Cross-layer bidirectional connection system for automatic stereoscopic warehouse

Abstract

The invention provides a cross-layer two-way connection system for an automatic stereoscopic warehouse. The cross-layer two-way connection system comprises a first elevator, a second elevator, a connection conveying platform and a dispatching control module. The connection conveying platform comprises a first connection conveying area and a fourth connection conveying area which are arranged in a bottom storeroom, and a second connection conveying area and a third connection conveying area which are arranged in an upper storeroom; the first elevator is used for lifting materials from the bottom layer to the upper layer; the second elevator is used for conveying the materials from the upper layer to the bottom layer; the scheduling control module is used for acquiring multi-source data of the first elevator, the second elevator and the connection conveying platform, and realizing cooperative scheduling of the first elevator, the second elevator and the connection conveying platform based on the multi-source data and a dynamic optimization strategy; wherein the multi-source data at least comprises the task feature data, the equipment state data and the transmission environment sensing data, and the efficiency and the reliability of the cross-layer bidirectional connection system of the automatic stereoscopic warehouse are improved.

Description

Technical Field

[0001] This invention relates to the field of warehousing and logistics, and in particular to a cross-level bidirectional connection system for automated storage and retrieval systems (AS / RS). Background Technology

[0002] With rising land costs and increasing demands for logistics efficiency, the adoption of multi-level automated warehouses has become a trend in modern warehousing. The rapid flow of materials between different floors (such as the outbound preparation area on the first floor and the storage / assembly area on the second floor) is crucial to ensuring overall operational efficiency.

[0003] Currently, most multi-story warehouses use a single vertical lift for cross-level transportation. This model has inherent and insurmountable drawbacks: Efficiency bottleneck: A single hoist can only perform a task in one direction (up or down) at a time. When there are concurrent tasks, the hoist must move back and forth between different floors, wasting a lot of time on idle travel, waiting and alignment. The overall utilization rate of the equipment is low, and it can easily become the bottleneck of the entire system during peak inbound and outbound periods.

[0004] Task conflicts and waiting: If there is material on the upper level that needs to be moved down, and at the same time there is material on the lower level that needs to be moved up, a single elevator system will inevitably cause one task to wait, increasing the material flow cycle time.

[0005] Poor system flexibility: High risk of single point of failure. Once the hoist fails, the entire cross-level logistics will be paralyzed.

[0006] Integration and transformation are difficult: When upgrading existing warehouses to be automated, the new cross-level systems often require large-scale and disruptive modifications to the original conveyor layout, which results in long construction periods, disruption to normal operations, and high costs.

[0007] To address the aforementioned problems, this invention provides a cross-level bidirectional connection system for automated storage and retrieval systems (AS / RS) to solve at least one of the problems mentioned above. Summary of the Invention

[0008] In order to solve the problems existing in the prior art, this invention innovatively proposes a cross-layer bidirectional connection system for automated storage and retrieval systems (AS / RS), which effectively solves the problem of low efficiency and reliability of cross-layer bidirectional connection systems for AS / RS caused by the prior art, and effectively improves the efficiency and reliability of cross-layer bidirectional connection systems for AS / RS.

[0009] The first aspect of this invention provides a cross-level bidirectional transfer system for an automated storage and retrieval system (AS / RS), comprising: a first elevator, a second elevator, a transfer conveying platform, and a scheduling and control module. The transfer conveying platform includes a first transfer conveying area and a fourth transfer conveying area located in the bottom-level warehouse, and a second transfer conveying area and a third transfer conveying area located in the upper-level warehouse. The bottom-level inlet / outlet of the first elevator connects to the first transfer conveying area of ​​the bottom-level warehouse, and the upper-level outlet / inlet of the first elevator connects to the third transfer conveying area of ​​the upper-level warehouse, for lifting materials from the bottom level to the upper level. The upper level of the second elevator... The inlet / outlet connects to the fourth connecting conveyor area of ​​the upper warehouse, and the bottom outlet / inlet of the second elevator connects to the second connecting conveyor area of ​​the bottom warehouse, for conveying materials from the upper level to the bottom level; the scheduling control module communicates with the first elevator, the second elevator, and the connecting conveyor platform respectively, for acquiring multi-source data of the first elevator, the second elevator, and the connecting conveyor platform, and realizing the coordinated scheduling of the first elevator, the second elevator, and the connecting conveyor platform based on multi-source data and dynamic optimization strategies; wherein, the multi-source data includes at least task characteristic data, equipment status data, and conveying environment perception data.

[0010] Optionally, the scheduling and control module includes a data acquisition submodule, a task decision submodule, and a control submodule. The data acquisition submodule is used to collect multi-source data from the first hoist, the second hoist, and the connecting conveyor platform. The task decision submodule is used to perform task allocation and path planning based on the multi-source data from the data acquisition submodule, using a dynamic priority scheduling algorithm combined with a conflict prediction model. The control submodule is used to control the first hoist and the second hoist to perform conveying operations based on the task allocation and path planning from the task decision submodule.

[0011] Furthermore, the data acquisition submodule includes a task feature data layer, an equipment status data layer, a conveying environment perception data layer, and a historical performance data layer. The task feature data layer is used to acquire multiple task attribute feature data. The equipment status data layer acquires multiple status data of a certain equipment. The conveying environment perception data layer is used to acquire the material density and task density of each floor's conveyor line. The historical performance data layer is used to acquire the historical distribution of task completion time, the statistics of historical equipment failure frequency, and the historical processing efficiency matrix of different material types.

[0012] Optionally, based on the multi-source data from the data acquisition submodule, a dynamic priority scheduling algorithm, combined with a conflict prediction model, is used for task allocation and path planning, specifically including: The priority of the task to be decided is determined based on the task attribute characteristics data, the factors affecting the task priority, and the corresponding weight coefficients. Based on the conflict prediction model of discrete event simulation and spatiotemporal state extrapolation, the simulation predicts the possibility of mission conflicts between the first elevator, the second elevator and the docking area in the future. Determine the level of task conflict and the corresponding conflict resolution strategy based on the likelihood of task conflict. Adjust task allocation and delivery paths according to conflict resolution strategies.

[0013] Furthermore, based on the conflict prediction model of discrete event simulation and spatiotemporal state extrapolation, the simulation prediction of the possibility of task conflicts between the first elevator, the second elevator and the connecting area in the future includes: adding all assigned and scheduled tasks to the simulation queue in chronological order. Each task includes: starting floor, ending floor, expected start time, expected end time, elevator and connecting area number used. A time-space occupancy table is set up for each hoist and each connecting area to record the time occupancy of each hoist and each connecting area in future time periods. When a new task arrives or the task status is updated, the conflict probability between tasks is calculated based on the simulation time window and simulation path cross-analysis. The specific calculation method for the conflict probability between tasks is: the product of the collision time overlap ratio × the equipment load coefficient × the path sharing factor. Wherein, the collision time overlap ratio is the ratio of the overlap time length to the total coverage time length; the equipment load coefficient is the sum of the initial value and the load rate ratio, where the load rate ratio is the product of the load sensitive parameter and the current load rate of the equipment; the path sharing factor is the ratio of the first weighted sum and the second weighted sum. The first weighted sum is the weighted sum of the number of identical nodes in the planned paths of the two tasks and the corresponding node weights, and the second weighted sum is the weighted sum of the number of non-overlapping nodes in the planned paths of the two tasks and the corresponding node weights.

[0014] Optionally, determining the task conflict level and corresponding conflict resolution strategy based on the likelihood of task conflict specifically includes: If the probability of a task conflict is less than the first preset probability threshold, the task conflict level is no conflict, and the assignment of the task to be assigned is executed directly. If the probability of a task conflict is not less than the first preset probability threshold and is less than the second preset probability threshold, the task conflict level is a minor conflict, and the task to be assigned will be delayed for a preset time before being assigned. If the probability of a task conflict is not less than the second preset probability threshold and is less than the third preset probability threshold, the task conflict level is medium conflict, and the task to be assigned will be re-optimized. If the probability of a task conflict is not less than the third preset probability threshold and less than the fourth preset probability threshold, the task conflict level is severe conflict, and the task to be assigned will be split. If the probability of a task conflict is not less than the fourth preset probability threshold, and the task conflict level is deadlock conflict, a system alarm will be issued for the tasks to be assigned.

[0015] Furthermore, re-optimizing the paths for tasks to be assigned specifically includes: Based on task priority, equipment availability time, and path travel time, an optimal time window is calculated for each task to be optimized. The optimal time window includes: the docking preparation period, the hoist occupancy period, and the exit docking period. Plan a unique path for each task to be optimized. The path includes: starting connection area number, elevator number used, target connection area number, and floor conveyor route. The conflict probability between tasks is recalculated based on the optimal time window and the planned path. If the conflict level is less than that of a moderate conflict, resources are locked and a scheduling instruction set is generated. If the conflict level is not less than that of a moderate conflict, the optimal time window and the planned path are optimized again. Before executing the task to be optimized, mark the time window of all nodes in the planned path; other tasks need to avoid locked path nodes when planning. Once the optimization task is completed, release all locked path nodes.

