Intelligent transport vehicle dispatching method and system

Abstract

The application discloses an intelligent transport vehicle scheduling method and system, and relates to the technical field of transport vehicle scheduling data processing. The intelligent transport vehicle scheduling method and system comprises the following steps: S1, collecting transport vehicle scheduling data in a task execution process, and performing denoising, completion, alignment and normalization processing on the transport vehicle scheduling data; S2, real-time monitoring of task state changes, identification of the single-insertion capability of a vehicle after task cancellation, and screening of candidate vehicles entering a single-insertion scheduling process according to a single-insertion threshold; S3, matching of a single-insertion task for the candidate vehicle, analysis of single-insertion task matching effects, and completion of task path reconstruction and single-insertion task marking; S4, execution of the single-insertion task and recording of a state, analysis of the deviation between single-insertion task scoring and actual execution, proposal of scoring structure updating suggestions, completion of feedback archiving and recovery of a vehicle scheduling state. The problem that waste is caused by the fact that an original path and a time window cannot be quickly recycled and reused when a task is cancelled is solved.

Description

Technical Field

[0001] This invention relates to the field of transportation vehicle scheduling data processing technology, specifically to an intelligent transportation vehicle scheduling method and system. Background Technology

[0002] Existing transportation vehicle dispatching systems mainly rely on preset task paths and static task allocation mechanisms. They typically adopt a centralized dispatching strategy, which assigns tasks to specific vehicles through initial planning and continuously tracks their status during task execution. This approach is suitable for routine transportation scenarios with large volumes and low variability.

[0003] For example, the invention with publication number CN118278844A discloses an intelligent dispatching method and system for freight transportation. It includes the following steps: A) Creating an order or entering upstream customer entrustment information through a customer service terminal and generating a pre-scheduled plan; B) Calculating and analyzing the route through the route analysis and processing unit of the dispatch center and matching vehicle-related information based on the capacity in the dispatch center's capacity pool; C) The order response system assigns the target vehicle. If the target vehicle accepts the assignment, the information is fed back to the customer service terminal; if the target vehicle refuses the assignment, matching with other vehicles continues.

[0004] For example, the invention with publication number CN115409308A relates to the field of transportation management, specifically to an intelligent dispatching system based on TMS transportation management. Its transportation operation management module includes a vehicle allocation section, which establishes an intelligent dispatching algorithm. This algorithm establishes a multi-order aggregation module based on threshold range search to aggregate orders that are close to each other in the same area into regional orders for unified delivery. It also establishes truck matching rules based on matching factors, assigning different weights to these factors, and assigning the corresponding regional order to the truck with the highest total weight.

[0005] However, in actual operation, due to various dynamic factors such as sudden changes in orders, fluctuations in traffic conditions, customer cancellations, and abnormal vehicle status, task cancellations or interruptions occur frequently. The existing system's resource response mechanism after task cancellation is relatively crude, often resulting in problems such as paths and time windows not being released in a timely manner, short-term vehicle stagnation, redundant resource allocation, and delayed response to inserted tasks, leading to decreased scheduling efficiency and increased energy waste.

[0006] Therefore, in order to address the above problems, there is an urgent need for an intelligent transportation vehicle scheduling method and system. Summary of the Invention

[0007] Technical problems to be solved

[0008] To address the shortcomings of existing technologies, this invention provides an intelligent transportation vehicle scheduling method and system, which solves the problem of waste caused by the failure to quickly reclaim and reuse the original path and time window when a task is canceled.

[0009] To achieve the above objectives, the present invention provides the following technical solution: an intelligent transport vehicle scheduling method and system, comprising the following steps: S1, collecting transport vehicle scheduling data during task execution, and performing noise reduction, completion, alignment, and normalization processing on the transport vehicle scheduling data; S2, monitoring task status changes in real time, identifying the vehicle's ability to insert orders after task cancellation, and filtering candidate vehicles to enter the order insertion scheduling process based on the order insertion threshold; S3, matching candidate vehicles with tasks to be inserted based on path, time, and load conditions, analyzing the matching effect of the order insertion tasks, and completing task path reconstruction and order insertion task marking; S4, executing the order insertion task and recording the status, analyzing the deviation between the order insertion task score and the actual execution, proposing a score structure update suggestion, completing feedback archiving, and restoring the vehicle scheduling status.

[0010] Further, the specific steps for collecting transportation vehicle scheduling data during task execution are as follows: Transportation vehicle scheduling data is collected in real time through the integrated geographic positioning unit, path inertial measurement module, energy consumption sensor, and vehicle execution status feedback unit. This data includes vehicle task status, scheduling distance, vehicle speed, vehicle speed change rate, scheduling energy consumption, remaining transportation fuel, remaining scheduling time, and empty capacity ratio. The task lifecycle management module in the scheduling system obtains the vehicle task status in real time, marking canceled tasks as 1 and tasks currently in progress as 0. The high-precision satellite positioning system integrated into the transportation vehicle control unit obtains the original task starting point's latitude and longitude coordinates, the vehicle's current latitude and longitude coordinates, and the starting point's latitude and longitude coordinates of the task to be inserted, thus obtaining the original scheduling distance and the distance to be inserted. The vehicle speed is obtained by a speed sensor installed in the vehicle's main control board; the vehicle's linear acceleration and angular velocity are continuously measured by an inertial measurement module installed in the main control board, and the rate of change of speed over a continuous period is calculated to obtain the vehicle's speed change rate; the energy release rate of the vehicle per unit distance is measured by current and voltage sensors deployed in the battery management system, and the scheduling energy consumption is obtained by combining the drive motor load parameters; the remaining fuel volume is obtained by collecting the current percentage of remaining fuel through the onboard fuel management system, and by combining the remaining task distance in the vehicle's task execution list with historical unit fuel consumption data; the remaining scheduling time is obtained by reading the current time through the main control task unit and comparing it with the earliest execution time of the next task in the task scheduling table; and the empty capacity ratio is obtained by retrieving the ratio between the vehicle's current loading information and its maximum loading capacity.

[0011] Further, the specific steps for denoising, completing, aligning, and normalizing the transportation vehicle dispatching data are as follows: A sliding median filter combined with the interquartile range method is used to identify and remove local outliers in data such as dispatching distance, vehicle speed, and energy consumption; bidirectional timestamp interpolation and local trend extrapolation are used to complete short-term missing data in transportation vehicle dispatching data caused by data delays and interruptions; by aligning vehicle numbers with positioning times, the time of multi-source data such as dispatching distance, speed, fuel level, and task status under the same dispatching trajectory is unified, achieving data synchronization; exponential weighted average and sliding multinomial regression methods are used to denoise the transportation vehicle dispatching data, preserving trends while reducing fluctuations; and a combination of Z-score standardization and logarithmic scaling is used to normalize the transportation vehicle dispatching data, unifying the units and compressing the numerical range.

[0012] Furthermore, the specific steps for real-time monitoring of task status changes, identifying vehicle order insertion capabilities after task cancellation, and filtering candidate vehicles for entry into the order insertion scheduling process based on the order insertion threshold are as follows: Real-time monitoring and detection of the status of all executing tasks; when the system identifies a vehicle's task status changing from 0 to 1, immediately mark the vehicle as having a scheduling interruption, simultaneously suspend the execution of the vehicle's original scheduling path, and deregister the corresponding scheduling record in the system task queue; after removing the vehicle from the current scheduling queue, mark its status as pending order insertion and write it to the scheduling status buffer; simultaneously, the vehicle is stripped of scheduling master control in the task control center and transferred to the order insertion scheduling module, the original path segment resources are released and removed from the path allocation table; query the current transport vehicle scheduling data, reading vehicle speed, remaining transport fuel, scheduling energy consumption, remaining scheduling time, original scheduling distance, empty capacity ratio, and vehicle speed. The change rate is calculated as follows: The remaining transport fuel is divided by the dispatch energy consumption, and the vehicle speed is divided by 1 plus the vehicle speed change rate. The product of these two ratios is used as the base, with the capacity weight as the exponent. This power function is calculated as the first part. The remaining dispatch time is divided by the sum of the dispatch distance and the remaining dispatch time. The ratio is used as the base, with the time weight as the exponent. This power function is calculated as the second part. The empty capacity ratio is multiplied by the empty weight, and then 1 is added as the third part. These three parts are multiplied sequentially to obtain the order insertion capability assessment value. The obtained order insertion capability assessment value is normalized and compared with the order insertion threshold: when the order insertion capability assessment value is greater than or equal to the order insertion threshold, the vehicle is pushed into the order insertion candidate queue and a new dispatch entry is assigned to it; when the order insertion capability assessment value is less than the order insertion threshold, the vehicle is pushed into the standby buffer queue and does not participate in this round of order insertion matching.

