A hierarchical hybrid multi-objective scheduling method, system, equipment, and media for four-way shuttles.

Abstract

This invention discloses a hierarchical hybrid multi-objective scheduling method, system, device, and medium for four-way shuttles. The method includes: acquiring the status of the four-way shuttle, the status of the elevator, the inventory status, and the inbound / outbound task queue; performing a global optimal match between the tasks to be assigned in the inbound / outbound task queue and the four-way shuttle based on the four-way shuttle status, elevator status, inventory status, and the NSGA-II multi-objective optimization algorithm; and, upon receiving the matched task, performing scheduling based on spatiotemporal constraints. A The algorithm plans the path for executing the task; it continuously checks whether the next path segment meets the safety conditions. If not, it first waits; if the waiting time exceeds a threshold, it triggers an active backoff and replanning mechanism; if yes, it continues to execute the original task path; when the task is completed or the vehicle status changes, it releases expired nodes, edges, and elevator reservation resources, updates the status of the four-way shuttle, elevator, and inventory, and enters the next rolling scheduling cycle. This invention improves scheduling efficiency in multi-task concurrent scenarios.

Description

Technical Field

[0001] This invention relates to the field of intelligent warehousing and automated logistics scheduling technology, specifically to a hierarchical hybrid multi-objective scheduling method, system, equipment, and medium for four-way shuttle vehicles. Background Technology

[0002] With the development of high-density warehousing systems, four-way shuttles are widely used in multi-level warehousing, pallet storage, and bin retrieval scenarios because they can run in both longitudinal and transverse directions within a rack level and can achieve cross-level transportation via elevators. In such systems, multiple inbound and outbound tasks typically arrive concurrently, and multiple shuttles and elevators need to share limited tracks, connection points, and vertical layer-changing resources.

[0003] Existing scheduling methods typically fall into two categories: centralized global scheduling and local distributed scheduling. Centralized global scheduling can achieve better task allocation results at the macro level, but as the number of tasks, vehicles, and layers increases, the computational load of centralized planning becomes large, and it is difficult to respond quickly to local congestion and changes in equipment status. Local distributed scheduling offers flexible responses, but without unified spatiotemporal resource constraints, it is prone to problems such as node occupancy conflicts, opposing edge exchange conflicts, elevator connection conflicts, and close-range vehicle interference.

[0004] In terms of task allocation, simple first-come-first-served, nearest-vehicle-first, or single-objective shortest-path strategies struggle to simultaneously consider task completion time, energy consumption, vehicle load balancing, and conflict risks. Regarding path planning, traditional A / B methods... Algorithms typically only consider spatial distance or static obstacles, failing to express spatiotemporal constraints such as "a vehicle occupying a node or road segment at a certain time." At the execution level, if vehicles wait for each other on narrow tracks or at elevator docking points, local deadlocks can easily occur. For deeply obstructed goods in high-density storage, the lack of automatic relocation logic will also reduce the continuity and automation of outbound tasks.

[0005] Therefore, it is necessary to propose a four-way shuttle rolling scheduling method with multi-objective optimization as the core, spatiotemporal resource reservation as the constraint, and proactive retreat and replanning as the safety guarantee. Summary of the Invention

[0006] This invention addresses the shortcomings of existing technologies by providing a hierarchical hybrid multi-objective scheduling method, system, equipment, and medium for four-way shuttle vehicles. This solves the problems of insufficient multi-objective coordination, unpredictable track resource conflicts, easy congestion at elevator connections, and insufficient ability to resolve local deadlocks in existing multi-vehicle warehouse scheduling methods under multi-task concurrent scenarios.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A hierarchical hybrid multi-objective scheduling method for four-way shuttles, implemented on a multi-layer warehouse grid map model, includes the following steps: Obtain the status of the four-way shuttle, the status of the elevator, the inventory status, and the queue of inbound and outbound tasks to be executed; In the current rolling scheduling cycle, based on the status of the four-way shuttle, the elevator, and the inventory, the improved NSGA-II multi-objective optimization algorithm is used to perform a global optimal match between all the tasks to be assigned in the queue of inbound and outbound tasks to be executed and all the four-way shuttles that are currently in an idle state. After receiving the matching task, the four-way shuttle is based on the spatiotemporal constraints. A The algorithm plans the path for executing the task; During the execution of a mission, the four-way shuttle continuously checks whether the next path meets the safety conditions. If it does not, the four-way shuttle first waits. If the waiting time exceeds the threshold, the four-way shuttle triggers an active backoff and replanning mechanism. If it meets the requirements, it continues to execute the original mission path. When a task is completed or the vehicle status changes, expired nodes, edges, and hoist reservation resources are released, the status of the four-way shuttle, hoist, and inventory are updated, and the next rolling scheduling cycle begins.

[0009] To optimize the above technical solution, the specific measures also include: Furthermore, the multi-layer warehouse grid map model divides the warehouse space into layers, with each layer divided into several grids, establishing a set of grid nodes. Sum of edges Based on the functional attributes of the grid nodes, they are divided into track nodes, shelf nodes, elevator connection nodes, elevator resource nodes, and restricted access nodes; edge sets Used to describe the driving relationship between adjacent passable nodes; Mesh nodes in three-dimensional coordinates Perform indexing, where For the floor number, For horizontal grid numbering, For vertical grid numbering.