[0016] Optionally, splitting the tasks to be assigned specifically includes: splitting tasks according to task characteristics and conflict types, using one or more of the following methods; wherein, the splitting methods include splitting by quantity, splitting by path, and splitting by time window. Splitting by quantity specifically involves splitting a batch of materials into multiple transportation sub-batches; splitting by path specifically involves decomposing the task into sub-tasks that sequentially pass through different physical paths or transfer areas; splitting by time window specifically involves splitting a time-consuming task into multiple sub-tasks that occupy short time windows. All child tasks originating from the same parent task are assigned the same task group ID, and the child tasks initially inherit the priority of the parent task.

[0017] Optionally, based on the multi-source data from the data acquisition submodule, a dynamic priority scheduling algorithm, combined with a conflict prediction model, is used for task allocation and path planning. Specifically, this also includes: The load rates of the first and second hoists are monitored in real time. The load difference is calculated based on the load rates of the first and second hoists. If the load difference exceeds the preset difference threshold, the task is optimized and adjusted according to the preset load balancing strategy.

[0018] Furthermore, optimizing and adjusting tasks according to the preset load balancing strategy specifically includes: If the difference between the load rate of the first hoist and the load rate of the second hoist is greater than a preset difference threshold, some of the uplink tasks will be transferred to the second hoist for execution. If the difference between the load rate of the second hoist and the load rate of the first hoist is greater than a preset difference threshold, a symmetrical processing strategy is adopted to transfer some downlink tasks to the first hoist for execution. The specific screening conditions for transferable tasks include: task attribute conditions, task current status conditions, and transfer contribution conditions. The task attribute conditions include low task priority, task that allows bidirectional scheduling, or task path selectability. The task current status conditions include the task that is currently assigned but has not started execution. The transfer contribution conditions include the task to be transferred having a high contribution to reducing the load of the original equipment and improving the utilization rate of the target equipment.

[0019] The technical solution adopted in this invention has the following technical effects: 1. In the technical solution of this invention, the first elevator is used to lift materials from the bottom layer to the upper layer; the second elevator is used to transport materials from the upper layer to the bottom layer; the scheduling and control module is used to acquire multi-source data from the first elevator, the second elevator, and the connecting conveyor platform, and to realize the coordinated scheduling of the first elevator, the second elevator, and the connecting conveyor platform based on the multi-source data and dynamic optimization strategies; wherein, the multi-source data includes at least task characteristic data, equipment status data, conveying environment perception data, and historical performance data, effectively solving the problem of low efficiency and reliability of cross-layer bidirectional connecting systems in automated storage and retrieval systems caused by existing technologies, and effectively improving the efficiency and reliability of cross-layer bidirectional connecting systems in automated storage and retrieval systems.

[0020] 2. In the technical solution of this invention, by setting up two elevators and clearly defining their functions (one for upward movement and one for downward movement), two independent, unidirectional material flow channels are created; the upward and downward tasks are truly executed concurrently; the process of material moving from the second floor to the first floor and from the first floor to the second floor can be carried out simultaneously without interference, completely eliminating task conflicts and waiting time at the physical level, and improving the system's cross-floor material throughput capacity.

[0021] 3. The data acquisition submodule in the technical solution of this invention includes a task feature data layer, an equipment status data layer, a conveying environment perception data layer, and a historical performance data layer. The task feature data layer is used to acquire multiple task attribute feature data. The equipment status data layer acquires multiple status data of a certain equipment. The conveying environment perception data layer is used to acquire the material density and task density of each floor's conveyor line. The historical performance data layer is used to acquire the historical distribution of task completion time, the statistics of historical equipment failure frequency, and the historical processing efficiency matrix of different material types. This constructs a multi-layer data acquisition architecture to ensure the comprehensiveness, real-time performance, and reliability of the scheduling and control module's decision-making.

[0022] 4. In the technical solution of this invention, the priority of the task to be decided is determined based on the task attribute characteristic data, task priority influencing factors and corresponding weight coefficients; based on the conflict prediction model of discrete event simulation and spatiotemporal state extrapolation, the possibility of task conflict between the first hoist, the second hoist and the connecting area in the future is simulated and predicted; the task conflict level and corresponding conflict resolution strategy are determined according to the possibility of task conflict; the task allocation and transportation path are adjusted according to the conflict resolution strategy, which ensures the reliability of task scheduling and improves the efficiency of task scheduling.

[0023] 5. In the technical solution of this invention, the task conflict level and corresponding conflict resolution strategy are determined according to the probability of task conflict, which further ensures the reliability of task scheduling and improves the efficiency of task scheduling.

[0024] 6. The technical solution of this invention uses multi-source data from the data acquisition submodule, employs a dynamic priority scheduling algorithm, and combines it with a conflict prediction model to perform task allocation and path planning. Specifically, it also includes: real-time monitoring of the load rates of the first and second hoists, calculating the load difference based on the load rates of the first and second hoists, and optimizing and adjusting the tasks according to a preset load balancing strategy if the load difference exceeds a preset threshold, thereby ensuring the efficient and reliable operation of the first and second hoists.

[0025] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a schematic diagram of the layout of the first floor (ground floor warehouse) in the system of Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the layout of the first floor (upper warehouse) in the system of Embodiment 1 of the present invention; Figure 3 This is a schematic diagram of the architecture of the scheduling and control module in Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of the task feature data layer in the system of Embodiment 1 of the present invention; Figure 5 This is a schematic diagram of the hoist state matrix in the system of Embodiment 1 of the present invention; Figure 6This is a schematic diagram of the connection area state vector in the system of Embodiment 1 of the present invention; Figure 7 This is a schematic diagram of the architecture of the task decision submodule in Embodiment 1 of the present invention. Figure 8 This is a schematic diagram showing the corresponding conflict probability, conflict level, and conflict resolution strategy in the system of Embodiment 1 of the present invention. Figure 9 This is a schematic diagram of the real-time load evaluation matrix in the system of Embodiment 1 of the present invention. Detailed Implementation

[0028] To clearly illustrate the technical features of this solution, the invention will be described in detail below through specific embodiments and in conjunction with the accompanying drawings. The following disclosure provides many different embodiments or examples for implementing different structures of the invention. To simplify the disclosure of the invention, components and arrangements of specific examples are described below. Furthermore, reference numerals and / or letters may be repeated in different examples. This repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed. It should be noted that the components illustrated in the drawings are not necessarily drawn to scale. Descriptions of well-known components, processing techniques, and processes are omitted in this invention to avoid unnecessarily limiting the invention.

[0029] Example 1 like Figures 1-2As shown, this invention provides a cross-level bidirectional connection system for an automated storage and retrieval system (AS / RS), comprising: a first hoist (first continuous hoist), a second hoist (second continuous hoist), a connection conveying platform, and a scheduling and control module. The connection conveying platform includes a first connection conveying area (first connection area) and a fourth connection conveying area (fourth connection area) located in the bottom warehouse, and a second connection conveying area (second connection area) and a third connection conveying area (third connection area) located in the upper warehouse. The bottom inlet / outlet of the first hoist connects to the first connection conveying area in the bottom warehouse, and the upper outlet / inlet of the first hoist connects to the third connection conveying area in the upper warehouse, for lifting materials from the bottom to the upper level; the upper outlet / inlet of the second hoist connects to the third connection conveying area in the upper warehouse. The outlet of the second elevator connects to the fourth connecting conveyor area of ​​the upper warehouse, and the bottom outlet / inlet of the second elevator connects to the second connecting conveyor area of ​​the bottom warehouse, for conveying materials from the upper level to the bottom level; the scheduling control module communicates with the first elevator, the second elevator, and the connecting conveyor platform respectively, for acquiring multi-source data from the first elevator (equipment PLC, sensors, etc.), the second elevator (equipment PLC, sensors, etc.), and the connecting conveyor platform (sensors, RFID readers, vision systems, etc.), and realizes coordinated scheduling of the first elevator, the second elevator, and the connecting conveyor platform based on multi-source data and dynamic optimization strategies; wherein, the multi-source data includes at least task characteristic data, equipment status data, and conveying environment perception data.

