Optimization method, device, medium and program product for agv scheduling system

Abstract

The application provides an AGV scheduling system optimization method, device, medium and program product. The method belongs to the technical field of logistics and comprises the following steps: obtaining operation data of an AGV scheduling system; simulating AGV scheduling of the AGV scheduling system according to the operation data, a scheduling strategy of the AGV scheduling system and an operation environment map through a digital twin model of the AGV scheduling system to obtain optimization suggestions for the scheduling strategy and / or the operation environment map; and optimizing the scheduling strategy and / or the operation environment map according to the optimization suggestions. According to the application, precise optimization suggestions are obtained through scheduling simulation of the digital twin model of the AGV scheduling system with high fidelity, intelligent optimization of the scheduling strategy and / or the operation environment map of the AGV scheduling system is better achieved, and therefore, a global optimal solution can be obtained as much as possible during AGV scheduling planning, path congestion and low efficiency in large-scale scheduling are avoided, and sudden situations can be responded to quickly.

Description

Technical Field

[0001] This application relates to the field of logistics technology, specifically to an optimization method, equipment, medium, and program product for an AGV scheduling system. Background Technology

[0002] Robotic control systems (RCS) play an important role in automated logistics and intelligent manufacturing, especially in automated guided vehicle (AGV) scheduling systems. As the core component of the AGV scheduling system, RCS performs real-time intelligent scheduling and full-process control of AGVs.

[0003] In AGV scheduling systems, RCS typically employs traditional algorithms such as greedy algorithms and genetic algorithms to decompose and optimize scheduling tasks when handling complex scheduling tasks, and then distributes these tasks to the AGVs for processing.

[0004] However, traditional greedy algorithms and genetic algorithms have many shortcomings when calculating task decomposition, including but not limited to: the inability to obtain a globally optimal solution, the susceptibility to path congestion and inefficiency in large-scale scheduling, and the difficulty in quickly responding to unexpected situations. Therefore, it is necessary to propose an optimization method for AGV scheduling systems. Summary of the Invention

[0005] In view of this, the present invention aims to provide an optimization method for an AGV scheduling system to solve problems such as the inability to obtain a globally optimal solution, easy path congestion, low scheduling efficiency, and difficulty in quickly responding to emergencies in the prior art.

[0006] According to a first aspect of the embodiments of this application, an optimization method for an AGV scheduling system is provided. The method includes: acquiring operational data of the AGV scheduling system; simulating AGV scheduling using a digital twin model of the AGV scheduling system, based on the operational data and a dynamic scheduling map, to obtain simulation results, wherein the dynamic scheduling map includes a scheduling strategy and an operational environment map of the AGV scheduling system, and the simulation results include optimization suggestions for the AGV scheduling system, wherein the optimization suggestions include optimization suggestions for the scheduling strategy and / or the operational environment map; and optimizing the scheduling strategy and / or the operational environment map based on the optimization suggestions for the AGV scheduling system.

[0007] According to a second aspect of the embodiments of this application, an electronic device is provided, including a memory and a processor; the memory is connected to the processor and is used to store a program; the processor is used to implement the optimization method of the AGV scheduling system as described in any one of the first aspects of the embodiments of this application by running the program in the memory.

[0008] According to a third aspect of the embodiments of this application, a computer program product is provided, including computer program instructions, which, when executed by a processor, cause the processor to perform an optimization method for an AGV scheduling system as described in any one of the first aspects of the embodiments of this application.

[0009] According to a fourth aspect of the embodiments of this application, a chip is provided, including a processor and a data interface. The processor reads and runs a program stored in a memory through the data interface to execute an optimization method for an AGV scheduling system as described in any one of the first aspects of the embodiments of this application.

[0010] According to a fifth aspect of the embodiments of this application, a storage medium is provided, on which a computer program is stored. When the computer program is run by a processor, it implements the optimization method of the AGV scheduling system as described in any one of the first aspects of the embodiments of this application.

[0011] According to the technical solution of this application, the operation data of the AGV scheduling system is acquired; using a digital twin model of the AGV scheduling system, based on the operation data, the scheduling strategy of the AGV scheduling system, and the operating environment map, the AGV scheduling system is simulated to obtain simulation results. The simulation results include optimization suggestions for the AGV scheduling system, including optimization suggestions for the scheduling strategy and / or the operating environment map; based on the optimization suggestions of the AGV scheduling system, the scheduling strategy and / or the operating environment map are optimized. According to this application, real-time monitoring, simulation, and optimization of the AGV scheduling system are achieved, improving the intelligence level, adaptability, and multi-objective optimization capability of the AGV scheduling system. This enables the AGV scheduling system to obtain a globally optimal solution when performing scheduling planning after optimizing the scheduling strategy and / or the operating environment map, avoiding path congestion and inefficiency in large-scale scheduling, and quickly responding to emergencies. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0013] Figure 1A flowchart illustrating the optimization method for the AGV scheduling system provided in this application embodiment;

[0014] Figure 2 This is a schematic diagram of the process for obtaining AGV scheduling system operation data provided in an embodiment of this application;

[0015] Figure 3 A flowchart illustrating the process of determining optimization suggestions for an AGV scheduling system, as provided in an embodiment of this application;

[0016] Figure 4 A schematic diagram illustrating the process of establishing a digital twin model provided in an embodiment of this application;

[0017] Figure 5 A schematic diagram of AGV trajectory deviation provided in an embodiment of this application;

[0018] Figure 6 A schematic diagram illustrating the upstream and downstream relationships between the various models provided in the embodiments of this application;

[0019] Figure 7 A flowchart illustrating the process of determining the fidelity of a digital twin model, provided for an embodiment of this application;

[0020] Figure 8 A flowchart illustrating the process of determining the parameter accuracy of a digital twin model, provided in an embodiment of this application;

