An automated air track scheduling method and system

Abstract

The application discloses an automatic air rail scheduling method and system, relates to air rail scheduling, and comprises the following steps: acquiring air rail data; calculating the residual load of each vehicle according to a vehicle data set, generating a vehicle set V to be distributed; performing priority sorting on a task data set, generating a task set U; generating a vehicle-task matching relationship according to the vehicle set V and the task set U; performing path planning on the vehicle-task matching relationship through an improved Dijkstra algorithm, and obtaining an optimal scheduling scheme of vehicle scheduling; the improved shortest path algorithm Dijkstra performs dynamic adjustment on path weight by introducing a track congestion degree coefficient; scheduling air rail vehicles according to the optimal scheduling scheme; and aiming at the low path planning efficiency caused by the dynamic change of air rail track congestion, the application improves the planning efficiency by constructing a vehicle-task bipartite graph and combining a path planning strategy of dynamic adjustment of track congestion degree.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of air track scheduling, in particular to an automatic air track scheduling method and system. BACKGROUND

[0002] The automatic air track system is the core infrastructure of modern intelligent factories and smart logistics, and is widely used in manufacturing production lines, large warehouse centers, port container transportation and other scenarios. Through the suspended track network, the system realizes efficient transportation and scheduling of materials and products in three-dimensional space, effectively solving the bottleneck problems of space limitation and fixed path of ground transportation system.

[0003] The air track network has the characteristics of limited capacity, and some key sections (such as intersections and nearby loading and unloading points) are prone to form "bottlenecks". The traditional scheduling method often uses a static shortest path planning strategy, without considering the dynamic congestion state of the track, resulting in multiple vehicles being assigned to the same section, causing path conflicts and system congestion, and seriously affecting the overall operation efficiency.

[0004] In addition, most existing technologies use a greedy algorithm for vehicle-task matching, which is based only on simple distance or time indicators, without considering the matching degree of vehicle load capacity and task material demand, as well as the current system congestion state, making it difficult to achieve optimal allocation of global resources. In particular, during the task-intensive period, the non-optimal scheduling caused by this simple matching strategy will cause problems such as decreased system throughput and prolonged response time. SUMMARY

[0005] In view of the low path planning efficiency caused by the dynamic change of air track congestion in the prior art, the present application provides an automatic air track scheduling method and system, which improves the planning efficiency by constructing a vehicle-task bipartite graph and combining a path planning strategy that dynamically adjusts the track congestion degree.

[0006] One aspect of the present application provides an automatic air track scheduling method, comprising: S1, obtaining air track data, the air track data comprising a vehicle data set and a task data set; the vehicle data set comprising the current coordinates and current load of each vehicle; the task data set comprising the material demand and task priority of each task; S2, calculating the remaining load of each vehicle according to the vehicle data set, and generating a set of vehicles to be allocated V; S3, sorting the task data set by priority, and generating a task set U; S4, generating a vehicle-task matching relationship according to the vehicle set V and the task set U; S5, performing path planning on the vehicle-task matching relationship by an improved Dijkstra algorithm to obtain an optimal scheduling scheme for vehicle scheduling; the improved shortest path algorithm Dijkstra dynamically adjusts the path weight by introducing a track congestion degree coefficient; S6, scheduling the air track vehicles according to the optimal scheduling scheme.

[0007] Further, S2 generates a set of vehicles to be assigned, V, including: extracting the maximum load limit and current load value of each vehicle from the vehicle dataset; calculating the remaining load by comparing the maximum load limit and the current load value; determining whether the remaining load is greater than the minimum load threshold required by the task; and for vehicles that meet the load requirements, recording the corresponding vehicle's identification number, current location coordinates, and remaining load to generate the vehicle set V.

[0008] Further, S4 generates vehicle-task matching relationships, including: calculating the Manhattan distance between each available vehicle and each task to be executed based on the vehicle set V and the task set U; calculating the resource matching degree based on the vehicle's remaining load and material requirements; constructing a vehicle-task bipartite graph based on the Manhattan distance and resource matching degree; and solving the vehicle-task bipartite graph to obtain the vehicle-task matching relationships.

[0009] Furthermore, based on Manhattan distance and resource matching degree, a vehicle-task bipartite graph is constructed, including: using the vehicle set V and the task set U as the two vertex sets of the bipartite graph, respectively; calculating v for each vehicle. i With each task t j The weighting coefficient w between ij , The calculation shows that, d ij For Manhattan distance, r ij For resource matching degree, α and β are preset balance coefficients; when the weight coefficient w ij When the value is greater than the threshold θ, in vehicle v i With task t j Establish a connection edge e between them ij and w ij As the weight values ​​of the corresponding edges; construct an adjacency matrix G as a bipartite graph, where, when there exists a connecting edge e ij When, matrix element g ij =w ij When there is no connecting edge e ij At that time, g ij =0.

[0010] Furthermore, the vehicle-task bipartite graph is solved to obtain the vehicle-task matching relationship, including: (1) constructing a time window matrix T based on the adjacency matrix G, where each element t of matrix T... ij Indicates vehicle v i Can it be done in task t? j (1) Complete the corresponding task within the time window; (2) Correct the adjacency matrix G according to the time window feasibility matrix T to obtain G'; where, when t ij When = 0, the corresponding g ij The value is set to the preset maximum value G. max When tij = 1, maintain the original g ij ; (3) normalize the rows and columns of the matrix G' to obtain G"; (4) mark the zero elements according to G", cover all zero elements using the minimum number of horizontal and vertical lines, and calculate the total number of covering lines, if the total number of covering lines is greater than or equal to the matrix order n, then the optimal matching scheme is found; if the number of covering lines is less than n, then step (5) is performed; (5) in the G" matrix, obtain the minimum value min val from all elements not covered by the lines val add min to all elements covered by two lines at the same time val , keep the values of other elements unchanged, and return to (4) for continuous execution; (6) according to the final zero element position in the matrix G" obtained in step (5), select different row and column zero elements for combination, and establish the matching relationship between the vehicle v i and the task t j according to the selected zero element position (i, j), to obtain the vehicle-task matching relationship.

