Large-scale automatic guided vehicle scheduling and path optimization method and system

Abstract

The invention belongs to the field of intelligent scheduling and optimization, and discloses a large-scale automatic guided vehicle scheduling and path optimization method and system, and the method comprises the steps: building a warehouse network model based on the actual layout and operation process of a warehouse, and obtaining an order distribution sequence of each automatic guided vehicle through combining with the order demands in the warehouse; analyzing task requirements of each automated guided vehicle, and determining a driving path of each automated guided vehicle from a starting point to a terminal point in combination with a warehouse network model and a breadth-first search algorithm with constraints; and according to the order distribution sequence and the driving path, executing parallel calculation on a plurality of processors by utilizing a progressive optimal distribution algorithm so as to determine the optimal order distribution sequence and the driving path. According to the method, intelligent allocation and collision-free path planning are realized through a rule-based scheduling algorithm and a breadth-first search algorithm with constraints, and meanwhile, a progressive optimal allocation algorithm and a parallel computing framework are introduced, so that the scheduling efficiency and global path planning are optimized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent scheduling and optimization, and particularly to a method and system for scheduling and path optimization of large-scale automatic guided vehicles. Background Art

[0002] With the continuous progress of intelligent warehousing technology, automatic guided vehicles (AGVs), as key transportation tools in intelligent logistics systems, have been widely used in large warehouses, manufacturing workshops, and automated distribution centers. By autonomously performing tasks such as cargo handling and order delivery, AGVs have significantly improved the intelligence level of warehousing operations, reduced manual intervention, and increased logistics efficiency. However, traditional AGV scheduling mainly relies on preset paths or manual instructions. In complex dynamic environments, its scheduling and path planning efficiency are difficult to meet the requirements of efficient operation. Especially with the expansion of warehousing scale and the dynamic changes in order demands, existing AGV scheduling and path optimization methods are overwhelmed when dealing with large-scale, multi-task concurrency.

[0003] Current AGV scheduling technologies mainly include rule-based methods, optimization algorithms, and machine learning methods. Rule-based methods such as first-come, first-served are suitable for small-scale applications. However, when faced with fluctuations in order volume and increased task complexity, they are difficult to dynamically adjust strategies, resulting in insufficient resource utilization and low efficiency. Optimization algorithms such as mixed integer programming and genetic algorithms can theoretically obtain optimal solutions, but they have high computational complexity and are particularly difficult to solve in real time in large-scale scenarios. In addition, these methods have poor adaptability to environmental uncertainty factors. In recent years, machine learning, especially deep reinforcement learning methods, have gradually attracted attention. They can achieve good dynamic scheduling, but problems such as long training time and high computational resource requirements limit their application in actual large-scale systems.

[0004] In terms of path optimization, common traditional search algorithms such as Dijkstra and A* can find the shortest path, but they do not fully consider the interaction and path conflicts between AGVs, which easily cause congestion and even deadlocks. The time window constraint method can improve the safety and feasibility of paths, but it is difficult to cope with the dynamic changes in the warehousing environment. Intelligent optimization methods attempt to predict and avoid congestion, but their computational cost is high and their generalization ability is limited, making it difficult to adapt to different warehousing environments and dynamic changes.

[0005] In summary, the existing AGV scheduling and path optimization technologies face challenges in large-scale intelligent warehousing: static rules or deterministic optimization methods lack flexibility and cannot effectively cope with dynamic environments; the scheduling and path optimization processes lack a coordination mechanism, reducing the utilization rate of AGV resources and increasing the risk of local congestion; traditional path planning algorithms fail to fully consider the impact of dynamic traffic flow, resulting in poor implementation effects of the solutions; traditional optimization algorithms have poor real-time performance, and although artificial intelligence technologies have certain advantages, their generalization capabilities are limited in different warehousing environments.

[0006] Therefore, how to provide a large-scale automatic guided vehicle scheduling and path optimization method and system is an urgent problem to be solved at present. Summary of the Invention

[0007] Embodiments of the present invention provide a large-scale automatic guided vehicle scheduling and path optimization method and system to solve the above technical problems existing in the prior art.

[0008] To provide a basic understanding of some aspects of the disclosed embodiments, a simple summary is given below. This summary part is not a general review, nor is it intended to identify key / important constituent elements or delineate the protection scope of these embodiments. Its sole purpose is to present some concepts in a simple form as a prelude to the detailed description that follows.

[0009] According to the first aspect of the embodiments of the present invention, a large-scale automatic guided vehicle scheduling and path optimization method is provided.

[0010] In one embodiment, a large-scale automatic guided vehicle scheduling and path optimization method includes: Based on the actual layout and operation process of the warehouse, construct a warehouse network model, and combine the order requirements in the warehouse to obtain the order allocation sequence for each automatic guided vehicle; Analyze the task requirements of each automatic guided vehicle, and combine the warehouse network model with the breadth-first search algorithm with constraints to determine the driving path of each automatic guided vehicle from the starting point to the ending point; According to the order allocation sequence and the driving path, use the progressive optimal allocation algorithm to perform parallel computing on several processors to determine the optimal order allocation sequence and driving path.

[0011] In one embodiment, the step of based on the actual layout and operation process of the warehouse, construct a warehouse network model, and combine the order requirements in the warehouse to obtain the order allocation sequence for each automatic guided vehicle includes: Obtain the actual layout and operation process of the warehouse, use the discrete event simulation algorithm to construct a warehouse network model, and set warehouse parameters to simulate the actual operating environment of the warehouse; Analyze the order requirements in the warehouse to determine the actual time, order set, and automated guided vehicle set; Input the determined actual time, order set, and automated guided vehicle set into the warehouse network model, and output the preliminary order allocation sequence for each automated guided vehicle through the warehouse network model; Utilize the adaptive dynamic scheduling mechanism to optimize the preliminary order allocation sequence, dynamically adjust the matching relationship between orders and automated guided vehicles, and obtain the order allocation sequence for each automated guided vehicle.

[0012] In one embodiment, the warehouse parameters include: the number of automated guided vehicles, order arrival rate, movement speed of automated guided vehicles, loading time, unloading time, total simulation time, grid size, and grid layout.

[0013] In one embodiment, the step of utilizing the adaptive dynamic scheduling mechanism to optimize the preliminary order allocation sequence, dynamically adjust the matching relationship between orders and automated guided vehicles, and obtain the order allocation sequence for each automated guided vehicle includes: Based on the preliminary order allocation sequence, divide the order set into orders with waiting time exceeding the preset waiting time and orders with waiting time not exceeding the preset waiting time according to the waiting time; Calculate the distance between each order and each automated guided vehicle, and dynamically adjust the starting point of the order according to the allocation status of the order; Select the automated guided vehicle with the optimal distance from the orders with waiting time exceeding the preset waiting time for priority allocation, and allocate automated guided vehicles to the orders with waiting time not exceeding the preset waiting time according to the distance and the preset capacity limit; Evaluate the order allocation situation, combine the distance and the number of orders, and adaptively and dynamically adjust the matching relationship between orders and automated guided vehicles to optimize the order sequence for each automated guided vehicle and obtain the order allocation sequence for each automated guided vehicle.

[0014] In one embodiment, the step of analyzing the task requirements of each automated guided vehicle and determining the driving path of each automated guided vehicle from the starting point to the end point in combination with the warehouse network model and the breadth-first search algorithm with constraints includes: Obtain and analyze the task requirements of each automated guided vehicle to determine the starting point position, end point position, and grid layout where the automated guided vehicle is located; Input the determined starting point position, end point position, and grid layout where the automated guided vehicle is located into the warehouse network model, and output the preliminary driving path of each automated guided vehicle from the starting point to the end point through the warehouse network model; Utilize the breadth-first search algorithm with constraints to optimize the preliminary driving path and obtain the driving path of each automated guided vehicle from the starting point to the end point.