[0021] Figure 9 A flowchart illustrating the process of determining the intuitiveness of a digital twin model, provided for embodiments of this application;

[0022] Figure 10 A schematic diagram illustrating the entire simulation process provided in the embodiments of this application;

[0023] Figure 11a A schematic diagram of the system architecture of the simulation system provided in the embodiments of this application;

[0024] Figure 11b A schematic diagram of the system architecture of the simulation system provided in the embodiments of this application;

[0025] Figure 12 This is a schematic diagram of the device provided in the embodiments of this application;

[0026] Figure 13 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0028] The technical terms used in the embodiments of this application are explained below:

[0029] Digital twin technology is a simulation process that integrates multiple disciplines, multiple physical quantities, multiple scales, and multiple probabilities. It reflects the entire life cycle of the corresponding physical equipment by completing mapping in virtual space.

[0030] AGV: Automated Guided Vehicle, is a transportation vehicle equipped with electromagnetic or optical automatic guidance devices, controlled by a computer, characterized by wheeled movement, self-powered or power conversion device, and capable of automatically traveling along a prescribed guidance path.

[0031] AGV scheduling system: Control software in the upper control system used for task scheduling, vehicle management and traffic management. It controls the operation of AGVs through wireless module communication, including operation management and monitoring.

[0032] RCS: It is the core part of the AGV scheduling system, responsible for real-time intelligent scheduling and full-process control of AGVs.

[0033] Before introducing the solution proposed in this application, the relevant technologies will first be introduced:

[0034] RCS (Responsive Control System) plays a crucial role in automated logistics and intelligent manufacturing, especially in AGV (Automated Guided Vehicle) scheduling systems. As the core component of an AGV scheduling system, RCS enables real-time intelligent scheduling and full-process control of AGVs. An efficient AGV scheduling system, or an optimized AGV scheduling system, possesses a high level of intelligence, adaptability, and multi-objective optimization capabilities, effectively improving logistics efficiency and automation levels.

[0035] When the RCS in the AGV scheduling system processes a scheduling task, it first decomposes and optimizes the scheduling task, and then sends the subtasks obtained after the decomposition and optimization of the scheduling task to each AGV in the AGV scheduling system. Each AGV responds to the received subtasks to perform scheduling operations and complete the scheduling task.

[0036] However, when the RCS in the AGV scheduling system typically uses traditional algorithms for task decomposition and optimization, there are still many problems that need to be solved. For example, it is impossible to obtain a globally optimal solution, path congestion and inefficiency are prone to occur in large-scale scheduling, and it is difficult to respond quickly to emergencies.

[0037] In view of this, the embodiments of this application aim to provide an optimization method, device, medium, and program product for an AGV scheduling system. By using a highly realistic digital twin model of the AGV scheduling system, and based on the operating data of the AGV scheduling system, the AGV scheduling system is simulated to perform AGV scheduling, which can obtain accurate simulation results, namely, accurate optimization suggestions for the scheduling strategy and / or operating environment map of the AGV scheduling system. In this way, based on the optimization suggestions, the scheduling strategy and operating environment map of the AGV scheduling system can be intelligently optimized, and the globally optimal solution can be obtained as much as possible when allocating scheduling tasks and planning paths after optimization. This avoids path congestion and inefficiency in large-scale scheduling and enables rapid response to emergencies.

[0038] The optimization method of the AGV scheduling system provided in the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0039] Exemplary methods

[0040] Figure 1 This is a flowchart illustrating the optimization method for the AGV scheduling system provided in this embodiment of the application. Please refer to [link / reference]. Figure 1 In an exemplary embodiment, the provided optimization method for the AGV scheduling system is applied to the RCS of the AGV scheduling system, and the optimization method for the AGV scheduling system includes the following steps:

[0041] Step 100: Obtain the operation data of the AGV scheduling system.

[0042] The AGV scheduling system's operating environment map includes the overall map layout, the location coordinates of each key storage location (including loading and unloading ports, charging areas, waiting areas, etc.), the possible routes, and equipment data. The overall map layout includes the terrain, obstacles, and restricted areas of the AGV's operating area. The equipment data includes attribute parameters such as model, dimensions, weight, load capacity, positioning accuracy, and battery capacity.

[0043] The operational data of the AGV scheduling system includes target task instructions, which are used to instruct multiple AGVs in the AGV scheduling system to perform target scheduling tasks.

[0044] Each target task instruction is used at different times to instruct an AGV in the AGV scheduling system to execute a decomposed task of the target scheduling task in order to complete the target scheduling task. In other words, the target task instructions are executed sequentially to complete the target scheduling task.

[0045] For example, the target scheduling task R is decomposed into task r1 and task r2 by RCS. For instance, target scheduling task R is to transport goods H from point Q to point Z, task r1 is for AGV 1 to transport goods H from point Q to point d, and task r2 is for AGV 2 to transport goods H from point d to point Z. The target task instructions include instruction a sent to the AGV scheduling system at time t1 and instruction b sent to the AGV scheduling system at time t2. Time t1 is before time t2. Instruction a instructs AGV 1 in the AGV scheduling system to execute task r1, and instruction b instructs AGV 2 in the AGV scheduling system to execute task r2.

[0046] Step 110: Using the digital twin model of the AGV scheduling system, based on the operating data and dynamic scheduling map, simulate the AGV scheduling system to obtain simulation results.

[0047] The dynamic scheduling map includes the scheduling strategy and operating environment map of the AGV scheduling system; in other words, the dynamic scheduling map consists of the scheduling strategy and the operating environment map.

[0048] The aforementioned digital twin models include AGV models, loading port models, unloading port models, etc. For detailed information on each model, please refer to other embodiments.

