Iot-based warehouse data comprehensive management system, method, device and medium

Abstract

The application relates to the technical field of Internet of Things warehouse management, and discloses a warehouse data comprehensive management system, method, equipment and medium based on the Internet of Things, which is characterized in that: firstly, warehouse basic environment and main body modeling are carried out, including establishing a warehouse grid map, an AGV cluster and a task set model; a global space-time occupation ST(v, t) table is constructed according to AGV empty running rate, conflicts, delays and energy consumption, and a multi-objective optimization function is constructed; the generated AGV carrying tasks are sorted according to priority and deadline, and the tasks are auctioned according to task application values, the tasks with larger application values obtain the tasks, and space-time resources are locked; lower local re-planning is carried out, the AGV obtaining the tasks travels along a pre-planned space-time path, and periodic detection is carried out during path execution, whether obstacles / conflicts / lockups or task changes appear is judged, and a new path is generated in combination with the global ST(v, t) table and real-time obstacles. The application solves the contradiction of multi-AGV collaborative task conflicts, and avoids the problem that traditional algorithms can only stop or invalidly bypass when encountering lockups, so that the overall efficiency is low.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) warehouse management technology, specifically to an IoT-based integrated warehouse data management system, method, equipment, and medium. Background Technology

[0002] Automated Guided Vehicles (AGVs) are the core execution units of modern intelligent warehousing systems, serving as crucial carriers for automating, digitizing, and intelligently managing the entire lifecycle of goods from inbound to outbound. They completely replace repetitive manual labor in the inbound, storage, heavy-duty handling, outbound, and picking stages, achieving a fully unmanned, closed-loop operation of goods. The core competition in the warehousing industry is essentially a competition of turnover efficiency and operating costs. AGVs fundamentally solve the core pain points of traditional manual warehousing, achieving long-term, sustainable cost reduction and efficiency improvement.

[0003] However, traditional warehouse AGVs still rely on traditional A* and Dijkstra algorithms and pre-set maps during operation, making them unable to dynamically adapt to temporary obstacles (such as goods stacking or equipment failure), leading to AGV deadlocks or detours. Warehouses suffer from rigid path planning and high AGV empty-running rates. At the same time, multi-AGV collaboration is inefficient, lacking global optimization capabilities. Task allocation conflicts and frequent path intersections between multiple AGVs result in low overall operational efficiency and make them unsuitable for the complex operation and control of AGVs in large warehouses.

[0004] Therefore, to address the above issues, we need a comprehensive warehouse data management system, methods, equipment, and media based on the Internet of Things. Summary of the Invention

[0005] The purpose of this invention is to provide a comprehensive warehouse data management system, method, equipment, and medium based on the Internet of Things (IoT). This invention overcomes the shortcomings of traditional decoupling in path planning by using conflict-free path pre-planning as a prerequisite for bidding, reducing path intersections and conflicts at the source of allocation, and directly optimizing empty-running rates by coupling the time dimension with the global dimension. The table not only adapts to dynamic temporary obstacles but also avoids global conflicts caused by distributed replanning, resolving the contradiction of multi-AGV collaborative task conflicts and avoiding the problem of traditional algorithms having to stop or ineffectively detour when encountering deadlocks, resulting in low overall efficiency.

[0006] This invention is implemented as follows: This invention provides a comprehensive warehouse data management method based on the Internet of Things: First, the warehouse's basic environment and main structure are modeled, including the creation of a warehouse grid map, AGV cluster, and task set model. Based on the warehouse grid map, AGV cluster, and task set model, the AGV empty run rate, conflict, delay, and energy consumption are calculated, and a global spatiotemporal occupancy table is constructed. And construct a multi-objective optimization function; The generated AGV transport tasks will be sorted by priority and deadline, and the AGVs will be sorted by idle status and battery level. In the upper-level global scheduling, the AGV first pre-plans the spatiotemporal path according to the transport task, then calculates the task request value for each task. In the upper-level global scheduling, the tasks are auctioned according to the task request value, and the one with the larger request value wins the task, locking the spatiotemporal resources and generating a task allocation scheme. The lower-level local replanning allows the AGV to travel along the pre-planned spatiotemporal path and periodically check during path execution to determine if there are obstacles / conflicts / deadlocks or task changes. If there are no abnormalities, it continues to travel; if there are abnormalities, it triggers replanning. The search proceeds simultaneously from the current point to the target point, constrained by the spatiotemporal node positions and time, and combined with global parameters. The table generates new paths based on real-time obstacles.

[0007] Furthermore, the warehouse grid map model is as follows: ; in, This is a 2D raster map model of the warehouse. For a set of grid nodes, Let be the set of edges between adjacent nodes; where node ∈ Corresponding coordinates ;side ∈ Connectivity ; The AGV cluster is: in, For AGV cluster collection, For the first Unit AGV; n is the total number of AGVs, and is defined as follows: For AGV cart The current position node, where ∈ . AGV (Automated Guided Vehicle) Maximum driving speed and minimum turning radius; Task set model is ; in, For a set of tasks, Let m be the j-th transportation task, and m be the total number of tasks. and define , Tasks Pick-up and unloading points , ∈ ; For the task The deadline; For the task Effective transportation mileage; ∆t is the time window granularity / AGV control cycle; ∆T is the global data synchronization cycle; Assign 0-1 variables to the AGV-task; where =1 indicates that the AGV (Automated Guided Vehicle) is... Assigned to task Otherwise, it is 0.