[0010] Furthermore, the improved NSGA-II multi-objective optimization algorithm is specifically as follows: The task-vehicle assignment chromosome is constructed using integer encoding, where the first chromosome is the first... The gene represents the first Each task is assigned to a four-way shuttle with a corresponding number; Randomly generate the initial population; For each task-vehicle assignment chromosome in the population, a physical constraint-based prediction and evaluation is performed. The prediction and evaluation process includes: based on the multi-layer warehouse grid map model, the kinematic parameters of the four-way shuttle, the layer-changing parameters of the elevator, the node time window reservation table, the edge time window reservation table, and the elevator time window reservation table, the path, waiting time, number of reversals, elevator usage time, energy consumption, and potential conflicts of each four-way shuttle when executing its assigned task are deduced, and the value of the multi-objective optimization function is calculated. The multi-objective optimization function includes at least the objectives of minimizing the maximum completion time, minimizing the total energy consumption, minimizing the vehicle load balancing, and minimizing the spatiotemporal conflict risk. Based on the value of the multi-objective optimization function for each individual, the individuals in the population are sorted in a non-dominated manner. The population is divided into multiple non-dominant classes based on dominance relationships, and crowding distance is calculated for individuals in the same class. By using tournament selection, crossover, mutation, and elite retention, we obtain the Pareto front solution set of the mission-vehicle allocation scheme. Based on the preset scheduling preference weights, a compromise solution is selected from the Pareto front solution set as the task-vehicle matching scheme for the current rolling scheduling cycle, and the task is assigned to the corresponding four-way shuttle.

[0011] Furthermore, the aforementioned spatiotemporal constraint-based A The algorithm plans the specific path for executing the task as follows: First, determine the starting node and target node according to the task type; for inbound tasks, the starting node is the elevator connection node or the inbound connection node, and the target node is the passable track node corresponding to the target cargo location; for outbound tasks, the starting node is the current node of the vehicle, and the target node is the passable track node or the elevator connection node corresponding to the target cargo location. Secondly, initialize the open list and the closed list, add the starting node to the open list, and calculate the comprehensive evaluation value of each node in the open list according to the comprehensive evaluation function; select the node with the smallest comprehensive evaluation value from the open list as the current expansion node; Next, reachability is determined for each of the adjacent nodes of the current extended node. The determination includes at least: whether the adjacent node is a passable track or a node that allows connection, whether it is in a restricted area, whether the vehicle needs to change direction when moving from the current node to the adjacent node, and whether the estimated travel time meets the vehicle speed and acceleration constraints. Then, based on the edge length from the current node to the adjacent node, the vehicle's maximum speed, maximum acceleration, and reversal time, calculate the estimated time when the vehicle enters the adjacent node, the estimated time when it leaves the adjacent node, and the estimated passage time window for passing through the current edge; based on the estimated passage time window, query the node time window reservation table and the edge time window reservation table. If the adjacent node has been reserved by other vehicles within the estimated entry time window, or if there is a conflict in the opposite edge exchange within the estimated passage time window, then increase the waiting time for the adjacent node or abandon the adjacent node. When the route involves cross-level transportation, the elevator time window reservation table is further queried. The availability of elevator resources is determined based on the estimated time of vehicle arrival at the elevator connection node, the start time of elevator service, and the end time of service. If the corresponding elevator service time window is occupied, the waiting cost is calculated and the candidate route is re-evaluated. Finally, when an adjacent node is not booked by other vehicles within the expected time window, and there is no conflict in the exchange of opposite edges on the passage from the current node to the candidate node, the adjacent node is added to the open list and its parent node is recorded. The node expansion process is repeated until the target node is selected or the open list is empty. If the target node is selected, the complete path is generated by backtracking according to the parent node pointer, and the path node sequence, edge passage sequence, node entry time, exit time, and elevator booking time window are output. If the open list is empty, the planning in this round is determined to be a failure and a waiting or replanning is triggered.

[0012] Furthermore, the safety conditions include: the next node is not occupied by other vehicles within the expected entry time window; there is no reciprocal exchange at the current edge; the center distance between the vehicle and other vehicles on the same level after entering the next node is not less than the preset minimum safety distance; and the hoist connection node is available within the corresponding time window.

[0013] Furthermore, the proactive retreat and replanning mechanism specifically refers to: The nearest available track nodes are obtained by breadth-first search, and are comprehensively scored according to the following factors: the length of the yield path, the degree of deviation from the target node, whether it is located on the main road, whether it is near the elevator connection point, and whether it is reserved by other vehicles. The nearest available track node with the lowest score is selected as the yield node. The pre-arranged retreat path is used; the four-way shuttle moves to the retreat node and re-establishes a spatiotemporal constraint based on its current position. A Algorithm execution path planning.