[0011] Further, a time window matrix T is constructed, including: obtaining the time window parameters of the task t j from the task data set, including the earliest start time TE j and the latest end time TL j ; according to the Manhattan distance d ij , the time TA i of the vehicle v j moving from the current position to the position of the task t ij is calculated; when TA ij ≤ TL j , t ij = 1 indicates that the vehicle v i can arrive within the time window of the task t j , and vice versa t ij = 0 indicates that the vehicle v i cannot complete the task t j within the specified time.

[0012] Further, S5, the vehicle-task matching relationship is path planned through the improved Dijkstra algorithm to obtain the optimal scheduling scheme of vehicle scheduling, including: according to the vehicle-task matching relationship, constructing an empty rail network topology graph G = (N, E), wherein N represents a node set, the node set including a vehicle current position node and a task position node; E represents an edge set, and the edge represents a passable track between nodes; for each edge e E in the empty rail network, the initial weight w e is calculated, w e = d e × (1 + γ × ce ), where d e c is the actual track distance corresponding to edge e. e Here, γ is the track congestion coefficient, and γ is the congestion impact factor; for each vehicle-task matching pair (v) in the vehicle-task matching relationship... i ,t j Starting from the vehicle's current position and ending at the task position, the shortest path P from the starting point to the destination is generated using an improved Dijkstra's algorithm. i and the corresponding total weight W i For all vehicle-task matching pairs (v i ,t j The set of shortest paths {P1, P2, ..., P} n}, based on total weight

[0013] {W1,W2,.....,W n Sort the data; based on the weighted sorting results, confirm each vehicle-task matching pair in turn (v i ,t j The path allocation yields the optimal scheduling scheme, which includes the path planning sequence and the task execution order.

[0014] Track congestion coefficient c e The calculation formula is: c e =n e / u e , where n e denoted as the number of vehicles currently allocated on edge e, and ue as the maximum capacity threshold for edge e.

[0015] Furthermore, using an improved Dijkstra's algorithm, the shortest path P from the starting point to the ending point is generated. i and the corresponding total weight W i This includes: initializing the distance array D, setting the starting distance to 0 and the distances of other nodes to infinity; initializing the visited node set S to an empty set; initializing the predecessor node array P to record the predecessor node of each node in the shortest path; and performing the following operations in a loop until the endpoint is added to set S: i. Selecting the node u with the smallest distance value from the unvisited nodes and adding it to set S; ii. For each neighboring node v of node u, if D[u] + w uv If <D[v], then update D[u]+w uv =D[v]; iii. Update the congestion coefficient c of the relevant edges based on the currently planned path. e , iv. Dynamically adjust the weights of related edges, w e '=d e ×(1+γ×c e); according to the predecessor node array P, backtracking from the end point to the start point to construct the shortest path P i ; total weight W i equals the value corresponding to the end point in the distance array D.

[0016] Another aspect of the present application also provides an automated air track scheduling system for performing an automated air track scheduling method of the present application.

[0017] Compared with the prior art, the present application has the advantages of:

[0018] Due to the interweaving of the limited track capacity and the task time window constraint, the path planning of the air track is difficult to adapt to the dynamically changing system congestion state, resulting in low vehicle scheduling efficiency and insufficient resource utilization. In the prior art, static path planning strategies or simple greedy algorithms are generally used, which cannot effectively respond to the dynamic changes of track congestion degree, and often separate the vehicle-task matching and path planning, resulting in poor overall system performance.

[0019] The present application, on the one hand, constructs a vehicle-task bipartite graph, combines weight coefficients The distance and resource matching degree are comprehensively considered, and the Hungarian algorithm is applied to solve it, which ensures the global optimality of the matching and avoids the local optimal trap of the traditional greedy algorithm. On the other hand, the time window constraint is converted into a feasibility matrix T, and through matrix transformation technology (G→G'→G”), the infeasible matching is naturally excluded from the optimal solution, ensuring the time feasibility of the scheduling scheme and improving the calculation efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0020] The present application will be further described in the form of exemplary embodiments, which will be described in detail through the accompanying drawings. These embodiments are not limiting, and in these embodiments, the same numbers represent the same structures, wherein:

[0021] Figure 1 is an exemplary flowchart of an automated air track scheduling method according to some embodiments of the present application;

[0022] Figure 2 is an exemplary application scenario schematic diagram of an automated air track scheduling system according to some embodiments of the present application;

[0023] Figure 3 is an exemplary flowchart of vehicle residual load calculation according to some embodiments of the present application;

[0024] Figure 4 is an exemplary flowchart of vehicle-task matching according to some embodiments of the present application;

[0025] Figure 5is a topological graph of an aerial rail network according to some embodiments of the present application. DETAILED DESCRIPTION

[0026] The method and system provided by the embodiments of the present application are described in detail below with reference to the drawings.

[0027] As shown in Figure 1 and as shown in Figure 2 , the embodiment obtains aerial rail data from the central control system, including the position coordinates and load conditions of 5 vehicles, and the material requirements and priorities of 3 task points. The residual load of each vehicle is calculated, and three available vehicles V1, V2 and V5 are selected to form a set V. The tasks T1, T2 and T3 are sorted according to priority to generate a task set U. According to the residual load of the vehicle and the material requirement of the task, a vehicle-task bipartite graph is established, and the matching relationship is obtained: V1→T1, V2→T3, V5→T2. The improved Dijkstra algorithm is used for path planning to avoid congested tracks, and the optimal path scheme (solid line part) is generated. The control system issues a scheduling instruction to control the three vehicles to travel to the target task point according to the planned path.

[0028] The three-dimensional effect of the scheduling system is shown in detail. Specifically, the vehicle set V information is as follows: V1: aerial rail vehicle 1, residual load: 200 kg; V2: aerial rail vehicle 2, residual load: 280 kg; V3: aerial rail vehicle 3, state: in task; V4: aerial rail vehicle 4, state: in task; V5: aerial rail vehicle 5, residual load: 120 kg. The task set U information is as follows: T1: finished product loading area, material requirement: 150 kg, priority: high. T2: finished product loading area, material requirement: 80 kg, priority: medium. T3: assembly area, material requirement: 40 kg, priority: low.

[0029] S1, obtaining aerial rail data, in this embodiment, the aerial rail scheduling system obtains the aerial rail system state data from the central database of the aerial rail control center in real time through a unified data interface. The specific implementation mode is as follows: the scheduling system establishes a secure connection channel with the central database, and adopts a data subscription mode, when the system state changes, the related data will be automatically pushed to the scheduling system.