[0015] In one embodiment, optimizing the preliminary driving path by using a breadth-first search algorithm with constraints to obtain the driving path of each automated guided vehicle from the starting point to the end point includes: According to the preliminary driving path and the actual time, obtain the set of all automated guided vehicles during the movement, and calculate the estimated arrival time of each automated guided vehicle; Based on the occupancy of the automated guided vehicle and the estimated arrival time, define the constraint conditions, where the constraint conditions include a time interval constraint set and a target position constraint; Combined with the defined constraint conditions, use the breadth-first search algorithm with constraints to start from the starting point and search for the target driving path to the target node; According to the target driving path, the automated guided vehicle moves according to the local path request and performs dynamic avoidance based on the priority of the request time; According to the result of dynamic avoidance, determine the driving path of each automated guided vehicle from the starting point to the end point.

[0016] In one embodiment, the combined with the defined constraint conditions, using the breadth-first search algorithm with constraints to start from the starting point and search for the target driving path to the target node includes: Create an empty queue, add the starting point to the queue, and mark the starting point as the visited state; Select a position node from the queue, check whether all adjacent nodes of the position node meet the constraint conditions, whether they have been visited, and whether they are the target node, and add the adjacent nodes that meet all the check conditions to the queue and mark them as the visited state; Based on the target node obtained from the inspection result, backtrack from the target node to the starting point, and generate the target driving path according to the search path.

[0017] In one embodiment, according to the order allocation sequence and the driving path, using the progressive optimal allocation algorithm to perform parallel computing on several processors to determine the optimal order allocation sequence and the driving path includes: According to the order allocation sequence and the driving path, determine the set of design schemes of the automated guided vehicle, and allocate an initial simulation budget for each design scheme based on the conjugate prior distribution to quantify the performance evaluation; Use the allocated initial simulation budget to perform simulation experiments on each design scheme, and update the performance evaluation of each design scheme through the Bayesian update formula according to the simulation results to obtain the posterior distribution; Evaluate the value obtained by allocating additional simulation budget to each design scheme through the value calculation formula, and optimize the allocation of the simulation budget among different design schemes in combination with the progressive optimal allocation algorithm; Allocate the optimized design scheme to several processors, and use the parallel computing framework to accelerate the execution of the simulation experiment; Summarize the simulation results of each processor, identify the optimal design scheme, and determine the optimal order allocation sequence and driving route.

[0018] In one embodiment, the formula of the progressive optimal allocation algorithm is: ; In the formula, A t+1 ( ε t ) represents the allocation decision at the ( t + 1)-th moment; t represents the t -th moment; ε t represents the information set obtained by simulating the budget allocation up to the t -th moment; (1) t represents the optimal design scheme at the t -th moment; ( h ) t represents the set H (t+1) containing the design scheme corresponding to the index label at the t -th moment; represents the value obtained by allocating additional simulation budget; H (t+1) represents the set of index labels of candidate schemes at the ( t + 1)-th moment.

[0019] According to the second aspect of the embodiments of the present invention, a large-scale automatic guided vehicle scheduling and path optimization system is provided.

[0020] In one embodiment, the large-scale automatic guided vehicle scheduling and path optimization system includes: A model construction and order allocation module, configured to construct a warehouse network model based on the actual layout and operation process of the warehouse, and combine the order requirements in the warehouse to obtain the order allocation sequence of each automatic guided vehicle; A driving path determination module, configured to analyze the task requirements of each automatic guided vehicle, and combine the warehouse network model with a breadth-first search algorithm with constraints to determine the driving path of each automatic guided vehicle from the starting point to the end point; A scheduling and path optimization module, configured to use the progressive optimal allocation algorithm to perform parallel calculations on a plurality of processors according to the order allocation sequence and the driving path, so as to determine the optimal order allocation sequence and the driving path.

[0021] According to the third aspect of the embodiments of the present invention, a computer device is provided.

[0022] In some embodiments, the computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the above method are implemented.

[0023] According to a fourth aspect of the embodiments of the present invention, a computer-readable storage medium is provided.

[0024] In one embodiment, a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0025] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: 1. Through the rule-based scheduling algorithm and the breadth-first search path optimization algorithm with constraints, the present invention calculates the distance between the order and the automatic guided vehicle in real time, dynamically adjusts the priority and adopts an obstacle avoidance strategy to achieve intelligent allocation and collision-free path planning. Furthermore, the progressive optimal allocation algorithm and the parallel computing framework are introduced to optimize the scheduling and global path planning efficiency.

[0026] 2. Through simulation budget allocation, the present invention optimizes the probability of selecting the optimal solution in scheduling and path planning under the condition of limited computing resources, ensures the global optimality and stability of the automatic guided vehicle scheduling, can adjust the scheduling strategy in real time in a dynamic environment, effectively reduces the average order cycle time, and improves the order completion rate.

[0027] 3. The present invention introduces an adaptive scheduling mechanism, which can dynamically adjust the task allocation of the automatic guided vehicle according to the real-time changes of the system state, avoids the local optimal problem caused by fixed rules, enhances the flexibility and adaptability of the scheduling strategy, enables it to have stronger dynamic adjustment ability, and effectively reduces system congestion and path conflicts.

[0028] 4. The present invention uses the breadth-first search algorithm with constraints and combines real-time traffic flow prediction to ensure that the driving path of the automatic guided vehicle has good obstacle avoidance ability and real-time responsiveness, effectively prevents path conflicts and deadlock problems when multiple automatic guided vehicles operate in parallel, and thus optimizes the overall throughput and efficiency of the warehousing system.

[0029] 5. The present invention optimizes the allocation of simulation resources through parallel computing, greatly improves the computing efficiency of scheduling and path optimization, supports multiple simulation tasks to be executed in parallel in a multi-core computing environment, accelerates the optimization process of scheduling and path planning, and can quickly calculate the optimal automatic guided vehicle scheduling scheme and path planning strategy in a highly dynamic and multi-task concurrent intelligent warehousing environment, thereby improving the throughput rate and operation efficiency of the entire warehousing system.

[0030] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and should not limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings herein are incorporated into and constitute a part of this specification, showing embodiments consistent with the present invention, and are used together with the specification to explain the principles of the present invention.

[0032] Figure 1 is a flowchart of a large-scale automated guided vehicle scheduling and path optimization method shown according to an exemplary embodiment; Figure 2 is a schematic block diagram of a large-scale automated guided vehicle scheduling and path optimization system shown according to an exemplary embodiment; Figure 3 is a schematic structural diagram of a computer device shown according to an exemplary embodiment; Figure 4 is a grid representation diagram of an automated guided vehicle grid of a large-scale automated guided vehicle scheduling and path optimization method shown according to an exemplary embodiment; Figure 5 is a schematic mechanism diagram of an automated guided vehicle movement simulation of a large-scale automated guided vehicle scheduling and path optimization method shown according to an exemplary embodiment; Figure 6 is a schematic internal mechanism diagram of a partial path request sub-module of a large-scale automated guided vehicle scheduling and path optimization method shown according to an exemplary embodiment; Figure 7 is a schematic diagram of head-on collision (left) and same-cell collision (middle and right) of a large-scale automated guided vehicle scheduling and path optimization method shown according to an exemplary embodiment; Figure 8 is a schematic diagram of implementing a progressive optimal allocation algorithm in a parallel computing environment for a large-scale automated guided vehicle scheduling and path optimization method shown according to an exemplary embodiment; Figure 9 is a schematic simulation environment diagram of a large-scale automated guided vehicle scheduling and path optimization method in Embodiment 1 shown according to an exemplary embodiment; Figure 10 is a schematic diagram of an automated guided vehicle scheduling process in Embodiment 1 of a large-scale automated guided vehicle scheduling and path optimization method shown according to an exemplary embodiment; Figure 11 is a schematic simulation environment diagram of a large-scale automated guided vehicle scheduling and path optimization method in Embodiment 2 shown according to an exemplary embodiment; Figure 12It is a schematic diagram of the automatic guided vehicle scheduling process in the second embodiment of a large-scale automatic guided vehicle scheduling and path optimization method shown according to an exemplary embodiment; Figure 13 It is a schematic diagram of a simulation environment in the third embodiment of a large-scale automatic guided vehicle scheduling and path optimization method shown according to an exemplary embodiment; Figure 14 It is a schematic diagram of the automatic guided vehicle scheduling process in the third embodiment of a large-scale automatic guided vehicle scheduling and path optimization method shown according to an exemplary embodiment. Specific implementation manners