[0014] Furthermore, the method also includes: for outbound tasks, checking whether there are any obstructing goods between the target storage location and the nearest accessible track before the vehicle enters the picking action; the detection method is to scan grid by grid from the target storage location along at least one preset direction; if a shelf node with stored goods is encountered before reaching the accessible track, the shelf node is identified as an obstructing goods node; searching for reachable and vacant candidate storage locations on the same floor, and calculating the candidate storage location score. :

[0015] Where s represents the candidate storage location. To block the cargo Distance to the candidate storage location To move the candidate storage location to the original target storage location distance, For candidate storage location accessibility penalty items, , and All are weighting coefficients; Select the candidate storage location with the lowest score as the target storage location for the transfer, and control the four-way shuttle to first block the transfer of goods, and then continue to execute the original outbound task.

[0016] This invention also proposes a hierarchical hybrid multi-objective scheduling system for four-way shuttles, implemented based on a multi-layer warehouse grid map model, including: The data acquisition module is used to obtain the status of the four-way shuttle, the status of the elevator, the inventory status, and the queue of inbound and outbound tasks to be executed. The global rolling scheduling layer is used to perform a global optimal match between all the tasks to be assigned in the queue of inbound and outbound tasks and all the currently idle four-way shuttles, based on the status of the four-way shuttle, the status of the elevator, and the status of the inventory, using an improved NSGA-II multi-objective optimization algorithm in the current rolling scheduling cycle. The distributed execution layer is used by the four-way shuttle to execute tasks based on spatiotemporal constraints after receiving a matched task. A The algorithm plans the path for executing the task; The execution monitoring and safety replanning layer is used to continuously detect whether the next path meets the safety conditions during the execution of the four-way shuttle. If it does not meet the conditions, the vehicle first waits. If the waiting time exceeds the threshold, the vehicle triggers the active backoff replanning mechanism. If it meets the conditions, it continues to execute the original task path. When the task is completed or the vehicle status changes, expired nodes, edges and hoist reservation resources are released, the status of the four-way shuttle, hoist, and inventory are updated, and the next rolling scheduling cycle begins.

[0017] The present invention also proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the four-way shuttle hierarchical hybrid multi-objective scheduling method as described above.

[0018] The present invention also proposes a computer-readable storage medium storing a computer program that enables a computer to execute the four-way shuttle hierarchical hybrid multi-objective scheduling method described above.

[0019] The beneficial effects of this invention are as follows: First, by combining multi-objective task allocation with equipment physical parameters, path time windows, and elevator time windows, the task allocation results are more consistent with actual execution conditions. Second, by describing the resource occupancy relationship of multiple vehicles through three types of time window reservation tables (nodes, edges, and elevators), potential conflicts can be predicted before path execution. Third, by using the minimum safe distance between vehicles as a hard constraint at the execution layer, the weakening of safety constraints due to relying solely on optimization objectives is avoided. Fourth, by employing an active backoff and replanning mechanism when waiting timeouts or paths become impassable, the probability of local deadlocks can be reduced. Fifth, by using deep obstruction detection and automatic warehouse relocation processes, the continuity of high-density warehouse outbound tasks is improved. Attached Figure Description

[0020] Figure 1 This is an overall flowchart of the method of the present invention; Figure 2 This is a top view of the rasterized schematic diagram of the warehouse dynamic map of the present invention; Figure 3 This is a side view of a rasterized schematic diagram of the warehouse dynamic map of the present invention; Figure 4 This invention provides an improved flowchart for the multi-objective task allocation in NSGA-II. Figure 5 This is a flowchart of the proactive retreat and replanning mechanism of the present invention; Figure 6 This is a flowchart of the deep obstruction detection and automatic storage transfer process of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0022] Example 1 This invention proposes a hierarchical hybrid multi-objective scheduling method for four-way shuttles, the method flow is as follows: Figure 1As shown. This method is implemented in a multi-layer warehouse grid map model, which divides the warehouse space into layers, each layer into several grids, and establishes a set of grid nodes. Sum of edges Based on the functional attributes of the grid nodes, they are divided into track nodes, shelf nodes, elevator connection nodes, elevator resource nodes, and restricted access nodes; edge sets Used to describe the driving relationship between adjacent passable nodes; Mesh nodes in three-dimensional coordinates Perform indexing, where For the floor number, For horizontal grid numbering, For vertical grid numbering.

[0023] Top and side views of the rasterized schematic diagram of the warehouse dynamic map, as shown below. Figure 2 and Figure 3 As shown.

[0024] The method includes the following steps: S1. Obtain the status of the four-way shuttle, the status of the elevator, the status of the inventory, and the queue of inbound and outbound tasks to be executed; the status of the four-way shuttle includes at least the floor number, grid coordinates, busy / idle status, load status, maximum speed, maximum acceleration, and current task status. S2. Establish a spatiotemporal resource table for multi-vehicle collaborative scheduling. The spatiotemporal resource table shall include at least a node time window reservation table, a side time window reservation table, and a hoist time window reservation table. Among them, the node time window reservation form Record the time interval during which a vehicle occupies a grid node:

[0025] in, It is a grid node Node time window reservation table, For vehicle identification, For the expected entry time, For the expected departure time, This is the corresponding task identifier.