[0013] Further, the specific steps for matching candidate vehicles with tasks to be inserted based on path, time, and load conditions are as follows: Read the current latitude and longitude coordinates, remaining scheduling time, insertion capability assessment value, current task sequence, and scheduling status of the vehicles to be inserted sequentially from the insertion candidate queue, and initialize a path attempt status flag for each vehicle; query all unassigned tasks in the task pool, extract the estimated execution time of the task to be inserted, the required load percentage of the task to be inserted, and the insertion scheduling distance, and load them into a set of optional tasks; for each candidate vehicle, traverse each optional task, obtain the insertion scheduling distance, and compare the insertion scheduling distance with the set maximum insertion distance; if it exceeds the limit, skip the current task; for tasks with path distances within acceptable ranges, further determine whether the start and end ranges of the task time window intersect with the vehicle's remaining scheduling time; if there is no intersection, skip the task; then determine whether the required load percentage of the task to be inserted is within the vehicle's current empty capacity ratio; if it exceeds the limit, skip the task; for all tasks that meet the path distance, time window overlap, and load constraints, generate a task insertion attempt record.

[0014] Further, the specific steps for analyzing the order insertion task matching effect and completing task path reconstruction and order insertion task marking are as follows: Obtain the scheduling distance to be inserted, the maximum order insertion distance, the estimated execution time of the task to be inserted, the load ratio required by the task to be inserted, the remaining scheduling time, the idle capacity ratio, and the order insertion capability evaluation value. Calculate the order insertion task matching evaluation value for the task to be inserted: Divide the scheduling distance to be inserted by the maximum order insertion distance, and subtract this ratio from 1. Multiply this difference by the matching weight and use the product as the first part; Divide the estimated execution time of the task to be inserted by the remaining scheduling time and add a minimum term; Subtract the load ratio required by the task to be inserted from the vehicle idle capacity ratio, and then divide by the idle capacity ratio and add a minimum term. Combine the above two ratios... The first part is the product of the value multiplied by the insertion capability assessment value and 1 minus the matching weight. The product is then added to the second part to obtain the insertion task matching assessment value. The calculated insertion task matching assessment values ​​are sorted in descending order, and the task with the highest insertion task matching assessment value is extracted as the current vehicle path reconstruction attempt object, generating a new task sequence structure. The reconstructed path is checked for scheduling legality, including path distance, time window overlap, load constraints, etc. If the check passes, the insertion task is marked as a proposed execution state, written into the vehicle task structure, and the path is numbered. If the check fails, the vehicle will not participate in insertion in this round, is marked as standby and unassigned, and waits for the next cycle of scheduling.

[0015] Further, the specific steps for executing the insertion task and recording its status are as follows: Extract the insertion task marked as to be executed, and synchronously write the task's path structure, the scheduling distance to be inserted, the estimated execution time of the insertion task, the load ratio required by the insertion task, and the task number into the task execution list of the corresponding vehicle, and generate an updated scheduling chain based on the vehicle's current task sequence; send the new task chain after the insertion task is inserted to the vehicle scheduling unit, update the execution mark of the insertion task, and record the structural changes between the current insertion position and the original scheduling path; the vehicle starts executing the insertion task, and the scheduling module receives the task execution status in real time; after the insertion task is completed, read the actual execution result of the insertion task. If the task is completed on time and not canceled, it is recorded as 1; otherwise, it is recorded as 0.

[0016] Further, the specific steps for analyzing the deviation between the order insertion task score and the actual execution are as follows: Calculate the average order insertion capability assessment value of the current order insertion batch. For each order insertion task, obtain the actual execution result of the order insertion task, the order insertion task matching assessment value, the dispatch distance to be inserted, the original dispatch distance, the estimated execution time of the order insertion task, the load ratio required by the order insertion task, the remaining dispatch time, the empty capacity ratio, and the order insertion capability assessment value. Calculate the order insertion task residual assessment value: Square the difference between the order insertion task matching assessment value and the actual execution result of the order insertion task as the first part; divide the estimated execution time of the order insertion task by the remaining dispatch time and add a minimum term; subtract the load ratio required by the order insertion task from the vehicle empty capacity ratio and then divide by the empty capacity ratio and add a minimum term; divide the original dispatch distance by the dispatch distance to be inserted; divide the order insertion capability assessment value by the absolute value of the order insertion capability assessment value minus the average order insertion capability assessment value of the current order insertion batch and add a minimum term; multiply the above four ratios, and the product is the second part; multiply the first part and the second part to obtain the order insertion task residual assessment value.

[0017] Furthermore, the following specific steps are proposed to update the scoring structure, complete feedback archiving, and restore the vehicle dispatch status: The residual evaluation value of the inserted task is used as the basis for adjusting the dispatch strategy. The residual evaluation value of the inserted task is compared with the task matching threshold in real time. If the residual evaluation value of an inserted task is greater than or equal to the task matching threshold, it is determined to be a strategy deviation sample. All parameters of this sample, including the inserted task matching evaluation value, inserted capability evaluation value, remaining dispatch time, original dispatch distance, waiting dispatch distance, empty capacity ratio, and the load ratio required by the waiting task, are written into the strategy deviation sample table and marked as a high residual sample. Such samples will be analyzed in detail in the subsequent scoring function structure optimization. If the residual evaluation value of an inserted task is less than the task matching threshold, it is determined to be a scoring stable sample. This sample will still record the residual score and execution parameters to maintain the integrity of the sample data, but it will not enter the deviation sample buffer. This data serves only as a reference for subsequent stability assessments of the scoring function and does not participate in weight adjustments. After all insertion tasks in this round are completed, the scheduling module summarizes the strategy deviation samples and scoring stable samples to form an insertion sample set. Statistical analysis is performed on high residual samples to identify variable combinations that occur frequently in the high residual samples and mark them as abnormal parameter combinations. Based on the abnormal parameter combinations, structural adjustment suggestions are proposed for the relevant variable terms in the insertion task matching evaluation value formula, generating an updated draft scoring function as a candidate scoring structure for the next cycle. The draft scoring function, along with the current cycle's insertion task residual evaluation value distribution and deviation sample ratio, is written into the scoring function version control table to generate a version number for use in the next cycle's scheduling decision-making. After all insertion task feedback records are completed, the vehicle scheduling status is updated, the insertion control is released, the vehicle is rewritten into the main scheduling queue, the system enters the next scheduling cycle, and the insertion process is completed in a closed loop.

[0018] The second aspect of this invention provides an intelligent transport vehicle scheduling system, comprising: a transport vehicle scheduling data acquisition and preprocessing module, a task cancellation detection and resource status modeling module, an order insertion task screening and path reconstruction module, and an order insertion task execution and feedback correction module. The transport vehicle scheduling data acquisition and preprocessing module is used to collect transport vehicle scheduling data during task execution and to perform denoising, completion, alignment, and normalization processing on the transport vehicle scheduling data. The task cancellation detection and order insertion vehicle screening module is used to monitor task status changes in real time, identify the order insertion capability of vehicles after task cancellation, and screen candidate vehicles to enter the order insertion scheduling process based on the order insertion threshold. The order insertion task screening and path reconstruction module is used to match candidate vehicles with tasks to be inserted based on path, time, and load conditions, analyze the matching effect of the order insertion tasks, and complete task path reconstruction and order insertion task marking. The order insertion task execution and feedback correction module is used to execute the order insertion task and record the status, analyze the deviation between the order insertion task score and the actual execution, propose a score structure update suggestion, complete feedback archiving, and restore the vehicle scheduling status.