[0033] The following description and the accompanying drawings fully illustrate the specific implementation manners herein, enabling those skilled in the art to practice them. Parts and features of some embodiments may be included in or replaced with parts and features of other embodiments. The scope of the embodiments herein includes the entire scope of the claims and all available equivalents of the claims. In this document, the terms "first", "second", etc. are only used to distinguish one element from another, and do not require or imply any actual relationship or order between these elements. In fact, the first element can also be called the second element, and vice versa. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a structure, device or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such structure, device or equipment. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the structure, device or equipment including the said element. The embodiments herein are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other.

[0034] The terms "longitudinal", "transverse", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. in this document indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings. They are only for the convenience of describing this document and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation to the present invention. In the description of this document, unless otherwise specified and defined, the terms "mounted", "connected", "coupled" should be understood in a broad sense. For example, it can be a mechanical connection or an electrical connection, or it can be the internal connection of two elements. It can be directly connected or indirectly connected through an intermediate medium. For those of ordinary skill in the art, the specific meanings of the above terms can be understood according to specific circumstances.

[0035] In this text, unless otherwise specified, the term "a plurality of" means two or more than two.

[0036] In this text, the character " / " indicates that the objects before and after are in an "or" relationship. For example, A / B means: A or B.

[0037] In this text, the term "and / or" is an associative relationship describing an object, indicating that there can be three relationships. For example, A and / or B means: A or B, or, the three relationships of A and B.

[0038] It should be understood that although the steps in the flowchart are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this text, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the figure may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0039] Each module in the device or system of this application can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned each module can be embedded in the processor in the computer device in the form of hardware or be independent of it, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0040] Without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0041] Figure 1 An embodiment of a method for scheduling and path optimization of a large-scale automatic guided vehicle according to the present invention is shown.

[0042] In this alternative embodiment, the method for scheduling and path optimization of a large-scale automatic guided vehicle includes: Step S101: Based on the actual layout and operation process of the warehouse, construct a warehouse network model, and combine the order requirements in the warehouse to obtain the order allocation sequence of each automatic guided vehicle. Step S102: Analyze the task requirements of each automatic guided vehicle, and combine the warehouse network model with the breadth-first search algorithm with constraints to determine the driving path of each automatic guided vehicle from the starting point to the end point. Step S103: According to the order allocation sequence and the driving route, use the progressive optimal allocation algorithm to perform parallel computing on several processors to determine the optimal order allocation sequence and driving route.

[0043] It should be added that based on the actual layout and operation process of the warehouse, a warehouse network model is constructed, and combined with the order requirements in the warehouse, the order allocation sequence for each automatic guided vehicle includes: Construct a simulation environment for an automatic guided vehicle (AGV) warehousing system based on discrete event simulation. Key factors such as order arrival distribution, AGV driving status, warehousing road network structure, and path congestion are accurately described through the simulation model. Through this simulation environment, the AGV scheduling and path optimization processes under different warehousing system layouts, different numbers of AGVs, and different order densities can be comprehensively simulated, thus providing a reliable test platform for the development and verification of optimization algorithms. On this basis, through an adaptive scheduling mechanism, the AGV task allocation strategy can be dynamically adjusted according to factors such as real-time order changes, warehouse goods storage distribution, and the current operating status of AGVs, rather than using fixed scheduling rules or static optimization schemes based on historical data. This can effectively balance the AGV task load, avoid some AGVs running at a high load for a long time while other AGVs are idle, thereby improving the system throughput rate, reducing the overall order completion time, and enhancing the operating efficiency of the warehousing system. The scheduling mechanism has stronger adaptability and can be dynamically adjusted according to various information such as order arrival situations, the current positions of AGVs, and the load conditions of warehousing areas. This dynamic adjustment ability avoids the local optimization problem brought by fixed scheduling rules, enabling the system to more flexibly respond to different operating environments. For example, in some high-density areas, AGVs can be preferentially assigned to areas with less task volume to avoid congestion in that area and ensure the overall fluency of the system. The adaptive scheduling mechanism can also adjust task allocation in real time according to the urgency and priority of tasks, enabling important orders to be processed in a timely manner, thereby improving the order completion efficiency and reducing the waiting time and driving conflicts of AGVs.

[0044] Analyze the task requirements of each automatic guided vehicle, and combine the warehouse network model and the breadth-first search algorithm with constraints to determine the driving route of each automatic guided vehicle from the starting point to the ending point, including: In terms of path optimization, through the breadth-first search algorithm with constraints, combined with real-time traffic flow prediction and dynamic path adjustment mechanism, it is ensured that the AGV can avoid congested areas during driving, reduce the probability of path conflicts, and ensure the stability of the overall operation. By analyzing information such as path historical data, current AGV operation status, and warehouse road occupancy, dynamic constraints are added during the path search process to ensure that the generated path achieves the best balance among feasibility, obstacle avoidance ability, and real-time performance. The path optimization algorithm not only considers the shortest path of the AGV but also fully takes into account the dynamic changes inside the warehouse and the prediction of real-time traffic flow, ensuring that the AGV can avoid highly congested areas during operation and avoid path conflicts and deadlock problems. Especially when multiple AGVs run in parallel, the path planning method of the present invention can predict the driving status of the AGV and plan an idle path for it in advance, thereby reducing path conflicts and ensuring the smoothness of the AGV's driving. This strategy greatly improves the throughput capacity of the warehousing system and ensures the successful completion of AGV tasks.

[0045] According to the order assignment sequence and driving path, using the progressive optimal assignment algorithm, parallel computing is performed on several processors to determine the optimal order assignment sequence and driving path, including: To further improve the execution efficiency of path planning, a parallel computing framework is introduced to simultaneously execute multiple path planning tasks in a multi-core computing environment, enabling the algorithm to still maintain a high computing efficiency in a large-scale AGV system and ensuring that the system can still operate efficiently under a high order load. By simultaneously allocating simulation tasks to multiple computing nodes and using the parallel computing framework to quickly evaluate various scheduling and path planning strategies, while reducing the computing time, the accuracy and efficiency of the optimization process are ensured. Parallel computing not only improves the optimization efficiency but also greatly reduces the computing cost, enabling the large-scale AGV scheduling problem to be solved within an acceptable time.