[0030] The system periodically collects (for example, every 500 milliseconds) the state information of all vehicles in the aerial rail network. The vehicle data set is stored in a structured table form, and each record contains the following fields: vehicle identification number (unique identification code), current track unit number, three-dimensional coordinate value (X, Y, Z), current vehicle load value (unit: kg), vehicle state flag (idle / in task / charging / failure), maximum load limit value (factory set value) and other basic information. These data are obtained through the cooperation of vehicle sensors and track monitoring points, and are transmitted to the central database through a wireless communication network.

[0031] Task data is issued by production management system (MES) or warehouse management system (WMS) to the overhead track scheduling system through standard data interface. Each task record contains: task identification number, material code, material demand (unit: kg), pick-up location coordinates, feeding location coordinates, task creation timestamp, earliest start time, latest completion time, task priority value (for example, 1-10, the larger the value, the higher the priority), and other information. The system saves these task data in the task queue data structure, preparing for subsequent processing.

[0032] As shown in Figure 3 The system first needs to filter out vehicles that can be used for task allocation and evaluate their load capacity. After the system obtains the vehicle data set from the central control database in real time, it executes the vehicle filtering process through a special data processing module. This process first filters out a subset of vehicles with a state of "idle" (state_flag = 1) according to the vehicle state flag, avoiding processing of vehicles that have been allocated tasks or are malfunctioning.

[0033] For each idle vehicle, the system reads its maximum load limit (max_load) and current load value (current_load) respectively. The maximum load limit is an inherent parameter of the vehicle, which is set according to the vehicle model and is usually stored in the system device parameter table. The current load value is obtained in real time by a high-precision weighing sensor installed on the vehicle, with an accuracy of ±0.5 kg and an update interval of every 5 seconds.

[0034] The system uses a floating-point number calculation engine to calculate the remaining load value: remaining_load = max_load - current_load. For example, the maximum load of No. 3 overhead track vehicle is 300 kg, and the current load is 75 kg, so its remaining load is 225 kg. The calculation result is rounded to one decimal place and stored in a temporary data buffer.

[0035] Next, the system performs load threshold judgment. The system determines the minimum load threshold (min_threshold) by analyzing the material demand of all pending tasks in the current task queue. This threshold also takes into account a safety margin coefficient (usually set to 1.1), i.e.: min_threshold = min_material_demand × 1.1. The system only retains vehicle records with remaining_load > min_threshold.

[0036] For vehicles that meet the load requirements, the system creates a structured vehicle set V. This set is implemented using an optimized hash table data structure, with the vehicle identifier as the key, and includes the following fields: vehicle_id: unique vehicle identifier (e.g., "AGV-2023-C056"); position_X, position_Y, position_Z: the vehicle's current three-dimensional coordinates; remaining_load: the calculated remaining load value; battery_level: the percentage of remaining battery power (as auxiliary decision-making information); last_maintenance_time: the timestamp of the last maintenance. Set V is indexed and optimized, supporting fast multi-dimensional queries by remaining load value or position coordinates, providing efficient data access for subsequent matching algorithms. Simultaneously, the system sorts the vehicles in set V in descending order of remaining load, facilitating the priority allocation of vehicles with strong load capacity in scenarios with high load requirements. Specifically, in this embodiment, the original vehicle data is detailed in Table 1, and the idle vehicle filtering results are detailed in Table 2.

[0037] Table 1 Original Vehicle Data

[0038]

[0039] Table 2: Results of Idle Vehicle Screening

[0040]

[0041]

[0042] As shown in Tables 1 and 2, in this embodiment, the maximum load of the No. 3 monorail car is 300kg, and the current load is 75kg. Therefore, its remaining load is: REMAINING_LOAD = 300kg - 75kg = 225kg. If the minimum material requirement in the current task queue is 100kg, then the minimum load threshold after considering the safety margin is: MIN_THRESHOLD = 100kg × 1.1 = 110kg.

[0043] The vehicle set V uses an optimized hash table data structure, with the vehicle ID as the key:

[0044] AGV-2023-C147:{VEHICLE_ID:"AGV-2023-C147",POSITION_X:125.4,POSITION_Y:78.2,POSITION_Z: 0.0,REMAINING_LOAD:350.0,BATTERY_LEVEL:78,LAST_MAINTENANCE_TIME:"2025-03-1508:30:00"};

[0045] AGV-2023-C056: {VEHICLE_ID: "AGV-2023-C056", POSITION_X: 210.6, POSITION_Y: 156.3, POSITION_Z: 0.0, REMAINING_LOAD: 250.0, BATTERY_LEVEL: 85, LAST_MAINTENANCE_TIME: "2025-04-02 14:15:00"} ;

[0046] AGV-2023-C003: {VEHICLE_ID: "AGV-2023-C003", POSITION_X: 56.8, POSITION_Y: 98.4, POSITION_Z: 0.0, REMAINING_LOAD: 225.0, BATTERY_LEVEL: 90, LAST_MAINTENANCE_TIME: "2025-03-25 10:45:00"} ;

[0047] AGV-2023-C021: {VEHICLE_ID: "AGV-2023-C021", POSITION_X: 185.2, POSITION_Y: 42.7, POSITION_Z: 0.0, REMAINING_LOAD: 200.0, BATTERY_LEVEL: 95, LAST_MAINTENANCE_TIME: "2025-04-08 09:20:00"}.

[0048] S3, in this embodiment, the system adopts a multi-factor comprehensive evaluation method to prioritize tasks, generating a task set U; the system first extracts all task records with a status of "to be allocated" from the task queue. Each task record contains a base priority value (base_priority), which is set by the upper-layer production management system and ranges from 1 to 10 (10 being the highest priority).

[0049] The sorting process uses an improved priority calculation model that dynamically adjusts the actual priority of tasks by considering multiple factors: time urgency assessment: the system calculates the deadline time margin of the task, with the formula:

[0050] time_urgency = (deadline_time - current_time) / estimated_execution_time; The smaller the time_urgency value, the more urgent the task. When this value is less than 1.5, the system will increase the base priority.

[0051] Production line dependency assessment: The system queries the production line dependency table to determine whether the task is related to a critical production line. If the task's materials are required by a critical production line, its priority is increased.