[0019] Beneficial effects

[0020] The present invention has the following beneficial effects:

[0021] (1) This intelligent transport vehicle scheduling method and system constructs an order insertion capability assessment model, taking remaining transport fuel, scheduling energy consumption, vehicle speed, speed change rate, remaining scheduling time, and empty capacity ratio as input variables, and uses an exponential power function structure to output an order insertion capability assessment value, which is used to measure the possibility of a vehicle undertaking a new task. This assessment mechanism is based on the actual scheduling status for dynamic calculation, making the selection of order insertion candidates more accurate and efficient.

[0022] (2) This intelligent transport vehicle scheduling method and system designs an order insertion task matching evaluation function, which combines the scheduling distance to be inserted, the expected execution time of the task to be inserted, the ratio of the load required by the task to be inserted to the empty capacity ratio, the remaining scheduling time and the order insertion capability evaluation value to score all insertable tasks, and prioritizes them. This scoring structure unifies and quantifies spatial adaptation, time coordination and resource carrying capacity, improving the rationality of order insertion path reconstruction and scheduling flexibility.

[0023] (3) This intelligent transport vehicle scheduling method and system constructs a residual scoring expression for insertion tasks, combining the deviation between the predicted score and the actual execution result with the proportion of task execution time, load proportion, path distance proportion, and the degree of deviation of insertion capability to form a residual index that can be used to correct the scoring structure. This mechanism can identify key samples of scoring inaccuracies and generate suggestions for optimizing the scoring function structure, thereby realizing the continuous evolution of the scoring model and strategy adjustment.

[0024] (4) This intelligent transport vehicle scheduling method and system monitors the task status field of transport vehicles in real time. When the task is detected to have changed from being executed to canceled, the execution of the original scheduling path is immediately stopped, the corresponding path segment and time window resources are released, and the vehicle status is switched to the pending order status, thus completing the transfer of scheduling control. This mechanism effectively ensures the cleanup of resources and the connection of vehicle scheduling status after task interruption, providing basic resource space for subsequent order matching.

[0025] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0026] Figure 1 Here is a flowchart of an intelligent transportation vehicle scheduling method;

[0027] Figure 2 This is a structural diagram of an intelligent transport vehicle dispatching system;

[0028] Figure 3 Radar chart driven by the evaluation value of order insertion capability;

[0029] Figure 4 A bar chart showing the evaluation values ​​for the inserted task. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] Please see Figures 1-4 This invention provides a technical solution: an intelligent transport vehicle scheduling method and system, comprising the following steps: S1, collecting transport vehicle scheduling data during task execution, and performing noise reduction, completion, alignment, and normalization processing on the transport vehicle scheduling data; S2, monitoring task status changes in real time, identifying the vehicle's ability to insert orders after task cancellation, and filtering candidate vehicles to enter the order insertion scheduling process based on the order insertion threshold; S3, matching candidate vehicles with tasks to be inserted based on path, time, and load conditions, analyzing the matching effect of the order insertion tasks, and completing task path reconstruction and order insertion task marking; S4, executing the order insertion task and recording the status, analyzing the deviation between the order insertion task score and the actual execution, proposing a score structure update suggestion, completing feedback archiving, and restoring the vehicle scheduling status.

[0032] Specifically, the steps for collecting transportation vehicle scheduling data during task execution are as follows: Transportation vehicle scheduling data is collected in real time through the integrated geographic positioning unit, path inertial measurement module, energy consumption sensor, and vehicle execution status feedback unit. This data includes vehicle task status, scheduling distance, vehicle speed, vehicle speed change rate, scheduling energy consumption, remaining transportation fuel, remaining scheduling time, and empty capacity ratio. The task lifecycle management module in the scheduling system obtains the vehicle task status in real time, marking canceled tasks as 1 and tasks currently in progress as 0. The high-precision satellite positioning system integrated into the transportation vehicle control unit obtains the original task starting point's latitude and longitude coordinates, the vehicle's current latitude and longitude coordinates, and the starting point's latitude and longitude coordinates of the task to be inserted, thus obtaining the original scheduling distance and the distance to be inserted. The vehicle speed is obtained by a speed sensor installed in the vehicle's main control board; the vehicle's linear acceleration and angular velocity are continuously measured by an inertial measurement module installed in the main control board, and the rate of change of speed over a continuous period is calculated to obtain the vehicle's speed change rate; the energy release rate of the vehicle per unit distance is measured by current and voltage sensors deployed in the battery management system, and the scheduling energy consumption is obtained by combining the drive motor load parameters; the remaining fuel volume is obtained by collecting the current percentage of remaining fuel through the onboard fuel management system, and by combining the remaining task distance in the vehicle's task execution list with historical unit fuel consumption data; the remaining scheduling time is obtained by reading the current time through the main control task unit and comparing it with the earliest execution time of the next task in the task scheduling table; and the empty capacity ratio is obtained by retrieving the ratio between the vehicle's current loading information and its maximum loading capacity.

[0033] This implementation plan achieves complete acquisition of transportation vehicle scheduling data through multi-module collaborative data collection, ensuring real-time perception and quantitative expression of key parameters such as vehicle task status, scheduling distance, vehicle speed, vehicle speed change rate, scheduling energy consumption, remaining transportation fuel, remaining scheduling time, and empty capacity ratio. This process integrates hardware modules such as geolocation, inertial measurement, energy consumption detection, task management, and vehicle feedback to ensure the spatiotemporal synchronization and physical consistency of various scheduling data. This provides a stable data foundation for subsequent order insertion capability assessment, task matching determination, and route reconstruction, improving the scheduling process's response accuracy to changes in transportation status and its support capability for structural modeling.

[0034] Specifically, the steps for denoising, completing, aligning, and normalizing transport vehicle dispatching data are as follows: First, a sliding median filter combined with the interquartile range method is used to identify and remove local outliers in data such as dispatching distance, vehicle speed, and dispatching energy consumption, effectively avoiding interference from instantaneous errors in subsequent calculations. To address potential delays or instantaneous interruptions during data transmission, bidirectional timestamp interpolation and local trend extrapolation are used to complete short-term missing data for key indicators such as dispatching distance, vehicle speed, and vehicle speed change rate, restoring the continuous time series. Subsequently, by precisely aligning vehicle numbers with positioning times, dispatching distances and vehicle speeds from different sources are reconciled. Multi-source asynchronous data, such as remaining fuel volume and vehicle task status, are mapped onto a unified scheduling trajectory timeline to achieve time-series synchronization. To further reduce the impact of measurement noise and external disturbances, exponential weighted average and moving multinomial regression methods are used to dynamically denoise the vehicle scheduling data, extracting stable trends and smoothing local abrupt changes. Finally, by using a combination of Z-score standardization and logarithmic scaling, all vehicle scheduling data are uniformly normalized, eliminating dimensional differences in the numerical scale of each scheduling data point and compressing its dynamic fluctuation range, providing standardized input for the stable modeling of subsequent order insertion capability assessment values ​​and order insertion task matching assessment values.