[0046] Combining the simulation optimization method with the artificial intelligence scheduling strategy, and optimizing the selection of scheduling parameters through an improved simulation budget allocation method to further enhance the optimization effect of the system. In a complex intelligent warehousing environment, AGV scheduling and path optimization involve a large number of uncertain factors, such as the randomness of order arrival, the dynamic adjustment of warehouse layout, the change of the AGV's own operating state, etc. Through the improved simulation budget allocation method, under the limited simulation budget, the computing resources are intelligently allocated to optimize the scheduling and path planning parameters, enabling the algorithm to maintain better performance in different warehousing environments; through this efficient allocation of resources, the optimization algorithm can increase the probability of selecting the optimal solution for scheduling and path planning in a complex dynamic environment, thereby ensuring the global optimality and system stability of the AGV scheduling strategy. Using this method, the system can significantly reduce the average order cycle time after scheduling and optimization, improve the order completion rate, effectively reduce the vehicle waiting time, and enhance the overall operating efficiency of the system. By introducing a parallel simulation computing mechanism, the computing efficiency of scheduling and path optimization is significantly improved. In a large-scale intelligent warehousing environment, the dynamic demand and high-concurrency orders of the system require the system to be able to respond and make adjustments in real time. Through the parallel computing framework, simulation tasks can be executed simultaneously on multiple computing nodes, and scheduling and path optimization strategies can be quickly explored in different simulation models. This not only greatly reduces the computing time but also ensures that the system can still maintain high processing capabilities when facing a large number of AGVs and complex tasks, respond to environmental changes in real time, and improve the overall adaptability of the system. At the same time, a multi-objective optimization strategy is adopted. While optimizing the AGV driving path, the scheduling fairness, system throughput, and average task completion time are comprehensively considered, making the AGV scheduling more reasonable and the warehousing operation more efficient.

[0047] In addition, by jointly optimizing scheduling and path planning, the resource utilization efficiency of AGVs is maximized. In terms of task scheduling, by dynamically adjusting the allocation of AGV tasks, the idleness of some AGV resources or the overload of some AGVs is avoided, thereby improving the resource utilization rate and reducing the congestion phenomenon in the system. In terms of path planning, through real-time path optimization and conflict detection, the path conflicts and congestion problems between AGVs are effectively reduced, making the path selection of each AGV smoother, reducing the unnecessary waiting time, and further improving the overall throughput of the warehousing system. Combining the comprehensive optimization of scheduling and path planning enables the system not only to complete orders quickly but also to achieve efficient resource allocation, shorten the order processing time, and improve the overall efficiency of the system.

[0048] In this alternative embodiment, constructing a warehouse network model based on the actual layout and operation process of the warehouse, and combining the order requirements in the warehouse to obtain the order allocation sequence for each automatic guided vehicle includes: Obtain the actual layout and operation process of the warehouse, construct a warehouse network model using the discrete event simulation algorithm, and set warehouse parameters to simulate the actual operating environment of the warehouse; Analyze the order requirements in the warehouse, and determine the actual time, order set, and automated guided vehicle set; Input the determined actual time, order set, and automated guided vehicle set into the warehouse network model, and output the preliminary order allocation sequence of each automated guided vehicle through the warehouse network model; Use the adaptive dynamic scheduling mechanism to optimize the preliminary order allocation sequence, dynamically adjust the matching relationship between orders and automated guided vehicles, and obtain the order allocation sequence of each automated guided vehicle.

[0049] In this alternative embodiment, the warehouse parameters include: the number of automated guided vehicles, order arrival rate, movement speed of automated guided vehicles, loading time, unloading time, total simulation time, grid size, and grid layout.

[0050] In this alternative embodiment, the step of using the adaptive dynamic scheduling mechanism to optimize the preliminary order allocation sequence, dynamically adjust the matching relationship between orders and automated guided vehicles, and obtain the order allocation sequence of each automated guided vehicle includes: Based on the preliminary order allocation sequence, divide the order set into orders with waiting time exceeding the preset waiting time and orders with waiting time not exceeding the preset waiting time; Calculate the distance between each order and each automated guided vehicle, and dynamically adjust the starting point of the order according to the allocation status of the order; Select the automated guided vehicle with the optimal distance from the orders with waiting time exceeding the preset waiting time for priority allocation, and allocate automated guided vehicles for the orders with waiting time not exceeding the preset waiting time according to the distance and the preset capacity limit; Evaluate the order allocation situation, and adaptively and dynamically adjust the matching relationship between orders and automated guided vehicles by combining the distance and the number of orders to optimize the order sequence of each automated guided vehicle and obtain the order allocation sequence of each automated guided vehicle.

[0051] In this alternative embodiment, the step of analyzing the task requirements of each automated guided vehicle and determining the driving path of each automated guided vehicle from the starting point to the end point by combining the warehouse network model and the breadth-first search algorithm with constraints includes: Obtain and analyze the task requirements of each automated guided vehicle, and determine the starting point position, end point position, and grid layout where the automated guided vehicle is located; Input the determined starting point position, end point position, and grid layout where the automated guided vehicle is located into the warehouse network model, and output the preliminary driving path of each automated guided vehicle from the starting point to the end point through the warehouse network model; Optimize the preliminary driving path using a breadth - first search algorithm with constraints to obtain the driving path of each automatic guided vehicle from the starting point to the ending point.

[0052] In this alternative embodiment, the step of optimizing the preliminary driving path using a breadth - first search algorithm with constraints to obtain the driving path of each automatic guided vehicle from the starting point to the ending point includes: According to the preliminary driving path and the actual time, obtain the set of all automatic guided vehicles during movement and calculate the estimated arrival time of each automatic guided vehicle; Based on the occupancy situation and the estimated arrival time of the automatic guided vehicles, define the constraint conditions, where the constraint conditions include a time - interval constraint set and a target - position constraint; Combined with the defined constraint conditions, use a breadth - first search algorithm with constraints to start from the starting point and search for the target driving path to the target node; According to the target driving path, the automatic guided vehicle moves according to the local path request and performs dynamic avoidance based on the priority of the request time; According to the result of dynamic avoidance, determine the driving path of each automatic guided vehicle from the starting point to the ending point.

[0053] In this alternative embodiment, the step of combined with the defined constraint conditions, using a breadth - first search algorithm with constraints to start from the starting point and search for the target driving path to the target node includes: Create an empty queue, add the starting point to the queue, and mark the starting point as the visited state; Select a position node from the queue, check whether all adjacent nodes of the position node meet the constraint conditions, whether they have been visited, and whether they are the target node, and add the adjacent nodes that meet all the check conditions to the queue and mark them as the visited state; Based on the target node obtained from the check result, trace back from the target node to the starting point and generate the target driving path according to the search path.

[0054] In this alternative embodiment, the step of according to the order - allocation sequence and the driving path, using a progressive - optimal - allocation algorithm to perform parallel computing on several processors to determine the optimal order - allocation sequence and the driving path includes: According to the order - allocation sequence and the driving path, determine the set of design schemes of the automatic guided vehicle, and allocate an initial simulation budget for each design scheme based on the conjugate prior distribution to quantify the performance evaluation; Use the allocated initial simulation budget to perform simulation experiments on each design scheme, and update the performance evaluation of each design scheme through the Bayesian update formula according to the simulation results to obtain the posterior distribution; Evaluate the value obtained by allocating additional simulation budgets to each design solution through a value calculation formula, and optimize the allocation of simulation budgets among different design solutions in combination with the progressive optimal allocation algorithm; Allocate the optimized design solutions to a number of processors, and use a parallel computing framework to accelerate the execution of simulation experiments; Summarize the simulation results of each processor, identify the optimal design solution, and determine the optimal order allocation sequence and driving route.

[0055] In this alternative embodiment, the formula of the progressive optimal allocation algorithm is: ; In the formula, A t+1 ( ε t ) represents the allocation decision at the ( t + 1)-th moment; t represents the t -th moment; ε t represents the information set obtained by allocating the simulation budget up to the t -th moment; (1) t represents the optimal design solution at the t -th moment; ( h ) t represents the set H (t+1) contains the design solution corresponding to the index label at the t -th moment; represents the value obtained by allocating additional simulation budgets; H (t+1) represents the set of index labels of candidate solutions at the ( t + 1)-th moment.