[0030] The first elevator's upper inlet / outlet connects to the third connecting conveyor area of ​​the upper warehouse, used to lift materials from the bottom to the top. The second elevator's upper inlet / outlet connects to the fourth connecting conveyor area of ​​the upper warehouse, and its lower inlet / outlet connects to the second connecting conveyor area of ​​the lower warehouse, used to transport materials from the top to the bottom. By setting up two elevators and clearly defining their functions (one for upward movement and one for downward movement), two independent, unidirectional material flow channels are created, enabling true concurrent execution of upward and downward tasks. The process of materials moving from the second floor to the first floor and from the first floor to the second floor can occur simultaneously without interference, completely eliminating task conflicts and waiting time at the physical level. Theoretically, the system's cross-floor logistics throughput capacity can be increased by 100%.

[0031] Furthermore, the interfaces of the two elevators at the bottom level (the first and second connecting conveyor areas) are designed to directly connect with the existing main conveyor line and belt extension conveyor in the warehouse. This system can be embedded into the existing warehouse logistics system as an independent "plug-in" module without requiring large-scale modifications to the original conveyor layout, achieving "plug and play" functionality, maximizing the protection of existing investments, and reducing renovation costs and construction complexity.

[0032] Moreover, the dual-machine system provides natural redundancy; when one hoist fails and needs maintenance, the other hoist can be temporarily switched to bidirectional emergency mode (efficiency is reduced, but the system does not stop) under the coordination of the scheduling system, which greatly improves the reliability and availability of the entire warehousing system.

[0033] Specifically, such as Figure 3 As shown, the scheduling and control module (unified scheduling and control system) includes a data acquisition submodule (data input layer), a task decision submodule (task decision layer), and a control submodule (output control layer). The data acquisition submodule is used to collect multi-source data from the first hoist, the second hoist, and the connecting conveyor platform. The task decision submodule is used to perform task allocation and path planning based on the multi-source data from the data acquisition submodule, using a dynamic priority scheduling algorithm combined with a conflict prediction model. The control submodule is used to control the first hoist and the second hoist to perform conveying operations based on the task allocation and path planning from the task decision submodule.

[0034] Specifically, the data acquisition submodule includes a task feature data layer, an equipment status data layer, a conveying environment perception data layer, and a historical performance data layer. The task feature data layer is used to acquire multiple task attribute feature data. The equipment status data layer acquires multiple status data of a certain equipment. The conveying environment perception data layer is used to acquire the material density and task density of each floor's conveyor line. The historical performance data layer is used to acquire the historical distribution of task completion time, the statistics of historical equipment failure frequency, and the historical processing efficiency matrix of different material types.

[0035] The scheduling and control module collects the following types of data in real time: Task data: Task instructions from the upper-level warehouse management system (WMS), including material origin, destination, priority, material type, size, weight, urgency, etc.

[0036] Material status data: Material identification, current location, and flow status obtained through RFID, barcode, or visual recognition systems.

[0037] Equipment status data includes the operating status (idle, running, faulty, under maintenance) of the two hoists, their current position, speed, and load; the occupancy / idle status, conveying speed, and alignment status of each connecting conveyor zone.

[0038] System status data: congestion level of the main conveyor line on each floor, task queue length, and estimated waiting time.

[0039] The multi-layer data acquisition architecture specifically includes: (1) Task feature data layer: such as Figure 4As shown, the task characteristic data includes task ID, starting floor, ending floor, material type, material size, material weight, task priority, task urgency, estimated task processing time, and associated task group registration. (2) Equipment status data layer: such as Figure 5 As shown, the hoist status matrix includes attribute information such as: equipment ID, equipment operating position, equipment operating direction, equipment operating speed, equipment load, equipment operating temperature, equipment operating vibration value, equipment fault code, and equipment maintenance countdown.

[0040] like Figure 6 As shown, the connection zone status vector includes: connection zone area ID, connection zone occupancy status, connection zone alignment error, connection zone conveying speed, connection zone queue length, and connection zone identification status.

[0041] (3) Environmental perception data layer: Floor congestion index: calculated based on the material density of the conveyor lines on each floor; Time window pressure value: The number of tasks expected to arrive within the next 5 minutes; Energy consumption monitoring data: Real-time power consumption of each device; (4) Historical performance data layer: Historical distribution of task completion times; Equipment failure frequency statistics; Processing efficiency matrix for different material types; (5) External related data layer: WMS system order urgency; Production cycle synchronization requirements; The next process is pending.

[0042] Among them, such as Figure 7 As shown, in the task decision submodule, based on multi-source data from the data acquisition submodule, a dynamic priority scheduling algorithm is used in conjunction with a conflict prediction model to perform task allocation and path planning. Specifically, this includes: S11. Determine the priority of the task to be decided based on the task attribute characteristic data, task priority influencing factors and corresponding weight coefficients. Specifically, a dynamic priority scheduling algorithm can be used, combined with a conflict prediction model and load balancing strategy, for task allocation and path planning. The dynamic priority calculation for each task can be dynamically derived using the following formula: Total Priority Score = W1 × Task Urgency Coefficient (0.1-1.0) + W2 × Waiting Penalty Coefficient (0-1.5) + W3 × Material Value Coefficient (0.8-2.0) + W4 × Path Optimization Coefficient (0.5-1.2) W5×Conflict Risk Coefficient (0-1.0) + W6×System Equilibrium Coefficient (0.8-1.2); The data source for the task urgency coefficient includes: the "urgency level" field (0.1~1.0) marked when the task is issued by the WMS; the normalized urgency value passed in by the WMS can be used directly; if there is no explicit urgency value, it will be mapped according to the task type and preset rules. If the task type ∈ {replenishment, urgent orders, production line downtime risk} → the task urgency coefficient UF = 0.9~1.0; If the task type ∈ {routine in / out, batch transfer} → the task urgency factor UF = 0.3~0.7; If the task type ∈ {inventory, warehouse relocation, maintenance and handling} → the task urgency coefficient UF = 0.1~0.3.

[0043] Waiting penalty coefficient: The time the task enters the scheduling queue, the current system time, and the preset standard waiting time. Calculation logic: Waiting time = Current time - Task entry time; Standard waiting time = Preset according to task type (e.g., 300 seconds for normal tasks, 60 seconds for urgent tasks); Then the waiting penalty coefficient = min(1.5, Waiting time / Standard waiting time × 1.2); The upper limit of the waiting penalty coefficient is set to 1.5 to prevent extremely waiting tasks from excessively crowding out new tasks.

[0044] Material value factor: This factor is derived from the material master data, including material type, weight, dimensions, and value class. Calculation logic: Basic value grade (corresponding coefficients of 1.5 / 1.2 / 0.8 for A / B / C class materials); Weight adjustment factor: 1 + (current material weight / maximum load capacity of the elevator) × 0.2; 1 + (current material weight / maximum load capacity of the elevator) × 0.2; Size adjustment factor: If it is an extra-large size (length + width + height > 2.5m), multiply by 1.1; the final coefficient is normalized in the range of 0.8 to 2.0.

[0045] Example: For a certain type A material, with a weight percentage of 0.6 and not an oversized material, the material value factor MVF = 1.5 × (1 + 0.6 × 0.2) = 1.68.

[0046] Path optimization coefficient (optional): Path optimization coefficient = distance from the current position of the elevator to the starting point of the task + distance from the starting point of the task to the end point, and finally normalized within the range of 0.5 to 1.2.

[0047] Conflict Factor in Transport Path (PCF): Calculates the probability of conflict and the weight of conflict impact in conflict prediction.

[0048] Calculation logic: Conflict probability Pc: derived from spatiotemporal resource collision detection (0~1); Conflict impact weight Wc: Preset (0.5~2.0) based on the type of conflicting equipment and the importance of the time period. The collision factor PCF of the delivery path is calculated as ∑(Pc×Wc) (which is the sum of the potential conflicts between this task and all other tasks in the queue). It should be noted that the transport path conflict coefficient is a negative factor; the higher the value, the greater the risk of system conflict caused by executing the task.

[0049] System load balancing coefficient: Real-time monitored load rates of the two hoists (uplink load rate Lu, downlink load rate Ld). Calculation logic: Load difference = |Lu - Ld|; System balance coefficient = 1 / (1 + load difference) (output range 0.5~1.0, the smaller the difference, the closer to 1); used to encourage the system to allocate tasks to the less loaded hoists, promoting balanced equipment utilization.

[0050] W1, W2, W3, W4, W5, and W6 are the weight coefficients corresponding to each coefficient. The weights can be automatically adjusted according to the operating mode, supporting adaptive learning optimization. Peak mode (focusing on throughput and emergency response): W1 (0.4), W2 (0.3), W6 (0.3) (other weights set to 0); Balanced mode (comprehensive optimization of efficiency and risk): W1(0.25), W3(0.25), W5(0.25), W6(0.25) (other weights set to 0); Energy-saving mode (focusing on path optimization and equipment balancing): W4 (0.4), W6 (0.6) (other weights set to 0); The trigger condition for switching operating modes can be based on time rules (such as setting peak mode from 9:00 to 11:00) or on system status (such as automatically entering peak mode when the task queue is continuously >10 and the device utilization rate is >85%). This embodiment does not impose any restrictions here.

