An AI-based vehicle dispatching system

Abstract

The application relates to the technical field of vehicle intelligent scheduling, in particular to a vehicle scheduling system based on AI, which comprises a demand analysis module used to determine whether to include a current scheduling task into a scheduling queue based on demand urgency and task complexity; a queue optimization module used to determine whether to trigger replacement analysis based on cargo stacking interference degree, and to determine task combination based on time reference constraint degree and vehicle adaptability; a point location laying module used to determine a verification point setting mode based on path stability; and a point location analysis module used to determine a point location analysis mode based on node connection balance degree of a path topological network. The application effectively improves vehicle scheduling efficiency.

Description

Technical Field

[0001] This invention relates to the field of intelligent vehicle dispatching technology, and in particular to an AI-based vehicle dispatching system. Background Technology

[0002] With the acceleration of urbanization and the rapid development of the logistics industry, the demand for vehicle dispatching is growing and becoming increasingly complex. Traditional vehicle dispatching mainly relies on manual scheduling or simple rule matching. However, in actual operation, vehicle dispatching faces multiple constraints such as dynamic changes in demand, real-time fluctuations in traffic conditions, limited vehicle resources, and diverse task priorities. Traditional dispatching methods are difficult to achieve optimal allocation of vehicle resources in complex and ever-changing scenarios.

[0003] Existing vehicle dispatching technologies mainly adopt the following approaches: rule-based static dispatching routes, which allocate vehicles according to preset fixed rules. This approach ignores factors such as real-time traffic conditions and dynamic changes in orders, resulting in rigid dispatching schemes that cannot adapt to complex and ever-changing actual operating environments; single-cycle dispatching routes based on optimization algorithms, which use optimization methods such as genetic algorithms and particle swarm optimization to generate the optimal dispatching scheme at a certain moment. This type of scheme usually only performs a single optimization and lacks a mechanism for post-event evaluation and dynamic adjustment of dispatching effects; and dynamic dispatching routes based on real-time data, which collect real-time data such as vehicle location and traffic flow and adjust the dispatching scheme accordingly. However, existing schemes mostly focus on path optimization itself and do not fully consider the multi-level logic of dispatching decisions, lacking a systematic hierarchical decision-making architecture.

[0004] Chinese patent publication CN119831238A discloses an AI-optimized automotive market logistics and delivery scheduling system, including a data acquisition module, an order management module, a traffic flow analysis module, and a scheduling decision module. It generates scheduling schemes by combining real-time traffic prediction, order priority ranking, and route optimization algorithms. However, this scheme lacks a necessity analysis mechanism before scheduling decisions are made, and it also lacks an evaluation and feedback adjustment mechanism after the scheduling scheme is executed. It lacks adaptive adjustment capabilities when facing scenarios such as abnormal orders or sudden vehicle malfunctions. In summary, the existing technology has the following core technical problems: lack of necessity analysis for scheduling task requirements, resulting in indiscriminate scheduling optimization of all scheduling requests and wasted computational resources; lack of a hierarchical evaluation mechanism for vehicle and task compatibility, making it impossible to make differentiated matching decisions based on vehicle status and task characteristics; and lack of a feasibility verification and dynamic adjustment mechanism for scheduling schemes, preventing continuous optimization of scheduling effectiveness. Summary of the Invention

[0005] Therefore, the purpose of this invention is to overcome the technical deficiencies in the prior art, such as the lack of necessity analysis of task requirements, the lack of hierarchical evaluation of vehicle and task adaptability, and the lack of feedback and adjustment of scheduling effect in vehicle scheduling. It provides an AI-based vehicle scheduling system that realizes intelligent and adaptive management of the entire vehicle scheduling process through a multi-layered progressive logical judgment architecture.

[0006] To achieve the above objectives, the present invention provides an AI-based vehicle dispatching system, comprising: The requirements analysis module is used to obtain the requirements characteristic parameters of the tasks to be scheduled, and to determine whether to include the current task in the scheduling queue based on the urgency of the requirements and the complexity of the task. The queue optimization module is used to determine whether to trigger replacement analysis based on the cargo stacking interference, and to determine the task combination based on the time reference constraint and vehicle adaptability. The point deployment module is used to determine whether to set up verification points uniformly or based on node connectivity, based on path stability. The point analysis module is used to determine the point analysis method based on the node connectivity balance of the path topology network, which is either time window analysis or path depth analysis.

[0007] Furthermore, the demand characteristic parameters include: the urgency of the demand determined based on the difference between the task deadline and the current time; the task load determined based on the volume or weight of the item to be transported; and the task path complexity determined based on the road network topology complexity between the starting point and the destination.

[0008] Furthermore, when the cargo stacking interference exceeds the preset cargo stacking interference, the queue optimization module determines to trigger task replacement analysis.

[0009] Furthermore, the queue optimization module identifies transportation tasks with both time reference constraints and vehicle suitability within a preset range as a matching combination and prioritizes their allocation to the same transportation vehicle.

[0010] Furthermore, the path analysis module determines the path priority coefficient based on the path length and path congestion, and records the path with the highest path priority coefficient as the global optimal path; The path priority coefficient is negatively correlated with both path length and path congestion.