[0052] Task waiting time compensation: To avoid low-priority tasks not being executed for a long time, the system calculates the waiting time factor of the task: waiting_factor = (current_time - creation_time) / average_waiting_time; when the waiting_factor exceeds the set threshold (usually 2.0), the task priority is appropriately increased.

[0053] Taking all the above factors into account, the system calculates the overall priority value for each task:

[0054] `final_priority` = `base_priority` + `urgency_bonus` + `production_line_bonus` + `waiting_bonus`. The system uses a stable merge sort algorithm to sort tasks in descending order of `final_priority`. After sorting, the system stores the sorted results in a double-ended queue data structure, forming a task set U.

[0055] like Figure 4 As shown, in step S4, vehicle-task matching relationships are generated based on the vehicle set V and the task set U. In this embodiment, the air-rail scheduling system first constructs the distance matrix between vehicles and tasks. Specific key parameter settings are detailed in Table 3.

[0056] Table 3 Key Parameter Settings

[0057] Parameter name Value Explanation α 0.4 Distance factor weight coefficient, control the influence degree of distance on matching β 0.6 Resource matching degree weight coefficient, control the influence degree of load matching on result θ 0.3 Weight threshold, determine whether an effective connection is established between the vehicle and the task v 2 meters / second Standard vehicle driving speed of the empty rail system Tdelay 10 seconds Dispatching delay compensation time, to cope with system communication delay and other factors λ 0.8～1.5 Dynamic congestion coefficient, reflecting the influence of rail congestion on driving time Gmax 999999 Alternative value of time unfeasible matching, to ensure that it will not be selected by the algorithm

[0058] The system uses a three-dimensional Manhattan distance calculation method, which is suitable for reflecting the actual movement characteristics of the air-rail system in three-dimensional space. For each vehicle v in the vehicle set V... i and each task t in the task set U j The system calculates its Manhattan distance d. ij :d ij =|x i -x j |+|y i -y j |+|z i -z j |, where (x) i ,y i ,z i ) for vehicle v i The current spatial coordinates, (x j ,yj z j ) is the space coordinate of the pick-up point of task t j . The system uses an efficient vectorized calculation method to process the distance matrix, and stores the calculation result in the distance matrix D. The calculation result of the embodiment is shown in Table 4. The three-dimensional Manhattan distance calculation method is used in the embodiment to reflect the actual movement characteristics of the aerial rail system in three-dimensional space. When the vehicle vi is located at the position of the task tj (dij = 0), the system sets 1 / dij to be a preset maximum value 10.0 to avoid numerical overflow.

[0059] Table 4 Manhattan distance matrix

[0060] Vehicle / task Task T1 Task T2 Task T3 Vehicle V1 120 85 210 Vehicle V2 175 150 65 Vehicle V3 95 180 145 Vehicle V4 160 190 110

[0061] At the same time, the system calculates the resource matching degree matrix R to evaluate the matching degree of each vehicle and each task in terms of load: wherein RL i is the remaining load of the vehicle v i , MD j is the material demand of the task t j , and MD j,max is the maximum material demand in all tasks. The formula ensures that the resource matching degree is in the interval [0, 1], and the value closer to 1 indicates a higher matching degree. When the remaining load of the vehicle just meets the demand of the task (without wasting load capacity), the matching degree is the highest. The resource matching result of the embodiment is shown in Table 5. The application evaluates the matching degree by comparing the remaining load of the vehicle and the material demand of the task. When the remaining load of the vehicle just meets the demand of the task, the matching degree is the highest, avoiding the waste of load capacity.

[0062] Table 5 Resource matching degree matrix

[0063] Vehicle / task Task T1 (150 kg) Task T2 (80 kg) Task T3 (40 kg) V1 (200 kg) 1 0.83 0.75 V2 (280 kg) 1 0.54 0.95 V3 (225 kg) 1 0.65 0.67 V4 (120 kg) 0.8 0.67 0.5

[0064] In the preferred embodiment of the application, the aerial rail scheduling system uses a dedicated graph processing module to construct a vehicle-task bipartite graph. The module receives the distance matrix D and the resource matching degree matrix R as input, and outputs an adjacency matrix G representing the association relationship between vehicles and tasks.

[0065] In specific implementation, the system first allocates a two-dimensional floating-point array of mxn in memory as a data container for the adjacency matrix G, where m represents the number of vehicles in the vehicle set V, and n represents the number of tasks in the task set U. The system uses a continuous storage mode to improve memory access efficiency, and initializes all matrix elements to 0, indicating that there is no initial connection relationship between all vehicle-task pairs.

[0066] For each vehicle-task combination (v i , t j), the system performs comprehensive weight calculation. During the weight calculation process, the system sets two key balance coefficients: a and β. In this embodiment, it is experimentally verified that the system achieves the best balance between distance optimization and resource matching when a = 0.4 and β = 0.6. The weight calculation formula is: When d ij = 0 (the vehicle is exactly at the task location), the system sets to a predetermined maximum value (10.0 in this embodiment) instead of infinity to avoid numerical overflow problems. The weight coefficient matrix of this embodiment is shown in Table 6. This embodiment combines the distance factor (the closer the better) and the resource matching degree (the better the match the better), with a = 0.4 and β = 0.6, which are experimentally verified to achieve the best balance between distance optimization and resource matching.

[0067] Table 6 Weight coefficient matrix

[0068] Vehicle / task Task T1 Task T2 Task T3 Vehicle V1 0.63 0.72 0.49 Vehicle V2 0.63 0.41 0.81 Vehicle V3 0.76 0.47 0.54 Vehicle V4 0.58 0.44 0.47

[0069] The system introduces a weight threshold θ for connection screening, which is determined to be 0.3 through offline simulation and actual operation data analysis. When the calculated weight coefficient w ij > θ, the system sets g ij = w ij in the adjacency matrix G to establish an effective connection relationship between the vehicle v i and the task t j ; when w ij ≤ θ, g ij = 0 is maintained, indicating that the vehicle-task pair does not participate in subsequent matching calculation, thereby reducing the calculation amount and avoiding unreasonable matching.