[0049] Step 120: Optimize the scheduling strategy and / or the operating environment map based on the optimization suggestions of the AGV scheduling system.

[0050] Specifically, based on the optimization suggestions of the AGV scheduling system, the operating environment map and / or scheduling strategy of the AGV scheduling system are adjusted to obtain the adjusted operating environment map and / or scheduling strategy. Then, based on the adjusted operating environment map and scheduling strategy, the optimized dynamic scheduling map is determined.

[0051] More specifically, after obtaining the adjusted runtime environment map and scheduling strategy, the adjusted runtime environment map and scheduling strategy are verified. If the verification passes, the adjusted runtime environment map and scheduling strategy are determined as the optimized runtime environment map and scheduling strategy, and the dynamic scheduling map formed by the optimized runtime environment map and scheduling strategy is determined as the optimized dynamic scheduling map. Conversely, if the verification fails, step S110 is repeated to redetermine the optimization suggestions.

[0052] When validating the adjusted operating environment map and scheduling strategy, the digital twin model of the AGV scheduling system is adjusted in the simulation system according to the adjusted operating environment map and scheduling strategy. Using the adjusted digital twin model of the AGV scheduling system, and based on operating data, the adjusted operating environment map, and scheduling strategy, the AGV scheduling system is simulated to perform AGV scheduling. The data from this simulation is analyzed to determine whether the scheduling efficiency of the adjusted digital twin model of the AGV scheduling system has improved. If yes, the validation passes; otherwise, the validation fails.

[0053] Among them, optimizing and adjusting the runtime environment map and / or scheduling strategy can be achieved based on user commands. These user commands can be triggered by the user based on the relevant user interface and optimization suggestions for the runtime environment map and / or scheduling strategy.

[0054] In other words, the user interface displays the AGV scheduling system's operating environment map, scheduling strategy, and optimization suggestions for the operating environment map and / or scheduling strategy. User instructions are generated based on the user's operations in the user interface, making it convenient for the user to adjust the scheduling strategy and / or operating environment map according to the optimization suggestions or other actual needs.

[0055] Furthermore, the user interface displays the operating status of the digital twin model of the AGV scheduling system. This enables monitoring of the entire AGV scheduling process, allowing users to understand whether the model's operating status is abnormal and control the operation of the digital twin model through the user interface, thus realizing interaction between the user and the digital twin model of the AGV scheduling system.

[0056] For example, in the simulation optimization process, a simulation configuration file is input into the simulation system. This configuration file includes configuration information such as scheduling strategies, overall map layout, location coordinates of key storage locations (including loading and unloading ports, charging areas, waiting areas, etc.), walkable routes, and vehicle information. Based on the simulation configuration file, an operating environment map, including the site environment, is generated. Based on the scheduling strategy and the operating environment map, the AGV scheduling system is simulated using a digital twin model of the AGV scheduling system to perform AGV scheduling, and simulation results are output. If the simulation results include optimization suggestions for the scheduling strategy and the operating environment map, the scheduling strategy is modified according to these suggestions, and the walkable route map and vehicle parameters are modified according to the optimization suggestions for the operating environment map, such as optimization suggestions for walkable routes and vehicle parameters. Using the modified scheduling strategy and operating environment map, the AGV scheduling system is re-simulated to perform AGV scheduling, resulting in new simulation results. If the new simulation results are empty, meaning the new simulation results indicate that the modified scheduling strategy and operating environment map meet the requirements, but there are no optimization suggestions for the modified scheduling strategy and operating environment map, the process ends.

[0057] Specifically, in the simulation system, the UE experimental module is used to optimize / adjust various models in the digital twin model of the AGV scheduling system. This UE experimental module includes functions such as optimization, calibration, Monte Carlo simulation, and sensitivity analysis. The UE experimental module can be constructed using the UE platform model.

[0058] During model optimization, the parameters to be optimized for each model are input into the UE experimental module. The UE experimental module selects at least one optimization method based on the parameters and optimization suggestions, and performs multiple iterations of optimization based on the selected method, outputting the optimized parameters for each model. The optimization methods available in the UE module experiment include calibration, Monte Carlo, sensitivity analysis, or custom methods. The optimization method used in each iteration can be the same or different.

[0059] According to the technical solution of this embodiment, the operation data of the AGV scheduling system is obtained; through the digital twin model of the AGV scheduling system, based on the operation data, the operation environment map, and the scheduling strategy, the AGV scheduling system is simulated to obtain simulation results, including optimization suggestions for the AGV scheduling system; then, based on the optimization suggestions, the operation environment map and / or scheduling strategy in the dynamic scheduling map of the AGV scheduling system are optimized. According to this application, by using the digital twin model of the AGV scheduling system to simulate AGV scheduling, and based on the optimization suggestions obtained from the simulation, the operation environment map and / or scheduling strategy are optimized, realizing real-time monitoring, simulation, and optimization of the AGV scheduling system. This improves the intelligence level, adaptability, and multi-objective optimization capability of the AGV scheduling system, enabling the AGV scheduling system to obtain the globally optimal solution as much as possible after its operation environment map and / or scheduling strategy are optimized, thereby avoiding path congestion and inefficiency in large-scale scheduling and enabling rapid response to emergencies.

[0060] To obtain the data required for simulation, i.e., to achieve step 100, when the AGV scheduling system's operating data includes the target task instructions, this application provides an optional embodiment, such as... Figure 2 As shown, step 100 can be specifically implemented through steps 200 and 210, or steps 200 and 220.

[0061] Step 200: Obtain the operating environment map of the AGV scheduling system.

[0062] Specifically, RCS obtains the simulation parameter configuration file, reads the overall map layout data, key storage location coordinate data, and walkable route data from the simulation parameter configuration file, and generates the operating environment map of the AGV scheduling system according to the read data.

