from the road segment-task occupancy matrix. Since the matrix uses road segments as row indices, all matrix elements of the target road segment represent the occupancy time intervals of all transportation tasks on that road segment. Through this filtering operation, all transportation tasks scheduled to pass through the target road segment or target node within the event's impact time window can be quickly located, forming a set of conflicting tasks. This enables accurate identification of affected vehicles and avoids unnecessary scheduling of unaffected vehicles.

[0034] Based on the timeliness attribute data of each transportation task in the conflict task set, the conflict task set is divided into high priority level and low priority level according to the preset priority classification rules.

[0035] The preset priority ranking rules are as follows: Obtain the timeliness attribute data for each of the aforementioned transportation tasks, wherein the timeliness attribute data includes the task type identifier and the agreed delivery time limit.

[0036] When the task type is identified as cold chain transportation, pharmaceutical transportation, fresh produce transportation, or emergency relief supplies transportation, the corresponding transportation task will be marked as first priority. These task types have strict time requirements. Cold chain transportation involves the preservation of temperature-controlled goods, pharmaceutical transportation involves drug safety, fresh produce transportation involves commodity spoilage, and emergency relief supplies transportation involves public interest. Delayed delivery will result in damage to goods, safety risks, or loss of social utility; therefore, they should receive the highest priority.

[0037] When the task type is identified as general cargo transportation and the time difference between the agreed delivery deadline and the current time is less than a first preset time threshold, the corresponding transportation task is marked as second priority. The agreed delivery deadline for this type of task is approaching; although its urgency is not as high as that of first-priority tasks, there is a risk of default due to timeout, and it needs to be prioritized.

[0038] When the task type is identified as general cargo transportation and the time difference between the agreed delivery time limit and the current time is greater than or equal to the first preset time threshold, the corresponding transportation task is marked as the third priority. This type of task has a relatively generous delivery time limit, can tolerate a certain degree of detour delays, and is suitable for bearing the scheduling adjustment costs incurred due to the passage of high-priority tasks.

[0039] The high priority level consists of transportation tasks of the first priority and the second priority, and the low priority level consists of transportation tasks of the third priority.

[0040] The number of transport tasks included in the high-priority tier is less than or equal to the remaining capacity of the target road segment or node after the capacity reduction event. This quantitative constraint ensures that the total traffic demand of high-priority tasks does not exceed the actual carrying capacity of the target road segment or node after the event, making subsequent capacity allocation physically feasible.

[0041] The remaining traffic capacity is calculated as follows: The historical number of vehicles passing through the target road segment or node per unit time before the occurrence of the capacity reduction event is obtained as the baseline capacity. The baseline capacity reflects the capacity baseline of the road segment or node under normal conditions.

[0042] Based on the number of lanes in the target road segment or node, the event type of the capacity reduction event, and the capacity reduction coefficient corresponding to the event type, the predicted number of vehicles passing per unit time after the event occurs is calculated as the event capacity. Different event types have different degrees of impact on capacity. For example, traffic accidents may lead to the closure of some lanes, severe weather may cause vehicles to slow down, and traffic control may lead to a complete closure. By setting a capacity reduction coefficient corresponding to the event type, the actual impact of various events on capacity can be quantitatively characterized.

[0043] The difference between the baseline traffic capacity and the event traffic capacity is determined as the traffic capacity loss of the target road segment or target node, and the product of the event traffic capacity and a preset safety redundancy coefficient is determined as the remaining traffic capacity. The introduction of the safety redundancy coefficient provides a buffer margin for uncertainties at the execution level, avoiding the failure of the scheduling scheme during execution due to the deviation between the theoretical capacity and the actual executable capacity.

[0044] For each of the transportation tasks in the high priority level, its planned path remains unchanged, and the remaining traffic capacity is allocated to each of the transportation tasks in the high priority level according to the time interval occupied by each of the transportation tasks.

[0045] High-priority transportation tasks retain their original planned routes, meaning these tasks do not require route replanning and their routes remain the pre-planned optimal paths. This approach has two advantages: First, high-priority tasks are often time-sensitive, and maintaining the original routes avoids extra travel time and uncertainty caused by detours, ensuring their efficiency. Second, since high-priority tasks do not participate in route replanning, the number of tasks requiring route recalculation is significantly reduced, effectively lowering computational costs.