[0008] Furthermore, the multi-objective optimization function includes the total cost function of empty mileage. Total cost function of path conflicts Total cost function of task delay Total energy cost function ;in, The expression is as follows: in, ; This refers to the empty mileage. The expression is as follows: in, for and ; The cost of a single conflict; Total cost function of task delay The expression is as follows: in, For the task Priority coefficient, ∈ The larger the value, the higher the priority. For the task The actual completion time; For the task The deadline for completion; Total energy cost function The expression is as follows: in, For the task Effective transportation mileage; ; in, For AGV cart The normalized value of the remaining battery power. , For AGV cart Energy consumption coefficient per unit mileage; The global multi-objective optimization objective function for the multi-objective optimization task is calculated as follows: in, , , , Let be the weights of each sub-objective, and + + =1.

[0009] Furthermore, in the aforementioned upper-level global scheduling, the global state is first initialized and data is preprocessed; the inputs include the warehouse map G, the AGV cluster state A, the task set T, and the weight coefficients. - The time granularity is Δt; then the map is preprocessed to mark restricted areas, one-way passages, and shelving areas, generating an adjacency matrix M; M ij = For connectivity, M ij If the sum is +∞, then the connection is not connected; Task sorting is based on priority. Descending order, with the same priority ordered by Ascending order, generate a sorting task queue. AGVs are sorted by idle status and remaining battery power. Descending order, generate a queue of available AGVs. .

[0010] Furthermore, in the upper-level global scheduling, tasks are auctioned based on their request values, with the AGV with the highest request value winning the task. A valid request value is only given when the AGV can plan a conflict-free spatiotemporal path. The expression is as follows: in, The cost of pre-planned path conflicts, The cost of task delay; in, As a unit of conflict cost, for implement The pre-planned spatiotemporal path; The task completion time for the AGV to plan a conflict-free spatiotemporal path is as follows: in, For AGV The average driving speed.

[0011] The auction based on task application value begins with auction initialization, setting the number of rounds. =0, maximum number of rounds =100; Initial price vector for the task =0; Allocation matrix =0, winning set ; Task broadcasting and AGV bidding; right Each unassigned task ,Towards Broadcast full task information for each AGV After receiving the task, first based on the current situation... The table uses the spatiotemporal A* algorithm to plan from... Then Conflict-free spacetime path If the plan fails... =-∞, abandon the bid; among the valid bids for AGVs, select the highest bidder. The corresponding AGV is ; Among them, if > Update task application value = - ; Temporarily Assigned to ;renew ; and will Node-time pairs in The column marked 2 in the table represents pre-occupancy. like ≤ The task failed to sell in this round and will proceed to the next round. Repeat the task auction process based on task request value in the upper-level global scheduling until all tasks are allocated; the auction will terminate if all available AGVs are allocated, the maximum number of rounds is reached, or no task is won in 3 consecutive rounds. When a newly added AGV completes its task and becomes idle, an incremental auction is triggered. This process, which auctions tasks based on their request values ​​in the upper-level global scheduling, only applies to newly added tasks and idle AGVs. No global recalculation is required, ensuring real-time scheduling. The updated allocation results are consistent with... The table is synchronized to the entire system.

[0012] Furthermore, the aforementioned lower-level local replanning involves the AGV performing a perception and judgment every Δt cycle while traveling along the pre-planned path. Nodes ahead on the pre-planned path are detected as temporary obstacles (cargo stacking, equipment malfunction, personnel intrusion) by onboard sensors (laser / vision). Globally... After the table is updated, if a node / edge conflict occurs in the pre-planned path, or a deadlock risk is detected (one-way channel travel in opposite directions, loop waiting); or the AGV's position deviates from the pre-planned path by more than the threshold δ (taken as 0.5m), and it cannot return to the original path; or if the task status changes (cancellation, priority adjustment, destination change), a replanning of the path is triggered. The replanning signal is triggered, outputting the local obstacle map Glocal(k) (a 5m grid map around the AGV) and the current global obstacle map. Table, AGV Current Status , Replanning target points (Pick-up / unloading point).

[0013] Furthermore, this invention provides a comprehensive warehouse data management system based on the Internet of Things (IoT). The method for integrated warehouse data management based on the Internet of Things, as described above, is used to execute any one of the above-mentioned methods; it includes a map modeling and system initialization module, which provides a standardized environment model and initial parameters for the entire system; The global spatiotemporal state management module receives data input from all modules in the system and status update requests from control modules, and synchronizes the latest data. The table extends to all modules across the entire system; The global task optimization and auction scheduling module obtains the latest information from the spatiotemporal state module. The table obtains real-time status and sensor data from the AGV control module, reports the newly planned path to the global auction module and the spatiotemporal status module, and synchronizes it to the AGV control module for execution. The global conflict detection and deadlock resolution module is used to implement a hierarchical conflict and deadlock resolution mechanism in the algorithm of this invention, and its core solution is to address path conflicts and cyclic deadlock problems in multi-AGV collaboration. The warehousing business system interface module serves as the interface entry point between the scheduling system and the warehousing business system, enabling the scheduling method of this invention to be integrated and executed with the warehousing business process data.