[0035] This implementation plan effectively improves the continuity, accuracy, and comparability of key data such as dispatch distance, vehicle speed, vehicle speed change rate, dispatch energy consumption, remaining fuel, and vehicle task status by systematically denoising, completing, aligning, and normalizing the transportation vehicle dispatch data. Based on eliminating local outliers, repairing short-term missing values, and achieving time alignment of multi-source data, this processing flow, through dynamic denoising and scale normalization, provides dispatch data with a unified time benchmark and numerical standard. This provides high-quality, low-noise data support for the stable calculation of order insertion capability assessment values ​​and order insertion task matching assessment values, ensuring the executability of subsequent dispatch calculation logic and the reliability of the evaluation structure.

[0036] Specifically, the steps for real-time monitoring of task status changes, identifying vehicle order insertion capabilities after task cancellation, and filtering candidate vehicles for order insertion scheduling based on order insertion thresholds are as follows: Real-time monitoring and detection of all executing task statuses; when the system detects a vehicle's task status changing from 0 to 1, immediately mark the vehicle as interrupted, simultaneously suspend the execution of the vehicle's original scheduling path, and deregister the corresponding scheduling record in the system task queue; after removing the vehicle from the current scheduling queue, mark its status as pending order insertion and write it to the scheduling status buffer; simultaneously, the vehicle is stripped of scheduling master control in the task control center and transferred to the order insertion scheduling module, the original path segment resources are released and removed from the path allocation table; query current transport vehicle scheduling data, reading vehicle speed, remaining transport fuel, scheduling energy consumption, remaining scheduling time, original scheduling distance, empty capacity ratio, and vehicle speed change rate; divide the remaining transport fuel by scheduling energy consumption, divide the vehicle speed by 1 and add the vehicle speed change rate, multiply these two ratios as the base, and use the capability weight as the most important indicator. The process involves three steps: First, calculating the value of the power function. Second, dividing the remaining scheduling time by the sum of the scheduling distance and the remaining scheduling time, using the ratio as the base and the time weight as the exponent, and calculating the value of the power function again. Third, multiplying the empty capacity ratio by the empty weight and adding 1, resulting in the third part. The process also involves fitting successful and failed order insertion samples from historical order insertion data using the least mean square error method. A residual model is established and trained using the scheduling energy consumption, vehicle speed change rate, remaining scheduling time, empty capacity ratio, and task completion status before order insertion. This model yields the capability weight, time weight, and empty weight, all within the range of [0,1]. These three parts are then multiplied sequentially to obtain the order insertion capability assessment value. This assessment value is then normalized and compared with the order insertion threshold. If the assessment value is greater than or equal to the threshold, the vehicle is added to the order insertion candidate queue and assigned a new scheduling entry. If the assessment value is less than the threshold, the vehicle is added to the standby buffer queue and does not participate in this round of order insertion matching.

[0037] The specific formula for calculating the order insertion capability assessment value is as follows:

[0038] ;

[0039] In the formula, This indicates the evaluation value of the order insertion capability. Indicates the remaining transport fuel volume. Indicates the energy consumption for dispatching. Indicates vehicle speed. Indicates the rate of change of vehicle speed. Indicates the remaining scheduling time. Indicates the original scheduling distance. Indicates the proportion of unloaded capacity. Indicates capability weights. Indicates time weighting, Indicates the unloaded weight.

[0040] In this embodiment, as shown in Table 1, five sets of vehicle dispatch status samples were recorded. The data included seven parameters: remaining transport fuel, dispatch energy consumption, vehicle speed, vehicle speed change rate, remaining dispatch time, original dispatch distance, and empty capacity ratio. Based on these parameters, the order insertion capability assessment value corresponding to each set of samples was calculated. In Sample 1, the remaining transport fuel was 0.37, dispatch energy consumption was 0.62, vehicle speed was 0.09, vehicle speed change rate was 0.39, remaining dispatch time was 0.97, original dispatch distance was 0.00, and empty capacity ratio was 0.84, corresponding to an order insertion capability assessment value of 7.14. In Sample 2, the remaining transport fuel increased to 0.98, dispatch energy consumption decreased to 0.10, vehicle speed was 0.49, vehicle speed change rate was 0.27, remaining dispatch time was 0.69, original dispatch distance was 0.67, and empty capacity ratio was 0.93, corresponding to an order insertion capability assessment value of 7.14. The value is 18.64; in sample 3, the vehicle speed is 0.00, the vehicle speed change rate is 0.83, the original dispatch distance is 0.31, the empty capacity ratio is 0.33, and the corresponding order insertion capability assessment value is 6.31; in sample 4, the vehicle speed is 0.94, the vehicle speed change rate is 0.36, the remaining dispatch time is 0.04, the empty capacity ratio is 0.10, and the corresponding order insertion capability assessment value is 12.49; in sample 5, the remaining transport fuel is 0.14, the original dispatch distance is 0.98, the empty capacity ratio is 0.23, and the corresponding order insertion capability assessment value is 5.79.

[0041] Table 1. Data Table of Order Placement Capability Assessment Values

[0042]

[0043] like Figure 3 As shown, this is a radar chart driven by the single-item insertion capability assessment value, as illustrated in Table 1 and... Figure 3 It can be seen that there are significant differences in the order-jumping capabilities of different vehicles. Vehicle 2 performs well across multiple key variables such as fuel level, remaining time, and empty capacity, with relatively high values, resulting in the highest order-jumping capability assessment value and significant potential for priority scheduling. While vehicles 1 and 4 excel in certain dimensions, such as vehicle 1's strong remaining time and capacity, and vehicle 4's high speed, their overall capabilities are only average due to high energy consumption or insufficient time resources. Vehicles 3 and 5 perform poorly across multiple dimensions such as speed, time, and capacity, resulting in the lowest order-jumping capability assessment values. The thickness and color intensity of the radar lines in the graph directly reflect the strength of capabilities, while the extent and balance of the graph reveal the scheduling resource structure of each vehicle, helping to intuitively judge scheduling priority and task matching suitability.

[0044] This implementation scheme achieves dynamic evaluation of the ability of transport vehicles to fill in for missed tasks after task cancellation by real-time monitoring of vehicle task status and identification of scheduling interruptions. This process is based on vehicle speed, remaining fuel, scheduling energy consumption, remaining scheduling time, original scheduling distance, vehicle speed change rate, and empty capacity ratio from transport vehicle scheduling data. A task-filling capability evaluation model quantifies the resource status of each interrupted vehicle, generating a task-filling capability evaluation value. After normalization, this value is compared with a task-filling threshold to accurately distinguish whether a vehicle has the scheduling potential to participate in task-filling. Vehicles with task-filling capability are then pushed into the task-filling candidate queue, constructing a stable and controllable scheduling entry screening mechanism, improving resource recovery efficiency and scheduling response flexibility in task cancellation scenarios.

[0045] Specifically, the steps for matching candidate vehicles with tasks to be inserted based on path, time, and load conditions are as follows: First, sequentially read the current latitude and longitude coordinates, remaining scheduling time, insertion capability assessment value, current task sequence, and scheduling status of each vehicle to be inserted from the insertion candidate queue. Then, initialize a path attempt status flag for each vehicle to record subsequent changes in the scheduling path. Simultaneously, query all unassigned tasks in the task pool, extract the estimated execution time of each task to be inserted, the required load percentage of each task, and the insertion scheduling distance, and load these into the current batch of optional tasks. For each candidate vehicle, iterate through the optional tasks one by one, obtain its corresponding insertion scheduling distance, and compare it with the set maximum insertion distance. Each task is compared individually. If the distance to be inserted for the current task exceeds the maximum range, the task is skipped and marked as a path mismatch. For tasks with path distances within the acceptable range, it is further determined whether there is an intersection between the start and end range of its time window and the remaining scheduling time of the vehicle. If the time windows do not overlap, the task is skipped. Next, it is determined whether the load percentage required by the task to be inserted is less than or equal to the current empty capacity percentage of the vehicle. If it exceeds this, it is considered a load conflict and the current task is skipped. Finally, for all tasks that simultaneously meet the path distance condition, time window condition, and load condition, a single task insertion attempt record is generated, and its task number and candidate vehicle association information are marked as input data for subsequent matching evaluation and path reconstruction.