[0051] Adaptive learning optimization: Priority parameters can also include equipment health coefficient (based on real-time vibration, temperature, and historical fault prediction reliability of equipment; equipment with low health has a lower weight when assigning tasks; equipment health coefficient DHF = 0.5 + 0.5 × health score (0~1)), energy efficiency optimization coefficient (in the case of time-of-use electricity pricing, tasks are encouraged to be executed during off-peak hours; energy efficiency optimization coefficient EOF = electricity price factor × expected energy consumption of task), and order association coefficient (if multiple tasks belong to the same production order, their overall priority is increased to promote order completeness; order association coefficient ORF = 1 + 0.2 × number of uncompleted tasks in the same order), etc.

[0052] S12, a conflict prediction model (conflict detector) based on discrete event simulation and spatiotemporal state extrapolation, simulates and predicts the possibility of mission conflicts between the first elevator, the second elevator and the docking area in the future. Specifically, step S12 includes: S121, add all assigned and pending tasks to the simulation queue in chronological order. Each task includes: starting floor, ending floor, estimated start time, estimated end time, elevator used, and connection area number. Specifically, task queue modeling: all assigned and scheduled tasks are entered into the simulation queue according to time sequence. Each task includes: starting floor, ending floor, estimated start time, estimated end time, elevator used (up / down), connection area number, and path node sequence.

[0053] S122, set up a time-space occupancy table for each hoist and each connecting area to record the time occupancy of each hoist and each connecting area in future time periods; Specifically, a spatiotemporal state matrix of the equipment is constructed, and a time-space occupancy table is maintained for each hoist and each connecting area, recording the time occupancy of each resource (hoist and connecting area) within a future period (e.g., the next 5 minutes). S123, when a new task arrives or the task status is updated, the conflict probability between tasks is calculated based on the simulation time window and simulation path cross-analysis between tasks. The specific calculation method for the conflict probability between tasks is: the product of the collision time overlap ratio × the equipment load coefficient × the path sharing factor. The collision time overlap ratio is the ratio of the overlap time length to the total coverage time length; the equipment load coefficient is the sum of the initial value and the load rate ratio, where the load rate ratio is the product of the load-sensitive parameter and the current load rate of the equipment; the path sharing factor is the ratio of the first weighted sum to the second weighted sum, where the first weighted sum is the weighted sum of the number of identical nodes in the planned paths of the two tasks and the corresponding node weights, and the second weighted sum is the weighted sum of the number of non-overlapping nodes in the planned paths of the two tasks and the corresponding node weights.

[0054] When a new task arrives or a task status is updated, the simulation proceeds sequentially according to the task time window, starting from the current time; the execution process of each task is simulated; the status changes of each device are recorded; it is detected whether two or more tasks are competing for the same device or path at the same time; the conflict time point, conflicting device, and conflict type are output, and the following steps are performed: Time window collision detection: Check whether the hoist and docking area planned for this task are already occupied during the corresponding time period; Path intersection analysis: For tasks that use the same floor or adjacent paths simultaneously, determine whether there is spatial intersection; One method for calculating the probability of conflict is: Conflict probability = Collision time overlap ratio × Equipment load coefficient × Path sharing factor; The collision probability is calculated by multiplying the collision time overlap ratio, the equipment load coefficient, and the path sharing factor (optional).

[0055] 1. Collision Time Overlap Ratio: This refers to the degree of overlap in the time periods when two tasks are scheduled to use the same equipment. Calculation method: Collision time overlap ratio = overlap time length / total coverage time length. The result is between 0 and 1, where 1 indicates complete overlap.

[0056] Set a "time buffer threshold" (e.g., 5 seconds). If the calculated overlap time is less than this threshold, it is considered that there is no actual conflict, and the ratio is treated as 0. During peak periods with heavy workloads, the system will automatically reduce this buffer threshold (e.g., adjust it to 3 seconds) to make more precise use of time gaps. During off-peak periods with sparse workloads, the system will increase the buffer threshold (e.g., adjust it to 8 seconds) to make scheduling more relaxed and stable.

[0057] 2. Equipment Load Factor: Reflects the current workload of the equipment. The busier the equipment, the greater the impact of a conflict. Calculation Method: Equipment load factor = 1.0 + (load sensitivity parameter × current load rate).

[0058] The "load sensitivity parameter" is an adjustable value, with a default value of 0.5. For example, if the equipment load rate is 80% (i.e., 0.8) and the load sensitivity parameter is 0.5, then the equipment load factor = 1.0 + 0.5 × 0.8 = 1.4. If the equipment is in a fault or maintenance state, the equipment load factor will be set to a higher fixed value (e.g., 2.0).

[0059] Continuously monitor the queue length and response speed of the equipment. If the equipment is under high load for an extended period and the queue is backlogged, the "load sensitivity parameter" will be automatically increased (e.g., from 0.5 to 0.7) to increase the load factor, thereby providing earlier and stronger warnings in conflict prediction.

[0060] In "Energy Saving Mode", this parameter will be lowered to encourage tasks to be assigned to low-load devices, and the risk of computational conflicts will be lower even if the load rate is the same.

[0061] 3. Path Sharing Factor: This refers to the proportion of identical equipment nodes (such as specific connection areas or conveyor segments) shared in the physical paths of two task plans. Calculation method: Decompose the planned path of each task into a series of sequentially traversed device nodes; identify the nodes that are exactly the same in the paths of two tasks.

[0062] The path sharing factor is calculated as: First weighted sum / Second weighted sum. The first weighted sum is the product of the number of identical nodes and the weight of each node. The second weighted sum is the product of the total number of non-overlapping nodes in the two task paths and the corresponding node weights. The result is between 0 and 1, with 1 indicating that the paths are completely identical.

[0063] The scheduling and control module assigns different "conflict weights" to different types of path nodes. For example, the weight of elevator entrances and main road intersections is greater than that of ordinary conveyor sections. During calculation, "weighted shared node count" is used instead of simple "identical node count," thus more accurately reflecting the conflict risk of critical nodes. If a certain path frequently experiences congestion or conflicts, the system will dynamically increase the weight of nodes on that path, making subsequent task planning more inclined to avoid this path segment.

[0064] Another way to calculate the probability of conflict is: conflict_prob = 0.7 × time_overlap + 0.3 × space_overlap; this probability value is also constrained to be between 0 and 1, and the higher the value, the greater the risk of conflict.

[0065] Conflict is predicted by analyzing the degree of overlap between two tasks in the temporal and spatial dimensions. Each task is modeled as an object with the following properties: scheduled_window: The scheduled time window (start time, end time); planned_path: The planned physical path (device node sequence); priority_score: Dynamic priority score for the task; material_value: Material value coefficient; in the formula for calculating the probability of conflict, time_overlap: Time window overlap (0-1). space_overlap: Spatial path overlap (0-1); the weighting coefficients of 0.7 and 0.3 are empirical values ​​derived from historical data analysis, indicating that temporal overlap has a higher weight in the occurrence of conflicts than spatial overlap.

[0066] Spatial path overlap calculation method: This method quantifies the degree of conflict between two tasks in physical space through path edge overlap analysis. The path (device node sequence) of each task is converted into an edge sequence (i.e., consecutive node pairs), thus capturing the directionality and sequence of the path. Subsequently, the intersection (number of shared edges) and union (total number of unique edges) of the two task edge sequences are calculated. Spatial path overlap is defined as the ratio of the number of shared edges to the total number of unique edges. This method considers not only whether nodes are identical but also the connections between nodes, thus more accurately reflecting the nature of path conflicts. The final result is also normalized to between 0 and 1.

[0067] S13, Determine the task conflict level and corresponding conflict resolution strategy based on the probability of task conflict; Specifically, such as Figure 8 As shown, the calculated task conflict probabilities (between 0 and 1) are linearly mapped to an integer scale of 1 to 5. This scale provides a direct basis for subsequently selecting conflict resolution strategies of different severity levels. Specifically, If the probability of a task conflict is less than the first preset probability threshold (0.2), the task conflict level is no conflict, and the assignment of the task to be assigned is carried out directly. If the probability of a task conflict is not less than the first preset probability threshold and less than the second preset probability threshold (0.4), the task conflict level is a minor conflict, and the task to be assigned will be delayed for a preset time (e.g., 30 seconds) before being assigned. If the probability of a task conflict is not less than the second preset probability threshold and less than the third preset probability threshold (0.6), the task conflict level is moderate. After a delay of 60 seconds, the path of the task to be assigned will be re-optimized. If the probability of a task conflict is not less than the third preset probability threshold and less than the fourth preset probability threshold (0.8), the task conflict level is severe conflict, and the task to be assigned will be split. If the probability of a task conflict is not less than the fourth preset probability threshold, and the task conflict level is deadlock conflict, a system alarm will be issued for the tasks to be assigned.