[0046] The remaining traffic capacity is allocated according to the time interval occupied by each high-priority task, that is, the right of way is allocated in order of the time when each task plans to enter the target road segment. After the allocation is completed, each high-priority task obtains a definite passage permission on the target road segment or target node, and its passage time has been included in the post-event remaining capacity for scheduling, ensuring the orderly allocation of right of way.

[0047] The time interval and route segment occupied by each transportation task in the high priority level are determined as spatiotemporal obstacle constraints.

[0048] The physical meaning of spatiotemporal obstacle constraints is: within a specific time interval, a specific road segment is occupied by a high-priority task, and other tasks are not allowed to enter the road segment during that time interval, otherwise a spatiotemporal conflict will occur. Specifically, for each transportation task in the high-priority level, its occupied time interval on the target road segment and the target road segment together constitute a "road segment-time" tuple, and the set of tuples for all high-priority tasks is the spatiotemporal obstacle constraint.

[0049] The technical significance of defining the time intervals and road segments occupied by high-priority tasks as spatiotemporal obstacle constraints lies in the fact that high-priority tasks have already obtained passage permission and maintain their original paths; their passage behavior is deterministic and uninterruptible. Therefore, any alternative paths planned for low-priority tasks must not intersect with these determined spatiotemporal occupancy limits. Otherwise, it would cause low-priority and high-priority tasks to arrive at the same road segment at the same time, resulting in new conflicts. By setting the spatiotemporal occupancy of high-priority tasks as hard constraints, secondary conflicts between scheduling schemes are avoided from the outset.

[0050] Based on the spatiotemporal obstacle constraints, alternative routes are replanned for each of the transportation tasks in the low priority level.

[0051] The alternative path avoids the spatiotemporal region defined by the spatiotemporal obstacle constraint. Specifically, the process of replanning the alternative path is as follows: Starting from the current location of the target transportation task in the low-priority level and ending at its final destination, candidate paths are searched within the road network. The search space for candidate paths excludes paths encompassing the spatiotemporal region defined by the spatiotemporal obstacle constraints. In other words, any path that overlaps with the time interval occupied by the high-priority task in the temporal dimension and with the road segment occupied by the high-priority task in the spatial dimension is excluded from the search space. This exclusion operation ensures that all searched candidate paths do not intersect with the passage of the high-priority task in the spatiotemporal dimension.

[0052] The alternative path is selected from the candidate paths based on the path cost, which is weighted by the total travel distance, the total estimated travel time, and the real-time congestion index of the road segments traversed by the path. The weighting method involves multiplying the total travel distance, the total estimated travel time, and the real-time congestion index by their respective weighting coefficients and then summing the results. By introducing the real-time congestion index, the path cost reflects the real-time operating status of the current road network, avoiding the planning of short but inefficient paths for low-priority tasks.

[0053] The total estimated travel time of the proposed route shall not exceed a second preset percentage threshold of the total travel time of the original planned route for the transportation task. This constraint ensures that the increase in travel time of the alternative route relative to the original route is controlled within an acceptable range, preventing the timeliness requirements of the alternative route for low-priority tasks from being violated due to excessive detours. If no route satisfying all the above constraints is found, the system generates a prompt message and reports it to the dispatch center for manual intervention.

[0054] The passage permits for each transportation task in the high-priority level and the alternative paths for each transportation task in the low-priority level are respectively sent to the corresponding terminal devices.

[0055] Specifically, based on the terminal device identifiers of each transportation task in the high-priority tier, the passage permission for each transportation task is encapsulated into a first data packet and sent to the corresponding terminal device. The first data packet contains the permitted passage time interval for the transportation task on the target road segment or target node. Based on the terminal device identifiers of each transportation task in the low-priority tier, the alternative path for each transportation task is encapsulated into a second data packet and sent to the corresponding terminal device. The second data packet contains the identifiers of each road segment traversed by the alternative path and the predicted time sequence of arrival at each road segment. The first data packet and the second data packet are encrypted and transmitted using an encryption key pre-negotiated with the terminal device to ensure the security and integrity of the scheduling instructions during transmission and to prevent the instructions from being tampered with or leaked.