[0046] In this implementation plan, a task insertion screening mechanism is established by extracting and filtering the transportation vehicle scheduling data of vehicles in the candidate queue for insertion based on criteria. The mechanism is centered on constraints such as the distance to be inserted, remaining scheduling time, and the proportion of empty capacity. This process, based on path range judgment, time window overlap analysis, and load matching verification, ensures the consistency between the task to be inserted and the current vehicle status in terms of scheduling trajectory, task timeliness, and resource load. By generating a record of attempted insertion tasks that meet the constraints of path distance, time window, and load, a set of tasks with clear objectives and boundaries is provided for subsequent matching evaluation value calculation and path reconstruction, effectively improving the logical rigor and execution efficiency of the task insertion process in the task screening stage.

[0047] Specifically, the steps for analyzing the matching effect of insertion tasks and completing task path reconstruction and insertion task marking are as follows: Obtain the scheduling distance to be inserted, the maximum insertion distance, the estimated execution time of the task to be inserted, the load percentage required by the task to be inserted, the remaining scheduling time, the idle capacity ratio, and the insertion capability assessment value. Calculate the insertion task matching assessment value for the task to be inserted: Divide the scheduling distance to be inserted by the maximum insertion distance, and subtract this ratio from 1. Multiply this difference by the matching weight, and the product is the first part. Divide the estimated execution time of the task to be inserted by the remaining scheduling time plus a minimum term. Subtract the load percentage required by the task to be inserted from the vehicle idle capacity ratio, and then divide by the idle capacity ratio plus a minimum term. Multiply the above two ratios, the insertion capability assessment value, and 1 minus the matching weight, and the product is the second part. Among these, a grid search combined with a cross-validation algorithm is used to analyze historical data. In the insertion task, the estimated execution time of the task to be inserted, the proportion of the load required by the task to be inserted, the scheduling distance to be inserted, and the scoring error between the insertion capability evaluation value and the actual successful result are minimized to obtain the matching weight. The value range of the matching weight is [0,1]. The first part and the second part are added to obtain the insertion task matching evaluation value. The calculated insertion task matching evaluation values ​​are sorted in descending order, and the task with the highest insertion task matching evaluation value is extracted as the current vehicle path reconstruction attempt object to generate a new task sequence structure. The scheduling legality of the reconstructed path is checked, including path distance, time window overlap, load constraints, etc. If the check passes, the insertion task is marked as the proposed execution state, written into the vehicle task structure and numbered. If the check fails, the vehicle will not participate in the insertion in this round, is marked as standby and unassigned, and waits for the next cycle of scheduling.

[0048] The specific formula for calculating the evaluation value of the inserted task matching is as follows:

[0049] ;

[0050] In the formula, This indicates the evaluation value for matching the inserted task. Indicates the distance to be inserted and scheduled. Indicates the maximum insertion distance. This indicates the estimated execution time of the task to be inserted. This indicates the percentage of payload required by the task to be inserted. Indicates the remaining scheduling time. Indicates the proportion of unloaded capacity. Indicates minterms, This indicates the evaluation value of the order insertion capability. This indicates the matching weight.

[0051] In this embodiment, as shown in Table 2, the scheduling parameter combinations of five groups of insertion tasks and their corresponding matching evaluation values ​​are recorded. Each group of tasks includes six data items: the scheduling distance to be inserted, the estimated execution time of the task to be inserted, the remaining scheduling time, the load ratio required by the task to be inserted, the empty capacity ratio, and the insertion capability evaluation value. The insertion task matching evaluation value of the task under the current vehicle status is calculated based on the aforementioned scoring model. In Task 1, the waiting scheduling distance is 4.58, the estimated execution time of the task to be inserted is 44.21, the remaining scheduling time is 116.69, the required load percentage of the task to be inserted is 0.35, the idle capacity percentage is 0.79, the insertion capability assessment value is 13.39, and the corresponding insertion task matching assessment value is 2.93. In Task 2, the waiting scheduling distance increases to 11.91, the estimated execution time of the task to be inserted is 41.35, the remaining scheduling time is 101.38, the required load percentage of the task to be inserted is 0.55, the idle capacity percentage is 0.90, the insertion capability assessment value is 24.83, and the corresponding insertion task matching assessment value is 2.01. In Task 3, the waiting scheduling distance is 10.21, the estimated execution time of the task to be inserted is 25.07, and the remaining scheduling time is 124.8. 1. The required load percentage for the task to be inserted is 0.27, the idle capacity percentage is 0.91, the insertion capability assessment value is 18.52, and the insertion task matching assessment value is 2.49; 2. The insertion scheduling distance for task 4 is 6.01, the estimated execution time for the task to be inserted is 66.41, the remaining scheduling time is 95.74, the required load percentage for the task to be inserted is 0.41, the idle capacity percentage is 0.59, the insertion capability assessment value is 8.52, and the insertion task matching assessment value is 1.68; 3. In task 5, the insertion scheduling distance is 6.73, the estimated execution time for the task to be inserted is 53.98, the remaining scheduling time is 86.56, the required load percentage for the task to be inserted is 0.31, the idle capacity percentage is 0.71, the insertion capability assessment value is 20.52, and the insertion task matching assessment value is 3.86.

[0052] Table 2. Evaluation Values ​​of Inserted Task Matching

[0053]

[0054] like Figure 4The image shows a bar chart of the evaluation values ​​for the single task matching, combined with Table 2 and... Figure 4 As can be seen, there are significant differences in the scores among the five tasks. Task 5 has the highest matching evaluation value, indicating that it has the strongest overall adaptability in terms of insertion distance, task duration, empty capacity, and vehicle capability, and possesses the optimal scheduling matching conditions. Task 4 has the lowest matching evaluation value, possibly affected by factors such as long distance, insufficient time, or load mismatch, resulting in weaker overall adaptability. The scores of the remaining tasks are in the middle tier, indicating average adaptability. Overall, the scoring results in the figure effectively reflect the matching quality of each task under multiple scheduling parameters, providing an intuitive basis for scheduling ranking and insertion decisions.

[0055] This implementation plan constructs a task insertion matching evaluation value calculation process, comprehensively considering transportation vehicle scheduling data such as the pending scheduling distance, maximum insertion distance, estimated execution time of the pending task, load ratio required by the pending task, remaining scheduling time, empty capacity ratio, and insertion capability evaluation value. This enables a quantitative judgment of the compatibility between tasks and vehicles. Based on the ranking mechanism of the matching evaluation value, the pending tasks that best match the current state of the vehicles can be effectively screened. Furthermore, based on the legality verification of the scheduling path, the task path structure is reconstructed and the pending task is marked for execution. This ensures the consistency and executability of the insertion scheduling behavior in the three dimensions of resources, timing, and path structure, improving the scheduling success rate of insertion tasks and the structural rationality of the scheduling path.

[0056] Specifically, the steps for executing and recording the status of insertion tasks are as follows: Extract the insertion tasks marked as to be executed; synchronously write the task's path structure, the distance to be inserted, the estimated execution time, the load percentage required by the insertion task, and the task number into the corresponding vehicle's task execution list; and generate an updated scheduling chain based on the vehicle's current task sequence to ensure logical continuity of task order; simultaneously, send the new task chain structure after the insertion task to the vehicle scheduling unit, and update the insertion task's execution marker to clarify the current task's execution identity and control attribution; the system synchronously records the current insertion position and the original scheduling... The system tracks changes in the path structure, including task numbers before and after the insertion point, path reconstruction identifiers, and resource configuration change fields. During the period when a vehicle begins executing an insertion task, the scheduling module continuously receives task execution status feedback and monitors in real time whether its execution progress meets the task setting parameters. After the insertion task is completed, the system reads the actual execution result of the task. If the task is completed within the specified time window and is not canceled midway, the task execution result is marked as 1. If the task is not completed as planned or is canceled midway, it is marked as 0, and the execution status record is written to the task status history table for subsequent scheduling strategy correction and scoring function backtracking analysis.