[0068] S14, adjust task allocation and delivery path according to conflict resolution strategy.

[0069] This includes re-optimizing the paths of tasks to be assigned (using a time and space resource reservation mechanism to allocate a time window and physical path to each task, ensuring that each task has obtained the necessary equipment and path usage rights before execution), specifically including: S21. Calculate an optimal time window for each task to be optimized based on task priority, equipment availability time, and path travel time. The optimal time window includes: the connection preparation period, the hoist occupancy period, and the exit docking period. Specifically, based on task priority, equipment availability, and path travel time, an optimal time window is calculated for each task. The time window includes: the docking preparation period (materials enter the docking area), the elevator occupancy period (materials move within the elevator), and the exit docking period (materials leave the elevator and reach the target floor). The specific calculation steps for the optimal time window for each task can be as follows: 1. Collect all influencing factors: Task requirements: priority, latest completion deadline, and estimated processing time.

[0070] Equipment availability: When will the two hoists be idle in the near future (e.g., the next 10 minutes)? When will the connection areas on each floor be available?

[0071] Path traffic conditions: approximate time from the starting point to the end point, traversing each conveyor line and passing the elevator; which path segments are currently congested.

[0072] Overall system status: Are the loads on the two hoists balanced? How many tasks are expected to arrive in the next few minutes?

[0073] 2. Generate candidate time windows: The system will quickly identify several feasible time windows based on the task's start and end points and available devices. For example: Window A: Using the upward hoist, starting in 5 minutes and ending in 8 minutes.

[0074] Window B: Use the downhill hoist (detour), starting in 3 minutes and ending in 7 minutes.

[0075] Window C: Waiting for the busy connection area to be released, starting in 7 minutes and ending in 10 minutes.

[0076] 3. Score each candidate window: The system evaluates the merits of each window from four core dimensions and calculates a comprehensive cost (lower scores are better): Waiting Cost: How long the task will take before it begins. The longer the wait, the higher the cost.

[0077] Transportation cost: How long does it take to complete the entire route? The longer the time, the higher the cost.

[0078] Conflict risk cost: Based on a conflict detection algorithm, predict how many other tasks will conflict with the task being executed within this window, and the severity of the conflict. The higher the risk, the higher the cost.

[0079] Load balancing cost: Will selecting this window and device result in one hoist being extremely busy and the other extremely idle? The greater the imbalance, the higher the cost.

[0080] 4. Select and confirm the optimal window: Weighted calculation: Based on the current operating mode (such as peak mode, balanced mode, energy-saving mode), different weights are assigned to the above four costs, and then the total cost of each window is calculated.

[0081] Peak mode: prioritizes "waiting costs" and "transportation costs" and strives for rapid throughput.

[0082] Equilibrium model: Equilibrium considers four costs.

[0083] Energy-saving mode: It places more emphasis on the "load balancing cost" and may also include the "energy consumption cost".

[0084] Final decision: Select the candidate window with the lowest total cost and determine it as the "optimal time window" for the task.

[0085] 5. Dynamic adjustment mechanism: If the system state changes abruptly before the task is executed (such as a device failure), the system will quickly recalculate for the affected task and switch to a new "suboptimal time window".

[0086] S22, plan a unique path for each task to be optimized. The path includes: starting connection area number, elevator number used, target connection area number, and floor conveyor route. The system plans a unique path for each task, including: starting connection area number, elevator number used, target connection area number, floor conveyor route, and path locking mechanism; before the task is executed, the time window of all nodes of the path is marked in the resource occupancy table; S23, recalculate the conflict probability between tasks formed by the optimal time window and the planned path; if the task conflict level is less than medium conflict, then officially lock the resources and generate a scheduling instruction set; if the task conflict level is not less than medium conflict, then re-optimize the optimal time window and the planned path. S24, Before the task to be optimized is executed, mark the time window of all nodes in the planned path; other tasks need to avoid locked path nodes when planning. S25, After the optimization task is completed, release all locked path nodes.

[0087] Specifically, the complete path of materials from the starting point to the end point is planned, including the main conveyor line, the connecting area, the hoist, and the target floor conveyor line, and the path instructions are issued to each execution unit in real time.

[0088] Specifically, the avoidance and solutions to path planning and spatiotemporal conflicts are as follows: 1. Path Planning Process: Path planning is an optimization process based on spatiotemporal joint search. Its goal is to find a conflict-free passage scheme with a clear physical path and a defined time window for materials to travel from the starting point to the destination. The core steps are as follows: Step 1: Topology Modeling. Abstract the physical structure of the automated warehouse into a directed graph. Nodes include: connection areas between floors, elevator entrances / exits, and intersections of the main conveyor lines; edges represent conveyor paths, and weights include conveying time, energy consumption, etc.

[0089] Step 2: Spatiotemporal Status Query. Access the spatiotemporal resource reservation table to obtain the occupancy status of each node and edge within the planned time period.

[0090] Step 3: Candidate Path Generation. Using an improved A* algorithm or a bidirectional search algorithm, traverse the graph, avoiding already occupied "spatiotemporal obstacles," and generate several candidate paths that satisfy the basic constraints.

[0091] Step 4: Multi-objective evaluation and selection. For each candidate path, calculate its total time consumption, energy consumption, conflict risk value, and impact on system equilibrium. Combined with task priority, select the optimal path through a cost function.

[0092] Step 5: Resource reservation and instruction generation. Lock all device nodes and time windows involved in the selected path in the resource reservation table, and generate a detailed sequence of timestamped control instructions.

[0093] 2. Methods to avoid spatiotemporal conflicts in path planning: The core of conflict avoidance is to integrate the time dimension into path search to achieve "spatiotemporal joint planning".

[0094] Spatiotemporal dilation method: For each node in the path graph, in addition to its spatial coordinates, a temporal attribute is also added. During the search, "spatial nodes + time intervals" already occupied by other tasks are regarded as obstacles, and the algorithm automatically avoids them.

[0095] Resource reservation system: During the planning phase, a time window is requested for key resources along the route (such as specific connecting areas or elevator entrances). The system checks the global reservation table to ensure that the window is not occupied, thereby avoiding conflicts at the source.

[0096] Cost-driven approach based on conflict prediction: A conflict risk cost term is explicitly introduced into the cost function of path search. The algorithm tends to select paths with a low probability of spatiotemporal overlap with existing tasks.

[0097] 3. Unavoidable Conflicts and Solutions in Planning: If all candidate paths encounter unavoidable conflicts due to system congestion or high task density (e.g., essential nodes are occupied for extended periods), the system will initiate a dynamic conflict resolution process, which is deeply integrated with the planning process. Method 1: Planning-Feedback Iteration. The path planning module feeds back the "cannot be planned" result and conflict points to the scheduling decision layer. The decision layer activates the conflict resolution strategy library (such as using time offset for the current task, or replanning the path for tasks that consume resources), and then instructs the path planning module to retry planning.

[0098] Method 2: Introduce a compromise path. When the conflict of the optimal path cannot be resolved, the system can plan a suboptimal compromise path, such as: using a more distant backup connection area; or temporarily allowing the elevator to stop at a non-typical floor; or splitting the task and planning sub-paths separately.

[0099] Method 3: Collaborative Competition Decision-Making. For path conflicts among multiple high-priority tasks, the system initiates a micro-auction or negotiation mechanism to dynamically adjust the paths and timing of each task based on higher-order optimization objectives (such as overall order completion rate, production line uninterrupted operation).

[0100] Specifically, the process of splitting the tasks to be assigned includes: S31, splitting tasks according to task characteristics and conflict types using one or more of the following methods; the splitting methods include splitting by quantity, splitting by path, and splitting by time window. Splitting by quantity specifically involves splitting a batch of materials into multiple transportation sub-batches; splitting by path specifically involves decomposing the task into sub-tasks that sequentially pass through different physical paths or transfer areas; and splitting by time window specifically involves splitting a time-consuming task into multiple sub-tasks that occupy short time windows. The system intelligently splits tasks based on their characteristics and conflict types, from one or more of the following dimensions: Splitting by Quantity: This method splits a batch of materials into multiple transport sub-batches. The number of sub-batches after splitting is calculated as follows: min(ceil(total quantity or mass of materials in the batch / optimal single-batch transport quantity or mass of the equipment), maximum allowable number of splits); where min is the minimization function, ceil is the floor function, the optimal single-batch transport capacity of the equipment is dynamically determined by the equipment performance curve and the current system throughput pressure, and the maximum allowable number of splits is set to prevent excessive fragmentation.