[0056] After receiving the data packet, each terminal device decrypts the encrypted data and renders the access permit or alternative route on the electronic map layer for driver navigation, or directly inputs the data into the path tracking control module of the autonomous vehicle, allowing the vehicle to automatically execute the dispatch instructions. This forms a complete closed loop from data collection, conflict detection, hierarchical decision-making to instruction issuance.

[0057] After the dispatching instructions are issued, new changes may occur during the development of dynamic events and the execution of transportation tasks. This embodiment also provides an update mechanism: After determining the time interval and road segment occupied by each transportation task in the high priority level as spatiotemporal obstacle constraints, the real-time location data of each transportation task in the high priority level is continuously acquired, and the actual driving speed of each transportation task is calculated based on the real-time location data.

[0058] When the absolute value of the deviation between the actual travel speed and the predicted travel speed of any high-priority transportation task exceeds a preset deviation threshold, it indicates that the actual operating state of the task has deviated from the assumptions used in the prediction, and the actual arrival and departure times of its subsequent road segments will differ from the predicted values. In this case, the occupancy time interval of its subsequent road segments is re-predicted based on the current real-time location data of the transportation task, resulting in an updated occupancy time interval. The corresponding occupancy time interval in the spatiotemporal obstacle constraint is then updated based on the updated occupancy time interval. Based on the updated spatiotemporal obstacle constraint, alternative routes are replanned for each of the affected low-priority transportation tasks.

[0059] Through the aforementioned dynamic update mechanism, the system can continuously perceive the actual execution status of high-priority tasks and adjust the spatiotemporal obstacle constraints and alternative paths for low-priority tasks in a timely manner when they deviate from the predicted values, ensuring that the scheduling scheme remains effective and feasible throughout the entire transportation execution process.

[0060] In the process of acquiring multi-source data, embodiments of the present invention also provide a handling mechanism for interruption of positioning signals.

[0061] Acquire multi-mode positioning data for each of the aforementioned transportation tasks. This multi-mode positioning data includes at least two of the following: satellite positioning data, RFID gantry read records, and AIS (Automatic Identification System) ship data. Multiple positioning methods serve as backups for each other; if one positioning method fails, the continuity of positioning data can be maintained through other methods.

[0062] When the satellite positioning data for any transportation task has a signal strength below a preset signal threshold within the target area, the RFID gantry reading record or the AIS data is used instead of the satellite positioning data as the positioning data source for that transportation task. For example, in areas with satellite signal obstruction, such as ports or tunnels, the system automatically switches to RFID gantry reading records for positioning. When neither the RFID gantry reading record nor the AIS data can be obtained, the trajectory of the transportation task is calculated based on the historical occupancy time interval in the road segment-task occupancy matrix to generate a complete trajectory for the transportation task during the signal interruption. The trajectory calculation can use an interpolation method based on historical data, estimating the position at each intermediate moment according to the known position and time information before and after the signal interruption and the historical travel speed.

[0063] The multi-mode positioning and trajectory completion mechanism described above avoids the loss of tracking data due to the interruption of a single signal source, ensures the integrity of multi-source data and the data foundation for scheduling decisions are not damaged, and keeps the occupancy time interval in the road segment-task occupancy matrix always available and accurate.

[0064] Although the invention has been specifically shown and described in conjunction with preferred embodiments, those skilled in the art should understand that various changes in form and detail may be made to the invention without departing from the spirit and scope of the invention as defined in the appended claims, all of which shall be within the scope of protection of the invention.