[0070] After completing the construction of the basic bipartite graph, the system further considers the time window constraint, which is a key link to ensure that the task is completed on time. The system designs a dedicated time constraint evaluation module responsible for constructing and processing the time window feasibility matrix T. This module first extracts the time window parameters of each task t j from the task database, including: the earliest start time TE j (the time point at which the task can start execution earliest); the latest completion time TL j (the deadline time point at which the task must be completed); the system uses a physical model-based time prediction method to calculate the estimated time TA ij for the vehicle to reach the task location from the current location: where: v is the standard vehicle speed of the empty rail system, which is set through the system parameter configuration file and is set to 2 meters / second in this embodiment; T delayTo schedule delay compensation time, for coping with system communication delay, vehicle start acceleration time and other uncertain factors, the embodiment is set to 10 seconds. The traditional air track scheduling system often processes time window constraints in post-verification after path planning, resulting in a large number of planned paths being discarded due to failure to meet the time window, causing waste of computing resources and low scheduling efficiency. The application excludes time-unfeasible matches before bipartite graph solving, fundamentally avoiding waste of computing resources.

[0071] Specifically, the system further considers the influence of track congestion factors on time prediction, introduces a dynamic congestion coefficient λ (0.8≤λ≤1.5), and automatically adjusts it during peak hours as follows: The time feasibility judgment adopts an accurate comparison algorithm: when TA ij ≤TL j , the system determines that the vehicle v i can arrive before the specified latest completion time of task t j , sets the time window feasibility matrix element t ij =1; otherwise, sets t ij =0, indicating that the matching is not feasible in time. The time window feasibility matrix of the embodiment is shown in Table Seven. The embodiment excludes time-unfeasible matches before bipartite graph solving, fundamentally avoiding waste of computing resources. The dynamic congestion coefficient λ is also introduced during peak hours to enhance the accuracy of time prediction.

[0072] Table Seven Time Window Feasibility Matrix

[0073]

[0074] The system modifies the adjacency matrix G based on the time window feasibility matrix T to generate an adjacency matrix G' considering time constraints: when t ij =0, the corresponding g ij is replaced with a preset maximum value G max (the embodiment sets it to 9999), ensuring that time-unfeasible matches are automatically excluded in subsequent optimization; when t ij =1, the original g ij value remains unchanged, retaining time-feasible matching candidates. This processing ensures that time-unfeasible matches are never selected during the execution of the Hungarian algorithm, while avoiding the complexity of modifying the optimization algorithm itself. In the embodiment, the modified adjacency matrix is shown in Table Eight. For example, d31=|56.8-110.5|+|98.4-75.2|+|0-0|=53.7+23.2+0=76.9m; TA31=76.9 / 2+10=48.45s (calculated value)→57.5s (considering congestion coefficient λ=1.2); TL1=80s>TA31=57.5s, time feasible→t31=1.

[0075] Table Eight Modified Adjacency Matrix

[0076]

[0077]

[0078] The system employs an improved Hungarian algorithm to solve the bipartite graph optimal matching problem. The algorithm module is designed to efficiently handle large-scale matching problems and is specifically optimized for the empty track scheduling scenario. First, the system performs row and column normalization on the adjacency matrix G' considering time constraints: row normalization: for each row of the matrix G', find the minimum non-infinite value min row,i from each element in the row, subtract min row,i from each element in the row, ensuring that each row contains at least one zero element; column normalization: for the row-normalized matrix, perform the same operation on each column, find the column minimum value and subtract it, ensuring that each column contains at least one zero element; after normalization, the system obtains the optimized matrix G", where the zero elements represent potential optimal matching positions.

[0079] The system then performs the covering line algorithm to cover all zero elements in the matrix with the minimum number of straight lines: initial marking: the system first marks the zero elements in the matrix G" with an initial marking, using the improved star-prime marking algorithm (Star-Prime method), ensuring that each row and column has at most one star-marked zero element; covering operation: the system covers all columns containing star-zero elements and performs an iterative covering process; covering line count: the system calculates the minimum number of horizontal and vertical lines sum lines required to cover all zero elements.

[0080] The system determines the relationship between the number of covering lines and the order of the matrix n, where n = min(m, n) is the effective order of the matrix: when sum lines = n, it indicates that n independent zero elements have been found, i.e., the optimal matching solution; when sum lines < n, further adjustment of the matrix G" is needed.

[0081] The matrix adjustment process uses the incremental method: the system scans the matrix G" to find the minimum value min val from all elements not covered by the covering lines; subtract min val from all uncovered elements to reduce the weight of the uncovered area; add min val to all elements covered by two lines to maintain numerical balance; the remaining elements (covered by one line) remain unchanged; after adjustment, the system returns to the covering line step for iterative processing until the optimal matching solution is found.

[0082] The final matching selection adopts a row-first scanning strategy: the system scans the final optimized matrix G", and for each star zero element position (i, j) in a row, the system establishes a vehicle v i and a task t j matching relationship. The system records the matching result in a structured matching table, which contains the following fields: match id matching record unique identifier; vehicle id assigned vehicle identifier; task id corresponding task identifier; match weight original matching weight; estimated arrival,time estimated arrival time; expected complete,time estimated completion time; matching timestamp matching generation timestamp; the system ensures atomic update of the matching table through a transaction processing mechanism to avoid data inconsistency caused by concurrent operations. The matching result is also cached in a high-speed memory area for subsequent path planning module to quickly access. The final vehicle-task matching result obtained in this embodiment is shown in Table 9.

[0083] Table 9 Vehicle-task matching result

[0084]

[0085] S5, path planning is performed on the vehicle-task matching relationship to obtain an optimal scheduling scheme of vehicle scheduling. In the implementation process of the present application, the system constructs an air rail network topology graph G(N, E) based on the vehicle-task matching relationship generated in step S4. Unlike traditional methods, the present system does not construct a complete topology graph of the entire air rail network, but constructs a directed topology graph for the determined matching relationship, thereby improving the calculation efficiency.

[0086] In specific implementation, the system first obtains a vehicle-task matching relationship set from step S4, for example

[0087] {(v1, t3), (v2, t5), (v4, t1)}. For each matching pair (v i , t j ), the system creates two special nodes in the topology graph: the current position node of the vehicle v i and the position node of the task t j . For example, if the vehicle v2 is currently located at the factory coordinates (35, 42) and the task t5 is located at the coordinates (78, 63), these two positions are taken as key nodes in the topology graph.