[0011] Furthermore, when the path stability is greater than the preset path stability, the point deployment module determines that the verification point setting method is to uniformly set static verification points. When the path stability is less than or equal to the preset path stability, the verification point setting method is determined to be dynamic verification point setting based on node connectivity. The path stability is determined based on the frequency and clustering of anomalies at each node on the globally optimal path.

[0012] Furthermore, when the point deployment module performs static verification point setting, it determines the polling frequency of the verification point based on the expected travel time of the global optimal path; The polling frequency is positively correlated with the expected passage time.

[0013] Furthermore, when the point deployment module performs dynamic verification point setting, it determines the verification optimization coefficient based on the node connectivity and the frequency of the optimal path, and selects a preset number of nodes as dynamic verification points in descending order of the verification optimization coefficient.

[0014] Furthermore, when performing basic time window analysis, the location analysis module also includes: The delay deviation determination unit is used to obtain the actual time when the vehicle actually arrives at the current verification point, read the preset time window corresponding to the verification point, and calculate the delay deviation between the actual time and the preset time window. The time window determination unit is used to accumulate the delay deviation for verification points where the delay deviation is greater than the preset deviation. The delay trend determination unit is used to determine whether the current verification point is a potentially failed road segment if the cumulative delay value shows an increasing trend along the driving direction.

[0015] Furthermore, when the location analysis module performs path depth analysis, the location analysis module also includes: The cumulative delay rate determination unit is used to determine the cumulative delay rate of the current verification point based on the current verification point delay deviation and the actual transportation time. The risk section determination unit is used to determine whether the path between the current verification point and the next verification point is a high-risk section based on the cumulative delay rate of the current verification point. The alternative road determination unit is used to determine alternative roads based on the transportation routes that start from the end node of the high-risk road segment and do not contain the high-risk road segment in the road network.

[0016] Compared with existing technologies, the beneficial effect of this invention lies in that the demand analysis module in this invention determines to include the current scheduling task in the scheduling queue when the urgency of the demand is greater than a preset urgency or the task complexity is greater than a preset task complexity. By identifying time-sensitive scheduling tasks through demand urgency and task complexity through task complexity, and using both as the admission criteria for the scheduling queue, this avoids the problems in existing technologies where a single admission condition leads to urgent tasks being blocked and delayed by inefficient tasks, or complex path tasks lacking sufficient planning time due to late inclusion, resulting in frequent detours or timeouts during execution. In other words, by using a rule that includes either time urgency or path complexity as the entry criterion, this invention accurately filters out high-value tasks that truly require scheduling optimization intervention and prioritizes the allocation of computing and transportation resources, while excluding routine tasks with low urgency and low complexity to reduce the ineffective load on the scheduling system. Therefore, while ensuring the response speed of high-urgency tasks and the execution reliability of high-complexity tasks, it significantly improves the utilization efficiency of scheduling resources.

[0017] Furthermore, this invention characterizes the degree of spatial conflict on the existing stacking layout after the insertion of a new task by comprehensively calculating the differences in the three-dimensional dimensions of the cargo (length, width, and height). This avoids the problem of neglecting dimensional adaptability caused by prior art that only plans loading based on cargo weight or volume. That is, cargo with similar volume but large differences in length, width, and height ratios will have an unstable center of gravity and mismatched pressure surfaces when stacked, requiring multiple repositioning adjustments during loading and unloading. By identifying high-conflict tasks through stacking interference and triggering task replacement analysis, the execution order of tasks is dynamically rearranged, so that cargo with similar dimensions is relatively concentrated or optimally stacked in layers in the transportation sequence, reducing the number of ineffective repositioning operations and the need for manual intervention during loading and unloading. Thus, while significantly improving the utilization rate of vehicle loading space and loading and unloading efficiency, it reduces the risk of transportation loss caused by unstable stacking, enhances the transportation scheduling system's ability to finely adapt to the geometric characteristics of cargo, and thereby improves the effectiveness of vehicle scheduling.

[0018] Furthermore, the point deployment module of this invention determines different verification point setting methods based on the comparison results between path stability and preset path stability. Static verification points are fixed verification points uniformly deployed at equal intervals along the globally optimal path, while dynamic verification points are verification points whose deployment positions are dynamically determined based on node connectivity. By quantifying the cumulative intensity of historical abnormal events at each node of the path through anomaly frequency and quantifying the spatial distribution concentration of abnormal events on the path through anomaly aggregation, the two are jointly used to characterize the overall operational stability of the path. This avoids the monitoring resource mismatch problem caused by the uniform verification point deployment strategy for all paths in the prior art. That is, deploying too many verification points on stable paths causes resource redundancy and increases computation and communication overhead, while uniform deployment on unstable paths leads to insufficient monitoring density in high-incidence areas and the omission of key abnormal signals. By adaptively switching the deployment method based on path stability, uniform static deployment is used to achieve full coverage basic monitoring with low complexity when the path is stable, while dynamic deployment based on node connectivity focuses verification point resources on high-incidence nodes and their strongly correlated neighbors when the path is unstable, thus tilting monitoring resources towards high-risk areas. Therefore, while ensuring the effective capture rate of abnormal states along the route, the cost of setting up and maintaining verification points is significantly reduced, the resource utilization efficiency and the pertinence of abnormal response of the monitoring system are improved, and the vehicle dispatching efficiency is improved. Attached Figure Description