[0056] It should be noted that a discrete event simulation model is used to simulate the AGV scheduling and path optimization process in the intelligent warehousing environment. This model is based on the actual layout and operation process of the warehouse management system, and is mainly composed of a scheduling module and a path planning module, aiming to minimize the adjusted average order cycle time and improve the throughput rate and resource utilization rate of the warehousing system. The main parameters of the model include the number of AGVs N AGV , order arrival rate λ, AGV movement speed V AGV , loading and unloading times (D loading and D unloading ), total simulation time T simulation and grid size and layout. This model provides a dynamic test environment for the optimization algorithm by simulating the AGV scheduling and path planning process.

[0057] In the present invention, the warehouse is modeled as a grid system, where each grid cell represents a possible location and the AGV can move between these cells. The order arrival process is assumed to be a Poisson process with an arrival rate parameter of λ. Each order contains a picking rack and a delivery location, and the task of the AGV is to pick up the goods from the picking rack and transport them to the delivery point. The objective of the model is to minimize the adjusted average order cycle time, defined as: ; where, E represents the expectation; represents the total cycle time of the completed orders, represents the penalty parameter associated with the uncompleted orders caused by deadlocks; f represents completion; N f represents the number of completed orders; u represents uncompleted; N u represents the number of uncompleted orders; N = N u + N f represents the total number of orders received by the simulation model. The objective of this model is to minimize the average order cycle time value, that is, while ensuring the system throughput rate, reduce the order waiting time and reduce the deadlock problem caused by improper AGV scheduling.

[0058] As Figure 4 shown, it shows the grid representation of the AGV traffic network. Each grid cell can accommodate only one AGV at a time, and the AGV can only move horizontally or vertically. The AGV may be in one of the following four states at any time: (1) picking state - moving to the picking rack to pick up goods; (2) delivery state - delivering the goods to the delivery point; (3) parking state - not assigned a new order after unloading; (4) idle state - parked or waiting for a newly assigned order.

[0059] As Figure 5 shown, it illustrates the mechanism of the AGV motion simulation, including the current position of the AGV, the target position, and the constraints involved in the path planning process. When a new order is received by the warehouse system or the state of the AGV becomes "parked", the AGV scheduling module is triggered. After being triggered, all available orders in the simulation model (including unassigned orders and assigned but not yet started picking orders) are assigned to the AGVs that are not in the "parked" state. In the l-th call of the scheduling module, define as the set of available orders, For the set of all available AGVs (excluding AGVs in the "parked" state). The scheduling module uses a scheduling algorithm to assign orders to AGVs. The AGV path planning module is triggered after the scheduling module, AGV loading, and AGV unloading. This path planning module consists of a partial path request sub-module, aiming to dynamically adjust the driving trajectory of the AGV and avoid path conflicts. At the th call of the scheduling module, define to represent the j current position of the vehicle, and to represent the j destination position of the vehicle. When the path planning module is triggered, the system calculates the complete path from to based on the current environmental state. Subsequently, the partial path request sub-module is repeatedly triggered to continuously evaluate the path feasibility of the AGV until the system calculates an optimal driving route to ensure the AGV reaches the destination smoothly.

[0060] The present invention proposes a partial path request sub-module for optimizing the driving path of the AGV and avoiding path conflicts and deadlocks. As Figure 6 shown, the working mechanism of this module is demonstrated. Let N rf represent the number of square units specified by the partial path request sub-module. If the remaining path length is less than N rf , then request the subsequent path; if there are no obstacles in the requested path, the AGV moves according to the planned path; if the path is occupied, the AGV waits at the current position until the path is available; when multiple AGVs compete for the same path, the AGV with the earliest application time is given priority. During the driving process of the AGV, this sub-module is repeatedly triggered until the AGV reaches the target position smoothly and releases the occupied grid cells.

[0061] To achieve efficient AGV scheduling and path planning in a discrete event simulation model, the present invention has developed two deterministic algorithms: a rule-based dynamic scheduling algorithm and a constraint-based breadth-first search (BFS) path planning algorithm with constraints. These algorithms improve the rationality of AGV task allocation and path planning efficiency by optimizing parameters, thereby enhancing the overall performance and throughput rate of the warehousing system.

[0062] The core of the scheduling algorithm lies in dynamically allocating order tasks based on the real-time distance between the order and the AGV. The algorithm first divides the orders into two subsets according to the waiting time of the order and the current position of the AGV: the set of urgent orders, that is, the orders with a longer waiting time and the set of non-urgent orders . For each order , the algorithm calculates the Manhattan distance between it and all available AGVs , the nearest AGV within the allowed search range r is selected as the allocation object. When allocating orders, priority is given to orders with longer waiting times to reduce the delay risk caused by long-term backlogged orders. In addition, during each scheduling process, reallocation of orders that have been allocated but not yet executed is allowed to further optimize the AGV task scheduling and improve the task completion rate. The Manhattan distance calculation formula between an order and an AGV is d ( w 1, w 2)=| x 1 - x 2| + | y 1 - y 2|, where ( x 1, y 1) and ( x 2, y 2) represent the position coordinates of order w 1 and automated guided vehicle w 2 respectively, and d represents the distance symbol. Through this calculation method, the algorithm can dynamically adjust the AGV tasks to make the task allocation more balanced and avoid resource waste caused by unreasonable AGV positions. To prevent a specific AGV from receiving too many tasks, the system introduces a maximum task capacity constraint c , that is, the maximum number of orders that can be allocated to a single AGV, to ensure reasonable task allocation, avoid over-scheduling of some AGVs, and reduce the situation of inefficient path selection. The present invention proposes to optimize the overall performance of the scheduling algorithm by adjusting parameters c and r to make it adapt to the requirements of different warehousing scenarios and improve the task execution efficiency.

[0063] The path planning algorithm of the present invention is based on the breadth-first search (BFS) method with constraints and introduces constraint conditions to avoid path conflicts and deadlock problems between AGVs. As Figure 7 shown, it shows two main AGV collision situations: head-on collision: when two AGVs move simultaneously in opposite directions along the same path, a frontal collision may occur; same-cell collision: when multiple AGVs simultaneously try to occupy the same cell, a conflict may occur. For example: AGV3 and AGV4 move towards the same cell simultaneously; AGV6 tries to enter a cell already occupied by AGV5, resulting in a cell conflict.

[0064] To effectively avoid these collisions, the present invention introduces a constraint set in the BFS search space to predict the future positions of other AGVs and ensure that the driving strategy can be dynamically adjusted during the path search process to avoid potential conflicts. The core steps of the BFS path planning algorithm are as follows: (1) Introduce time constraints to prevent AGVs from entering cells occupied by other AGVs during certain time intervals: An automated guided vehicle cannot move into an occupied cell during the time interval t o - s 1, t o + s 1 + 1], where t o represents the time when an AGV occupies a certain cell, and s 1 is an additional safety buffer time in the constraint.

[0065] (2) Ensure that an AGV cannot enter the target cell during the unloading period: An automated guided vehicle cannot move into the target cell during the time interval where represents the time when the AGV reaches the target cell; D unloading represents the unloading time of the AGV, and s 2 represents an additional safety buffer time in the constraint. The above constraints ensure that an AGV will not cause a path conflict due to incorrect path planning during unloading. Since the size of the constraint set directly affects the search space of the BFS algorithm, the constraint parameters s 1 and s 2 need to be optimized: If the constraint set is set too large, it may cause the BFS algorithm to fail to find a feasible solution, affecting the driving efficiency of the AGV; if the constraint set is set too small, it may not be able to completely avoid collisions between AGVs. The present invention proposes to adjust the constraint parameters s 1 and s 2 through a simulation budget allocation method, so that the path planning algorithm can improve the search efficiency while ensuring the driving safety of the AGV, and maximize the applicability of path planning. Optimize the performance of the path planning algorithm.

[0066] Combining the above scheduling and path planning algorithms, the present invention gives the specific implementation steps of the path planning module and the scheduling module.