[0014] Furthermore, the present invention provides an IoT-based warehouse data integrated management device, the management device including at least one processor; and a memory communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the IoT-based integrated warehouse data management method as described above.

[0015] Furthermore, the present invention provides a computer storage medium storing a computer program that can run on a processor, wherein the processor executes the computer program to implement the IoT-based integrated warehouse data management method described above.

[0016] Compared with the prior art, the beneficial effects of the present invention are: It overcomes the shortcomings of traditional decoupling from path planning by using conflict-free path pre-planning as a prerequisite for bidding, thereby reducing path intersections and conflicts from the source of allocation and greatly optimizing the empty running rate of AGVs.

[0017] The bidirectional spatiotemporal dynamic search of this invention, performed using the algorithm of this invention, significantly improves the search speed while simultaneously coupling the time dimension and the global dimension. The table not only adapts to dynamic temporary obstacles, but also avoids global conflicts caused by distributed replanning.

[0018] This invention achieves end-to-end data closed-loop operation through global spatiotemporal occupancy. The table enables seamless data integration between upper-level global optimization and lower-level local replanning, balancing global operational efficiency with local dynamic adaptability, and resolving the contradiction between global optimization and real-time response in multi-AGV collaboration.

[0019] The hierarchical deadlock resolution mechanism of this invention covers the handling of all scenarios from minor conflicts to severe deadlocks, avoiding the problem that traditional algorithms can only stop or make ineffective detours when encountering deadlocks, resulting in low carrying efficiency. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system structure diagram of the present invention; Figure 3 This is a flowchart of the method according to Embodiment 2 of the present invention. Figure 4 This is a structural diagram of the management device of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to describe selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figure 1 This invention provides a comprehensive warehouse data management method based on the Internet of Things: Example 1: The specific method flow of this example is as follows: I. Initialization Phase: Establish warehouse grid map, AGV cluster, and task set model, and then construct global spatiotemporal occupancy. The table is used to construct a multi-objective optimization function, which includes empty run rate, conflict, delay, and energy consumption.

[0024] Second, the upper-level global scheduling sorts tasks by priority and deadline, sorts AGVs by idle status and battery level, pre-plans the time and space path for AGVs, calculates their value, and conducts an ascending-bid auction, with the highest bidder winning the task, locking time and space resources, and generating a task allocation scheme.

[0025] 3. AGV Path Execution and Real-time Monitoring: The AGV travels along the pre-planned path, and periodic checks are performed during the journey to determine if obstacles / conflicts / deadlocks / task changes occur; if no abnormalities are found, the journey continues; if abnormalities are found, replanning is triggered. IV. Lower-level local replanning involves simultaneous bidirectional searches from the current point and the target point, constrained by spatiotemporal node positions and time, and combined with global considerations. The table generates new paths based on real-time obstacles.

[0026] V. Global Conflict and Deadlock Resolution: For Level 1 conflicts, low-priority AGVs wait; for Level 2 conflicts, AGVs detour and replan; for Level 3 deadlocks, priority is avoided and local rescheduling is implemented; extreme cases trigger global task reassignment.

[0027] VI. Iterative closed-loop update of global spacetime occupancy The table shows how to release resources when a task is completed, and how to continue execution / replanning if the task is not completed, forming a data processing loop of global planning, local dynamic adjustment, and global synchronization.

[0028] The method flow of this embodiment includes: First, the warehouse's basic environment and main structure are modeled, including the creation of a warehouse grid map, AGV cluster, and task set model. The warehouse grid map model is as follows: ; in, Warehouse 2D raster map model For a set of grid nodes, Let be the set of edges between adjacent nodes; where node ∈ Corresponding coordinates ;side ∈ Connectivity ; The AGV cluster and task set model is as follows: in, For AGV cluster collection, For the first Unit AGV; n is the total number of AGVs, and is defined as follows: For AGV cart The current position node, where ∈ . AGV (Automated Guided Vehicle) Maximum driving speed and minimum turning radius; Task set model is ; in, For a set of tasks, Let m be the j-th transportation task, and m be the total number of tasks. and define , Tasks Pick-up and unloading points , ∈ ; For the task The deadline; For the task The effective transportation mileage; ∆t is the time window granularity / AGV control cycle, usually taken as 100ms, matching the frequency of the AGV on-board controller; ∆T is the global data synchronization cycle, usually taken as 100ms, aligned with ∆t; Assign 0-1 variables to the AGV-task; where =1 indicates that the AGV (Automated Guided Vehicle) is... Assigned to task Otherwise, it is 0.