[0019] Figure 1 This is a module connection diagram of an AI-based vehicle dispatching system according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating whether task replacement analysis is triggered based on cargo stacking interference in an embodiment of the present invention. Figure 3 This is a flowchart illustrating how to determine the setting method of verification points based on path stability, as described in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0021] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0022] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0023] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0024] Please see Figure 1 The diagram shown is a module connection diagram of an AI-based vehicle dispatching system according to an embodiment of the present invention. The present invention provides an AI-based vehicle dispatching system, comprising: The requirements analysis module is used to obtain the requirements characteristic parameters of the tasks to be scheduled, and to determine whether to include the current task in the scheduling queue based on the urgency of the requirements and the complexity of the task. The queue optimization module, which is connected to the demand analysis module, is used to determine whether to trigger replacement analysis based on the cargo stacking interference degree, and to determine the task combination based on the time reference constraint degree and vehicle adaptability degree. The point deployment module is connected to the demand analysis module and the queue optimization module respectively, and is used to determine the verification point setting method based on path stability, either by uniformly setting static verification points or by determining dynamic verification points based on node connectivity. The point analysis module, which is connected to the point deployment module, is used to determine the point analysis method based on the node connectivity balance of the path topology network, namely, basic time window analysis or path depth analysis.

[0025] Specifically, the demand characteristic parameters include: the urgency of the demand determined based on the difference between the task deadline and the current time, and the task complexity determined based on the number of paths between the starting point and the destination.

[0026] In this embodiment of the invention, the method for obtaining the urgency of the demand includes: obtaining the task time limit and the current time of the current scheduled task, calculating the difference between the task time limit and the current time, recording this difference as the remaining response time, and normalizing the reciprocal of the remaining response time as the demand urgency. It can be understood that the demand urgency is used to characterize the urgency of the current scheduled task in the time dimension. The higher the demand urgency, the shorter the remaining response time of the task, and the more likely it is to be prioritized in the scheduling queue and allocated transportation resources as soon as possible.

[0027] In this embodiment of the invention, the method for obtaining the task complexity includes: obtaining the road network topology between the starting point and the ending point of the current scheduling task, and recording the number of all optional paths between the starting point and the ending point as the task complexity. It can be understood that task complexity is used to determine the navigation and execution difficulty of the current scheduling task in the path dimension. The greater the task complexity, the more complex the transportation path of the task and the more uncertainties that may be encountered during the journey.

[0028] In this embodiment of the invention, the demand analysis module determines to include the currently scheduled task in the scheduling queue when the urgency of the demand is greater than a preset urgency or the task complexity is greater than a preset task complexity. It is understood that a high urgency indicates that the task is time-sensitive and needs to be scheduled first; a high task complexity indicates that the task path is complex and needs to be included in the scheduling plan in advance to avoid execution delays. By jointly determining urgency and task complexity, the system can distinguish between high-priority and low-priority tasks, preventing low-value tasks from consuming scheduling computing resources. It is understood that the greater the user's sensitivity to scheduling resource utilization efficiency, the higher the values ​​of the preset urgency and preset task complexity, so that the system only performs scheduling optimization on tasks with high urgency or complexity.

[0029] Specifically, the queue optimization module determines to trigger task replacement analysis when the cargo stacking interference exceeds the preset cargo stacking interference.

[0030] In this embodiment of the invention, the process of obtaining the cargo stacking interference degree includes: acquiring the length, width, and height dimensions of the cargo in the current task, and simultaneously acquiring the dimension data of cargo in existing tasks of the transport vehicle. For each existing task cargo, the dimensional differences between the current task cargo and existing task cargo in the three dimensions of length, width, and height are calculated. The normalized dimensional differences in each dimension are summed and then averaged to obtain the cargo stacking interference degree between the current task cargo and the corresponding existing task cargo. It can be understood that the smaller the dimensional difference, the higher the cargo stacking interference degree.

[0031] In this embodiment of the invention, the setting method for the preset cargo stacking interference degree includes: collecting historical scheduling samples containing different cargo stacking states, using different values ​​as cargo stacking interference degree thresholds to trigger replacement analysis tests, using the improvement in vehicle loading and unloading efficiency after replacement as the evaluation index, and selecting the average value of the cargo stacking interference degree thresholds corresponding to each simulation result that meets the user's needs as the preset cargo stacking interference degree. It can be understood that the cargo stacking interference degree is used to determine the degree of interference caused to the existing cargo stacking layout of the vehicle after the current task to be assigned is inserted into the existing task sequence of the transport vehicle. The larger the cargo stacking interference degree, the more significant the conflict between the current task and the existing task at the cargo stacking level, requiring task replacement analysis to optimize the task sequence. The larger the preset cargo stacking interference degree, the higher the system's tolerance for stacking interference, triggering replacement analysis only when the interference level is severe; the smaller the preset cargo stacking interference degree, the more sensitive the system is to stacking interference, triggering replacement analysis even when the interference level is low. Therefore, the greater the user's sensitivity to loading and unloading efficiency, the smaller the value of the preset cargo stacking interference degree.