[0088] The system also connects these special nodes with the actual air track network. In practice, the system obtains the track connection information from a pre-stored air track system physical layout database. For each vehicle or task location, the system identifies the nearest track access point and establishes a temporary connection edge. For example, if the nearest track access point of v2 is node N 14 (40, 45), the system creates a virtual edge from v2 to N 14 (40, 45); similarly, if the nearest track access point of t5 is N 28 (75, 60), the system creates a virtual edge from N 28 (75, 60) to t5.

[0089] In addition, the system also allocates network resources according to the priority of the matching relationship. High-priority matching pairs are given priority of passage in the topology graph, which is achieved by introducing a priority coefficient in the edge weight calculation. For example, if the priority of (v1, t3) is higher than that of (v4, t1), the edge weight on the possible path of v1 will be reduced to a certain extent in the calculation, increasing the likelihood of these edges being selected.

[0090] Based on the special nodes and virtual connection edges created in the above steps, the system further constructs a complete topology graph G. The final topology graph G contains the following components: a comprehensive node set N: contains three types of nodes; the current location nodes of all matching vehicles, such as v1(15, 20), v2(35, 42), v4(50, 30); the location nodes of all matching tasks, such as t3(60, 25), t5(78, 63), t1(45, 80). The track nodes related to the above locations in the actual air track network, including intersection points, branch points and access points, such as N 14 (40, 45), N 22 (55, 45), N 23 (70, 50), N 28 (75, 60), etc. The air track network topology structure diagram of this embodiment is shown in Figure 5 .

[0091] A comprehensive edge set E: contains three types of edges; actual track edges: represent the actual existing track segments in the air track system, such as E 45 (N 14 → N 22 ), E 62 (N 22 → N 23 ); vehicle access edges: virtual edges from the current location of the vehicle to the nearest track access point, such as E v2 (v2 → N 14 ); task access edges: virtual edges from the nearest track access point to the task location, such as E t5 (N 28 → t5).

[0092] The system represents this topological graph using an adjacency matrix or an adjacency list. In actual implementation, considering that the aerial-rail network is usually sparse (i.e., there are fewer connections between nodes), the system prefers to use the adjacency list representation to save storage space. For example, for a node N 22 , its adjacency list entry can be: N 22 → {(N 14 , E 45 , 15m), (N 23 , E 62 , 20m), (N 19 , E 51 , 18m)} indicating that N 22 is connected to three nodes through edges E 45 , E 62 and E 51 , respectively, with corresponding rail distances of 15m, 20m and 18m.

[0093] To improve the efficiency of path planning, the system also constructs a spatial indexing structure, such as quadtree or R-tree, for the topological graph. This enables the system to quickly locate spatially adjacent nodes. For example, when finding the nearest rail access point for a vehicle v2, the system can use the spatial indexing structure to find N 14 in O(log n) time complexity without having to traverse all rail nodes.

[0094] In actual operation, the topological graph G is updated in real time to reflect changes in system state. For example, when a vehicle has moved from its initial position to a certain position on the rail, its position node will be updated in the topological graph; when a rail is temporarily closed for maintenance, the corresponding edge will be marked as unavailable or temporarily removed from the graph.

[0095] The final constructed topological graph G not only contains spatial and connection information, but also contains weight information for each edge. The initial weight is calculated by the formula w e = d e × (1 + γ × c e ) and adjusted according to the priority of the matching pair. For example, for path edges related to high-priority task t3, the system can apply a priority coefficient λ = 0.8 to multiply the original weight, making these edges more likely to be selected in path planning.

[0096] In traditional aerial-rail scheduling systems, path planning usually only considers physical distance factors, i.e., using a static weight calculation method, simply setting the edge weight as the actual rail distance d eThis method is simple and direct in calculation, but has obvious limitations: when multiple vehicles are assigned to the same track segment at the same time, local track congestion is easily caused due to the weight calculation not considering real-time load. The path planning of each vehicle is independent of each other and cannot perceive the planning results of other vehicles, resulting in unbalanced allocation of system resources.

[0097] Therefore, for each edge e E in the empty rail network, the system adopts a weight calculation formula that comprehensively considers physical distance and congestion condition: w e = d e × (1 + γ × c e ). In practical applications, the physical distance d e of the edge is usually obtained from the layout database of the empty rail system. For example, the actual length of track segment N 22 to N 23 is 15 meters, so d e = 15. The congestion influence factor γ is set according to the overall operation strategy of the system, and can be set to a larger value such as 1.8 during high-traffic periods and reduced to 0.5 during low-peak periods to reflect the sensitivity to congestion under different conditions. In this embodiment, the calculation results of the track congestion degree coefficient are shown in Table Ten, and the calculation results of the edge weight are shown in Table Eleven.

[0098] Table Ten Track Congestion Degree Coefficient

[0099]

[0100] Table Eleven Edge Weight Calculation Results

[0101]

[0102] The calculation of the track congestion degree coefficient c e uses the formula In the implementation process, the system tracks the usage of each edge in real time. For example, the maximum capacity u 14 of the key track segment N 22 to N e of a certain factory is set to 4 (indicating that this track segment allows a maximum of 4 vehicles to safely pass at the same time). If there are currently 2 vehicles assigned to use this track segment, then n e = 2, and the congestion degree coefficient c e = 2 / 4 = 0.5.

[0103] The system maintains a dynamic edge usage record table to record the current allocation of each edge, such as edge E 45 (N 14 → N 22 ): vehicles v1 and v3 have been assigned, the current n e = 2, u e = 4, and c e = 0.5; edge E62 (N 22 →N 23 ) : allocated vehicle v5, current n e = 1, u e = 3, c e = 0.33; these congestion data are updated in real-time with the execution of the vehicle scheduling scheme, ensuring that path planning is based on the latest network status.

[0104] For each vehicle-task matching pair (v i , t j ) determined in step S4, the system uses an improved Dijkstra algorithm to plan the shortest path from the vehicle's current location to the task location. This method is based on the traditional Dijkstra algorithm and has been innovatively improved, mainly for the dynamic multi-vehicle scheduling scenario specific to the air track system.

[0105] The traditional Dijkstra algorithm, as a classic single-source shortest path algorithm, performs well in static road network environments. Its core feature is that the edge weight remains unchanged during the algorithm execution process, and the path planning of each vehicle is executed independently and does not affect each other. In the traditional implementation, the algorithm is usually executed with the following steps: initialize the distance array; select the node with the smallest distance that has not been visited each time; update the distances of the neighbor nodes of this node; repeat until the target node is found.