[0063] Step 210: Obtain the historical task instructions when the AGV scheduling system executes the target scheduling task, and determine the historical task instructions as the target task instructions.

[0064] Specifically, RCS acquires AGV call data and AGV status data from the AGV scheduling system, analyzes and organizes the AGV call data and AGV status data to obtain historical task instructions, and determines the historical task instructions as target task instructions.

[0065] The AGV call data includes the task identifier, initiation time, end time, start address, end address, and associated cage information. The AGV status data includes the AGV's current position, previous position, current speed, current load status, current orientation, current status (obstacle avoidance, waiting, charging, etc.), previous orientation, and current battery level at each specific moment. The AGV call data can be stored in a database as a table, named, for example, "Task Call Table."

[0066] It is understandable that the AGV call data and AGV status data mentioned above are historical data generated by the AGV scheduling system when executing historical scheduling tasks. This historical data can be obtained from the production system and stored in a database, such as MySQL, TiDB, Redis, or Kafka.

[0067] AGV status data can be acquired through sensors, RFID technology, I / O devices, etc., and integrated into a database.

[0068] More specifically, based on the different calling tasks, AGV call data and AGV status data are divided into time slices. The AGV call data and AGV status data corresponding to different time slices are analyzed and organized to obtain historical task instructions. By dividing the time slices, the time slices corresponding to each calling task with shorter time lengths are obtained. This allows the data corresponding to traffic control points that cannot be simulated and have a significant impact on the simulation to be extracted from the AGV call data and AGV status data, resulting in relatively pure task execution instructions, i.e., historical task instructions, as well as data status.

[0069] It is understandable that the execution time of a call task is one time slice.

[0070] Based on the batch of calling tasks in the AGV calling data, the task handling time, the utilization rate of each storage location, and the efficiency of personnel can be determined.

[0071] By using historical task instructions as target task instructions, scheduling simulations can be performed on historical tasks using the target task instructions, resulting in better simulation effects. This makes it easier to identify the shortcomings of the digital twin model and make timely adjustments.

[0072] Step 220: Obtain the task data of the target scheduling task from the production system, and generate the target task instruction based on the scheduling strategy of the AGV scheduling system, the operating environment map, and the task data.

[0073] The task data includes waybill data, parcel data, and even parcel arrival time, transit point, and location.

[0074] The scheduling strategy is used for task allocation, path planning, and traffic control for each AGV in the AGV scheduling system. That is, the scheduling strategy includes scheduling algorithms, traffic control algorithms, and path planning algorithms.

[0075] The scheduling algorithm decomposes the scheduling task and assigns the decomposed tasks to different AGVs. The path planning algorithm plans the path for each AGV to execute its assigned task. The traffic control algorithm sets the passage order of AGVs when their paths intersect, in order to avoid collisions and deadlocks and ensure collision-free path planning for AGVs.

[0076] Specifically, the production system and the AGV scheduling system are connected via RCS communication. RCS obtains task data for the target scheduling task from the production system. Based on the AGV scheduling system's scheduling strategy, operating environment map, and task data, RCS allocates tasks to each AGV in the AGV scheduling system and generates corresponding task instructions, which are then designated as the target task instructions. This enables the integration of the production system and the AGV scheduling system, allowing the digital twin model to communicate and exchange data effectively with actual hardware devices (such as AGVs and robotic arms) and upper-level information systems (such as WMS and ERP). This enables online and even real-time simulation of task data, facilitating timely detection and adjustment of scheduling strategy issues.

[0077] To obtain optimization suggestions for the AGV scheduling system through simulation, this application provides an optional embodiment that analyzes simulation data obtained from AGV scheduling in a simulated AGV scheduling system to obtain optimization suggestions for the AGV scheduling system. For example... Figure 3 As shown, its specific implementation includes steps 300, 310 and 320.

[0078] Step 300: Using the digital twin model of the AGV scheduling system, simulate the AGV scheduling system according to the scheduling strategy on the operating environment map based on the operating data to obtain simulation data.

[0079] Specifically, RCS sends the target task instructions to the simulation system in the execution order. Within the simulation system, the digital twin model of the AGV scheduling system simulates the execution of the target scheduling task on the runtime environment map according to the scheduling strategy. This simulates the AGV scheduling system performing AGV scheduling, obtaining data from the digital twin model during the execution of the target scheduling task—the simulation data. The execution order refers to the sequence in which the AGVs in the AGV scheduling system execute each target task instruction.

[0080] Specifically, the simulation system includes a simulation environment and a digital twin model, used for simulation within the simulation environment using the digital twin model. During simulation, a simulation environment configuration file is first obtained, including operating system configuration, central processing unit (CPU) configuration, memory configuration, and network configuration, and the simulation environment is configured according to this configuration file. After receiving the target task instruction, the simulation system sends it to the digital twin model of the AGV scheduling system. Within the configured simulation environment, the digital twin model of the AGV scheduling system simulates the AGV scheduling system performing AGV scheduling according to the scheduling strategy on the runtime environment map to complete the target scheduling task.

[0081] In addition, during the execution of the target scheduling task through the digital twin model of the AGV scheduling system, the AGV call data and AGV status data generated during the process are recorded.

[0082] For example, in the simulation process, at the start of the simulation, the simulation configuration file is loaded / read to generate the AGV operation map, i.e., the aforementioned operation environment map, including environmental objects; the target task instruction is obtained according to method one or method two. Method one: read the task call table and AGV status data from the database and continuously send the target task instruction to the simulation system; Method two: obtain task data from the production system, simulate the scheduling strategy to generate the target task instruction and output it to the simulation system; the simulation system calls the digital twin model to execute the target scheduling task according to the instruction; the simulation system provides feedback on the operation data when executing the target scheduling task; the feedback operation data is analyzed and the simulation results are output; the simulation ends.