[0032] In this embodiment of the invention, the task substitution analysis is an optimization process that adjusts the execution order of tasks in the current task sequence to reduce cargo stacking interference and minimize invalid cargo handling during loading and unloading. In this embodiment, the task substitution analysis generates candidate task sequences by exchanging adjacent tasks or inserting tasks, evaluates the cargo stacking interference of each candidate sequence, and selects the candidate sequence with the lowest interference as the optimized task sequence.

[0033] Specifically, the queue optimization module identifies transportation tasks with both time reference constraints and vehicle suitability within a preset range as a matching combination and prioritizes their allocation to the same transportation vehicle.

[0034] In this embodiment of the invention, the preset range refers to the acceptable value intervals corresponding to the time reference constraint degree and the vehicle adaptability degree, respectively. The upper and lower limits of this interval are pre-defined based on the statistical distribution of successfully combined samples in the historical task execution records, or can be flexibly set by the user according to the tightness of the current scheduling strategy. The preset range of the time reference constraint degree is used to determine whether the time windows of two transportation tasks have the compatibility condition of being executed continuously on the same vehicle without time conflicts. Its value standard is determined with reference to the distribution law of the time reference constraint degree of task pairs with no delays and smooth switching in historical working conditions. When the overlap of the task time windows is sufficiently low, both are considered to fall within the preset range. The preset range of the vehicle adaptability degree consists of two parts: basic adaptability degree and additional adaptability degree. The basic adaptability degree requires that the total load of the tasks to be combined does not exceed the rated load and cargo volume of the candidate vehicle. The additional adaptability degree requires that the special transportation requirements proposed by each task can be fully met by the vehicle's existing equipment configuration or qualification conditions, and accumulates matching scores based on the type and quantity of special requirements. This score must reach the qualification benchmark established based on historical scheduling effect feedback. Furthermore, the preset range is not fixed, but is dynamically adjusted according to the remaining time of the current system scheduling response and the overall vehicle load status: when the response time is tight, the acceptance range of the time reference constraint is appropriately tightened to prioritize matching highly compatible tasks and avoid timeouts caused by waiting; when there are enough idle vehicles in the system, the acceptance range of the two types of parameters is appropriately widened to expand the range of combinable tasks, thereby increasing the vehicle loading rate and reducing empty mileage.

[0035] In this embodiment of the invention, the process of obtaining the time reference constraint degree includes: the time reference constraint degree is the degree of matching between each transportation task in the time window dimension. For any two transportation tasks, the corresponding time reference constraint degree is the reciprocal of the overlap duration of the time windows of the two transportation tasks. The smaller the overlap duration of the time windows of two transportation tasks, that is, when the two transportation tasks are compatible, the higher the time reference constraint degree, indicating that the two tasks can be executed continuously on the same vehicle without time conflict. The higher the time reference constraint degree, the stronger the time feasibility of combining the two tasks for execution. Specifically, for a single transportation task, the estimated execution time and the latest completion time of the transportation task are used to determine the time window of the transportation task.

[0036] In this embodiment of the invention, the process of obtaining vehicle adaptability includes: obtaining the task load and special requirements of the task to be assigned, and obtaining the rated load and cargo volume of the candidate transport vehicles. A basic adaptability is determined based on the degree of matching between the task load and the vehicle's rated load and cargo volume; an additional adaptability is determined based on the degree of matching between the special requirements of the task and the vehicle's equipment configuration; and the sum of the basic adaptability and the additional adaptability is recorded as the vehicle adaptability between the task to be assigned and the candidate transport vehicles. The vehicle adaptability represents the degree of matching between each transport task and the transport vehicle in terms of load capacity and cargo volume. A higher vehicle adaptability indicates a better match between the task's requirements for the vehicle and the vehicle's actual capabilities, and a stronger feasibility for executing the task on that vehicle.

[0037] In this embodiment of the invention, the task load is the normalized sum of the total weight and total volume of the task to be transported. The basic fit of the vehicle is positively correlated with the matching degree of the task load, the vehicle's rated load capacity, and the vehicle's cargo volume. Specifically, when the total weight of the task to be transported is less than or equal to the vehicle's rated load capacity and the total volume is less than or equal to the vehicle's cargo volume, the basic fit is one, indicating that the vehicle meets the task requirements in terms of load-bearing capacity. When the total weight of the task is greater than the vehicle's rated load capacity or the total volume is greater than the vehicle's cargo volume, the basic fit is zero, indicating that the vehicle cannot carry the current task. It can be understood that the basic fit is used to screen candidate vehicles that meet the basic requirements of the task in terms of physical load-bearing capacity, ensuring that the task is not assigned to an overloaded or insufficiently spaced vehicle.

[0038] In this embodiment of the invention, the special requirements of the task include, but are not limited to, special placement requirements. The additional adaptability score of the vehicle is positively correlated with the matching degree between the special requirements of the task and the vehicle's equipment configuration. Specifically, for each special requirement of the task, it is determined whether the candidate vehicle possesses the corresponding equipment configuration or qualifications. If it does, the additional adaptability score is accumulated; otherwise, it is not. It can be understood that the additional adaptability score is used to further screen vehicles that match the special requirements of the task in terms of equipment configuration and qualifications, provided that the basic carrying capacity is met, to ensure the feasibility of the task execution under special transportation conditions.

[0039] In this embodiment of the invention, by considering both time-dimension compatibility and vehicle-dimension adaptability when combining tasks, the system can aggregate and allocate multiple tasks suitable for continuous execution on the same vehicle, reducing vehicle idling and task switching overhead, and improving vehicle loading rate and scheduling efficiency.