[0106] However, this static path planning method has obvious shortcomings in highly dynamic multi-vehicle environments such as air track systems: multiple vehicles independently selecting "shortest paths" will inevitably lead to congestion on some popular routes, while other routes are idle, causing uneven allocation of system resources and overall efficiency decline. For example, in the factory material distribution scenario, if multiple vehicles need to go from the raw material warehouse to the production line at the same time, the traditional algorithm will cause all vehicles to concentrate on the same trunk road, forming a "virtual traffic jam" and thus prolonging the overall distribution time.

[0107] Unlike the traditional algorithm, the improved version of the present invention realizes two core innovations: dynamic weight adjustment within a single path planning, and global multi-path priority ranking confirmation. These two innovations enable the system to dynamically perceive and respond to changes in congestion conditions, achieving global resource optimization allocation.

[0108] In the specific embodiments of the present application, the system first initializes three key arrays: the distance array D (which records the shortest distance from the starting point to each node), the set of visited nodes S, and the predecessor node array P. For example, for the matching pair (v2, t5), the system sets the distance of the v2 location node to 0 and the distance of all other nodes to infinity. The path ranking and optimal scheduling scheme in this embodiment are shown in Table XII, and the final scheduling scheme is shown in Table XIII.

[0109] Table 12 Path Sequencing and Optimal Scheduling Scheme

[0110]

[0111] Table Thirteen

[0112]

[0113] During algorithm execution, the system selects the unvisited node u with the smallest distance value to add to set S in each iteration. For example, the first iteration will necessarily select the node at position v2 itself; subsequently, it may select the nearest track access point N. 14 And so on. For each selected node, the system checks all its neighboring nodes v. If the total distance to v via node u is less than the currently known distance to v, the distance value and the predecessor node record are updated.

[0114] The first key innovation of this invention lies in dynamically updating the congestion status. Unlike traditional algorithms that update the network state only after the path is determined, this algorithm updates the congestion status in real time during a single path calculation. Specifically, whenever it is determined that node v will be reached via node u (i.e., D[u]+w), the congestion status is updated in real time. uv When <D[v] is true), the system immediately updates the congestion coefficient of edge (u, v). For example, when edge E is determined... 45 (N 14 →N 22 When the vehicle v2 is used, the system calculates the updated congestion level: (Adding 1 new vehicle to the existing 2 vehicles), and adjusting the edge weights accordingly: w e =15×(1+1.5×0.75)=31.875.

[0115] This real-time dynamic adjustment mechanism has significant technical implications: even during the calculation of the same path, if multiple possible paths share a common edge, once the algorithm determines to use that edge to reach a node, the weight of that edge immediately increases. This may cause the algorithm to "change its mind" in subsequent iterations and choose an alternative path to reach the destination. This achieves adaptive congestion avoidance in bicycle path planning, which is fundamentally different from traditional algorithms.

[0116] The single-path planning algorithm ends when the endpoint node (task location) is added to set S. The system constructs a complete path by backtracking through the predecessor node array P. Specifically, starting from the endpoint, it finds the previous node of each node step by step through the predecessor node array until it backtracks to the starting point, and then reverses the path to obtain an ordered path sequence Pi from the starting point to the endpoint. At the same time, the distance value D [endpoint] corresponding to the endpoint is the total weight Wi of the path.

[0117] The second key innovation of the present application is the global sequencing and order confirmation mechanism. After completing the path planning of all matching pairs, the system does not simply let all vehicles execute their respective paths at the same time, but rather globally sequences the path set {P1, P2,..., Pn} according to the total weight. Paths with smaller weight values represent lower passing costs and are executed first. For example, if the weight of path P1 is 85.6, the weight of P2 is 104.2, and the weight of P3 is 92.8, the execution order after sequencing is P1→P3→P2.

[0118] This sequencing mechanism naturally implements differentiated processing of task priorities: vehicles matching urgent tasks generally obtain lower weights and are executed first; while ordinary tasks are executed in turn as long as system resources permit. In an actual application in a car manufacturing plant, this mechanism ensures that the supply of key production line materials is prioritized, reducing the response time for key material distribution from an average of 12 minutes to 4 minutes.

[0119] Based on the sequencing results, the system confirms the path assignments of each vehicle-task matching pair in turn, rather than simultaneously. This sequential confirmation mechanism is another technical innovation of the present application. For each confirmed path, the system immediately updates the global congestion state table, enabling subsequent path planning to avoid areas that will soon be congested. For example, after P1 is confirmed, the ne values of all edges passed by P1 are increased by 1, and the corresponding congestion coefficients ce and edge weights we are updated accordingly. In this way, even if P3 does not foresee potential conflicts with P1 during planning, it can still avoid the congested sections already occupied by P1 during actual execution.

[0120] Finally, the system generates a complete scheduling scheme, containing the following key information: the precise path sequence of each vehicle, such as the path of vehicle v2: v2 position→N14→N22→N23→N28→t5 position; the estimated passing time at each path point, accurate to the second; the specific operation to be performed after reaching the task location (such as loading or unloading specific materials); and the next instruction after task completion (such as returning to the standby area or executing the next task).

[0121] Through this path planning method based on the improved Dijkstra algorithm, the system can generate an efficient and robust vehicle scheduling scheme while considering resource limitations and congestion avoidance, significantly improving the intelligent level of scheduling and overall operating efficiency of the air rail system.

[0122] S6, scheduling air rail vehicles according to the optimal scheduling scheme. After obtaining the optimal scheduling scheme, the system converts the scheme into specific control instructions and issues them to each air rail vehicle. In specific implementation, the system first converts the path planning sequence of each vehicle into a series of path point coordinates and time points, forming the movement trajectory of the vehicle. Then, the system sends these instructions to the corresponding vehicle control unit through the communication network.

[0123] After receiving the dispatching instruction, the vehicle control unit controls the start, acceleration, deceleration, turning and stop of the vehicle according to the planned path and time arrangement, so as to realize the accurate movement of the vehicle from the current position to the task position. At the same time, the system continuously monitors the actual running state of the vehicle, and if the planned path is deviated or a sudden situation occurs, an emergency handling mechanism will be triggered, and the path planning will be re-executed if necessary.

[0124] When the vehicle reaches the task position, the corresponding material loading and unloading operation is performed, and after completing the task, it returns to the standby state or executes the next assigned task. In this way, the full-automatic scheduling control of the air track system is realized, and the system running efficiency is improved.