[0026] Side Time Window Appointment Form Record the time interval for vehicles to travel between adjacent nodes on road segments:

[0027] In the formula, It is the edge time window reservation table for the edge between node u and node v. For vehicle identification, For the expected entry time, For the expected departure time, The duration for which the edge can be traversed.

[0028] Hoist time window reservation form Record the service time range of the hoisting machine resources:

[0029] in, This is the hoist time window reservation form for the hth hoist. and These are the initial layer and the target layer, respectively. and These represent the start and end times of the hoist service, respectively.

[0030] S3. In the current rolling scheduling cycle, based on the status of the four-way shuttle, the elevator, and the inventory, the improved NSGA-II multi-objective optimization algorithm is used to perform a global optimal match between all the tasks to be assigned in the queue of inbound and outbound tasks to be executed and all the four-way shuttles that are currently in an idle state. The improved NSGA-II multi-objective optimization algorithm is as follows: The task-vehicle chromosome assignment is constructed using integer encoding, and the number of tasks is [number missing]. The number of candidate vehicles is Among them, the first chromosome The gene represents the first Each task is assigned to a four-way shuttle with a corresponding number; Task-Vehicle Assignment Chromosome Expressed mathematically as follows:

[0031] in, It is the first One gene, The value of is the sequence number of the candidate vehicle, indicating the first . The task is assigned to the first A number of candidate vehicles.

[0032] Randomly generate the initial population; For each task-vehicle assignment chromosome in the population, a physical constraint-based prediction and evaluation is performed. The prediction and evaluation process includes: based on the multi-layer warehouse grid map model, the kinematic parameters of the four-way shuttle, the layer-changing parameters of the elevator, the node time window reservation table, the edge time window reservation table, and the elevator time window reservation table, the path, waiting time, number of reversals, elevator usage time, energy consumption, and potential conflicts of each four-way shuttle when executing its assigned task are deduced, and the value of the multi-objective optimization function is calculated. The multi-objective optimization function includes at least the objectives of minimizing the maximum completion time, minimizing the total energy consumption, minimizing the vehicle load balancing, and minimizing the spatiotemporal conflict risk. The maximum completion time refers to the completion time of the vehicle with the longest completion time among all vehicles, expressed by the formula:

[0033] In the formula, It is the maximum completion time, and the candidate vehicle set is... Assigned to vehicles The task set is , For vehicles Execute the task The predicted completion time. Including route travel time Additional time for reversing Picking and placing goods action time Waiting time Hoist connection time and the time for changing floors by the hoist The formula is expressed as:

[0034] In the formula, For the number of floors, For the number of reversals, It is the time required for a single reversal.

[0035] Total energy consumption The expression is as follows:

[0036] In the formula, The energy consumption coefficient per unit of unloaded driving. This refers to the distance the vehicle traveled without a load. The energy consumption coefficient per unit of cargo travel. For the distance traveled while carrying goods, The energy consumption coefficient for a single commutation is... For the number of reversals, This refers to the energy consumption coefficient for changing floors using a single-level hoist. For the number of floors, The unit waiting energy consumption coefficient, For waiting time.

[0037] Vehicle load The expression is as follows:

[0038] in, For vehicles The number of tasks assigned This represents the average number of tasks. For vehicles Predicted working hours To predict the average working time, and These are the weighting coefficients.

[0039] Spacetime conflict risk The expression is as follows:

[0040] in, The number of node time window conflicts. The number of collisions for exchanging opposite edges. To increase the number of time window overlaps for the booster, Penalties for congestion on main roads The minimum safe distance penalty item, , , , and These are the corresponding penalty weights.

[0041] Based on the value of each individual's multi-objective optimization function, the individuals in the population are non-dominated and ranked; if the individual Not inferior to individuals in all objectives And superior to individuals in at least one objective. Then the individual Dominant Individual .

[0042] The population is divided into multiple non-dominant classes based on dominance relationships, and crowding distance is calculated for individuals in the same class. By using tournament selection, crossover, mutation, and elite retention, we obtain the Pareto front solution set of the mission-vehicle allocation scheme. Based on the preset scheduling preference weights, a compromise solution is selected from the Pareto front solution set as the task-vehicle matching scheme for the current rolling scheduling cycle, and the task is assigned to the corresponding four-way shuttle.

[0043] Let the first The normalization target value for each individual is The compromise solution can be chosen according to the following formula:

[0044] in, For the first The preference weights of each objective, satisfying:

[0045] A compromise solution is selected based on actual operational preferences. For example, when the warehousing system is in peak periods, the weights of maximum completion time and spatiotemporal conflict risk can be increased; when the warehousing system is in off-peak periods or energy-saving operation mode, the weights of total energy consumption and vehicle load balancing can be increased. The selected compromise solution is the task-vehicle matching scheme for the current rolling cycle.

[0046] S4. After receiving the matched task, the four-way shuttle is based on the spatiotemporal constraints. A The algorithm plans the path for executing the task; and updates the node time window reservation table and the edge time window reservation table according to the expected arrival time and departure time respectively; when the path planning involves cross-level transportation, the time window reservation table of the corresponding hoist is updated.