[0029] Based on the warehouse grid map, AGV cluster, and task set model, the AGV empty run rate, conflict, delay, and energy consumption are calculated, and a global spatiotemporal occupancy table is constructed. And construct a multi-objective optimization function; Multi-objective optimization functions include the total cost function of empty mileage. Total cost function of path conflicts Total cost function of task delay Total energy cost function ;in, The expression is as follows: in, ; This refers to the empty mileage. The expression is as follows: in, for and ; The cost of a single conflict; Total cost function of task delay The expression is as follows: in, For the task Priority coefficient, ∈ The larger the value, the higher the priority. For the task The actual completion time; For the task The deadline for completion; Total energy cost function The expression is as follows: in, For the task Effective transportation mileage; ; in, For AGV cart The normalized value of the remaining battery power. , For AGV cart Energy consumption coefficient per unit mileage; The global multi-objective optimization objective function for the multi-objective optimization task is calculated as follows: in, , , , Let be the weights of each sub-objective, and + + =1.

[0030] The generated AGV transport tasks are sorted by priority and deadline, and the AGVs are sorted by idle status and battery level for upper-level global scheduling. The AGV first pre-plans the spatiotemporal path according to the transport task, then calculates the task request value for each task. In the upper-level global scheduling, the tasks are auctioned according to the task request value. The one with the larger request value wins the task, locks the spatiotemporal resources, and generates a task allocation scheme. In this embodiment, the core solution to the problems of multiple AGV task allocation conflicts, global path intersections, and high empty running rates in the upper-level global scheduling is to use the pre-planning of spatiotemporally conflict-free paths as a prerequisite for AGV bidding, thereby avoiding the shortcomings of traditional methods that only allocate tasks without considering path feasibility.

[0031] First, the global state is initialized and data is preprocessed; inputs include the warehouse map G, the AGV cluster state A, the task set T, and the weight coefficients. - The time granularity is Δt; then the map is preprocessed to mark restricted areas, one-way passages, and shelving areas, generating an adjacency matrix M; M ij = For connectivity, M ij If the sum is +∞, then the connection is not connected; Task sorting is based on priority. Descending order, with the same priority ordered by Ascending order, generate a sorting task queue. ; Among them, AGVs are sorted by idle status and remaining power. Descending order, generate a queue of available AGVs. In this step, the initial global spacetime occupancy is generated. The table, with all values ​​initially set to 0, along with the preprocessed map adjacency matrix M, sorted task and AGV queues, serves as the basic input data for subsequent auctions and path planning in this embodiment.

[0032] In this embodiment, the task is auctioned based on the task request value in the upper-level global scheduling. The AGV with the higher request value obtains the task. A valid request value is only given when the AGV can plan a conflict-free spatiotemporal path. The expression is as follows: in, The cost of pre-planned path conflicts, The cost of task delay; in, As a unit of conflict cost, for implement The pre-planned spatiotemporal path; The task completion time for the AGV to plan a conflict-free spatiotemporal path is as follows: in, For AGV The average driving speed.

[0033] In this embodiment, the auction based on task application value first involves auction initialization, setting the number of rounds. =0, maximum number of rounds =100; Initial price vector for the task =0; Allocation matrix =0, winning set ; Task broadcasting and AGV bidding; right Each unassigned task ,Towards Broadcast full task information for each AGV After receiving the task, first based on the current situation... The table uses the spatiotemporal A* algorithm to plan from... Then Conflict-free spacetime path If the plan fails... =-∞, abandon the bid; among the valid bids for AGVs, select the highest bidder. The corresponding AGV is ; Among them, if > Update task application value = - ; Temporarily Assigned to ;renew ; and will Node-time pairs in The column marked 2 in the table represents pre-occupancy. like ≤ The task failed to sell in this round and will proceed to the next round. In this embodiment, the above process is repeated until all tasks are assigned; the auction terminates when all available AGVs are assigned, the maximum number of rounds is reached, or no tasks are won in three consecutive rounds; through the above process, the final allocation matrix X is locked, and the winning paths are... The table status is updated from 2 to 1 (officially occupied), the pre-occupancy status of the failed auction task is rolled back to 0, and the assigned task and pre-planned spatiotemporal path Path(k) are issued to each AGV. The table is synchronized to all AGVs, completing the data distribution for the upper-level and subsequent lower-level global planning.

[0034] In this embodiment, when a newly added AGV completes its task and becomes idle, an incremental auction is triggered. The auction process described above is only performed on newly added tasks and idle AGVs; no global recalculation is required, ensuring real-time scheduling. The updated allocation result is consistent with... The table is synchronized to the entire system.

[0035] In this embodiment, the lower-level local replanning obtains the task's AGV to travel along the pre-planned spatiotemporal path, and periodically checks during path execution to determine whether there are obstacles / conflicts / deadlocks or task changes. If there are no abnormalities, it continues to travel; if there are abnormalities, it triggers replanning. The search proceeds simultaneously from the current point to the target point, constrained by the spatiotemporal node positions and time, and combined with global parameters. The table generates new paths based on real-time obstacles.