[0057] In this implementation plan, a complete closed loop for inserting tasks is constructed by writing the path structure, updating the scheduling chain, monitoring the execution status, and recording the results for tasks marked as to be executed. During this process, key scheduling parameters such as the scheduling distance to be inserted, the estimated execution time of the task to be inserted, and the proportion of the load required by the task to be inserted are synchronously written into the task execution list and form an updated task sequence, ensuring the sequential integrity and path structure continuity of tasks in the scheduling chain. Simultaneously, by receiving the execution status of the inserting tasks in real time and recording the actual execution results, a correlation mechanism between the task execution process and the results is effectively established, providing reliable data support for subsequent scoring function optimization, strategy deviation identification, and scheduling logic correction.

[0058] Specifically, the steps for analyzing the deviation between the score and actual execution of the order insertion task are as follows: Calculate the average order insertion capability assessment value of the current order insertion batch as a reference benchmark for batch scheduling capability; for each order insertion task, obtain complete scheduling data including the actual execution result, the order insertion task matching assessment value, the scheduling distance to be inserted, the original scheduling distance, the estimated execution time of the task to be inserted, the load percentage required by the task to be inserted, the remaining scheduling time, the idle capacity ratio, and the order insertion capability assessment value, and construct a score-execution analysis sample for that task; when calculating the residual assessment value of the order insertion task, first, the square of the difference between the order insertion task matching assessment value and the actual execution result of the order insertion task is used as the first part to reflect the relationship between the score and the execution result. The first step is to determine the degree of direct deviation between the results. The second step involves calculating the following four ratios in sequence: dividing the expected execution time of the task to be inserted by the remaining scheduling time and adding a minimum term; subtracting the required load ratio from the vehicle's empty capacity ratio and then dividing by the empty capacity ratio and adding a minimum term; dividing the original scheduling distance by the scheduling distance to be inserted; dividing the insertion capability assessment value by the absolute value of the difference between itself and the mean of the current insertion batch's insertion capability assessment values ​​and adding a minimum term; multiplying these four ratios together, the product is used as the second part, representing the indirect coupling relationship between task parameters and system state; finally, multiplying the first and second parts together yields the insertion task residual assessment value, which is used to measure the degree of deviation of the insertion task matching score under the current scheduling state.

[0059] The specific formula for the residual evaluation value of the inserted task is as follows:

[0060] ;

[0061] In the formula, This represents the residual evaluation value of the inserted task. This indicates the actual execution result of the insertion task. This indicates the evaluation value for matching the inserted task. Indicates the distance to be inserted and scheduled. Indicates the original scheduling distance. This indicates the estimated execution time of the task to be inserted. This indicates the percentage of payload required by the task to be inserted. Indicates the remaining scheduling time. Indicates the proportion of unloaded capacity. Indicates minterms, This indicates the evaluation value of the order insertion capability. This represents the average value of the current batch of order insertion capability assessment.

[0062] In this implementation plan, a calculation process for the residual evaluation value of the insertion task is constructed. Based on transportation vehicle scheduling data such as the actual execution result of the insertion task, the insertion task matching evaluation value, the insertion capability evaluation value, the scheduling distance to be inserted, the original scheduling distance, the estimated execution time of the task to be inserted, the load ratio required by the task to be inserted, the remaining scheduling time and the empty capacity ratio, a quantitative expression of the deviation between the scoring result and the actual execution performance is achieved. The residual evaluation value comprehensively considers the coupling strength between the scoring error and the task parameters, providing key error feedback basis for the scheduling system in identifying strategy deviations, optimizing the scoring model structure, and improving the accuracy of task selection, thereby enhancing the adaptability and correction capability of the scheduling logic to the actual operating state.

[0063] Specifically, the steps for proposing scoring structure updates, completing feedback archiving, and restoring vehicle dispatch status are as follows: The residual evaluation value of the inserted task is used as the basis for dispatch strategy correction. The residual evaluation value of the inserted task is compared with the task matching threshold in real time. If the residual evaluation value of an inserted task is greater than or equal to the task matching threshold, it is determined to be a strategy deviation sample. All parameters of this sample, including the inserted task matching evaluation value, inserted capability evaluation value, remaining dispatch time, original dispatch distance, waiting dispatch distance, empty capacity ratio, and the load ratio required by the waiting task, are written into the strategy deviation sample table and marked as a high residual sample. Such samples will be analyzed in detail in subsequent scoring function structure optimization. If the residual evaluation value of an inserted task is less than the task matching threshold, it is determined to be a scoring stable sample. This sample will still record the residual score and execution parameters to maintain sample data integrity, but will not enter the deviation sample buffer. This data serves only as a reference for subsequent stability assessments of the scoring function and does not participate in weight adjustments. After all insertion tasks in this round are completed, the scheduling module summarizes the strategy deviation samples and scoring stable samples to form an insertion sample set. Statistical analysis is performed on high residual samples to identify variable combinations that occur frequently in the high residual samples and mark them as abnormal parameter combinations. Based on the abnormal parameter combinations, structural adjustment suggestions are proposed for the relevant variable terms in the insertion task matching evaluation value formula, generating an updated draft scoring function as a candidate scoring structure for the next cycle. The draft scoring function, along with the current cycle's insertion task residual evaluation value distribution and deviation sample ratio, is written into the scoring function version control table to generate a version number for use in decision-making before the next cycle's scheduling. After all insertion task feedback records are completed, the vehicle scheduling status is updated, the insertion control is released, the vehicle is rewritten into the main scheduling queue, the system enters the next scheduling cycle, and the insertion process is completed in a closed loop.

[0064] In this implementation plan, a feedback mechanism for identifying scheduling strategy deviations and correcting scoring structures is established by comparing the residual evaluation values ​​of insertion tasks with the task matching thresholds in real time. After the insertion task is executed, the system divides the samples into strategy deviation samples and scoring stable samples based on the residual evaluation values ​​of the insertion task, and records key transportation vehicle scheduling data such as the insertion task matching evaluation value, insertion capability evaluation value, remaining scheduling time, original scheduling distance, waiting scheduling distance, empty capacity ratio, and the load ratio required by the waiting task. Through centralized statistical analysis and variable combination analysis of high residual samples, the system identifies the sources of scoring deviations, proposes structural adjustment suggestions for the insertion task matching evaluation value formula, and then generates a draft scoring function and manages version archiving. After completing the feedback records for all insertion tasks, the system synchronously updates the vehicle scheduling status, releases the insertion control, and writes the vehicles back to the main scheduling queue, realizing a closed loop of insertion task execution and scheduling logic correction, and providing a traceable and optimizable basis for the evolution of the scoring structure for the next cycle.

[0065] like Figure 2 As shown, the second aspect of this invention provides an intelligent transport vehicle scheduling system, comprising: a transport vehicle scheduling data acquisition and preprocessing module, a task cancellation detection and resource status modeling module, an order insertion task screening and path reconstruction module, and an order insertion task execution and feedback correction module. The transport vehicle scheduling data acquisition and preprocessing module is used to collect transport vehicle scheduling data during task execution and perform denoising, completion, alignment, and normalization processing on the transport vehicle scheduling data. The task cancellation detection and order insertion vehicle screening module is used to monitor task status changes in real time, identify the order insertion capability of vehicles after task cancellation, and screen candidate vehicles to enter the order insertion scheduling process based on the order insertion threshold. The order insertion task screening and path reconstruction module is used to match candidate vehicles with tasks to be inserted based on path, time, and load conditions, analyze the matching effect of the order insertion tasks, and complete task path reconstruction and order insertion task marking. The order insertion task execution and feedback correction module is used to execute the order insertion task and record the status, analyze the deviation between the order insertion task score and the actual execution, propose a score structure update suggestion, complete feedback archiving, and restore the vehicle scheduling status.