[0040] Specifically, the path analysis module determines the path priority coefficient based on the path length and path congestion, and records the path with the highest path priority coefficient as the global optimal path; The path priority coefficient is negatively correlated with both path length and path congestion.

[0041] In this embodiment of the invention, the path length is the total travel distance of the candidate path between the transportation starting point and the destination where the last transportation task is completed. The path congestion degree is the average travel speed of the candidate path in historical traffic. The higher the average travel speed, the lower the path congestion degree, i.e., the smoother the path. In this embodiment, the globally optimal path is the candidate path with the smallest sum of path length and path congestion degree after scaling. It can be understood that the globally optimal path comprehensively considers distance and time factors, achieving a balance between shorter path and smoother traffic.

[0042] Specifically, when the path stability is greater than the preset path stability, the point deployment module determines that the verification point setting method is to uniformly set static verification points. When the path stability is less than or equal to the preset path stability, the verification point setting method is determined to be dynamic verification point setting based on node connectivity. The path stability is determined based on the frequency and clustering of anomalies at each node on the globally optimal path.

[0043] It is easy to understand that each transport vehicle corresponds to a globally optimal path, and a topology graph is constructed. The nodes in the graph include, but are not limited to, path intersections and toll stations.

[0044] In this embodiment of the invention, the process of obtaining path stability includes: path stability is the sum of the normalized anomaly frequency and anomaly clustering degree of each node on the globally optimal path. The anomaly frequency is the sum of the number of times each node experiences abnormal events such as traffic congestion, accidents, and traffic control within a historical period; the anomaly clustering degree is the spatial concentration of abnormal events on the path.

[0045] The anomaly clustering degree is defined as the ratio of the smallest circular area that can cover the anomaly node to the target area, obtained from the nodes that are abnormal within a historical period.

[0046] In this embodiment of the invention, the process of setting the historical time period includes: the longer the historical time period, the more reference data is used to determine the path stability, thereby improving the accuracy of path stability determination. However, an excessively long historical time period leads to a large amount of computation, which in turn leads to a waste of resources. Therefore, users can set the historical time period according to the balance between resource consumption and determination accuracy.

[0047] In this embodiment of the invention, the preset path stability setting method includes: collecting historical scheduling samples containing different path states, conducting verification point setting mode switching tests with different values ​​as path stability thresholds, using the verification point's capture rate of abnormal path states as the evaluation index, and recording the average value of the path stability thresholds corresponding to each simulation result that meets the user's requirements as the preset path stability. It can be understood that the smaller the preset path stability, the more lenient the system's judgment of path instability, switching to dynamic verification point setting when the path state is still acceptable to improve the sensitivity to monitoring potential anomalies; the larger the preset path stability, the more lenient the system is in judging path instability, and the more sensitive it is to path anomaly monitoring, the more sensitive the system is to path anomaly monitoring, and the smaller the value of the preset path stability. Path stability is used to determine the state stability of the globally optimal path during historical operation. The larger the path stability, the lower the frequency of abnormal events at each node on the path, i.e., the smoother the overall path operation.

[0048] In this embodiment of the invention, the static verification points are fixed verification points evenly distributed at equal intervals along the globally optimal path, used to collect path traffic status data at a fixed frequency when the path status is stable. The dynamic verification points are verification points whose placement is dynamically determined based on node connectivity, used to focus on monitoring areas with high anomaly incidence when the path status is unstable. It can be understood that when the path stability is greater than a preset path stability, it indicates that the path is running smoothly, and the monitoring requirements can be met by uniformly setting static verification points, with low deployment complexity; when the path stability is less than or equal to the preset path stability, it indicates that there are areas with high anomaly incidence on the path, and verification points need to be dynamically set based on node connectivity to focus on monitoring abnormal areas and improve the effectiveness of anomaly monitoring.

[0049] Specifically, when the point deployment module performs static verification point setting, it determines the polling frequency of the verification point based on the expected travel time of the global optimal path. The polling frequency is positively correlated with the expected passage time.

[0050] In this embodiment of the invention, the process of obtaining the polling frequency includes: the polling frequency is the product of the reciprocal of the expected travel time and the adjustment coefficient, wherein the expected travel time is the estimated time required for a vehicle to travel from the starting point to the destination along the globally optimal path. The polling frequency of the verification point is the number of times each verification point collects path status data per unit time. The longer the expected travel time, the higher the polling frequency of the verification point, so as to maintain continuous monitoring coverage of the path status during long-term travel.