[0047] Based on spatiotemporal constraints A The algorithm plans the specific path for executing the task as follows: First, determine the starting node and target node according to the task type; for inbound tasks, the starting node is the elevator connection node or the inbound connection node, and the target node is the passable track node corresponding to the target cargo location; for outbound tasks, the starting node is the current node of the vehicle, and the target node is the passable track node or the elevator connection node corresponding to the target cargo location. Secondly, initialize the open list and the closed list, add the starting node to the open list, and calculate the comprehensive evaluation value of each node in the open list according to the comprehensive evaluation function; select the node with the smallest comprehensive evaluation value from the open list as the current expansion node; The comprehensive evaluation function is expressed as follows:

[0048] in, From the starting node to the node The cumulative path cost is obtained by adding up the travel time, reversal time, and waiting time of each road segment already traversed; For nodes The heuristic distance to the target node can be calculated from the Manhattan distance or the grid distance; This refers to the waiting cost incurred by candidate nodes due to the time windows of nodes, edges, or lifts being occupied. Penalties for scheduling conflicts; Penalties for congestion on main roads; , , These are the corresponding weighting coefficients. The above comprehensive evaluation function is based on the traditional A... cumulative path cost of the algorithm and heuristic distance Based on this, factors such as waiting, appointment conflicts, and main road congestion are added to obtain a solution that ensures the path search results simultaneously meet the requirements of shortest distance and availability of time and space resources.

[0049] The reservation conflict penalty can be calculated by weighting node conflicts, edge conflicts, and booster conflicts as follows:

[0050] in, , and These represent the conflict indicators for node time windows, edge time windows, and hoist time windows, respectively; a value of 1 is used when the corresponding resource experiences a conflict within the expected time window, and a value of 0 otherwise. , and These are the penalty weights for node conflicts, edge conflicts, and booster conflicts, respectively.

[0051] Next, reachability is determined for each of the adjacent nodes of the current extended node. The determination includes at least: whether the adjacent node is a passable track or a node that allows connection, whether it is in a restricted area, whether the vehicle needs to change direction when moving from the current node to the adjacent node, and whether the estimated travel time meets the vehicle speed and acceleration constraints. Then, based on the edge length from the current node to the adjacent node, the vehicle's maximum speed, maximum acceleration, and reversal time, calculate the estimated time when the vehicle enters the adjacent node, the estimated time when it leaves the adjacent node, and the estimated passage time window for passing through the current edge; based on the estimated passage time window, query the node time window reservation table and the edge time window reservation table. If the adjacent node has been reserved by other vehicles within the estimated entry time window, or if there is a conflict in the opposite edge exchange within the estimated passage time window, then increase the waiting time for the adjacent node or abandon the adjacent node. Among them, the current node to adjacent nodes The travel time can be estimated based on the side length, maximum vehicle speed, maximum acceleration, and reversing time:

[0052] in, For nodes With nodes The side lengths between and vehicles Maximum speed and maximum acceleration This is a reversing indication; when a vehicle moves from a node... Move to node Use 1 when a reversal is needed, otherwise use 0. This represents the time required for a single turnaround. This formula is used to estimate travel time based on the vehicle's physical movement capabilities, and can also be replaced by actual vehicle speed curves or control system feedback data.

[0053] The node occupancy time window and edge passage time window mentioned above are respectively expressed as follows:

[0054]

[0055] in, Indicates adjacent nodes The node occupancy time window, Indicates that the vehicle starts from the current node. Travel to adjacent nodes The time window for edge passage; and These are the expected entry and exit points for the vehicle. At that moment, and These are the expected start and end points of the vehicle's passage. The time window mentioned above is determined by the vehicle's current departure time and travel time. The minimum time occupied by the node is determined together and used for subsequent queries of the node time window reservation table. Side Time Window Appointment Form .

[0056] When node time windows overlap or opposite edge travel time windows overlap, it is determined that a reservation conflict exists; the corresponding conflict condition can be expressed as:

[0057]

[0058] Among them, superscript This indicates the corresponding time window that other vehicles have already booked; Indicates other vehicles Reserved nodes Occupying time window, Indicates other vehicles Passing through the edge in the opposite direction The system defines the time windows for edge travel. When the intersection is not empty, it indicates that the corresponding nodes or road segments have overlapping reservation times. If the node time windows overlap, it means that two vehicles may occupy the same node simultaneously. If the time windows for travel on opposite edges overlap, it means that two vehicles may meet head-on on the same road segment. For candidate nodes with conflicts, the system prioritizes increasing the waiting time and recalculating the time window; if the constraints still cannot be met after waiting, the candidate node is abandoned.

[0059] When the route involves cross-level transportation, the elevator time window reservation table is further queried. The availability of elevator resources is determined based on the estimated time of vehicle arrival at the elevator connection node, the start time of elevator service, and the end time of service. If the corresponding elevator service time window is occupied, the waiting cost is calculated and the candidate route is re-evaluated. The service time window of the hoist can be represented as:

[0060] in, Indicates hoist Service hours window and These are the service start time and the service end time, respectively. The estimated time required for this hoist service. The service start time is the estimated time when the vehicle arrives at the hoist connection point. Current availability of the hoist Jointly determined:

[0061] The expression for the time required for hoist service is as follows:

[0062] in, The time required for the hoist to align with the floor where the vehicle is located. The time it takes for a vehicle to enter the hoist. For the single-level hoist running time, The time it takes for the vehicle to leave the hoist. This is for connection gate control or safety confirmation time. The start and end times of the elevator service are determined based on the elevator's service hours, and both are entered into the elevator time window reservation table. This can reduce congestion caused by multiple vehicles competing for the same elevator simultaneously.