[0036] First, when the AGV travels along the pre-planned path, it performs a perception and judgment operation every Δt period. At nodes ahead on the pre-planned path, it is detected by onboard sensors. For example, if a temporary obstacle is detected by laser / vision, the data is transmitted via the Internet of Things to the global conflict detection and deadlock resolution module in the system server. Examples of obstacles include cargo stacking, equipment failure, and personnel intrusion. After the table is updated, if a node / edge conflict occurs in the pre-planned path, or a deadlock risk is detected (e.g., one-way travel in opposite directions, looping and waiting), or if the user's position deviates from the pre-planned path by more than a threshold δ (0.5m), making it impossible to return to the original path, or if the task status changes (e.g., cancellation, priority adjustment, destination change), then a replanning path is triggered. A replanning signal is triggered, and a local obstacle map is output. (Grid map within 5m radius of AGV), Current Global Table, AGV Current Status , Replanning target points (Pick-up / unloading point).

[0037] When replanning the route, each search node Coupled spatial location With arrival time This avoids spatiotemporal conflicts at the root; the bidirectional search architecture allows for a forward search F starting from the current position. To the target point Search, reverse search B from Towards The search terminates when the forward and reverse searches meet; the search space is only N of a unidirectional A*, and the speed is increased by more than 2 times.

[0038] The total search cost is expressed as follows: in, From the starting point to the node The actual cumulative passage cost includes edge weight, turning cost, and waiting cost; For nodes The minimum time cost to reach the target point, a forward heuristic function, satisfying admissibility and guaranteeing path optimality, is expressed as follows: The cost of spacetime conflict is expressed as follows: in, Obstacle markers are used for local map mapping, with 0 indicating no obstacles and 1 indicating obstacles. If the value is +∞, entry into the obstacle node is prohibited.

[0039] Then perform a reverse search; the cost function expression is as follows: in, To start from the target point To the node The cumulative cost to peers; The cost of spacetime conflict; The reverse heuristic function is expressed as follows: ; Then perform a bidirectional search, starting with the forward search and opening the list first. Close list , the starting point join in , ; Reverse search first open list Close list , target point join in , ; Initialize the meeting node Minimum total cost .

[0040] Then, a bidirectional alternating expansion search is performed, with selections made in each round. and Expand in directions with fewer nodes to balance search progress: During the forward expansion, from take out smallest node ,join in ; For the determination of meeting in bidirectional search, if Calculate the total cost: like < ,renew = , =n; The judgment will terminate upon encounter; if ,and The search will terminate if the condition is met. During the search process, the adjacent nodes are expanded to include 8 neighborhood nodes and nodes waiting in place (v remains unchanged, t+Δt). Kinematic constraints (turning radius, velocity) and obstacle constraints are checked, the cost of new nodes is calculated, and the results are updated. With the parent node; It then performs a reverse expansion, symmetrical to the forward expansion, extending from the target point towards the starting point and updating... and ; Then, path generation and smoothing are performed. If the search is successful, then from Forward backtracking Get the path Backtracking Get the path After merging and removing duplicate nodes, the spatiotemporal replanning path is obtained, expressed as follows: And perform kinematic smoothing to remove redundant inflection points; If the search fails, a deadlock alarm is triggered, which is reported to the global scheduling system and the deadlock resolution process begins. In this embodiment, the AGV will perform path synchronization and execution. Report to the global scheduling system for system updates. Table: Release the original path's occupied state, set to 0, and set the new path's nodes and time to 1.

[0041] Updated The table is synchronized to all AGVs via a global data bus to avoid cross-AGV path conflicts.

[0042] AGV Press The process repeats the above steps in a loop to achieve real-time dynamic adaptation. After replanning... The table update is synchronized to the upper-level global scheduler, and the latest global state is used directly when a new task is auctioned, forming a closed loop.

[0043] In this embodiment, a global-local closed-loop feedback and conflict resolution mechanism is established, with a cycle time of [missing information] for the upper and lower layers of the entire embodiment. The real-time data bus synchronizes all core data and serves as the sole data source for all modules, ensuring data consistency. Includes real-time position, speed, task status, and remaining battery power of all AGVs; global time and space occupancy. The table shows the latest status; the execution status of tasks, information on adding / canceling tasks; AGV replanning requests, conflict alarms, deadlock requests, and a global real-time data synchronization bus is established.

[0044] In this embodiment, after establishing a global real-time data synchronization bus, deadlock detection is performed on global conflicts, specifically including each A full scan is performed periodically. The full scan includes: Node conflict detection, inspection Does the table contain a single (v,t) value occupied by two or more AGVs? Edge conflict detection checks whether two AGVs occupy the same edge in opposite directions during the same time window. Deadlock detection involves constructing a directed graph D of AGV waiting nodes, where each node represents an AGV. to ,express The path was If a cycle exists in D, it is marked as deadlocked.