[0066] This implementation plan constructs a complete scheduling system consisting of a transportation vehicle scheduling data acquisition and preprocessing module, a task cancellation detection and vehicle insertion screening module, a task insertion screening and route reconstruction module, and a task insertion execution and feedback correction module. The transportation vehicle scheduling data acquisition and preprocessing module standardizes and cleans scheduling data during task execution, providing high-quality input for subsequent processing. The task cancellation detection and vehicle insertion screening module can perceive changes in task status in real time, identify interrupted resources, and quantify insertion capabilities, ensuring the timeliness and accuracy of scheduling responses. The task insertion screening and route reconstruction module filters tasks and reconstructs path structures under path distance, time window, and load constraints, improving the rationality of task insertion matching and scheduling continuity. The task insertion execution and feedback correction module records task status, evaluates execution deviations, and corrects the scoring structure, achieving dynamic closed-loop and adaptive optimization of the scheduling process. These four modules work together to form a complete chain of task insertion scheduling based on transportation vehicle scheduling data, enhancing scheduling flexibility, stability, and strategy adjustability in complex scenarios.

[0067] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0068] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for scheduling intelligent transport vehicles, characterized in that, Includes the following steps: S1 collects transportation vehicle scheduling data during the task execution process, and performs noise reduction, completion, alignment and normalization processing on the transportation vehicle scheduling data. S2, monitor task status changes in real time, identify the vehicle's order insertion capability after task cancellation, mark its status as pending order insertion, and filter candidate vehicles to enter the order insertion scheduling process according to the order insertion threshold. The analysis of the vehicle's order insertion capability is as follows: by querying the current transportation vehicle scheduling data, reading the vehicle speed, remaining transportation fuel, scheduling energy consumption, remaining scheduling time, original scheduling distance, empty capacity ratio and vehicle speed change rate, the order insertion capability evaluation value is obtained. S3. Based on path, time, and load conditions, match candidate vehicles with tasks to be inserted, analyze the matching effect of the insertion tasks, and complete task path reconstruction and insertion task marking. The analysis of the matching effect of the insertion tasks is as follows: obtain the scheduling distance to be inserted, the maximum insertion distance, the estimated execution time of the task to be inserted, the load ratio required by the task to be inserted, the remaining scheduling time, the idle capacity ratio, and the insertion capability evaluation value to obtain the insertion task matching evaluation value. Sort the calculated insertion task matching evaluation values ​​in descending order, extract the task with the highest insertion task matching evaluation value as the current vehicle path reconstruction attempt object, and generate a new task sequence structure. S4, execute the order insertion task and record its status, analyze the deviation between the order insertion task score and the actual execution, propose a score structure update suggestion, complete feedback archiving and restore the vehicle dispatch status. The analysis of the deviation between the order insertion task score and the actual execution is as follows: calculate the average order insertion capability assessment value of the current order insertion batch, and for each order insertion task, obtain the actual execution result of the order insertion task, the order insertion task matching assessment value, the dispatch distance to be inserted, the original dispatch distance, the estimated execution time of the task to be inserted, the load ratio required by the task to be inserted, the remaining dispatch time, the empty capacity ratio and the order insertion capability assessment value, and calculate the order insertion task residual assessment value.

2. The intelligent transport vehicle scheduling method according to claim 1, characterized in that: The specific steps for collecting transportation vehicle scheduling data during the task execution process are as follows: By integrating a geolocation unit, a path inertial measurement module, an energy consumption sensor, and a vehicle execution status feedback unit, real-time transportation vehicle dispatch data is collected. This data includes vehicle task status, dispatch distance, vehicle speed, vehicle speed change rate, dispatch energy consumption, remaining fuel, remaining dispatch time, and empty capacity ratio. The task lifecycle management module in the dispatch system obtains vehicle task status in real time, marking canceled tasks as 1 and tasks in progress as 0. A high-precision satellite positioning system integrated into the transport vehicle control unit acquires the original task starting point's latitude and longitude coordinates, the vehicle's current latitude and longitude coordinates, and the starting point's latitude and longitude coordinates of the task to be inserted, thus obtaining the original dispatch distance and the distance to be inserted. A speed sensor installed in the vehicle's main control board further contributes to the data. The system obtains vehicle speed; continuously measures vehicle linear acceleration and angular velocity using an inertial measurement module installed in the vehicle's main control board, calculates the rate of change of speed over continuous periods, and obtains the vehicle speed change rate; measures the energy release rate of the vehicle per unit travel distance using current and voltage sensors deployed in the battery management system, and obtains the scheduling energy consumption by combining it with drive motor load parameters; collects the current remaining fuel percentage through the onboard fuel management system, and obtains the remaining transport fuel by combining it with the remaining task distance in the vehicle task execution list and historical unit fuel consumption data; reads the current time through the main control task unit and compares it with the earliest execution time of the next task in the task scheduling table to obtain the remaining scheduling time; and obtains the empty capacity ratio by retrieving the ratio between the vehicle's current loading information and its maximum loading capacity.

3. The intelligent transport vehicle scheduling method according to claim 1, characterized in that: The specific steps for denoising, completing, aligning, and normalizing the transport vehicle dispatching data are as follows: A sliding median filter combined with the interquartile range method is used to identify and remove local outliers in dispatch distance, vehicle speed, and energy consumption data. Two-way timestamp interpolation and local trend extrapolation are used to fill in short-term gaps in transport vehicle dispatch data caused by data delays and interruptions. By aligning vehicle numbers with positioning times, the time of dispatch distance, speed, fuel consumption, and task status data from multiple sources under the same dispatch trajectory is unified, achieving data synchronization. Exponential weighted average and sliding multinomial regression methods are used to denoise the transport vehicle dispatch data, preserving trends while reducing fluctuations. Finally, a combination of Z-score standardization and logarithmic scaling is used to normalize the transport vehicle dispatch data, unifying the units and compressing the numerical range.

4. The intelligent transport vehicle scheduling method according to claim 1, characterized in that: The specific steps for real-time monitoring of task status changes, identifying vehicle order insertion capabilities after task cancellation, and filtering candidate vehicles to enter the order insertion scheduling process based on the order insertion threshold are as follows: The system monitors and detects the status of all executing tasks in real time. When the system detects that the vehicle task status of a vehicle changes from 0 to 1, it immediately marks the vehicle as having a scheduling interruption, stops the execution of the original scheduling path of the vehicle, and cancels the corresponding scheduling record in the system task queue. After the vehicle is removed from the current scheduling queue, it is written to the scheduling status buffer. At the same time, the vehicle is stripped of its scheduling master control in the task control center and transferred to the order insertion scheduling module. The original path segment resources are released and the vehicle is removed from the path map allocation table. Query the current vehicle dispatch data to read vehicle speed, remaining fuel, dispatch energy consumption, remaining dispatch time, original dispatch distance, empty capacity ratio, and vehicle speed change rate. Divide the remaining fuel by the dispatch energy consumption, divide the vehicle speed by 1 and add the vehicle speed change rate, multiply these two ratios as the base, use the capacity weight as the exponent, and calculate the value of this power function as the first part. Divide the remaining dispatch time by the sum of the dispatch distance and the remaining dispatch time, use the ratio as the base, use the time weight as the exponent, and calculate the value of this power function as the second part. Multiply the empty capacity ratio by the empty weight and add 1 to get the third part. Multiply the above three parts in sequence to obtain the order insertion capacity assessment value. The obtained order insertion capability assessment value is normalized and compared with the order insertion threshold: when the order insertion capability assessment value is greater than or equal to the order insertion threshold, the vehicle is pushed into the order insertion candidate queue and a new scheduling entry is assigned to it; when the order insertion capability assessment value is less than the order insertion threshold, the vehicle is pushed into the standby buffer queue and does not participate in the order insertion matching process in this round.