[0083] Step 310: Determine the simulation operation indicators based on the simulation data.

[0084] The simulation operation indicators cover multiple dimensions, including full-process simulation, single task execution, AGV positioning, and driving route.

[0085] Specifically, simulation operation metrics across different dimensions can be categorized into different levels based on their varying degrees of influence on their respective dimensions. For example, simulation operation metrics within the same dimension can be divided into primary and secondary metrics based on their importance. Furthermore, based on different practical needs, simulation operation metrics within the same dimension can be further subdivided into even more levels.

[0086] In the full-process simulation dimension, based on the AGV call data generated during simulation, the time slices allocated to each target scheduling task are divided into time slices for unbranched routes. The AGV call data and AGV status data within each time slice are then analyzed to obtain simulation operation indicators for the full-process simulation dimension. The associated models for this dimension include the AGV model, the loading port model, and the unloading port model. In other words, the simulation operation indicators for the full-process simulation dimension are obtained by processing the data generated by the AGV model, the loading port model, and the unloading port model during task execution.

[0087] For the single-task execution dimension, the simulation performance metrics are obtained by averaging the data from multiple runs (i.e., multiple executions of the call task) during simulation. The associated model for this dimension includes the AGV model; that is, the simulation performance metrics for the single-task execution dimension are obtained by processing the data generated by the AGVs executing tasks.

[0088] For the AGV positioning dimension, the simulated position of the AGV is determined based on the AGV status data generated during simulation, resulting in simulation operation indicators for this dimension. The associated model for this dimension includes the AGV model; that is, the simulation operation indicators for the AGV positioning dimension are obtained by processing the data generated by the AGV performing tasks.

[0089] For the travel route dimension, the simulated travel route, or operating trajectory, is obtained based on the simulated travel routes formed by the simulated positions of each AGV. The associated model for this dimension includes the operating trajectory model, that is, the simulated operating indicators under the travel route dimension are obtained by processing the AGV routes in the operating trajectory model.

[0090] For example, the simulation operation indicators under the full-process simulation dimension include primary indicators and secondary indicators. Primary indicators include production capacity, while secondary indicators include total mileage, AGV utilization rate, AGV attendance rate, average congestion rate (congestion rate caused by traffic control, congestion rate caused by obstacle avoidance), AGV non-idle time (or average task execution time), and average waiting time for pickup in the area. AGV utilization rate is the ratio of the time AGVs spend performing tasks (excluding waiting or congestion) to the total time AGVs spend performing tasks. AGV attendance rate is the ratio of the number of AGVs performing tasks to the total number of AGVs in the AGV scheduling system. Average congestion rate is the ratio of the time AGVs spend waiting or congested while performing tasks to the total time AGVs spend performing tasks.

[0091] For example, the simulation operation metrics under the single task execution dimension include primary metrics and secondary metrics. For instance, primary metrics include task duration, and secondary metrics include task distance (length), segmented route travel time, pickup time, and delivery time. Task duration is the average duration of executing multiple tasks.

[0092] For example, the simulation operation indicators under the AGV positioning dimension include primary indicators, such as the simulated position of the AGV.

[0093] For example, the simulation operation indicators under the driving route dimension include primary indicators and secondary indicators. For instance, the primary indicators include the number of times the AGV passes through the driving route or the number of AGVs passing through the driving route, and the secondary indicators include the heat map of the driving route of each AGV, which represents the number of times the AGV passes through each driving route.

[0094] Depending on the actual needs, other simulation operation indicators can be added, such as acceleration and velocity as simulation operation indicators under other dimensions.

[0095] Step 320: Analyze the scheduling strategy and the operating environment map based on the simulation operation indicators to obtain the optimization suggestions.

[0096] The optimization suggestions include suggestions for optimizing the runtime environment map and / or the scheduling strategy.

[0097] The optimization suggestions for the environmental operation map include suggestions for optimizing the number of AGVs, optimizing AGV parameters, and / or optimizing the site layout. An AGV quantity optimization suggestion might be, for example, adding one AGV. An AGV parameter optimization suggestion might be, for example, adjusting the value of a certain AGV parameter from 'a' to 'b'. Site layout optimization suggestions include optimization suggestions for the location of each device and optimization suggestions for the distribution of storage locations, such as moving a device from location 'a' to location 'b', or adjusting the location of a material unloading area from area 1 to area 2.

[0098] Specifically, for simulation operation indicators of different dimensions, the evaluation function corresponding to that dimension is called. Based on the evaluation function and the value of each simulation operation indicator under the corresponding dimension, each dimension is evaluated to obtain the evaluation score of each dimension. Based on the evaluation score of each dimension, optimization suggestions for the AGV scheduling system are obtained.

[0099] To establish a digital twin model of an AGV scheduling system, this application provides an optional embodiment that can be used to establish a digital twin model. For example... Figure 4 As shown, its specific implementation is as follows, including steps 400 and 410:

[0100] Step 400: Based on digital twin technology, construct the loading port model, unloading port model, multiple AGV models, and scene model.

[0101] The AGV model is used to simulate the driving and picking / placing functions of AGVs in actual production. It maps the actual driving and picking / placing of AGVs to the simulation environment, so as to verify the scheduling strategy adopted by the AGV scheduling system and the scheduling effect of the operating environment map in the simulation environment.

[0102] The loading port model is used to simulate the process from unloading to scanning the code at the loading port to generate a call task in a real environment, which can reflect the frequency of AGV calls on the site.

[0103] The unloading port model is used to simulate the process from unloading port to loading in a real environment. It can simulate the actual loading time on site based on the unloading port and the number of personnel.