[0101] Path-based splitting: The task is decomposed into subtasks that sequentially pass through different physical paths or connecting areas. Calculation criteria: When there is a persistently congested or faulty node on the task path (such as a connecting area), the task is split into "pre-congestion segment" and "post-congestion segment" in time or space, and scheduled separately using detour or waiting strategies.

[0102] Time-Window Splitting: A continuous, long-running task is broken down into multiple subtasks occupying short time windows, which are then executed intermittently. Calculation Standard: Based on a time-slice round-robin algorithm, several non-contiguous time windows are allocated to the original task, with other smaller tasks inserted in between to smooth out device utilization.

[0103] S32, all child tasks originating from the same parent task are assigned the same task group ID, and the child tasks initially inherit the priority of the parent task.

[0104] Specifically, the subtask management mechanism after splitting (as shown below) ensures that the splitting will not lead to confusion in task logic or a decrease in efficiency.

[0105] Logical association binding: All child tasks originating from the same parent task are assigned the same task group ID, and the system maintains their execution order dependencies (if any).

[0106] Priority inheritance and adjustment: Subtasks initially inherit the priority of their parent tasks. During execution, minor adjustments can be made based on the actual progress of the subtasks and the system status, but the highest priority within the group will not be lower than the original task.

[0107] Resource allocation strategy: Subtasks can: Parallel execution: If different devices are used and there are no path conflicts, then execute in parallel as much as possible.

[0108] Pipeline execution: If the same equipment is required, it is arranged in a compact time sequence to form a pipeline.

[0109] Integrity monitoring and recovery: The system monitors the completion status of the entire task group. If a subtask fails, it can trigger a rescheduling or partial retry within the group without requiring a complete rollback.

[0110] For example: 10 items need to be transported from the 1st floor to the 2nd floor, but the target Z2_F2 transfer area on the 2nd floor is currently occupied by a long-term maintenance operation (expected to take 5 minutes to release), and if this task is carried out in its entirety, it will block the upward hoist for up to 4 minutes.

[0111] Triggering split: Conflict prediction shows that the conflict probability of the Z2_F2 docking area is 0.9 (deadlock conflict), and the task duration is too long.

[0112] Calculation by splitting: A combination of splitting by quantity and splitting by time window is used.

[0113] The 10 boxes were split into two sub-batches: T100-1 (5 boxes) and T100-2 (5 boxes).

[0114] Allocate an immediate execution time window for T100-1 and use the uplift hoist to transport it to the backup docking area on the 2nd floor for temporary storage.

[0115] The execution of T100-2 is delayed until the Z2_F2 docking area is released before it is transported to the target area.

[0116] Meanwhile, during the gap between the execution of T100-1 and T100-2, other high-priority small tasks are inserted.

[0117] This embodiment avoids prolonged equipment blockage by splitting tasks, alleviating floor congestion. The overall task completion time may only increase by 1 minute compared to waiting for maintenance, but the overall system throughput and responsiveness are greatly improved.

[0118] Preferably, the step of allocating tasks and planning paths based on multi-source data from the data acquisition submodule, using a dynamic priority scheduling algorithm combined with a conflict prediction model, further includes: S15, monitor the load rates of the first and second hoists in real time, calculate the load difference based on the load rates of the first and second hoists, and optimize and adjust the task according to the preset load balancing strategy (executed by the load balancer) if the load difference is greater than the preset difference threshold (25%).

[0119] Specifically, a real-time load assessment matrix can be established, such as Figure 9 As shown, the uplink hoist utilization rate (load rate) U_up and downlink hoist utilization rate U_down are obtained in real time. The load difference is calculated as: load_diff = abs(U_up - U_down); abs is an absolute value function.

[0120] Optimizing and adjusting tasks according to the preset load balancing strategy specifically includes: If the difference between the load rate of the first hoist and the load rate of the second hoist is greater than a preset difference threshold, some of the uplink tasks will be transferred to the second hoist for execution. That is, Case A: U_up > U_down + 20 (uplink load is significantly too high); Strategy adopted: Transfer some of the uplink tasks to the downlink hoist for execution.

[0121] If the difference between the load rate of the second hoist and the load rate of the first hoist is greater than a preset difference threshold, a symmetrical processing strategy is adopted to transfer some downlink tasks to the first hoist for execution. The specific screening conditions for transferable tasks include: task attribute conditions, task current status conditions, and transfer contribution conditions. The task attribute conditions include low task priority, task that allows bidirectional scheduling, or task path selectability. The task current status conditions include the task that is currently assigned but has not started execution. The transfer contribution conditions include the task to be transferred having a high contribution to reducing the load of the original equipment and improving the utilization rate of the target equipment.

[0122] That is, scenario B: U_down > U_up + 20 (downlink load is significantly too high); A symmetrical processing strategy is adopted: some downlink tasks are transferred to the uplink hoist for execution.

[0123] Task transfer and rescheduling require independent safety constraint checks before any transfer operation. Task selection aims to identify tasks that minimize system disruption and maximize transfer benefits. Selection criteria (ranked by priority): Task direction flexibility: Only select tasks that allow "bidirectional scheduling" or are marked "path optional". For example, some materials can be reached to the target floor via either an up or down hoist.

[0124] Current task status: Prioritize tasks in the "Assigned but not yet started" state. Tasks already in the physical execution process are generally not transferred.

[0125] Task Attributes: Low-priority tasks prioritized: Tasks with lower dynamic priority scores are prioritized for transfer, while maintaining flexibility. Short-duration tasks prioritized: Tasks with short estimated execution times have a smaller impact on the load of the new equipment after transfer. Non-urgent related tasks: Tasks not belonging to urgent production orders requiring complete delivery.

[0126] Transfer Benefit Assessment: Evaluate the contribution of transferring this task to reducing the load on the original equipment and improving the utilization rate of the target equipment, and select the task with the highest contribution. For each selected task, the following conditions are checked in sequence: security constraint check, path feasibility verification, equipment capability compatibility, no introduction of new conflicts, and time window compliance.

[0127] The following steps are executed sequentially during task redistribution: Release the original allocation: In the scheduling plan, release the task from the original hoist, and release the time window and path resources originally occupied.

[0128] Allocate new resources: Reinsert the task into the target elevator's task queue, and calculate its new start time, end time, and detailed path based on its priority and current queue status.

[0129] Update task attributes: Modify fields such as assigned_lift, scheduled_window, and planned_path within the task object.

[0130] Update the schedule and synchronize instructions (update_schedule): Integrate the results of all reassigned tasks to generate a new, consistent global schedule plan.

[0131] The updated time window and path instructions are sent to the relevant equipment controllers (lifting machines, connecting conveyors).

[0132] Update the internal status tracking table.

[0133] Symmetrical processing refers to the system adopting an adjustment strategy that is logically symmetrical but opposite in direction to the one used when the downward hoist load is too high.

[0134] 1. Logical symmetry: The judgment conditions are symmetrical: if the uplink is too high, the judgment is U_up>U_down + 20; if the downlink is too high, the judgment is U_down>U_up + 20; the processing objectives are symmetrical: when the uplink is too high, the objective is to "reduce the uplink load and increase the downlink load"; when the downlink is too high, the objective is to "reduce the downlink load and increase the uplink load".

[0135] 2. Opposite Direction: Task Flow Direction: The up_to_down direction transfers upward tasks (starting from a lower floor, ending at a higher floor), changing them to be executed by the downward elevator. This usually means planning a different, potentially longer path (e.g., up first then down or using an alternative path). Conversely, the down_to_up direction transfers downward tasks (starting from a higher floor, ending at a lower floor), changing them to be executed by the upward elevator.

[0136] The symmetric processing mechanism ensures the completeness and fairness of the load balancing algorithm. It enables the system to dynamically respond to load surges in any direction (upstream or downstream), always maintaining the optimal overall efficiency of the dual-machine system, and is a key design feature for improving system robustness and adaptability.

[0137] The control submodule is used to control the first hoist and the second hoist to perform transport operation scheduling and real-time coordination according to the task allocation and path planning of the task decision submodule; according to the decision of the task decision submodule, the upward task is assigned to the first continuous hoist (dedicated to upward) and the downward task is assigned to the second continuous hoist (dedicated to downward).

[0138] Preferably, in order to ensure the scheduling execution effect, the scheduling control module can perform time coordination: through a precise time synchronization mechanism, it ensures that the alignment and handover of materials in the receiving area and the elevator inlet and outlet are seamless, reducing waiting and idle time.

[0139] Specifically, a hybrid synchronization architecture of "global unified clock + hierarchical timestamp + forward-looking motion control" can be adopted: 1. Global unified clock source: The scheduling and control module deploys a high-precision Network Time Protocol (NTP) server or uses the IEEE 1588 (PTP) precision clock protocol to provide a microsecond-level synchronized time reference for all lower-level device controllers (PLCs, robot controllers) and sensors.