Claims

1. A logistics cargo transportation tracking and scheduling method based on multi-source data fusion, characterized in that, Includes the following steps: Acquire multi-source data on logistics and cargo transportation, including real-time location data, planned route data, timeliness attribute data, and road network dynamic event data for each of the multiple transportation tasks; Based on the real-time positioning data and the planned route data, the time interval for each of the transportation tasks to occupy each road segment within the future time window is predicted. A road segment-task occupancy matrix is ​​generated based on each of the occupancy time intervals. The road segment-task occupancy matrix uses each road segment in the road network as the row index and each of the transportation tasks as the column index. The matrix elements are the occupancy time intervals of the corresponding transportation tasks on the corresponding road segments. In response to the road network dynamic event data indicating that a traffic capacity reduction event has occurred in a target road segment or target node, all matrix elements with the travel index of the target road segment or target node are filtered from the road segment-task occupancy matrix, and the transportation tasks corresponding to the filtered matrix elements are determined as a set of conflict tasks. Based on the timeliness attribute data of each transportation task in the conflict task set, the conflict task set is divided into a high-priority level and a low-priority level according to a preset priority classification rule, wherein the number of transportation tasks included in the high-priority level is less than or equal to the remaining traffic capacity of the target road segment or target node after the traffic capacity reduction event. For each of the transportation tasks in the high priority level, its planned path remains unchanged, and the remaining traffic capacity is allocated to each of the transportation tasks in the high priority level according to the time interval occupied by each of the transportation tasks. The time interval and route segment occupied by each transportation task in the high priority level are defined as spatiotemporal obstacle constraints; Based on the spatiotemporal obstacle constraints, alternative routes are replanned for each of the transportation tasks in the low priority level, wherein the alternative routes avoid the spatiotemporal region defined by the spatiotemporal obstacle constraints. The passage permits for each transportation task in the high-priority level and the alternative paths for each transportation task in the low-priority level are respectively sent to the corresponding terminal devices.

2. The logistics cargo transportation tracking and scheduling method based on multi-source data fusion according to claim 1, characterized in that, The step of predicting the time interval for each transportation task to occupy each road segment within a future time window based on the real-time positioning data and the planned route data includes the following steps: Based on the real-time positioning data, the location coordinates and driving speed of each transportation task at the current moment are extracted, and based on the positioning data sequence of each transportation task in the historical time period, the historical travel time sequence of each transportation task is generated. Based on the planned route data, the sequence of road segments to be traversed by each of the transportation tasks within the future time window is determined; The historical travel time series, current location coordinates, current driving speed, and real-time traffic flow parameters of each road segment in the sequence of road segments to be traversed are input into the pre-trained travel time prediction model. The travel time prediction model is a bidirectional long short-term memory network. The bidirectional long short-term memory network extracts the forward and backward temporal features of the transportation task in the historical travel time series, and corrects the forward and backward temporal features based on the current location coordinates, current driving speed and the real-time traffic flow parameters, and outputs the first predicted arrival time and the first predicted departure time of each of the transportation tasks to each segment in the sequence of road segments to be traversed. The first predicted arrival time and the first predicted departure time are defined as the occupied time interval.

3. The logistics cargo transportation tracking and scheduling method based on multi-source data fusion according to claim 1, characterized in that, The preset priority ranking rules include: Obtain the timeliness attribute data for each of the aforementioned transportation tasks, wherein the timeliness attribute data includes task type identifier and agreed delivery time limit; When the task type is identified as cold chain transportation, medical transportation, fresh food transportation and / or emergency relief supplies transportation, the corresponding transportation task will be marked as the first priority. When the task type is identified as general cargo transportation and the time difference between the agreed delivery time limit and the current time is less than the first preset time threshold, the corresponding transportation task will be marked as the second priority. When the task type is identified as general cargo transportation and the time difference between the agreed delivery time limit and the current time is greater than or equal to the first preset time threshold, the corresponding transportation task will be marked as the third priority. The high priority level consists of transportation tasks of the first priority and the second priority, and the low priority level consists of transportation tasks of the third priority.

4. The logistics cargo transportation tracking and scheduling method based on multi-source data fusion according to claim 1, characterized in that, The road network dynamic event data includes real-time traffic data, weather warning data, and traffic control data; the capacity reduction event is determined through one or more of the following methods: When the real-time traffic data indicates that the average traffic speed of the target road segment is lower than a first preset percentage threshold of the historical average speed of the same period of the road segment, it is determined that a traffic capacity reduction event has occurred in the target road segment. When the meteorological warning data indicates that the meteorological warning level in the area where the target road segment is located exceeds the preset level, it is determined that a traffic capacity reduction event has occurred on the target road segment; When the traffic control data indicates that traffic control should be implemented on the target road segment or target node, it is determined that a traffic capacity reduction event has occurred on the target road segment or target node.