[0104] The scenario model includes models of various devices and operators in the scenario, i.e., the operating environment map, where the AGV scheduling system is located, and can represent the scenario layout of the AGV scheduling system. For example, the scenario model includes a cage car model, a personnel model, and an obstacle model.

[0105] Specifically, using 3D modeling technology, a mechanistic model is created that represents the loading and unloading ports, multiple AGVs, and various devices and operators in the real world. This model includes the physical spatial structure of the AGV scheduling system, such as the loading and unloading ports, and the physical spatial structure of the AGVs and various devices and operators in the scene. It also reflects the status and performance of each device in the AGV scheduling system. Then, the operational and configuration data of the loading and unloading ports, multiple AGVs, and various devices and operators in the scene are used to construct a data model for the loading and unloading ports, multiple AGVs, and various devices and operators in the scene. By combining the mechanistic model and the data model, the production process of the loading and unloading ports, multiple AGVs, and various devices and operators in the scene is simulated, resulting in the loading port module, the unloading port model, the multiple AGV models, and the scene model.

[0106] The AGV model includes multiple execution modules. It simulates the driving and loading / unloading functions of AGVs in actual production, mapping the actual AGV driving and loading / unloading behavior to the simulation environment to verify the effectiveness of the scheduling strategy adopted by the AGV scheduling system.

[0107] The AGV model comprises multiple execution modules, including an obstacle avoidance component, a pick-and-place component, and a motion control component. The obstacle avoidance component handles forward, backward, and lateral obstacle avoidance, outputting the obstacle avoidance distance to the motion control component. The motion control component controls the AGV's movement based on its speed, acceleration, angular velocity, and angular acceleration, and controls whether the AGV decelerates based on the obstacle avoidance distance. The pick-and-place component controls the AGV's forklift arm to pick up and place goods.

[0108] The model parameters of the AGV model include ontology parameters, system parameters, and learning parameters. The data sources for the ontology parameters and system parameters can be suppliers, while the learning parameters are obtained through machine learning.

[0109] The system parameters of the AGV model include: key data of the obstacle avoidance component, such as obstacle avoidance distance and obstacle avoidance enable; key data of the motion control component, such as speed, angular velocity, power, starting position, acceleration, and friction; and key data of the picking and placing component, such as lifting time.

[0110] The ontological parameters (i.e., ontological attributes) of an AGV model include position, length, width, height, control status (obstacle avoidance, waiting, charging, etc.), battery level, obstacle avoidance distance, speed, and even weight, maximum speed, maximum acceleration, maximum angular velocity, turning radius, etc.

[0111] The input parameters of the AGV model include a list of travel routes and the permissible speed for each route. The output parameters include the AGV's operating status and its trajectory. The list of travel routes is determined based on the target task instructions executed by the AGV. The permissible speed is the AGV's maximum speed. The AGV's operating status includes attributes such as its position, speed, and acceleration. The AGV's trajectory represents the execution status of the target task instructions.

[0112] For example, motion control data (including speed and acceleration), obstacle avoidance data (including distances to obstacles in front, behind, left, and right), and collision avoidance data (collision avoidance signals) are input into the AGV model, and speed and acceleration are output.

[0113] The parameters that need to be learned in the AGV model, i.e., the learning parameters of the AGV model, are the AGV position information in the key data of the motion control components. The dimension of the simulation operation index to which these learning parameters belong is the travel route dimension. The AGV position can be obtained through machine training. Using historical AGV positions, the offset and correction data of the AGV during the travel process are learned to obtain the position deviation, and then the travel process is reproduced in the simulation environment based on the position deviation. Using the position, velocity, and acceleration of the AGV at each moment, combined with the task path information, the calculation method of the AGV's running trajectory, i.e., the calculation method of the AGV position, is trained. The formula for calculating the AGV position is, for example, s = ut + 1 / 2at. 2 +△t. u is the velocity vector, t is time, a is the acceleration vector, and △t is the position deviation vector determined by machine learning.

[0114] Compared to the time it takes for the cage car to generate its first task call after being unloaded from the truck and arriving at the loading port, the time it takes for the cage car to generate another task call after being transported away by the forklift AGV is more important, or rather, has a greater impact on the simulation.

[0115] Specifically, the learning parameters under different feature dimensions can be classified into parameter levels based on their importance. For example, the parameter corresponding to the feature dimension of the time it takes for the cage car to call a task again after being transported away by the AGV is a level 1 parameter, and the parameter corresponding to the feature dimension of the time it takes for the cage car to generate its first task call at the loading port after being unloaded from the truck is a level 2 parameter. Among them, the level 1 parameter is more important than the level 2 parameter.

[0116] For example, the deviation between the actual (running) trajectory of the AGV and the simulated (running) trajectory of the AGV can be as follows: Figure 5 As shown.

[0117] The model parameters for the feeding port model include system parameters and learning parameters. The data source for the system parameters can be the task status table in the Galaxy scheduling system, while the learning parameters are obtained through machine learning. The system parameters include key data for the feeding port, such as the number and location of the feeding ports, as well as key parameters for the operators, such as the number of operators.

[0118] The input parameters of the loading port model include vehicle information, cage data, and cage number. Vehicle information includes, for example, the time the vehicle arrives at the loading platform, vehicle identification number, and loading platform number. Cage data includes cage type, size, and weight. The cage number is used to query cage flow direction information. The output parameters of the loading port model include relevant data for calling tasks, such as the endpoint coordinates corresponding to the cage flow direction, the generation of calling tasks based on combined cage information, and the frequency of calling tasks.