Claims

1. An automated air track scheduling method, characterized by, Comprising: S1, obtaining empty rail data, the empty rail data comprising a vehicle data set and a task data set; The vehicle data set comprises the current coordinates and the current load of each vehicle; The task data set comprises the material requirements and the task priority of each task; S2, calculating the remaining load of each vehicle according to the vehicle data set, and generating a vehicle set V to be allocated; S3, priority sorting the task data set, and generating a task set U; S4, generating a vehicle-task matching relationship according to the vehicle set V and the task set U; Wherein, generating a vehicle-task matching relationship comprises: According to the vehicle set V and the task set U, calculating the Manhattan distance between each available vehicle and each task to be executed; According to the vehicle remaining load and the material requirement, calculating the resource matching degree; According to the Manhattan distance and the resource matching degree, constructing a vehicle-task bipartite graph; Solving the vehicle-task bipartite graph to obtain the vehicle-task matching relationship; Wherein, constructing a vehicle-task bipartite graph comprises: Taking the vehicle set V and the task set U as the two vertex sets of the bipartite graph respectively; Calculate each vehicle With each task Between the weight coefficient , Calculated, where, The Manhattan distance, The resource matching degree, α and β are the preset balance coefficient; When the weight coefficient is greater than the threshold value θ, the vehicle establishes a connection edge between the task and the task, and sets the weight value of the corresponding edge as 1. The adjacency matrix G is constructed as a bipartite graph, where the matrix element is one, when there is a connecting edge is zero, when there is no connecting edge ； ​ Wherein, solving the vehicle-task bipartite graph to obtain the vehicle-task matching relationship comprises: (1) According to the adjacency matrix G, a time window matrix T is constructed, each element of the matrix T represents whether the corresponding task can be completed within the time window of the task ​​​ (2) According to the time window feasibility matrix T, the adjacent matrix G is modified to obtain ; wherein, when , the corresponding value is set to a preset maximum value , when , the original value is maintained unchanged; (3) normalizing the rows and columns of the matrix to obtain ; (4) According to Marking zero elements, covering all zero elements with the minimum number of horizontal and vertical lines, and calculating the total number of covering lines. If the total number of covering lines is greater than or equal to the matrix order n, the optimal matching scheme is found. If the number of covering lines is less than n, step (5) is performed. (5) In the matrix, get the minimum value among the elements not covered by the line , subtract from all elements not covered by the line , add to all elements covered by both lines , leave other elements unchanged, and return (4) to continue (6) the matrix obtained in step (5) The final zero element position in the matrix is selected, and different row and column zero elements are combined. According to the selected zero element position (i, j), the matching relationship between the vehicle and the task is established, and the vehicle-task matching relationship is obtained. The final zero element position in the matrix is selected, and different row and column zero elements are combined. According to the selected zero element position (i, j), the matching relationship between the vehicle and the task is established, and the vehicle-task matching relationship is obtained.​ S5, path planning of the vehicle-task matching relationship by improved Dijkstra algorithm to obtain the optimal scheduling scheme of vehicle scheduling; the improved Dijkstra algorithm dynamically adjusts the path weight by introducing the track congestion degree coefficient; S6, scheduling the empty rail vehicle according to the optimal scheduling scheme.

2. The automatic empty rail scheduling method according to claim 1, wherein: S2, generating a vehicle set V to be allocated comprises: Extracting the maximum load limit and the current load value of each vehicle from the vehicle data set; Calculating the remaining load by comparing the maximum load limit and the current load value; Judging whether the remaining load is greater than the minimum load threshold required by the task; Recording the identification number, current position coordinates and remaining load of the corresponding vehicle for the vehicle meeting the load requirement, and generating the vehicle set V.

3. The automatic empty rail scheduling method according to claim 2, wherein: Constructing a time window matrix T comprises: obtaining a task from a task dataset a time window parameter including an earliest start time and a latest end time ; According to Manhattan distance , calculate the time for the vehicle to move from the current location to the task location; When Time, representing a vehicle is able to reach within the time window of a task , otherwise representing a vehicle is unable to complete the task within the specified time .

4. The automatic empty rail scheduling method according to claim 3, wherein: S5, path planning of the vehicle-task matching relationship by improved Dijkstra algorithm comprises: According to the vehicle-task matching relationship, constructing an empty rail network topology graph G=(N, E), wherein N represents a node set, the node set comprising vehicle current position nodes and task position nodes; E represents an edge set, the edge representing the passable rail between nodes; For each edge e E in the empty rail network, calculate the initial weight , , is the actual rail distance corresponding to the edge e, is the rail congestion degree coefficient, and γ is the congestion influence factor; each vehicle-task match pair in the vehicle-task matching relationship a shortest path from the start point to the end point is generated by an improved Dijkstra algorithm and a corresponding total weight ; Shortest path set for all vehicle-task matching pairs Ordered according to total weight ;​ The vehicle-task matching pairs are confirmed in turn according to the weight-ordered results The path assignment is obtained, and an optimal scheduling scheme containing path planning sequence and task execution sequence is obtained.

5. The automatic empty rail scheduling method according to claim 4, wherein: Track congestion degree coefficient The calculation formula is: Wherein, The number of vehicles allocated on the current edge e, ue is the maximum capacity threshold of the edge e.

6. The automatic empty rail scheduling method according to claim 5, wherein: By means of an improved Dijkstra algorithm, the shortest path from the start point to the end point is generated and the corresponding total weight comprising: Initializing a distance array D, setting the distance of the starting point to 0 and the distance of other nodes to infinity; Initializing a visited node set S as an empty set; Initializing a predecessor node array P for recording the predecessor node of each node in the shortest path; Looping the following operations until the end point is added to the set S: i. Selecting the node u with the minimum distance value from the unvisited nodes to join the set S; ii. For each neighbor v of node u, if then update ; iii. updating the congestion degree coefficient of the relevant edge according to the current planned path , ; iv. dynamically adjusting the weights of the relevant edges, ; constructing the shortest path from the end point back to the start point according to the predecessor node array P ; total weight is equal to the value in the distance array D corresponding to the end point.

7. An automated air track dispatching system, comprising: at least one processing unit for executing instructions to implement the automated air track dispatching method of any one of claims 1 to 6. ​

Images