[0063] Finally, when an adjacent node is not booked by other vehicles within the expected time window, and there is no conflict in the exchange of opposite edges on the passage from the current node to the candidate node, the adjacent node is added to the open list and its parent node is recorded. The node expansion process is repeated until the target node is selected or the open list is empty. If the target node is selected, the complete path is generated by backtracking according to the parent node pointer, and the path node sequence, edge passage sequence, node entry time, exit time, and elevator booking time window are output. If the open list is empty, the planning in this round is determined to be a failure and a waiting or replanning is triggered.

[0064] Once the planning is successful, the estimated entry and exit times for each node in the path are written into the node time window reservation table, and the travel time between adjacent nodes is written into the edge time window reservation table. If the task requires cross-level transportation, the arrival time of the elevator at the transfer point, the vehicle-elevator synchronous waiting time, and the floor-changing time are also written into the elevator time window reservation table. Through the above methods, subsequent vehicles can perceive the resource occupancy status in advance during path planning.

[0065] S5. During the execution of the mission, the four-way shuttle continuously checks whether the next path meets the safety conditions. If it does not meet the conditions, the four-way shuttle first waits. If the waiting time exceeds the threshold, the four-way shuttle triggers the active retreat and replanning mechanism. If the conditions are met, the original mission path is continued.

[0066] Safety conditions include: the next node is not occupied by other vehicles within the expected entry time window; there is no reciprocal exchange at the current edge; the center distance between the vehicle and other vehicles on the same level after entering the next node is not less than the preset minimum safety distance; and the hoist connection node is available within the corresponding time window.

[0067] The mechanism for proactively yielding and replanning is as follows: The nearest available track nodes are obtained by breadth-first search, and are comprehensively scored according to the following factors: the length of the yield path, the degree of deviation from the target node, whether it is located on the main road, whether it is near the elevator connection point, and whether it is reserved by other vehicles. The nearest available track node with the lowest score is selected as the yield node. Candidate yield node The scoring function is:

[0068] in, This represents the distance from the vehicle's current location to the candidate yield point. This represents the distance from the candidate yield node to the original target node. Penalties for main roads Penalty item for proximity to hoist docking port, For scheduling conflict penalties, to These are the weighting coefficients.

[0069] The pre-arranged retreat path is used; the four-way shuttle moves to the retreat node and re-establishes a spatiotemporal constraint based on its current position. A Algorithm execution path planning.

[0070] For outbound tasks, before the vehicle enters the pickup area, it checks whether there are any obstructions between the target storage location and the nearest accessible track. The detection method is to scan the grid one by one from the target storage location along at least one preset direction. If a shelf node with stored goods is encountered before reaching the accessible track, the shelf node is identified as an obstruction node. Within the same floor, accessible and vacant candidate storage locations are searched, and a score is calculated for each candidate storage location. :

[0071] Where s represents the candidate storage location. To block the cargo Distance to the candidate storage location To move the candidate storage location to the original target storage location distance, For candidate storage location accessibility penalty items, , and All are weighting coefficients; Select the candidate storage location with the lowest score as the target storage location for the transfer, and control the four-way shuttle to first block the transfer of goods, and then continue to execute the original outbound task.

[0072] S6. When a task is completed or the vehicle status changes, release expired nodes, edges, and elevator reservation resources, update the status of the four-way shuttle, elevator, and inventory, and enter the next rolling scheduling cycle. Compared with one-time static allocation, rolling scheduling can improve the system's adaptability to dynamic environmental changes while maintaining global multi-objective optimization capabilities.

[0073] Example 2 This invention proposes a hierarchical hybrid multi-objective scheduling system for four-way shuttles, corresponding to the method in Embodiment 1, implemented based on a multi-layer warehouse grid map model, comprising: The data acquisition module is used to obtain the status of the four-way shuttle, the status of the elevator, the inventory status, and the queue of inbound and outbound tasks to be executed. The global rolling scheduling layer is used to perform a global optimal match between all the tasks to be assigned in the queue of inbound and outbound tasks and all the currently idle four-way shuttles, based on the status of the four-way shuttle, the status of the elevator, and the status of the inventory, using an improved NSGA-II multi-objective optimization algorithm in the current rolling scheduling cycle. The distributed execution layer is used by the four-way shuttle to execute tasks based on spatiotemporal constraints after receiving a matched task. A The algorithm plans the path for executing the task; The execution monitoring and safety replanning layer is used to continuously detect whether the next path meets the safety conditions during the execution of the four-way shuttle. If it does not meet the conditions, the vehicle first waits. If the waiting time exceeds the threshold, the vehicle triggers the active backoff replanning mechanism. If it meets the conditions, it continues to execute the original task path. When the task is completed or the vehicle status changes, expired nodes, edges and hoist reservation resources are released, the status of the four-way shuttle, hoist, and inventory are updated, and the next rolling scheduling cycle begins.