[0045] The detected deadlocks are classified according to their conflict levels, and deadlock resolution strategies are then planned based on these conflict levels, as shown in Table 1: Table 1. Hierarchical Conflict Deadlock Resolution Strategies In this embodiment, if a temporary obstacle persists for more than 5 minutes, it is determined to be a semi-permanent obstacle. The adjacency matrix M of the global map G is updated, and the subsequent pre-planning of module 1 directly uses the updated map to avoid repeated replanning, improve scheduling efficiency, and iteratively update the global map.

[0046] In this embodiment, as Figure 2 This invention provides an IoT-based warehouse data integrated management system for executing the IoT-based warehouse data integrated management method described above; it includes a map modeling and system initialization module, which provides a standardized environment model and initial parameters for the entire system; The global spatiotemporal state management module receives data input from all modules in the system and status update requests from control modules, and synchronizes the latest data. The table extends to all modules across the entire system; The global task optimization and auction scheduling module obtains the latest information from the spatiotemporal state module. The table obtains real-time status and sensor data from the AGV control module, reports the newly planned path to the global auction module and the spatiotemporal status module, and synchronizes it to the AGV control module for execution. The global conflict detection and deadlock resolution module is used to implement a hierarchical conflict and deadlock resolution mechanism in the algorithm of this invention, which is the core solution to path conflicts and cyclic deadlock problems in multi-AGV collaboration.

[0047] The warehousing business system interface module serves as the interface entry point between the scheduling system and the warehousing business system, enabling the scheduling method of this invention to be integrated and executed with the warehousing business process data.

[0048] In this embodiment, as Figure 4 The present invention provides a warehouse data integrated management device based on the Internet of Things, the management device including at least one processor; and a memory communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the IoT-based integrated warehouse data management method as described above.

[0049] In this embodiment, the present invention provides a computer storage medium storing a computer program that can run on a processor. When the processor executes the computer program, it implements the IoT-based warehouse data integrated management method described above.

[0050] Example 2, as Figure 3 ; This embodiment is basically the same as the above embodiment, except that: in the upper-level global scheduling, the task auction is carried out according to the task request value using a genetic strategy; Specifically, a genetic strategy is used for AGV task allocation, retaining the path pre-planning prerequisites and ST table global constraints of the algorithm in Example 1, while simultaneously achieving a multi-objective global Pareto optimal solution. In this example, Input warehouse map G, AGV cluster status A, task set T, global spatiotemporal occupancy table ST, multi-objective weights λ1-λ4, and maximum number of generations G. max Population size N, crossover probability P c Probability of mutation P m .

[0051] First, initialize the population: randomly generate N integer-coded individuals that meet the hard constraints of task allocation to form the initial population P0, and set the number of generations t=0; Individual spatiotemporal feasibility verification and fitness calculation are performed. Specifically, the ind of each individual in the population is decoded to obtain the AGV-task allocation matrix X, clarifying the task sequence undertaken by each AGV; for each AGV, based on the current... The table calls the full-process conflict-free spatiotemporal path from Example 1. Map the AGV task allocation scheme to chromosome codes that can be processed by a genetic algorithm; the expression is as follows: in, ∈[1,n], representing the j-th task t j Assigned to number AGV ; Chromosome length L equals the total number of tasks m. Each gene position corresponds to one task, and the gene value corresponds to the AGV number that undertakes the task. For each AGV, a feasibility assessment is performed. If there is no feasible path for the AGV or the path has a spatiotemporal conflict, the individual is determined to be an illegal individual, and the fitness of its four sub-objectives is assigned to +∞, and it is directly eliminated.

[0052] Fitness calculation: For a legitimate individual, calculate the fitness values ​​for the four sub-objectives; the expression is as follows: Among them, for the sub-objective of total empty-run cost, the optimized empty-run rate is expressed as follows: in, = ; for AGV To task t j Empty mileage at the pickup point; Let Euclidean distance norm be 2D. Obtained by decoding chromosome ind; The sub-goal of total path conflict cost is expressed as follows: in, In this embodiment, the AGV and The number of pre-planned path conflicts, The cost of a single conflict; The sub-objective expression for the total cost of task delay is as follows: in, For task t in this embodiment... j The estimated completion time, For task t j Priority coefficient, The deadline for task completion; The expression for the total energy cost sub-objective is as follows: in, For task t j Effective transport mileage For AGV Energy consumption coefficient per unit mileage For AGV The normalized value of the remaining battery power. This is a penalty for low battery levels. In this embodiment, a single task must be assigned to one and only one AGV, with no duplicate assignments and no omissions, ensuring the business feasibility and spatiotemporal non-conflict of the assignment scheme, which matches the core constraints of Embodiment 1. The expression is as follows: Furthermore, a constraint is imposed to ensure that the number of tasks a single AGV can undertake does not exceed its maximum task capacity, in order to avoid overload. The expression is as follows: Furthermore, there are no conflict constraints on spatiotemporal paths. In this embodiment, the pre-planned path for each AGV is... All spatiotemporal nodes occupy global spatiotemporal space. The table shows an idle state with no pre-collision. In this embodiment, non-dominated sorting and crowding calculation are performed again for the current population. The non-dominated hierarchy is divided according to the Pareto dominance rule, and a value is assigned to each individual. .