[0051] In this embodiment of the invention, the adjustment coefficient is set as follows: the adjustment coefficient is a scaling parameter used to map the reciprocal of the expected travel time to the actual polling frequency dimension. It is determined by statistical calibration based on the mapping relationship between the expected travel time and the optimal polling frequency of valid samples whose data collection timeliness meets the path status monitoring requirements in historical operating conditions. The polling frequency characterizes the number of times a verification point collects path status data per unit time. It is positively correlated with the expected travel time; that is, the longer the expected travel time, the longer the vehicle travels on the path, and the higher the probability of road condition changes during this period, requiring a higher polling frequency to maintain continuous tracking and coverage of the path status. However, there is a difference in dimensions and a proportional offset between the reciprocal of the expected travel time and the polling frequency required by the actual system. Therefore, the theoretical frequency needs to be scaled to an engineering-executable collection frequency range using the adjustment coefficient. The adjustment coefficient is essentially a comprehensive factor that balances the performance limits of the data acquisition equipment at the verification point, the capacity of the communication bandwidth, and the user's requirements for the timeliness of path status updates. It ensures that the polling frequency matches the monitoring continuity requirements of long-term traffic scenarios without exceeding the real-time response limits of the hardware and network. Therefore, it serves as the benchmark for dimensional matching and range constraints in calculating the polling frequency. It is understandable that if the adjustment coefficient is set too high, the polling frequency will be too high, exceeding the equipment's acquisition capacity and bandwidth capacity limits, leading to data accumulation or packet loss, a surge in system overhead, and a decrease in the proportion of effective data. If the adjustment coefficient is set too low, the polling frequency will be too low, resulting in sparse path status data updates. During long-term traffic, it will be impossible to capture sudden congestion or interruption events in a timely manner, weakening the monitoring and early warning effectiveness of the verification point. The adjustment coefficient can be dynamically adjusted according to equipment performance upgrades, changes in communication conditions, and feedback on actual data timeliness; this will not be elaborated further.

[0052] Specifically, when the point deployment module performs dynamic verification point setting, it determines the verification optimization coefficient based on the node connectivity and the frequency of the optimal path, and selects a preset number of nodes as dynamic verification points in descending order of the verification optimization coefficient.

[0053] In this embodiment of the invention, the node connectivity is the number of nodes directly connected to the current node in the path topology network, used to determine the connectivity density of the node in the road network. It is easy to understand that the higher the node connectivity, the denser the connections of the node in the road network, indicating that the node is a key passage node.

[0054] In this embodiment of the invention, the optimal path frequency is the ratio of the number of shortest paths passing through the node to the total number of shortest paths in the path topology network. The optimal path frequency is used to determine the traffic hub role of the node in the road network. The higher the optimal path frequency, the more significant the mediating role of the node in the road network, that is, the greater the impact of the node's abnormal state on global traffic.

[0055] In this embodiment of the invention, the process of obtaining the verification optimization coefficient includes: for a single node, the normalized sum of the node connectivity and the optimal path frequency of that node is recorded as the verification optimization coefficient of that node. It is understood that dynamic verification points are preferentially deployed on key nodes in the road network that are densely connected and have significant hub functions, in order to maximize the monitoring coverage effect of the verification points on path status. Furthermore, during the process of setting dynamic verification points based on the verification optimization coefficient, the minimum distance between two adjacent dynamic verification points is greater than a preset distance to ensure that the verification points are evenly distributed in space and to avoid excessive concentration of multiple verification points in the same area.

[0056] In this embodiment of the invention, the preset distance setting method includes: collecting historical scheduling samples containing different road network structures, conducting dynamic verification point deployment tests with different values ​​as distance thresholds, and recording the average of the distance thresholds corresponding to the simulation results where the coverage rate of the verification points to the path status and the number of deployments meet the user's requirements as the preset distance. It is understood that a smaller preset distance results in denser verification point deployment and more comprehensive monitoring coverage, but increases deployment costs; a larger preset distance results in sparser verification point deployment, reduces costs, but may miss key monitoring areas.

[0057] In this embodiment of the invention, the preset quantity is a threshold used to determine the final number of dynamic verification points to be deployed in the path topology network, selected in descending order of verification optimization coefficients. This preset quantity is determined based on the statistical average of the deployment quantities corresponding to several simulation results where the verification point deployment coverage meets user requirements in historical operating conditions. The number of dynamic verification points directly determines the balance between the comprehensiveness of monitoring coverage and verification cost. More deployments result in better coverage of key traffic nodes and hub nodes in the road network, leading to higher completeness of path status monitoring. However, an excessive number introduces redundant verification points, increasing system overhead for data acquisition, transmission, and processing. Furthermore, the marginal monitoring gain decreases significantly when the spatial overlap between adjacent verification points is too high. Conversely, too few deployments may miss some key nodes that significantly impact the overall path status, resulting in insufficient sensitivity of verification points to abnormal events. Therefore, the preset quantity needs to find the optimal balance between monitoring coverage and system resource consumption to ensure that dynamic verification points can maximize the path status monitoring effect with a compact deployment scale. Thus, it serves as the quantitative benchmark for the dynamic verification point selection operation. Understandably, if the preset number is set too high, the dense deployment of verification points will waste computing and storage resources, and the small distance between adjacent verification points may conflict with the preset distance constraints. If the preset number is set too low, the coverage of verification points will be insufficient, and blind spots in the status of key nodes may lead to inaccurate path planning decisions. The preset number can be dynamically adjusted according to changes in the road network size, the total number of nodes, and feedback from actual monitoring coverage, which will not be elaborated further here.

[0058] Specifically, when the location analysis module performs basic time window analysis, it also includes: The delay deviation determination unit is used to obtain the actual time when the vehicle actually arrives at the current verification point, read the preset time window corresponding to the verification point, and calculate the delay deviation between the actual time and the preset time window. The time window determination unit is used to accumulate the delay deviation for verification points where the delay deviation is greater than the preset deviation. The delay trend determination unit is used to determine whether the current verification point is a potentially failed road segment if the cumulative values ​​show an increasing trend along the driving direction.