[0140] Hierarchical command issuance and timestamps: The scheduling system generates command sequences with absolute timestamps (e.g., "T1: 10:00:00.000, Conveyor line L1 starts; T2: 10:00:05.200, Hoist door opens on 1F"). Commands are issued to each equipment controller via industrial real-time Ethernet (e.g., PROFINET IRT, EtherCAT). The equipment controllers trigger local actions at precise absolute times based on the global clock.

[0141] Precise time stamping of status feedback: All trigger signals and status feedback from sensors (photoelectric, RFID readers, vision systems) are uploaded with precise source timestamps for the system to perform accurate status comparison and delay analysis.

[0142] 2. Key technologies to ensure seamless connection: Seamless connection refers to the smooth transfer of materials between the transfer area and the elevator inlet and outlet with "zero waiting time" or "extremely short time". The guarantee measures are as follows: (1) Feedforward and Coupling Control of Motion Trajectory: Speed ​​and Position Foresight: The connecting conveyor platform and the elevator do not move independently, but are controlled as a coordinated motion system. The meeting time and position of the handover point have been calculated in the planning stage. Speed ​​Curve Matching: When the material approaches the handover point, the speed of the connecting conveyor belt and the speed of the elevator car (or the speed of the inlet conveyor belt) are dynamically matched to ensure that the relative speed of the two is close to zero at the moment of handover.

[0143] (2) Real-time position fine-tuning and closed-loop compensation: High-precision position sensors (such as laser ranging and visual positioning) are deployed at key junctions (such as the outlet of the transfer area and the inlet of the elevator). When a millimeter-level deviation between the actual position and the expected position of the material is detected, the system calculates a small compensation amount of speed or time in real time and adjusts the action of the actuator through a fast control loop to ensure accurate alignment.

[0144] 3. Time-based state machine and handshake protocol: The handover process is controlled by a precisely time-synchronized state machine. For example: State S1 (Ready). Timestamp T1 Status S2 (Material arrived at the receiving area) Timestamp T2 State S3 (Lift door open, speed synchronize)... Each state transition relies on precise timing triggers and "ready" handshake signals from both devices. If either party is not ready, a predefined, very short dynamic wait or resynchronization process will be triggered.

[0145] 4. Network Latency Measurement and Compensation: The system periodically measures the network round-trip time (RTT) of control commands from the server to each terminal device. When issuing critical timestamp commands, advance compensation is performed based on the average network latency of the device to ensure that the execution time of the command at the target device is consistent with the system's expected time.

[0146] like Figure 7 As shown, the scheduling control module can also handle anomalies and make adaptive adjustments: when a hoist fails, the scheduling control module automatically switches to emergency scheduling mode, dynamically migrating the tasks of the failed hoist to another hoist and adjusting task priorities and paths to ensure continuous system operation (with a moderate reduction in efficiency). The scheduling control module has self-learning capabilities and can continuously optimize weight parameters and scheduling rules based on historical task data and equipment operation data to improve long-term operating efficiency.

[0147] Preferably, to increase the robustness of the system, the scheduling control module adopts a multi-level degradation strategy: Fully automatic intelligent scheduling mode (normal); Semi-automatic auxiliary scheduling mode (partial failure); Manual priority scheduling mode (critical fault); Emergency one-way passage mode (single machine operation); Preferably, such as Figure 7 As shown, the scheduling control module can also perform an adaptive learning mechanism based on historical performance data and optimize the cycle weekly: (1) collect the running data of the previous week (100,000+ task records); (2) analyze bottlenecks and abnormal patterns; (3) adjust the algorithm parameter weights; (4) simulate and verify the improvement effect; (5) deploy and update the scheduling rules.

[0148] To illustrate this solution more clearly, an example is provided below: Example of a scheduling system workflow: Task reception and parsing: WMS issues the task: "Power meter to the second floor assembly area, high priority".

[0149] The scheduling system analyzes task attributes: standard material size, light weight, and high urgency.

[0150] Data fusion and decision-making: The system query indicates that the first hoist is currently idle, while the second hoist is performing a descent task and is expected to release in 30 seconds.

[0151] The first connecting transport area has been detected as idle; the third connecting transport area is about to be released.

[0152] If the priority score of this task is higher than that of other tasks in the queue, it is decided to execute it immediately.

[0153] Command issuance and execution: The dispatch system sends instructions to the main transmission line on the first floor to guide the electricity meter box to the first connecting transmission area.

[0154] At the same time, a preparation command is sent to the first continuous hoist, with the target floor set to the second floor.

[0155] Once the elevator is in place, the connecting area connects with the elevator, and the material enters the elevator and moves upward.

[0156] Real-time monitoring and adjustment: During the lifting process, if a higher-priority downlink task arrives, the system can dynamically adjust the task order of the second lifter, but this will not affect the current uplink task.

[0157] The system records the execution time and device status of this task for subsequent scheduling and optimization.

[0158] Scenario: Mixed task scheduling during peak hours; Time: 10:00-11:00 AM (peak hours for warehouse entry and exit); Task Mixing: Upstream task: Electricity meter warehousing (medium priority), 30 boxes / hour; Downstream task: Finished product outbound (high priority), 40 boxes / hour; Urgent task: Replenish spare parts (highest priority), 5 boxes need to be processed immediately; Dispatch system response: 1. Identify urgent tasks → Immediately assign them the highest priority; 2. Analyze the current load: Upward hoist: 75% utilization, 3 tasks in queue; Downlink elevator: 82% utilization, 5 tasks in queue; 3. Conflict Prediction: A 45% probability of conflict was found within the 10:15 time window; 4. Adopt strategies: Emergency tasks should use the shortest path. Adjust the time offset of the three downlink tasks (delay by 2-3 minutes); Enable route optimization to route the two uplink tasks to the backup connection area; 5. Results: Emergency task completion time was reduced by 40%, and overall throughput increased by 18%.

[0159] In this invention, the first elevator is used to lift materials from the bottom layer to the top layer; the second elevator is used to transport materials from the top layer to the bottom layer; the scheduling and control module is used to acquire multi-source data from the first elevator, the second elevator, and the connecting conveyor platform, and to achieve coordinated scheduling of the first elevator, the second elevator, and the connecting conveyor platform based on the multi-source data and dynamic optimization strategies; wherein, the multi-source data includes at least task characteristic data, equipment status data, conveying environment perception data, and historical performance data, effectively solving the problem of low efficiency and reliability of cross-layer bidirectional connecting systems in automated storage and retrieval systems caused by existing technologies, and effectively improving the efficiency and reliability of cross-layer bidirectional connecting systems in automated storage and retrieval systems.

[0160] The technical solution of this invention creates two independent, unidirectional material flow channels by setting up two hoists and clearly defining their functions (one for upward movement and one for downward movement); it realizes the true concurrent execution of upward and downward tasks; the process of material moving from the second floor to the first floor and from the first floor to the second floor can be carried out simultaneously without interference, completely eliminating task conflicts and waiting time at the physical level, and improving the system's cross-floor material throughput capacity.

[0161] The data acquisition submodule in this invention includes a task feature data layer, an equipment status data layer, a conveying environment perception data layer, and a historical performance data layer. The task feature data layer is used to acquire multiple task attribute feature data; the equipment status data layer acquires multiple status data of a certain device; the conveying environment perception data layer is used to acquire the material density and task density of each floor's conveyor line; and the historical performance data layer is used to acquire the historical distribution of task completion time, the statistics of historical equipment failure frequency, and the historical processing efficiency matrix of different material types. This constructs a multi-layer data acquisition architecture, ensuring the comprehensiveness, real-time performance, and reliability of the scheduling and control module's decisions.

[0162] In this invention, the priority of a task is determined based on its task attribute characteristics, priority influencing factors, and corresponding weight coefficients. A conflict prediction model based on discrete event simulation and spatiotemporal state extrapolation is used to simulate and predict the likelihood of task conflicts between the first hoist, the second hoist, and the connecting area within a future period. The conflict level and corresponding conflict resolution strategy are determined based on the likelihood of conflict. The task allocation and transport path are adjusted according to the conflict resolution strategy, ensuring the reliability of task scheduling and improving its efficiency.

[0163] The technical solution of this invention determines the task conflict level and corresponding conflict resolution strategy based on the probability of task conflict, which further ensures the reliability of task scheduling and improves the efficiency of task scheduling.

[0164] The technical solution of this invention uses multi-source data from the data acquisition submodule, employs a dynamic priority scheduling algorithm, and combines it with a conflict prediction model to perform task allocation and path planning. Specifically, it also includes: real-time monitoring of the load rates of the first and second hoists, calculating the load difference based on the load rates of the first and second hoists, and optimizing and adjusting the tasks according to a preset load balancing strategy if the load difference exceeds a preset threshold, thereby ensuring the efficient and reliable operation of the first and second hoists.