[0074] The implementation methods of each module and its function in the system are completely consistent with the steps of the method in Implementation Example 1, so they will not be repeated here.

[0075] Example 3 This invention proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the four-way shuttle hierarchical hybrid multi-objective scheduling method as described in Embodiment 1.

[0076] Example 4 This invention proposes a computer-readable storage medium storing a computer program that enables a computer to execute the four-way shuttle hierarchical hybrid multi-objective scheduling method as described in Embodiment 1.

[0077] In the embodiments disclosed in this application, a computer storage medium may be a tangible medium that may contain or store programs for use by or in conjunction with an instruction execution system, apparatus, or device. The computer storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of computer storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, and portable compact disc read-only memory (CD). ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0078] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0079] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A hierarchical hybrid multi-objective scheduling method for four-way shuttles, characterized in that, The implementation on a multi-layered warehouse grid map model includes the following steps: Obtain the status of the four-way shuttle, the status of the elevator, the inventory status, and the queue of inbound and outbound tasks to be executed; In the current rolling scheduling cycle, based on the status of the four-way shuttle, the elevator, and the inventory, the improved NSGA-II multi-objective optimization algorithm is used to perform a global optimal match between all the tasks to be assigned in the queue of inbound and outbound tasks to be executed and all the four-way shuttles that are currently in an idle state. After receiving the matching task, the four-way shuttle is based on the spatiotemporal constraints. A The algorithm plans the path for executing the task; During the execution of a mission, the four-way shuttle continuously checks whether the next path meets the safety conditions. If it does not, the four-way shuttle first waits. If the waiting time exceeds the threshold, the four-way shuttle triggers an active backoff and replanning mechanism. If it meets the requirements, it continues to execute the original mission path. When a task is completed or the vehicle status changes, expired nodes, edges, and hoist reservation resources are released, the status of the four-way shuttle, hoist, and inventory are updated, and the next rolling scheduling cycle begins.

2. The four-way shuttle hierarchical hybrid multi-objective scheduling method as described in claim 1, characterized in that, The multi-layer warehouse grid map model divides the warehouse space into layers, each layer into several grids, and establishes a set of grid nodes. Sum of edges Based on the functional attributes of the grid nodes, they are divided into track nodes, shelf nodes, elevator connection nodes, elevator resource nodes, and restricted access nodes; edge sets Used to describe the driving relationship between adjacent passable nodes; Mesh nodes in three-dimensional coordinates Perform indexing, where For the floor number, For horizontal grid numbering, For vertical grid numbering.

3. The four-way shuttle hierarchical hybrid multi-objective scheduling method as described in claim 1, characterized in that, The improved NSGA-II multi-objective optimization algorithm is as follows: The task-vehicle assignment chromosome is constructed using integer encoding, where the first chromosome is the first... The gene represents the first Each task is assigned to a four-way shuttle with a corresponding number; Randomly generate the initial population; For each task-vehicle assignment chromosome in the population, a physical constraint-based prediction and evaluation is performed. The prediction and evaluation process includes: based on the multi-layer warehouse grid map model, the kinematic parameters of the four-way shuttle, the layer-changing parameters of the elevator, the node time window reservation table, the edge time window reservation table, and the elevator time window reservation table, the path, waiting time, number of reversals, elevator usage time, energy consumption, and potential conflicts of each four-way shuttle when executing its assigned task are deduced, and the value of the multi-objective optimization function is calculated. The multi-objective optimization function includes at least the objectives of minimizing the maximum completion time, minimizing the total energy consumption, minimizing the vehicle load balancing, and minimizing the spatiotemporal conflict risk. Based on the value of the multi-objective optimization function for each individual, the individuals in the population are sorted in a non-dominated manner. The population is divided into multiple non-dominant classes based on dominance relationships, and crowding distance is calculated for individuals in the same class. By using tournament selection, crossover, mutation, and elite retention, we obtain the Pareto front solution set of the mission-vehicle allocation scheme. Based on the preset scheduling preference weights, a compromise solution is selected from the Pareto front solution set as the task-vehicle matching scheme for the current rolling scheduling cycle, and the task is assigned to the corresponding four-way shuttle.