[0053] For individuals of the same non-dominant rank, calculate crowding degree. For two individuals in the population and It is true if and only if the following two conditions are met simultaneously. Dominate Condition 1: Under all sub-targets, Its fitness is not inferior to The expression is as follows: Condition 2: There exists at least one sub-objective. Its fitness is strictly better than The expression is as follows: The expression for non-dominant hierarchy is as follows: in, =1 represents the Pareto optimal frontier, which has the highest level. The larger the size, the less dominant an individual is, and all individuals that do not dominate each other are grouped into the same category.

[0054] The congestion level is calculated using the following expression: in, , The first The maximum and minimum values ​​of each sub-objective in the current non-dominated level; , In the current level, by number After sorting the sub-targets The next and previous individuals; in this embodiment, the crowding degree d=+∞ for the maximum and minimum values ​​of each sub-objective of the boundary individual, ensuring that the optimal boundary individual is retained. Output the non-dominated level and crowding degree of each individual to provide the core basis for individual selection; In this embodiment, an elite preservation strategy is adopted, specifically as follows: the current population... With offspring population Merging them yields a merged population of size 2N. ; according to From childhood to adulthood, When d is the same, the rule of descending d applies to merging populations. Sort; The top N individuals are selected based on the sorting results to form the next generation of parent population. This ensures that the best individuals are not lost while maintaining population diversity. The next generation of parent population is then output. This provides parental samples for genetic manipulation.

[0055] In this embodiment, the parent population Crossover and mutation genetic operations; Specifically, regarding the parent population According to the tournament selection rules, parent pairs of individuals are selected for breeding. Specifically, based on crossover probability For parent individuals, the execution order is crossed over. Generate offspring individuals to avoid illegal solutions; Specifically, according to the probability of mutation Perform bit mutation operations on offspring individuals to avoid premature convergence of the population; Specifically, all offspring individuals must satisfy the hard constraint of task allocation; invalid individuals are directly regenerated. The output is a offspring population of size N. Then, it enters the next evolutionary cycle.

[0056] In this embodiment, the termination condition is determined by updating the evolutionary generation t=t+1 and checking whether the termination condition is met. Condition 1: The number of generations reaches the maximum number of generations. ; Condition 2: The Pareto optimal frontier remains unchanged for 30 consecutive generations, and the population converges.

[0057] If the termination condition is not met, return to the individual spatiotemporal feasibility check and fitness calculation above, and enter the next round of evolutionary cycle; if the termination condition is met, stop evolution, output the final convergent population, and output the final population, where The set of individuals with a value of 1 is the Pareto optimal set.

[0058] In this embodiment, the optimal solution decision is made based on the Pareto optimal solution set. For individuals with a score of 1, a linear weighted summation is used to calculate the overall score for each individual, as shown in the following expression: Two individuals are randomly selected from the population, with priority given to those with lower non-dominant rank and greater crowding at the same rank, to enter the parent population.

[0059] In this embodiment, a bit mutation operation is performed again, expressed as follows: in, This represents the mutated gene value. This is the original gene value; It is a random integer between 1 and n. A random number between 0 and 1; The value represents the mutation probability, ranging from 0.01 to 0.05.

[0060] In this embodiment, the optimal solution decision is weighted, and the expression is as follows: Select the individual with the smallest S(ind) as the final globally optimal AGV task allocation scheme; lock the final allocation scheme and the pre-planned spatiotemporal path for each AGV.

[0061] In this embodiment, resource locking and output distribution involve sending the pre-planned spatiotemporal path of the final solution to the global spatiotemporal occupancy. The table requests a lock, updating the status of the corresponding spatiotemporal node from 0 to 1 (officially occupied). It also outputs the final AGV task allocation matrix X and the pre-planned spatiotemporal path for each AGV, with the output format completely consistent with Implementation 1. In this embodiment, the allocation results and pre-planned paths are sent to the AGV control module, and the updated... The table is synchronized to all modules of the entire system, connecting with the subsequent methods and processes of lower-level path replanning and conflict resolution in Implementation Example 1.

[0062] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations will be apparent to those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A comprehensive warehouse data management method based on the Internet of Things, characterized by: include: First, the warehouse's basic environment and main structure are modeled, including the creation of a warehouse grid map, AGV cluster, and task set model. Based on the warehouse grid map, AGV cluster, and task set model, the AGV empty run rate, conflict, delay, and energy consumption are calculated to construct the global spatiotemporal occupancy. The table is used to construct a multi-objective optimization function; the generated AGV transport tasks are sorted by priority and deadline, and the AGVs are sorted by idle status and battery level. After sorting, the upper-level global scheduling is carried out. The AGV first pre-plans the spatiotemporal path according to the transport task, then calculates the task request value for each task. In the upper-level global scheduling, the tasks are auctioned according to the task request value. The one with the larger request value wins the task, locks the spatiotemporal resources, and generates a task allocation scheme. The AGV that receives the task travels along the pre-planned spatiotemporal path and periodically checks during path execution to determine whether there are obstacles / conflicts / deadlocks or task changes. If there are no abnormalities, it continues to travel; if there are abnormalities, it triggers the lower-level local replanning. In the replanning process, a two-way simultaneous search is performed from the current point and the target point, constrained by the spatiotemporal node positions and time, and combined with the global context. The table generates new paths based on real-time obstacles.