5. The logistics cargo transportation tracking and scheduling method based on multi-source data fusion according to claim 4, characterized in that, The method for calculating the remaining traffic capacity includes: The historical number of vehicles passing through the target road segment or target node per unit time before the occurrence of the capacity reduction event is obtained as the baseline capacity. Based on the number of lanes in the target road segment or target node, the event type of the capacity reduction event, and the capacity reduction factor corresponding to the event type, the predicted number of vehicles passing through per unit time after the event occurs is calculated as the event capacity. The difference between the baseline traffic capacity and the event traffic capacity is determined as the traffic capacity loss of the target road segment or target node; The remaining capacity is determined by multiplying the event throughput capacity by a preset security redundancy coefficient.

6. The logistics cargo transportation tracking and scheduling method based on multi-source data fusion according to claim 1, characterized in that, The step of replanning alternative routes for each of the transportation tasks in the low-priority level based on the spatiotemporal obstacle constraints includes: Starting from the current location of the target transportation task in the low priority level, and ending at the final destination of the transportation task, a candidate path is searched in the road network. The search space for the candidate paths excludes paths that include the spatiotemporal region defined by the spatiotemporal obstacle constraints. The alternative path is selected from the candidate paths, and the path cost is calculated by weighting the total travel distance, the total estimated travel time, and the real-time congestion index of the road segments traversed by the path. The total estimated travel time of the route shall not exceed the second preset percentage threshold of the total travel time of the original planned route for the transportation task.

7. The logistics cargo transportation tracking and scheduling method based on multi-source data fusion according to claim 1, characterized in that, After determining the time interval and route segment occupied by each transportation task in the high-priority level as spatiotemporal obstacle constraints, the method further includes: Continuously acquire real-time location data of each of the transportation tasks in the high priority level, and calculate the actual driving speed of each of the transportation tasks based on the real-time location data; When the absolute value of the deviation between the actual driving speed of any of the high-priority transportation tasks and the predicted driving speed of the task exceeds a preset deviation threshold, the occupancy time interval of its subsequent road segments is re-predicted based on the current real-time positioning data of the transportation task, and the updated occupancy time interval is obtained. Update the corresponding occupancy time interval in the spatiotemporal obstacle constraint according to the updated occupancy time interval; Based on the updated spatiotemporal obstacle constraints, alternative routes are replanned for each transportation task in the affected low-priority level.

8. The logistics cargo transportation tracking and scheduling method based on multi-source data fusion according to claim 1, characterized in that, The step of issuing the passage permits for each transportation task in the high-priority level and the alternative paths for each transportation task in the low-priority level to the corresponding terminal devices includes: Based on the terminal device identifier of each transportation task in the high priority level, the passage permission of each transportation task is encapsulated into a first data packet and sent to the corresponding terminal device. The first data packet contains the passage time interval of the transportation task on the target road segment or target node. Based on the terminal device identifier of each transportation task in the low priority level, the alternative path of each transportation task is encapsulated into a second data packet and sent to the corresponding terminal device. The second data packet contains the identifier of each route segment of the alternative path and the predicted time sequence of arrival at each route segment. The first data packet and the second data packet are transmitted using encryption keys pre-negotiated with the terminal device.

9. The logistics cargo transportation tracking and scheduling method based on multi-source data fusion according to claim 1, characterized in that, The multi-source data also includes: Acquire multi-mode positioning data for each of the aforementioned transportation tasks, wherein the multi-mode positioning data includes at least two of the following: satellite positioning data, RFID gantry reading records, and AIS Automatic Identification System (AIS) ship data; When the signal strength of the satellite positioning data for any transportation task is lower than a preset signal threshold in the target area, the RFID gantry reading record or the AIS data shall be used instead of the satellite positioning data as the positioning data source for the transportation task. When neither the RFID gantry reading record nor the AIS data can be obtained, the trajectory of the transportation task is calculated based on the historical occupancy time interval in the segment-task occupancy matrix to generate the complete trajectory of the transportation task during the signal interruption.

Images