[0059] In this embodiment of the invention, for a single node, the number of other nodes connected to that node is denoted as the node's connectivity; the node connectivity balance is the variance of the node connectivity of each node. The node connectivity balance is used to determine the degree of balance in connectivity between nodes in the path topology network. When the node connectivity balance is greater than a preset node connectivity balance, it indicates that the road network structure is relatively balanced and there are no obvious structural bottlenecks. At this time, a full-path basic time window analysis is performed, that is, based on the expected travel time of the path and the current progress of the vehicle, it assesses whether the vehicle can complete the entire journey within the task time limit. The basic time window analysis does not involve local subdivision of the path, but rather judges the sufficiency of the path's timeliness from an overall perspective.

[0060] In this embodiment of the invention, the preset time window is the planned arrival time of the vehicle at each verification point, determined by the expected travel time of the globally optimal path and the vehicle's departure time. Each verification point corresponds to a preset time window, which serves as a benchmark for measuring whether the vehicle travels as planned. It is understood that the setting of the preset time window provides a time benchmark for the subsequent calculation of delay deviation, enabling the system to quantitatively assess the degree of deviation between the vehicle's actual travel status and planned progress on each road segment.

[0061] In this embodiment of the invention, the method for obtaining the delay deviation degree includes: obtaining the actual time when the vehicle actually arrives at the current verification point, reading the planned arrival time in the preset time window corresponding to the verification point, and calculating the difference between the actual time and the planned arrival time as the delay deviation degree. A positive delay deviation degree indicates that the vehicle arrives later than the planned time, resulting in a delay; a negative delay deviation degree indicates that the vehicle arrives earlier than the planned time, and traffic is smooth. It can be understood that the delay deviation degree is used to quantify the degree of deviation between the actual traffic status of the vehicle at the current verification point and the planned progress, and is a basic quantitative indicator for judging whether an anomaly has occurred in the path.

[0062] In this embodiment of the invention, the setting method for the preset deviation degree includes: collecting historical scheduling samples containing different delay levels, conducting delay judgment tests with different values ​​as preset deviation degrees, and recording the average value of the preset deviation degrees corresponding to each simulation result whose delay judgment accuracy meets user requirements as the preset deviation degree. It can be understood that the larger the preset deviation degree, the higher the system's tolerance for delays, and it only identifies delays as valid delays when they are relatively severe; the smaller the preset deviation degree, the lower the system's tolerance for delays, and it triggers cumulative judgment even when delays are still minor. The greater the user's sensitivity to changes in path status, the smaller the value of the preset deviation degree, so that the system begins cumulative monitoring when delays are not yet significant.

[0063] In this embodiment of the invention, the method for obtaining the accumulated delay deviation includes: for verification points where the delay deviation is greater than a preset deviation, the delay deviation of each verification point is accumulated in chronological order to obtain the accumulated delay value. It can be understood that the delay deviation of a single verification point only reflects the instantaneous state of that point, while the accumulated delay value reflects the degree of delay accumulation of the vehicle across multiple consecutive verification points, and is the basic data for determining whether the delay exhibits a systematic trend.

[0064] In this embodiment of the invention, the delay trend determination unit determines that the cumulative delay value shows an increasing trend along the driving direction. Specifically, it obtains the cumulative delay value corresponding to each verification point along the driving direction. For a single verification point, it obtains the previous and next verification points in the time sequence. If the cumulative delay value of the verification point and the previous verification point in the time sequence is less than the cumulative delay value of the verification point and the next verification point in the time sequence, then it is determined that the cumulative delay value shows an increasing trend along the driving direction. That is, by comparing the magnitude of the cumulative delay values ​​between adjacent verification points, if the cumulative delay values ​​along the driving direction show a continuous increasing trend, then the road segment where the current verification point is located is determined to be a potentially failed road segment. It can be understood that if the cumulative delay value only shows a single jump at a certain verification point and then tends to stabilize, it may be an occasional congestion rather than a systemic failure. If the cumulative delay value continuously increases at multiple consecutive verification points, it indicates that the delay is continuously accumulating and aggravating, and there is a very high possibility that there is a structural traffic obstacle on the road segment ahead, requiring timely warning and initiation of a detour decision. It is understood that the verification points include both static verification points and dynamic verification points.

[0065] In this embodiment of the invention, the process of setting the preset node connectivity balance degree includes: the preset node connectivity balance degree is a threshold for judging whether the degree of balance of connectivity between nodes in the path topology network has reached the structural prerequisite for performing full-path basic time window analysis. It is determined based on the statistical distribution of node connectivity balance degree of effective samples that highly match the timeliness assessment results of basic time window analysis in historical conditions with the actual traffic results. When the node connectivity balance degree is greater than the preset node connectivity balance degree, it means that the road network structure is relatively balanced and there are no obvious structural bottleneck nodes. At this time, the path travel time is mainly dominated by global factors such as road segment length and traffic flow, rather than being constrained by the capacity bottleneck of a certain local node. Therefore, using full-path basic time window analysis, that is, evaluating the matching degree between the expected path travel time and the task time limit from an overall perspective, is sufficient to accurately judge the timeliness adequacy, without the need for local subdivision analysis of the path. Conversely, if there is a structural bottleneck in the road network, a more refined local analysis of the path is required. Therefore, the preset node connectivity balance degree is used as the criterion for selecting the time window analysis mode. Understandably, if the threshold is set too low, full-path basic analysis will be activated prematurely when there are still local structural imbalances in the road network, ignoring the nonlinear drag effect of bottleneck nodes on traffic efficiency, leading to an overly optimistic timeliness assessment. If the threshold is set too high, local subdivision analysis will be forced even when the road network is already sufficiently balanced, introducing unnecessary computational overhead without obtaining additional assessment gains. The preset node connectivity balance can be dynamically adjusted according to the dynamic evolution of the road network topology, the accumulation of historical traffic data, and actual assessment feedback, which will not be elaborated further here.