[0067] Specific implementation steps of the scheduling module: The goal of the scheduling module is to assign orders to AGVs to optimize resource utilization and reduce order completion time. The present invention adopts a rule-based dynamic scheduling algorithm. The following are the specific implementation steps of the scheduling module: (1) Initialization. Input parameters: current time , the set of orders to be assigned , and the set of available AGVs 。Scheduling parameters r and c. Output: The order sequence assigned to each AGV.

[0068] (2) Order classification. Classify according to waiting time: Divide the order set into two subsets: is the order with a waiting time exceeding 300 seconds; is other orders.

[0069] (3) Calculate the distance between the order and the AGV. Distance calculation formula: For each order and each AGV , calculate the Manhattan distance | x i - x j | + | y i - y j |, where, ( x i , y i ) and ( x j , y j ) respectively represent the position coordinates of the order i and the automated guided vehicle j . Consider the order assignment status: If the order i has been assigned to the automated guided vehicle j in the previous scheduling, then calculate the distance from j to the order i after completing the current task. If the order i has not been assigned, directly calculate the distance from the starting point to the order i .

[0070] (4) Order assignment. Give priority to assigning the orders in : For the orders in , since their waiting time is longer, give priority to assigning them to the nearest AGV. Assign the orders in : For the orders in , assign them according to the distance d and the parameter r . If d ≤ r , then assign the order i to the automated guided vehicle j . Limit the maximum capacity of the AGV: The number of orders assigned to each AGV cannot exceed the parameter c . If the current number of orders of the automated guided vehicle j has reached c , then no new orders will be assigned.

[0071] (5) Dynamic adjustment. Reallocate orders: In each scheduling, reallocation of previously assigned orders is allowed. If a more suitable AGV is found, the order can be reallocated to the new AGV. Priority adjustment: If multiple AGVs can receive the same order, it is preferentially allocated to the AGV closest in distance. If the distances are the same, it is preferentially allocated to the AGV with fewer current tasks.

[0072] (6) Output the allocation result. Generate the allocation sequence: For each automatic guided vehicle j, generate the sequence of orders assigned to it. Update the system status: Update the status of the AGV, including its current location, assigned orders, and estimated completion time.

[0073] Specific implementation steps of the path planning module: The goal of the path planning module is to generate a collision-free path from the starting point to the ending point for each AGV. The present invention adopts a breadth-first search algorithm with constraints and introduces constraint conditions to avoid collisions. The following are the specific implementation steps of the path planning module: (1) Initialization. Input parameters: starting point location and ending point location . Current time , grid layout G . Constraint parameters s 1 and s 2 . Output: A collision-free path from the starting point to the ending point.

[0074] (2) Construct the constraint set. Predict the future positions of other AGVs: For the current time , obtain the set of all moving AGVs . For each automatic guided vehicle, the estimated time to reach the target cell is . Define the constraint conditions: Time interval constraint R ( t 0): The automatic guided vehicle cannot move to an occupied cell within the time interval t o - s 1, t o + s 1 + 1], where w t0， where, to represents the time when the AGV occupies a certain cell, s 1 is the additional safety buffer time in the constraint. Target location constraint : The automatic guided vehicle cannot move to the target cell within the time interval , where, Indicates the time when the AGV reaches the target cell; D unloading Indicates the unloading time of the AGV, s 2 Indicates the additional safety buffer time in the constraint.

[0075] (3) Breadth-First Search algorithm with constraints. a. Initialize the queue: Add the starting point to the queue and mark it as visited. b. Search process: Take out a node u from the queue. For all adjacent nodes u of the node v , check the following conditions: · Whether it is within the constraint set: Check whether v is within any time interval constraint R ( t 0) or the target location constraint . If so, skip this node. · Whether it has been visited: If v has been visited, skip this node. · Whether it is the target node: If v is the target node, return the path. · Add to the queue: Add v to the queue and mark it as visited. c. Path backtracking: Start from the target node and backtrack to the starting point to generate the complete path.

[0076] (4) Path request and local avoidance. Local path request: During actual operation, the AGV requests a local path each time (for example, requests a cell). If the local path is occupied, it waits until the path is available. Dynamic avoidance: If multiple AGVs request the same path simultaneously, the AGV with a higher priority (such as the AGV with the earliest request time) will obtain the path.

[0077] (5) Output path. Complete path: The path generated by the BFS algorithm is the complete path from the starting point to the end point. Local path request: During actual operation, the AGV requests and moves to the next cell of the local path each time until it reaches the end point.

[0078] In order to determine the optimal combination of adjustment parameters in the scheduling and path planning algorithms, the present invention adopts an improved Progressive Optimal Allocation Program (AOAP) method. This method can efficiently allocate simulation resources within a limited simulation budget to optimize system performance. Specifically, the AOAP method can dynamically adjust the simulation budget allocation according to the performance of different designs, so as to effectively explore the parameter space and determine the optimal combination of adjustment parameters.

[0079] Assume that each adjustment parameter has different candidate values, and there are a total of ka design scheme, and the unknown average order cycle time generated by each design is denoted as μ i , i = 1, ⋯, k , which can be estimated through an independent simulation budget , . At the initial stage of simulation budget allocation, an initial number of simulation replications n 0 is allocated to each candidate design to estimate the performance of each design. This process ensures that each candidate design has a basic performance evaluation, providing a basis for subsequent optimization.

[0080] Under the Bayesian framework, a conjugate prior distribution is adopted to quantify μ i the uncertainty, where N represents a normal distribution, represents the prior mean, represents the prior variance. As the simulation resources are gradually allocated, the posterior distribution of the unknown average order cycle time μ i is , where N represents a normal distribution, represents the posterior mean, represents the posterior variance. In this way, the improved AOAP algorithm of the present invention gradually allocates additional simulation budget times and updates the performance evaluation of each design using the Bayesian update formula to determine the optimal combination of adjustment parameters. The formula of the improved AOAP algorithm is as follows: ; In the formula, A t+1 ( ε t ) represents the allocation decision at the ( t +1)-th moment; t represents the t -th moment; ε t represents the information set obtained from the simulation budget allocation up to the t -th moment; (1) t represents the optimal design scheme at the t -th moment; ( h ) t represents the design scheme corresponding to the index label included in the set H (t+1) at the t -th moment; represents the value obtained from allocating additional simulation budget; H(t+1) Indicates the set of index labels of candidate solutions at the ([[]] t +1)-th moment.

[0081] Among them, Indicates the value obtained by allocating additional simulation budget to solution j The calculation formula is: ; In the formula, Indicates the value obtained by allocating additional simulation budget to solution j ; t Indicates the t -th moment; ε t Indicates the information set obtained by allocating simulation budget up to the t -th moment; j Indicates the design solution that obtains simulation resources; Indicates the posterior mean estimate of the optimal design solution evaluated at the t -th moment; Indicates the posterior mean estimate of the design solution that obtains simulation resources; Indicates the posterior standard deviation of the optimal design solution evaluated at the t -th moment; Indicates the posterior standard deviation after the update of the solution that obtains simulation resources; Indicates the t -th moment, the design solutions other than the evaluated optimal design solution and the design solution that obtains simulation resources; Indicates the t -th moment, the posterior mean estimate of the design solutions other than the evaluated optimal design solution and the design solution that obtains simulation resources; Indicates the t -th moment, the posterior standard deviation of the design solutions other than the evaluated optimal design solution and the design solution that obtains simulation resources.

[0082] Through this method, the algorithm can dynamically adjust the allocation of simulation resources, so that the optimal design can be quickly identified from the candidate design set, thereby efficiently exploring the parameter space and achieving the optimization goal.