[0165] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A cross-level bidirectional connection system for automated storage and retrieval systems, characterized in that, include: The system comprises a first hoist, a second hoist, a connecting conveyor platform, and a scheduling control module. The connecting conveyor platform includes a first and a fourth connecting conveyor area located in the bottom-level warehouse, and a second and a third connecting conveyor area located in the upper-level warehouse. The bottom-level inlet / outlet of the first hoist connects to the first connecting conveyor area in the bottom-level warehouse, and the upper-level outlet / inlet of the first hoist connects to the third connecting conveyor area in the upper-level warehouse, for lifting materials from the bottom level to the upper level. The upper-level inlet / outlet of the second hoist connects to the fourth connecting conveyor area in the upper-level warehouse, and the bottom-level outlet / inlet of the second hoist connects to the second connecting conveyor area in the bottom-level warehouse, for conveying materials from the upper level to the bottom level. The scheduling control module communicates with the first hoist, the second hoist, and the connecting conveyor platform to acquire multi-source data from the first hoist, the second hoist, and the connecting conveyor platform. Based on the multi-source data and dynamic optimization strategies, it achieves coordinated scheduling of the first hoist, the second hoist, and the connecting conveyor platform. The multi-source data includes at least task characteristic data, equipment status data, and conveying environment perception data.

2. The cross-level bidirectional connection system for automated storage and retrieval systems according to claim 1, characterized in that, The scheduling and control module includes a data acquisition submodule, a task decision submodule, and a control submodule. The data acquisition submodule is used to collect multi-source data from the first hoist, the second hoist, and the connecting conveyor platform. The task decision submodule is used to allocate tasks and plan paths based on the multi-source data from the data acquisition submodule, using a dynamic priority scheduling algorithm combined with a conflict prediction model. The control submodule is used to control the first hoist and the second hoist to perform conveying operations based on the task allocation and path planning from the task decision submodule.

3. A cross-level bidirectional connection system for an automated storage and retrieval system according to claim 2, characterized in that, The data acquisition submodule includes a task feature data layer, an equipment status data layer, a conveying environment perception data layer, and a historical performance data layer. The task feature data layer is used to acquire multiple task attribute feature data. The equipment status data layer acquires multiple status data of a certain equipment. The conveying environment perception data layer is used to acquire the material density and task density of each floor's conveyor line. The historical performance data layer is used to acquire the historical distribution of task completion time, the statistics of historical equipment failure frequency, and the historical processing efficiency matrix of different material types.

4. A cross-level bidirectional connection system for an automated storage and retrieval system according to claim 2, characterized in that, Based on multi-source data from the data acquisition submodule, a dynamic priority scheduling algorithm, combined with a conflict prediction model, is used for task allocation and path planning, specifically including: The priority of the task to be decided is determined based on the task attribute characteristics data, the factors affecting the task priority, and the corresponding weight coefficients. Based on the conflict prediction model of discrete event simulation and spatiotemporal state extrapolation, the simulation predicts the possibility of mission conflicts between the first elevator, the second elevator and the docking area in the future. Determine the level of task conflict and the corresponding conflict resolution strategy based on the likelihood of task conflict. Adjust task allocation and delivery paths according to conflict resolution strategies.

5. A cross-level bidirectional connection system for an automated storage and retrieval system according to claim 4, characterized in that, Based on the conflict prediction model of discrete event simulation and spatiotemporal state extrapolation, the simulation predicts the possibility of task conflicts between the first elevator, the second elevator and the connecting area in the future. Specifically, it includes adding all assigned and scheduled tasks to the simulation queue in chronological order. Each task includes: starting floor, ending floor, expected start time, expected end time, elevator and connecting area number used. A time-space occupancy table is set up for each hoist and each connecting area to record the time occupancy of each hoist and each connecting area in future time periods. When a new task arrives or the task status is updated, the conflict probability between tasks is calculated based on the simulation time window and simulation path cross-analysis. The specific calculation method for the conflict probability between tasks is: the product of the collision time overlap ratio × the equipment load coefficient × the path sharing factor. Wherein, the collision time overlap ratio is the ratio of the overlap time length to the total coverage time length; the equipment load coefficient is the sum of the initial value and the load rate ratio, where the load rate ratio is the product of the load sensitive parameter and the current load rate of the equipment; the path sharing factor is the ratio of the first weighted sum and the second weighted sum. The first weighted sum is the weighted sum of the number of identical nodes in the planned paths of the two tasks and the corresponding node weights, and the second weighted sum is the weighted sum of the number of non-overlapping nodes in the planned paths of the two tasks and the corresponding node weights.

6. A cross-level bidirectional connection system for an automated storage and retrieval system according to claim 4, characterized in that, Determining the task conflict level and corresponding conflict resolution strategies based on the likelihood of task conflicts specifically includes: If the probability of a task conflict is less than the first preset probability threshold, the task conflict level is no conflict, and the assignment of the task to be assigned is executed directly. If the probability of a task conflict is not less than the first preset probability threshold and is less than the second preset probability threshold, the task conflict level is a minor conflict, and the task to be assigned will be delayed for a preset time before being assigned. If the probability of a task conflict is not less than the second preset probability threshold and is less than the third preset probability threshold, the task conflict level is medium conflict, and the task to be assigned will be re-optimized. If the probability of a task conflict is not less than the third preset probability threshold and less than the fourth preset probability threshold, the task conflict level is severe conflict, and the task to be assigned will be split. If the probability of a task conflict is not less than the fourth preset probability threshold, and the task conflict level is deadlock conflict, a system alarm will be issued for the tasks to be assigned.

7. A cross-level bidirectional connection system for an automated storage and retrieval system according to claim 6, characterized in that, Re-optimizing the paths of tasks to be assigned specifically includes: Based on task priority, equipment availability time, and path travel time, an optimal time window is calculated for each task to be optimized. The optimal time window includes: the docking preparation period, the hoist occupancy period, and the exit docking period. Plan a unique path for each task to be optimized. The path includes: starting connection area number, elevator number used, target connection area number, and floor conveyor route. The conflict probability between tasks is recalculated based on the optimal time window and the planned path. If the conflict level is less than that of a moderate conflict, resources are locked and a scheduling instruction set is generated. If the conflict level is not less than that of a moderate conflict, the optimal time window and the planned path are optimized again. Before executing the task to be optimized, mark the time window of all nodes in the planned path; other tasks need to avoid locked path nodes when planning. Once the optimization task is completed, release all locked path nodes.

8. A cross-level bidirectional connection system for an automated storage and retrieval system according to claim 6, characterized in that, The process of splitting tasks to be assigned specifically includes: splitting tasks based on task characteristics and conflict types, using one or more of the following methods; these methods include splitting by quantity, splitting by path, and splitting by time window. Splitting by quantity specifically involves dividing a batch of materials into multiple transportation sub-batches; splitting by path specifically involves decomposing the task into sub-tasks that sequentially pass through different physical paths or transfer zones; and splitting by time window specifically involves dividing a time-consuming task into multiple sub-tasks that occupy short time windows. All child tasks originating from the same parent task are assigned the same task group ID, and the child tasks initially inherit the priority of the parent task.

9. A cross-level bidirectional connection system for an automated storage and retrieval system according to claim 4, characterized in that, Based on multi-source data from the data acquisition submodule, a dynamic priority scheduling algorithm, combined with a conflict prediction model, is used for task allocation and path planning. Specifically, this includes: The load rates of the first and second hoists are monitored in real time. The load difference is calculated based on the load rates of the first and second hoists. If the load difference exceeds the preset difference threshold, the task is optimized and adjusted according to the preset load balancing strategy.

10. A cross-level bidirectional connection system for an automated storage and retrieval system according to claim 9, characterized in that, Optimizing and adjusting tasks according to the preset load balancing strategy specifically includes: If the difference between the utilization rate of the first hoist and the utilization rate of the second hoist is greater than a preset difference threshold, some uplink tasks will be transferred to the second hoist for execution. If the difference between the utilization rate of the second hoist and the utilization rate of the first hoist is greater than a preset difference threshold, a symmetrical processing strategy is adopted to transfer some downlink tasks to the first hoist for execution. The specific screening conditions for transferable tasks include: task attribute conditions, task current status conditions, and transfer contribution conditions. The task attribute conditions include low task priority, task that allows bidirectional scheduling, or task path selectability. The task current status conditions include the task that is currently assigned but has not started execution. The transfer contribution conditions include the task to be transferred having a high contribution to reducing the load on the original equipment and improving the utilization rate of the target equipment.

Images