4. The four-way shuttle hierarchical hybrid multi-objective scheduling method as described in claim 1, characterized in that, The spatiotemporal constraint-based A The algorithm plans the specific path for executing the task as follows: First, determine the starting node and target node according to the task type; for inbound tasks, the starting node is the elevator connection node or the inbound connection node, and the target node is the passable track node corresponding to the target cargo location; for outbound tasks, the starting node is the current node of the vehicle, and the target node is the passable track node or the elevator connection node corresponding to the target cargo location. Secondly, initialize the open list and the closed list, add the starting node to the open list, and calculate the comprehensive evaluation value of each node in the open list according to the comprehensive evaluation function; select the node with the smallest comprehensive evaluation value from the open list as the current expansion node; Next, reachability is determined for each of the adjacent nodes of the current extended node. The determination includes at least: whether the adjacent node is a passable track or a node that allows connection, whether it is in a restricted area, whether the vehicle needs to change direction when moving from the current node to the adjacent node, and whether the estimated travel time meets the vehicle speed and acceleration constraints. Then, based on the edge length from the current node to the adjacent node, the vehicle's maximum speed, maximum acceleration, and reversal time, calculate the estimated time when the vehicle enters the adjacent node, the estimated time when it leaves the adjacent node, and the estimated passage time window for passing through the current edge; based on the estimated passage time window, query the node time window reservation table and the edge time window reservation table. If the adjacent node has been reserved by other vehicles within the estimated entry time window, or if there is a conflict in the opposite edge exchange within the estimated passage time window, then increase the waiting time for the adjacent node or abandon the adjacent node. When the route involves cross-level transportation, the elevator time window reservation table is further queried. The availability of elevator resources is determined based on the estimated time of vehicle arrival at the elevator connection node, the start time of elevator service, and the end time of service. If the corresponding elevator service time window is occupied, the waiting cost is calculated and the candidate route is re-evaluated. Finally, when an adjacent node is not booked by other vehicles within the expected time window, and there is no conflict in the exchange of opposite edges on the passage from the current node to the candidate node, the adjacent node is added to the open list and its parent node is recorded. The node expansion process is repeated until the target node is selected or the open list is empty. If the target node is selected, the complete path is generated by backtracking according to the parent node pointer, and the path node sequence, edge passage sequence, node entry time, exit time, and elevator booking time window are output. If the open list is empty, the planning in this round is determined to be a failure and a waiting or replanning is triggered.

5. The four-way shuttle hierarchical hybrid multi-objective scheduling method as described in claim 1, characterized in that, The safety conditions include: the next node is not occupied by other vehicles within the expected entry time window; there is no opposite exchange at the current edge; after the vehicle enters the next node, the center distance between the vehicle and other vehicles on the same level is not less than the preset minimum safety distance; and the hoist connection node is available within the corresponding time window.

6. The four-way shuttle hierarchical hybrid multi-objective scheduling method as described in claim 1, characterized in that, The proactive retreat and replanning mechanism is specifically as follows: The nearest available track nodes are obtained by breadth-first search, and are comprehensively scored according to the following factors: the length of the yield path, the degree of deviation from the target node, whether it is located on the main road, whether it is near the elevator connection point, and whether it is reserved by other vehicles. The nearest available track node with the lowest score is selected as the yield node. The pre-arranged retreat path is used; the four-way shuttle moves to the retreat node and re-establishes a spatiotemporal constraint based on its current position. A Algorithm execution path planning.

7. The hierarchical hybrid multi-objective scheduling method for four-way shuttles as described in claim 1, characterized in that, The method further includes: for outbound tasks, checking whether there are any obstructing goods between the target storage location and the nearest accessible track before the vehicle enters the picking action; the detection method is to scan grid by grid from the target storage location along at least one preset direction; if a shelf node with stored goods is encountered before reaching the accessible track, the shelf node is identified as an obstructing goods node; searching for reachable and vacant candidate storage locations on the same floor, and calculating the candidate storage location score. : Where s represents the candidate storage location. To block the cargo Distance to the candidate storage location To move the candidate storage location to the original target storage location distance, For candidate storage location accessibility penalty items, , and All are weighting coefficients; Select the candidate storage location with the lowest score as the target storage location for the transfer, and control the four-way shuttle to first block the transfer of goods, and then continue to execute the original outbound task.

8. A hierarchical hybrid multi-objective scheduling system for four-way shuttles, characterized in that, Implemented based on a multi-layer warehouse grid map model, including: The data acquisition module is used to obtain the status of the four-way shuttle, the status of the elevator, the inventory status, and the queue of inbound and outbound tasks to be executed. The global rolling scheduling layer is used to perform a global optimal match between all the tasks to be assigned in the queue of inbound and outbound tasks and all the currently idle four-way shuttles, based on the status of the four-way shuttle, the status of the elevator, and the status of the inventory, using an improved NSGA-II multi-objective optimization algorithm in the current rolling scheduling cycle. The distributed execution layer is used by the four-way shuttle to execute tasks based on spatiotemporal constraints after receiving a matched task. A The algorithm plans the path for executing the task; The execution monitoring and safety replanning layer is used to continuously detect whether the next path meets the safety conditions during the execution of the four-way shuttle. If it does not meet the conditions, the vehicle first waits. If the waiting time exceeds the threshold, the vehicle triggers the active backoff replanning mechanism. If it meets the conditions, it continues to execute the original task path. When the task is completed or the vehicle status changes, expired nodes, edges and hoist reservation resources are released, the status of the four-way shuttle, hoist, and inventory are updated, and the next rolling scheduling cycle begins.

9. An electronic device, characterized in that, include: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the hierarchical hybrid multi-objective scheduling method for four-way shuttles as described in any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, characterized in that, The computer program causes the computer to execute the four-way shuttle hierarchical hybrid multi-objective scheduling method as described in any one of claims 1-7.

Images