[0083] To further improve the computational efficiency, the present invention implements parallel computing on a multi-core central processing unit (CPU). The k candidate designs and T simulation budgets are evenly allocated to mA processor, each processor independently runs the improved AOAP algorithm, and performs simulation calculations in parallel on each processor. Subsequently, by integrating the calculation results of multiple processors, the optimal design is determined, thereby significantly reducing the calculation time. Through parallel computing, the present invention can significantly improve the efficiency of simulation calculations, shorten the execution time of the optimization process in large-scale simulation tasks, and ensure that the system can respond in real time to the evaluation results of different design schemes. This parallel computing framework not only effectively improves the optimization efficiency, but also greatly reduces the hardware cost and consumption of computing resources. As Figure 8 shown, it shows a schematic diagram of implementing the improved AOAP in a parallel computing environment, which illustrates how to improve the optimization efficiency through parallel computing, further accelerate the convergence speed of the optimal parameter combination, and reduce the calculation time under multi-task and multi-design evaluations.

[0084] A specific embodiment of a large-scale automatic guided vehicle scheduling and path optimization method is as follows: (1) Embodiment 1: Low-density orders, medium-scale grid As Figure 9 shown, it shows a simulation environment. The warehouse management system is modeled as an 18×18 grid, and a total of 8 AGVs are used for cargo transportation. The order arrival rate is set to 70 orders per hour, and this relatively low order arrival rate simulates the operation of the warehouse during off-peak hours. The movement speed of the AGV is 1 cell per second, the loading time is 3 seconds, the unloading time is 4 seconds, and the total simulation time is 1 hour. This scenario aims to evaluate the performance of the system under the condition of a relatively low order volume. To optimize the scheduling and path planning of the AGV, the method of the present invention introduces four key adjustment parameters and sets them respectively as: order allocation range = 1, 10, 19, ⋯, 100, the maximum capacity of the AGV = 1, 4, 7, 10, the constraint parameter in path planning = 1, 2, 3, 4, = 0, 1, 2, 3. Through the improved progressive optimal allocation program method, different combinations of these parameters are evaluated within a limited simulation budget to determine the optimal parameter combination.

[0085] After detailed simulation experiments, the method of the present invention has achieved significant performance improvement in Scenario 1. The specific results are as follows: The adjusted average order cycle time is reduced from 240.9 seconds of the default method to 86.3 seconds, a reduction of 64.2%. The order completion rate is increased from 83.8% of the default method to 96.1%, an increase of 14.7%. Although in some cases, the waiting time of the AGV has increased, this increase is to avoid collisions and congestion, thereby improving the overall performance of the system. For example, the waiting time of the AGV during loading and unloading processes is increased from 25.2 seconds and 5.8 seconds of the default method to 9.8 seconds and 6.7 seconds respectively. Compared with the existing default method, the method of the present invention shows obvious advantages in reducing the adjusted average order cycle time and increasing the order completion rate. In addition, by optimizing the parameter combination through the simulation budget allocation method, the method of the present invention can better adapt to the dynamics and randomness in the warehouse management system and improve the resource utilization efficiency. For example, in Example 1, although the default method optimized by the simulation budget allocation also achieved certain performance improvement, its adjusted average order cycle time was still 222.7 seconds and the order completion rate was 84.7%, both of which are lower than the performance of the method of the present invention. As Figure 10 shown, it shows the AGV scheduling process in Example 1 using the method of the present invention.

[0086] (2) Example 2: High-density orders, large-scale grid As Figure 11 shown, a more challenging warehouse operation environment is simulated, where the order arrival rate is increased to 120 orders per hour, the grid scale is expanded to 20×20, and the number of AGVs is increased to 10. This scenario aims to evaluate the performance of the method of the present invention under the conditions of high-density orders and large-scale grids. Other parameter settings are the same as those in Example 1, including the movement speed of the AGV, loading time, unloading time, and the adjusted parameter range.

[0087] In Example 2, the method of the present invention achieved significant performance improvement, especially in reducing the adjusted average order cycle time and increasing the order completion rate. The specific results are as follows: The adjusted average order cycle time was reduced from 687.4 seconds of the default method to 48.9 seconds, a decrease of 92.9%. The order completion rate was increased from 58.6% of the default method to 98.9%, an increase of 68.8%. In Example 2, the waiting times of the AGV during loading and unloading processes were increased from 16.3 seconds and 9.2 seconds of the default method to 33.0 seconds and 5.4 seconds respectively. This increase was mainly due to the introduction of more constraints in the path planning process of the method of the present invention to avoid collisions and congestion, thereby improving the overall performance of the system. Compared with the existing default method, the method of the present invention showed significant advantages in Scenario 2. For example, although the default method with simulated budget allocation optimization also achieved certain performance improvement, its adjusted average order cycle time was still 565.8 seconds and the order completion rate was 66.7%, both lower than the performance of the method of the present invention. In addition, the method of the present invention can better optimize the scheduling and path planning of AGVs under high-density order and large-scale grid conditions, improve resource utilization efficiency, and reduce system congestion. As Figure 12 shown, it shows the AGV scheduling process in Example 2 using the method of the present invention.

[0088] (3)Example 3: Low-density orders, medium-scale grid, different layouts As Figure 13 shown, Example 3 simulated a warehouse operation environment similar to that in Example 1, but with a different grid layout. Specifically, the grid scale in Example 3 was 18×18, the number of AGVs was 8, and the order arrival rate was 70 orders per hour. This scenario was designed to evaluate the performance of the method of the present invention under different grid layout conditions. Other parameter settings were the same as those in Example 1, including the movement speed, loading time, unloading time of the AGV, and the adjustment parameter range.

[0089] In Embodiment 3, the method of the present invention achieved significant performance improvement, especially in reducing the adjusted average order cycle time and increasing the order completion rate. The specific results are as follows: The adjusted average order cycle time decreased from 211.1 seconds of the default method to 43.1 seconds, a decrease of 79.6%. The order completion rate increased from 85.6% of the default method to 99.4%, an increase of 16.1%. In Embodiment 3, the waiting times of the AGV during loading and unloading increased from 9.8 seconds and 2.8 seconds of the default method to 10.8 seconds and 5.5 seconds, respectively. This increase is mainly due to the introduction of more constraints in the path planning process of the method of the present invention to avoid collisions and congestion, thereby improving the overall performance of the system. Compared with the existing default method, the method of the present invention showed significant advantages in Embodiment 3. For example, although the default method optimized by simulated budget allocation also achieved certain performance improvement, its adjusted average order cycle time was still 178.1 seconds and the order completion rate was 87.4%, both lower than the performance of the method of the present invention. In addition, the method of the present invention can better optimize the scheduling and path planning of AGVs under different grid layout conditions, improve resource utilization efficiency, and reduce system congestion. As Figure 14 shown, it shows the AGV scheduling process in Embodiment 3 using the method of the present invention.

[0090] Figure 2 Fig. shows an embodiment of a large-scale automatic guided vehicle scheduling and path optimization system of the present invention.

[0091] In this alternative embodiment, the large-scale automatic guided vehicle scheduling and path optimization system includes: A model construction and order allocation module 201, configured to construct a warehouse network model based on the actual layout and operation process of the warehouse, and combine the order requirements in the warehouse to obtain the order allocation sequence for each automatic guided vehicle; A driving path determination module 202, configured to analyze the task requirements of each automatic guided vehicle, and combine the warehouse network model with the breadth-first search algorithm with constraints to determine the driving path of each automatic guided vehicle from the starting point to the end point; A scheduling and path optimization module 203, configured to perform parallel computing on several processors according to the order allocation sequence and the driving path, using the progressive optimal allocation algorithm, to determine the optimal order allocation sequence and the driving path.

[0092] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as Figure 3As shown in the figure. The computer device includes a processor, a memory, and a network interface connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store static information and dynamic information data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, it implements the steps in the above method embodiments.

[0093] Those skilled in the art can understand that Figure 3 the structure shown in the figure is only a block diagram of some structures related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0094] In addition, the present invention also provides a computer device, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, it implements the steps in the above method embodiments.