2. The method for comprehensive warehouse data management based on the Internet of Things according to claim 1, characterized in that: The warehouse grid map model is as follows: ； in, This is a 2D raster map model of the warehouse. For a set of grid nodes, It is the set of edges between adjacent nodes; The AGV cluster is: in, For AGV cluster collection, For the first Unit AGV; n is the total number of AGVs. Task set model is ; in, For a set of tasks, Let m be the j-th transportation task, and m be the total number of tasks.

3. The method for comprehensive warehouse data management based on the Internet of Things according to claim 1, characterized in that: Multi-objective optimization functions include the total cost function of empty mileage. Total cost function of path conflicts Total cost function of task delay Total energy cost function ;in, The expression is as follows: in, ; This refers to the empty mileage. The expression is as follows: in, for and ; The cost of a single conflict; Total cost function of task delay The expression is as follows: in, For the task Priority coefficient, ∈ The larger the value, the higher the priority. For the task The actual completion time; For the task The deadline for completion; Total energy cost function The expression is as follows: in, For the task Effective transportation mileage; ; in, For AGV cart The normalized value of the remaining battery power. , For AGV cart Energy consumption coefficient per unit mileage; The global multi-objective optimization objective function for the multi-objective optimization task is calculated as follows: in, , , , Let be the weights of each sub-objective, and + + =1.

4. The method for comprehensive warehouse data management based on the Internet of Things according to claim 1, characterized in that: In the aforementioned upper-level global scheduling, the global state is first initialized and data is preprocessed; the inputs include the warehouse map G, the AGV cluster state A, the task set T, and the weight coefficients. - The time granularity is Δt; then the map is preprocessed to mark restricted areas, one-way passages, and shelving areas, and an adjacency matrix M is generated. Task sorting is based on priority. Descending order, with the same priority ordered by Ascending order, generate a sorting task queue. ; Among them, AGVs are sorted by idle status and remaining power. Descending order, generate a queue of available AGVs. .

5. The method for comprehensive warehouse data management based on the Internet of Things according to claim 1, characterized in that: In the upper-level global scheduling, tasks are auctioned based on their request values, with the AGV requesting the highest value winning the task. The auction is initialized first, and the number of rounds is set. =0, maximum number of rounds =100; Initial price vector for the task =0; Allocation matrix =0, winning set ; Then broadcast the task and bid for the AGV. Each unassigned task ,Towards Broadcast full task information for each AGV After receiving the task, first based on the current situation... The table uses the spatiotemporal A* algorithm to plan from... Then Conflict-free spacetime path If the plan fails... =-∞, abandon the bid; among the valid bids for AGVs, select the highest bidder. The corresponding AGV is ; Among them, if > Update task application value = - ; Temporarily Assigned to ;renew ; and will Node-time pairs in The column marked 2 in the table represents pre-occupancy. like ≤ The task failed to sell in this round and will proceed to the next round. The process of auctioning tasks based on task request values ​​in the upper-level global scheduling is repeated until all tasks are allocated; the auction is terminated when all available AGVs are allocated, the maximum number of rounds is reached, or no task is won in 3 consecutive rounds. When a newly added AGV completes its task and becomes idle, an incremental auction is triggered. This auction process, which is performed only on newly added tasks and idle AGVs, involves the task being auctioned based on the task request value in the upper-level global scheduling.

6. The method for comprehensive warehouse data management based on the Internet of Things according to claim 1, characterized in that: The aforementioned lower-level local replanning involves the AGV performing a perception and judgment operation every Δt period while traveling along the pre-planned path. Nodes ahead on the pre-planned path are detected as temporary obstacles by the onboard sensors; globally... After the table is updated, node / edge conflicts occur in the pre-planned path, or a deadlock risk is detected; or the position deviates from the pre-planned path by more than the threshold δ, making it impossible to return to the original path; This will trigger a rerouting process.

7. The method for comprehensive warehouse data management based on the Internet of Things according to claim 6, characterized in that: When replanning the route, each search node Coupled spatial location With arrival time A bidirectional search architecture, where the forward search F starts from the current position. To the target point Search, reverse search B from Towards The search terminates when the forward search and the reverse search meet.

8. A warehouse data integrated management system based on the Internet of Things, characterized in that: Used to perform the IoT-based integrated warehouse data management method as described in any one of claims 1-7.

9. A warehouse data integrated management device based on the Internet of Things, characterized in that: The management device includes at least one processor and a memory communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the IoT-based integrated warehouse data management method as described in any one of claims 1-7.

10. A computer storage medium storing a computer program executable on a processor, characterized in that, When the processor executes the computer program, it implements the IoT-based integrated warehouse data management method as described in any one of claims 1-7.

Images