5. The intelligent transport vehicle scheduling method according to claim 1, characterized in that: The specific steps for matching candidate vehicles with insertion tasks based on path, time, and load conditions are as follows: Read the current latitude and longitude coordinates, remaining scheduling time, order insertion capability assessment value, current task sequence and scheduling status of the vehicles to be inserted from the order insertion candidate queue, and initialize the path attempt status flag for each vehicle; query all unassigned tasks in the task pool, extract the estimated execution time of the tasks to be inserted, the load ratio required by the tasks to be inserted and the scheduling distance to be inserted, and load them into a set of optional tasks. For each candidate vehicle, iterate through each available task, obtain the scheduling distance to be inserted, and compare the scheduling distance to be inserted with the set maximum insertion distance. If it exceeds the maximum insertion distance, skip the current task. For tasks with path distance within the acceptable range, continue to determine whether there is an intersection between the start and end range of the task time window and the remaining scheduling time of the vehicle. If there is no intersection, skip the task. Then determine whether the load ratio required by the task to be inserted is within the current empty capacity ratio of the vehicle. If it exceeds the limit, skip the task. For all tasks that meet the path distance, time window overlap, and load constraints, generate a record of the task to be inserted attempt.

6. The intelligent transport vehicle scheduling method according to claim 1, characterized in that: The specific steps for analyzing the matching effect of the order insertion task and completing the task path reconstruction and order insertion task marking are as follows: To calculate the matching evaluation value of the task to be inserted, the following parameters are obtained: the distance to be inserted, the maximum insertion distance, the estimated execution time of the task to be inserted, the load percentage required by the task to be inserted, the remaining scheduling time, the idle capacity ratio, and the insertion capability assessment value. The first part is calculated by dividing the distance to be inserted by the maximum insertion distance and subtracting this ratio from 1. This difference is then multiplied by the matching weight. The second part is calculated by dividing the estimated execution time of the task to be inserted by the remaining scheduling time and adding a minimum term. The third part is calculated by subtracting the load percentage required by the task to be inserted from the vehicle idle capacity ratio and then dividing by the idle capacity ratio and adding a minimum term. The fourth part is calculated by multiplying the two ratios, the insertion capability assessment value, and 1 minus the matching weight. The product of these four values ​​is the second part. The first and second parts are then added together to obtain the matching evaluation value of the task to be inserted. The scheduling validity of the reconstructed path is checked, including path distance, time window overlap, and load constraints. If the check passes, the insertion task is marked as intended for execution, written into the vehicle task structure, and the path is numbered. If the check fails, the vehicle will not participate in the insertion in this round, and will be marked as standby and unassigned, waiting for the next cycle of scheduling.

7. The intelligent transport vehicle scheduling method according to claim 1, characterized in that: The specific steps for executing the order insertion task and recording its status are as follows: Extract the insertion tasks marked as to be executed, and synchronously write the task's path structure, the scheduling distance to be inserted, the estimated execution time of the insertion task, the load ratio required by the insertion task, and the task number into the task execution list of the corresponding vehicle. Generate an updated scheduling chain based on the vehicle's current task sequence. Send the new task chain after the insertion task is inserted to the vehicle scheduling unit, update the execution mark of the insertion task, and record the structural changes of the current insertion position and the original scheduling path. The vehicle begins to execute the order insertion task, and the scheduling module receives the task execution status in real time. After the order insertion task is completed, the actual execution result of the order insertion task is read. If the task is completed on time and is not canceled, it is recorded as 1; otherwise, it is recorded as 0.

8. The intelligent transport vehicle scheduling method according to claim 1, characterized in that: The specific steps for analyzing the deviation between the task scoring and actual execution are as follows: Calculate the average order insertion capability assessment value for the current order insertion batch. For each order insertion task, obtain the actual execution result of the order insertion task, the order insertion task matching assessment value, the dispatch distance to be inserted, the original dispatch distance, the estimated execution time of the order to be inserted, the load ratio required by the order to be inserted, the remaining dispatch time, the empty capacity ratio, and the order insertion capability assessment value. Calculate the order insertion task residual assessment value: Square the difference between the order insertion task matching assessment value and the actual execution result of the order insertion task as the first part; divide the estimated execution time of the order to be inserted by the remaining dispatch time and add a minimum term; subtract the load ratio required by the order to be inserted from the vehicle empty capacity ratio and then divide by the empty capacity ratio and add a minimum term; divide the original dispatch distance by the dispatch distance to be inserted; divide the order insertion capability assessment value by the absolute value of the order insertion capability assessment value minus the average of the order insertion capability assessment values ​​for the current order insertion batch and add a minimum term; multiply the above four ratios, and the product is the second part; multiply the first part and the second part to obtain the order insertion task residual assessment value.

9. The intelligent transport vehicle scheduling method according to claim 1, characterized in that: The specific steps for proposing scoring structure updates, completing feedback archiving, and restoring vehicle dispatch status are as follows: The residual evaluation value of the inserted task is used as the basis for adjusting the scheduling strategy. The residual evaluation value of the inserted task is compared with the task matching threshold in real time. If the residual evaluation value of a certain inserted task is greater than or equal to the task matching threshold, it is determined to be a strategy deviation sample. The inserted task matching evaluation value, inserted capability evaluation value, remaining scheduling time, original scheduling distance, waiting scheduling distance, idle capacity ratio, and the load ratio required by the waiting task are all written into the strategy deviation sample table and marked as high residual samples. Such samples will be analyzed in detail in the subsequent optimization of the scoring function structure. If the residual evaluation value of a certain insertion task is less than the task matching threshold, it is determined to be a stable scoring sample. This sample will still record the residual score and execution parameters to maintain the integrity of the sample data, but it will not enter the bias sample buffer. It will only be used as the reference data for subsequent scoring function stability evaluation and will not participate in weight correction. After all insertion tasks in this round are completed, the scheduling module summarizes the strategy deviation samples and the scoring stable samples to form an insertion sample set; it performs statistical analysis on the high residual samples to identify variable combinations that occur frequently in the high residual samples and marks them as abnormal parameter combinations; based on the abnormal parameter combinations, it proposes structural adjustment suggestions for the relevant variable terms in the insertion task matching evaluation value formula, and generates an updated scoring function draft as a candidate scoring structure for the next cycle; the scoring function draft, along with the current cycle's insertion task residual evaluation value distribution and deviation sample ratio, is written into the scoring function version control table to generate a version number for use in decision-making before the next cycle's scheduling; After all the order insertion task feedback records are completed, the vehicle dispatch status is updated, the order insertion control is released, the vehicle is rewritten into the main dispatch queue, the system enters the next dispatch cycle, and the order insertion process is completed.

10. An intelligent transport vehicle scheduling system, employing the intelligent transport vehicle scheduling method as described in any one of claims 1-9, characterized in that: include: The system includes a transportation vehicle dispatching data acquisition and preprocessing module, a task cancellation detection and resource status modeling module, an order insertion task filtering and route reconstruction module, and an order insertion task execution and feedback correction module, among which: The transport vehicle scheduling data acquisition and preprocessing module is used to collect transport vehicle scheduling data during the task execution process, and to perform noise reduction, completion, alignment and normalization processing on the transport vehicle scheduling data. The task cancellation detection and order insertion vehicle screening module is used to monitor task status changes in real time, identify the vehicle's order insertion capability after task cancellation, and filter out candidate vehicles to enter the order insertion scheduling process based on the order insertion threshold. The task insertion screening and path reconstruction module is used to match candidate vehicles with tasks to be inserted based on path, time and load conditions, analyze the task insertion matching effect, and complete task path reconstruction and task insertion marking. The order insertion task execution and feedback correction module is used to execute the order insertion task and record its status, analyze the deviation between the order insertion task score and the actual execution, propose a score structure update suggestion, complete feedback archiving, and restore the vehicle dispatch status.

Images