[0095] In addition, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by the processor, it implements the steps in the above method embodiments.

[0096] Those of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided by the present invention can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0097] The present invention is not limited to the structures that have been described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for dispatching and path optimization of large-scale automated guided vehicles, characterized in that The method includes: Based on the actual layout and operation process of the warehouse, construct a warehouse network model, and combine with the order requirements in the warehouse to obtain the order allocation sequence for each automated guided vehicle; Analyze the task requirements of each automated guided vehicle, and combine the warehouse network model with the breadth-first search algorithm with constraints to determine the driving path of each automated guided vehicle from the starting point to the end point; According to the order allocation sequence and the driving path, use the progressive optimal allocation algorithm to perform parallel computing on several processors to determine the optimal order allocation sequence and driving path.

2. The large-scale automatic guided vehicle scheduling and path optimization method according to claim 1, wherein The step of based on the actual layout and operation process of the warehouse, constructing a warehouse network model, and combining with the order requirements in the warehouse to obtain the order allocation sequence for each automated guided vehicle includes: Obtain the actual layout and operation process of the warehouse, use the discrete event simulation algorithm to construct a warehouse network model, and set the warehouse parameters to simulate the actual operation environment of the warehouse; Analyze the order requirements in the warehouse, and determine the actual time, order set and automated guided vehicle set; Input the determined actual time, order set and automated guided vehicle set into the warehouse network model, and output the preliminary order allocation sequence for each automated guided vehicle through the warehouse network model; Use the adaptive dynamic scheduling mechanism to optimize the preliminary order allocation sequence, dynamically adjust the matching relationship between orders and automated guided vehicles, and obtain the order allocation sequence for each automated guided vehicle.

3. A method for dispatching and path optimization of a large-scale automatic guided vehicle according to claim 2, characterized in that, The warehouse parameters include: the number of automated guided vehicles, order arrival rate, movement speed of automated guided vehicles, loading time, unloading time, total simulation time, grid size and grid layout.

4. A method for dispatching and path optimization of a large-scale automatic guided vehicle according to claim 3, characterized in that, The step of using the adaptive dynamic scheduling mechanism to optimize the preliminary order allocation sequence, dynamically adjust the matching relationship between orders and automated guided vehicles, and obtain the order allocation sequence for each automated guided vehicle includes: Based on the preliminary order allocation sequence, divide the order set into orders with waiting time exceeding the preset waiting time and orders with waiting time not exceeding the preset waiting time according to the waiting time; Calculate the distance between each order and each automated guided vehicle, and dynamically adjust the starting point of the order according to the allocation status of the order; Select the automated guided vehicle with the optimal distance from the orders with waiting time exceeding the preset waiting time for priority allocation, and allocate automated guided vehicles for the orders with waiting time not exceeding the preset waiting time according to the distance and the preset capacity limit; Evaluate the order allocation situation, and adaptively and dynamically adjust the matching relationship between orders and automated guided vehicles in combination with the distance and the number of orders to optimize the order sequence of each automated guided vehicle and obtain the order allocation sequence for each automated guided vehicle.

5. A method for dispatching and path optimization of large-scale automated guided vehicles according to claim 1, characterized in that, The step of analyzing the task requirements of each automated guided vehicle, and combining the warehouse network model with the breadth-first search algorithm with constraints to determine the driving path of each automated guided vehicle from the starting point to the end point includes: Obtain and analyze the task requirements of each automated guided vehicle, and determine the starting point position, end point position and grid layout where the automated guided vehicle is located; Input the determined starting point position, end point position and grid layout where the automated guided vehicle is located into the warehouse network model, and output the preliminary driving path of each automated guided vehicle from the starting point to the end point through the warehouse network model; Optimize the preliminary driving path using the breadth - first search algorithm with constraints to obtain the driving path of each automated guided vehicle from the starting point to the end point.

6. A method for dispatching and path optimization of a large-scale automatic guided vehicle according to claim 5, characterized in that, The step of optimizing the preliminary driving path using the breadth - first search algorithm with constraints to obtain the driving path of each automated guided vehicle from the starting point to the end point includes: According to the preliminary driving path and the actual time, obtain the set of all automated guided vehicles during the movement process, and calculate the estimated arrival time of each automated guided vehicle; Based on the occupancy situation and the estimated arrival time of the automated guided vehicles, define the constraint conditions, and the constraint conditions include the time interval constraint set and the target position constraint; Combined with the defined constraint conditions, use the breadth - first search algorithm with constraints to start from the starting point and search for the target driving path to the target node; According to the target driving path, the automated guided vehicle moves according to the local path request and performs dynamic avoidance based on the priority of the request time; According to the result of the dynamic avoidance, determine the driving path of each automated guided vehicle from the starting point to the end point.

7. A method for dispatching and path optimization of large-scale automated guided vehicles according to claim 6, characterized in that, The step of combined with the defined constraint conditions, using the breadth - first search algorithm with constraints to start from the starting point and search for the target driving path to the target node includes: Create an empty queue, add the starting point to the queue, and mark the starting point as the visited state; Select a position node from the queue, check whether all adjacent nodes of the position node meet the constraint conditions check, whether they have been visited, and whether they are the target node, and add the adjacent nodes that meet all the check conditions to the queue and mark them as the visited state; Based on the target node obtained in the inspection result, backtrack from the target node to the starting point, and generate the target driving path according to the search path.

8. A method for dispatching and path optimization of a large-scale automated guided vehicle according to claim 1, characterized in that The step of according to the order allocation sequence and the driving path, using the asymptotically optimal allocation algorithm, performing parallel computing on several processors to determine the optimal order allocation sequence and the driving path includes: According to the order allocation sequence and the driving path, determine the set of design schemes of the automated guided vehicle, and allocate an initial simulation budget for each design scheme based on the conjugate prior distribution to quantify the performance evaluation; Use the allocated initial simulation budget to perform simulation experiments on each design scheme, and update the performance evaluation of each design scheme through the Bayesian update formula according to the simulation results to obtain the posterior distribution; Evaluate the value obtained by allocating additional simulation budgets to each design scheme through the value calculation formula, and optimize the allocation of the simulation budget among different design schemes in combination with the asymptotically optimal allocation algorithm; Allocate the optimized design schemes to several processors, and use the parallel computing framework to accelerate the execution of the simulation experiments; Summarize the simulation results of each processor, identify the optimal design scheme to determine the optimal order allocation sequence and the driving path.

9. A method for dispatching and path optimization of large-scale automatic guided vehicles according to claim 8, characterized in that, The formula of the asymptotically optimal allocation algorithm is: ； In the formula, A t+1 ( ε t ) represents the allocation decision at the ( t + 1)-th moment; t represents the t -th moment; ε t represents the information set obtained by simulating budget allocation up to the t -th moment; (1) t represents the optimal design scheme at the t -th moment; ( h ) t represents the set H (t+1) The design scheme corresponding to the index label included in the t -th moment; represents the value obtained by allocating additional simulation budget; H (t+1) represents the set of index labels of candidate schemes at the ( t + 1)-th moment.

10. A large-scale automatic guided vehicle scheduling and path optimization system, characterized in that, The system includes: A model construction and order allocation module, which is used to construct a warehouse network model based on the actual layout and operation process of the warehouse, and combine the order requirements in the warehouse to obtain the order allocation sequence of each automated guided vehicle; A driving path determination module, which is used to analyze the task requirements of each automated guided vehicle, and combine the warehouse network model with the breadth-first search algorithm with constraints to determine the driving path of each automated guided vehicle from the starting point to the ending point; A scheduling and path optimization module, which is used to execute parallel computing on several processors according to the order allocation sequence and the driving path, using the progressive optimal allocation algorithm, to determine the optimal order allocation sequence and the driving path.

Images