[0119] The learning parameter for the loading port model is the call duration from the key data of the loading port. The frequency of calling tasks is related to the call duration, which can be obtained through machine learning. The feature dimensions of the call duration include the time it takes for the cage truck to call a task again after being moved by the forklift, and the time it takes for the cage truck to generate its first call task after being unloaded from the truck at the loading port. The time it takes for the cage truck to call a task again after being moved by the forklift is determined based on the AGV's previous task pickup time and subsequent task call time obtained from the data source; it is the time length between these two times. The time it takes for the cage truck to generate its first call task after being unloaded from the truck at the loading port is determined based on the vehicle unloading time and the cage truck call time inside the vehicle obtained from the data source; it is the time length between these two times.

[0120] The pickup time of the previous task and the call time of the next task of the AGV can be obtained by updating the time in the task status table of the Galaxy Dispatch System, and the vehicle unloading time can be obtained from the big data platform.

[0121] The model parameters of the feed port model include system parameters and learning parameters.

[0122] The system parameters of the discharge port model include: key parameters of the discharge port, such as the number and location of the discharge ports; and key parameters of the operators, such as the number of personnel. The key data for the discharge ports is sourced from the task status table in the Xinghe scheduling system, while the key data for the operators is sourced from site operation outputs. The learning parameters are obtained through machine learning.

[0123] Understandably, the loading port model, unloading port model, AGV model, and scenario model each include at least one sub-model. For example, the loading port model includes the vehicle entry model, unloading model, and task generation model; the unloading port model includes the loading model and vehicle exit model; the AGV model includes the picking and placing model, obstacle avoidance model, etc.; and the scenario model includes the running trajectory model, cage car model, personnel model, and obstacle model.

[0124] Specifically, the feed inlet model, discharge inlet model, AGV model, and scene model may each have at least one level of sub-models. There is a hierarchical relationship between the feed inlet model, discharge inlet model, AGV model, scene model, and their sub-models. Based on this hierarchical relationship, the feed inlet model, discharge inlet model, AGV model, and scene model can be defined as a first-level model; the next level of sub-models under the first-level model can be defined as second-level models; the next level of sub-models under the second-level model can be defined as third-level models, and so on, until the level of the lowest-level sub-model is determined.

[0125] Among them, the vehicle entry model, unloading model, task generation model, loading model, vehicle exit model, cargo retrieval and placement model, obstacle avoidance model, operation trajectory model, personnel model and obstacle model are secondary models, and the cage car model is a tertiary model.

[0126] The following section introduces some sub-models of the loading port model, unloading port model, AGV model, and scene model:

[0127] The unloading model is used to statistically analyze the vehicle unloading process. The loading model simulates loading cages from the loading area into trucks. The task call model (task generation model) simulates the process of personnel periodically using barcode scanners to generate task calls, including operations such as pulling cages, scanning codes, and clicking the barcode scanner to confirm the call. The interval is the time between the call time of any task and the pickup time of the previous task, where pickup time is the time it takes for the AGV to insert and retrieve the cage. The trajectory model records the AGV's trajectory to verify the AGV's operational status by comparing the data indicators of the simulated route with those of the actual route. The cage model simulates the manual loading and unloading of cages used in the previous AGV's transport process, to statistically analyze cage information, and to learn the distribution of cage types and cage distribution in the transfer area. The personnel model simulates the distribution of operators.

[0128] The sub-model also includes a human intervention model and a cage dismantling model, which can be divided into, for example, a three-level model according to the hierarchical relationship of the models. The human intervention model is used to simulate the situation of manually intervening to solve the AGV deadlock problem and outputs the duration of manual intervention. The cage dismantling model is used to simulate the process of dismantling and recombining cage cars of mixed cage types.

[0129] The unloading model's parameters include system parameters and learning parameters. These learning parameters can be obtained through machine learning, and the data source for these system parameters can be tables in a database, with table names such as dm_ops.super_flow_batch_stat_result. The system parameters include key vehicle data such as model, number of cages (including cage status), and unloading port, as well as key personnel parameters such as personnel data.

[0130] The input parameters of the unloading model include vehicle information and cage car information, such as the number and model of the cage cars. The output parameters of the unloading model include the time from the cage car to the unloading buffer and cage car information. The time from the cage car to the unloading buffer is obtained by calculating the time distribution based on the learned unloading time.

[0131] For example, vehicle data, unloading port data, and shift schedule are input into the unloading model, and unloading data and cage car data are output. Among them, vehicle data includes the unloading vehicle number and the total number of cages in the vehicle; unloading port data includes the unloading port number; shift schedule includes the personnel shift schedule for the current shift; unloading data includes the vehicle numbers and personnel deployment status of vehicles already unloaded in the current shift; and cage car quantity includes the unloading port number, vehicle number, personnel deployment status, and the time when the cage car enters the unloading line and the time when the cage car leaves the unloading line.

[0132] The learning parameters of the unloading model are the unloading time distribution in the key data of unloading time, and the feature dimension of the unloading time distribution is the time distribution of the cage truck from inside the truck to the barcode scanning position within the shift.

[0133] The model parameters of the loading vehicle include system parameters and learning parameters. The data source for the system parameters can be on-site observation or monitoring data, while the learning parameters can be obtained through machine learning. The system parameters include key data of the loading vehicle, such as its location, quantity, and volume, as well as key personnel parameters, such as the number of personnel.

[0134] The input parameters of the loading model include vehicle information and cage car information, such as the number of cage cars and their cage numbers. The output parameters of the loading model include the time it takes for the cage cars to be moved, i.e., the duration of the cage cars in the loading buffer zone.

[0135] For example, the loading port data, shift schedule, and cage car data are input into the unloading model, and the loading data and cage car data are output. The loading port data includes the loading port number, the shift schedule includes the personnel shift schedule for the current shift, the cage car data includes information on cage cars arriving at the loading port, the loading data includes the vehicle numbers and personnel deployment status of vehicles already loaded in the current shift, and the cage car data includes the loading port number, vehicle number, personnel deployment status, the time the cage car enters the loading line, and the time the cage car leaves the loading line.