[0066] Specifically, when the location analysis module performs path depth analysis, the location analysis module further includes: The cumulative delay rate determination unit is used to determine the cumulative delay rate of the current verification point based on the current verification point delay deviation and the actual transportation time. The risk section determination unit is used to determine whether the path between the current verification point and the next verification point is a high-risk section based on the cumulative delay rate of the current verification point. The alternative road determination unit is used to determine alternative roads based on the transportation routes that start from the end node of the high-risk road segment and do not contain the high-risk road segment in the road network.

[0067] In this embodiment of the invention, the cumulative delay rate is the ratio of the current verification point delay deviation to the actual transportation time, where the actual transportation time is the difference between the planned arrival time and the mission start time.

[0068] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. An AI-based vehicle dispatching system, characterized in that, include: The requirements analysis module is used to obtain the requirements characteristic parameters of the tasks to be scheduled, and to determine whether to include the current task in the scheduling queue based on the urgency of the requirements and the complexity of the task. The queue optimization module is used to determine whether to trigger replacement analysis based on the cargo stacking interference, and to determine the task combination based on the time reference constraint and vehicle adaptability. The point deployment module is used to determine whether to set up verification points uniformly or based on node connectivity, based on path stability. The point analysis module is used to determine the point analysis method based on the node connectivity balance of the path topology network, which is either time window analysis or path depth analysis.

2. The AI-based vehicle dispatching system according to claim 1, characterized in that, The demand characteristic parameters include: the urgency of the demand determined by the difference between the task deadline and the current time; the task load determined by the volume or weight of the items to be transported; and the task path complexity determined by the road network topology complexity between the starting point and the destination.

3. The AI-based vehicle dispatching system according to claim 1, characterized in that, When the cargo stacking interference exceeds the preset cargo stacking interference, the queue optimization module determines to trigger task replacement analysis.

4. The AI-based vehicle dispatching system according to claim 3, characterized in that, The queue optimization module identifies transportation tasks with both time reference constraints and vehicle suitability within a preset range as a matching combination and prioritizes their allocation to the same transportation vehicle.

5. The AI-based vehicle dispatching system according to claim 4, characterized in that, The path analysis module determines the path priority coefficient based on the path length and path congestion, and records the path with the highest path priority coefficient as the global optimal path. The path priority coefficient is negatively correlated with both path length and path congestion.

6. The AI-based vehicle dispatching system according to claim 5, characterized in that, When the path stability is greater than the preset path stability, the point deployment module determines that the verification point setting method is to uniformly set static verification points. When the path stability is less than or equal to the preset path stability, the verification point setting method is determined to be dynamic verification point setting based on node connectivity. The path stability is determined based on the frequency and clustering of anomalies at each node on the globally optimal path.

7. The AI-based vehicle dispatching system according to claim 6, characterized in that, When the point deployment module performs static verification point setting, it determines the polling frequency of the verification point based on the expected travel time of the global optimal path; The polling frequency is positively correlated with the expected passage time.

8. The AI-based vehicle dispatching system according to claim 6, characterized in that, When the point deployment module performs dynamic verification point setting, it determines the verification optimization coefficient based on the node connectivity and the frequency of the optimal path, and selects a preset number of nodes as dynamic verification points in descending order of the verification optimization coefficient.

9. The AI-based vehicle dispatching system according to claim 8, characterized in that, In response to the first preset condition, when the location analysis module performs basic time window analysis, the location analysis module also includes: The delay deviation determination unit is used to obtain the actual time when the vehicle actually arrives at the current verification point, read the preset time window corresponding to the verification point, and calculate the delay deviation between the actual time and the preset time window. The time window determination unit is used to accumulate the delay deviation for verification points where the delay deviation is greater than the preset deviation. The delay trend determination unit is used to determine whether the current verification point is a potentially failed road segment if the cumulative delay value shows an increasing trend along the driving direction.

10. The AI-based vehicle dispatching system according to claim 9, characterized in that, When the location analysis module performs path depth analysis, the location analysis module further includes: The cumulative delay rate determination unit is used to determine the cumulative delay rate of the current verification point based on the current verification point delay deviation and the actual transportation time. The risk section determination unit is used to determine whether the path between the current verification point and the next verification point is a high-risk section based on the cumulative delay rate of the current verification point. The alternative road determination unit is used to determine alternative roads based on the transportation routes that start from the end node of the high-risk road segment and do not contain the high-risk road segment in the road network.

Images