[0136] The task call model includes system parameters, the data of which are obtained from on-site observation or monitoring. System parameters include key data about the feeding port, such as its location and number, as well as key personnel data, such as the number of personnel. The output parameters of the task call model include the call task, which includes the endpoint coordinates corresponding to the cage flow direction, the information on combined cages that generates the call task, and the frequency of the call task.

[0137] The learning parameters of the task call model are key data of the task call interval, such as the call duration of personnel operating the barcode scanner. The corresponding feature dimensions include the time interval between the same operator calling the vehicle during peak hours: that is, the time length between the previous task pickup time and the subsequent task call time.

[0138] The input parameters for the trajectory model include the AGV's operating route and the travel time for each route segment. The AGV's operating route is the route indicated by the instructions issued by the RCS, and the travel time for each route segment is the travel time when passing through each sub-path of the operating route.

[0139] For example, the storage location information, waypoint information, rest point information, path information, and area information are input into the operation trajectory model, and the AGV operation map is output.

[0140] The output parameters of the operation trajectory model include the AGV density ratio in the operation route, the route frequency ratio, and the average travel time. In the operation route, the AGV density ratio is the ratio of the number of AGVs traveling on each path segment to the total number of AGVs when multiple AGVs are running. The route frequency ratio is the ratio of the number of times each path segment is traversed by multiple AGVs to the maximum number of times it is traversed. The average travel time is the average travel time when multiple AGVs travel on each path segment.

[0141] The model parameters for the dismantling cage model include ontological parameters and learned parameters. The ontological parameters are derived from manual on-site measurements, while the learned parameters are obtained through machine learning. Ontological parameters include key data about the cages, such as their location, quantity, length, width, and height.

[0142] The input parameters for the dismantling mixed cage model include the number of mixed cages and half-cages, and the number of new whole cages generated in that area can be calculated based on the ratio. The output parameters of the dismantling mixed cage model include the number of task calls, which is the number of task calls calculated based on the ratio of secondary calls to create whole cages.

[0143] The learning parameters for the cage dismantling model are the proportion of cages being converted into whole cages and the proportion of secondary call tasks generated in the key data of the cage car. The proportion of cages being converted into whole cages is the ratio of the number of cages that appear in a shift to the total number of whole cages created. The proportion of secondary call tasks generated is the ratio of the number of secondary tasks generated after cages are assembled to the total number of call tasks after cages are assembled.

[0144] The model parameters of the cage car model include system parameters, ontology parameters, and learning parameters. The system parameters are sourced from the PCM dataset stored in the database and include key cage car parameters such as flow direction, cage number, and weight. The ontology parameters are sourced from on-site observations and measurements and include key cage car data such as cage car type, length, width, and height. The learning parameters are obtained through machine learning.

[0145] The input parameters of the cage car model include the number of cages, and the output parameters include cage car information. The number of cages can be obtained from historical data, while the cage car information can be automatically generated in batches.

[0146] The learning parameters for the cage car model are the key parameters of the cage car, including the cage car's flow direction, cage number, and weight, as well as the type, flow direction, and weight distribution of all cage cars in the transfer yard where the AGV scheduling system is located.

[0147] The learning parameters of the human intervention model include key data on the duration of human intervention, such as the duration of human operation, including the distribution of intervention time for all human interventions.

[0148] The input parameters of the manual intervention model include the number of AGVs, and the output parameters include the deadlock resolution time. The number of AGVs refers to the number of AGVs affected (unable to move) during a deadlock. Deadlock situations can be detected through real-time monitoring of communication between AGVs.

[0149] It is understandable that the learning algorithm for the learning parameters in the above model can be a machine learning algorithm, and the specific algorithm type can be adjusted based on the different actual models.

[0150] In addition, the specific values ​​of the learning parameters for each model can be stored in the simulation configuration file. During simulation, the values ​​of the learning parameters for each model can be configured according to the parameter values ​​in the simulation configuration file.

[0151] The model constructed above needs to be updated in real time to accurately reflect the current status and historical behavior of the AGV. Moreover, the construction of the model must follow the "four-fold, four-fold, eight-fold" construction principle, namely, precision, standardization, lightweight, visualization, as well as interactivity, integration, reconfigurability, and evolution.

[0152] Step 410: Based on the spatial constraints and process coupling relationships between the loading port model, unloading port model, multiple AGV models, and scenario models, construct a digital twin model of the AGV scheduling system.

[0153] Among them, spatial constraint relationship refers to the spatial positional relationship between models in the operating environment map, and process coupling relationship refers to the association relationship between models in the job process of executing scheduling tasks.

[0154] The spatial constraints between the models can be determined based on the physical spatial structure of each model. The process coupling between the models can be determined based on the work flow when the AGV scheduling system executes scheduling tasks.

[0155] The workflow of the AGV scheduling system when executing scheduling tasks is as follows: the vehicle enters the loading port area of ​​the transfer yard where the AGV scheduling system is located and unloads the goods. The task is assigned according to the goods information, and the task execution instructions are sent to the AGV in sequence to control the AGV to execute the task. The goods are transported to the unloading port area according to the planned path. After the goods are loaded, the loading vehicle drives away from the transfer yard where the AGV scheduling system is located.

[0156] For example, the upstream and downstream relationships, i.e., the process coupling relationships, between the various models in the digital twin model of the simulation system can be as follows: Figure 6 As shown, the end-to-end data flow is as follows: Information from the vehicle information table and the cage car information table is input into the vehicle entry model. The vehicle entry model outputs the vehicle number, which is then input into the unloading model. The unloading model outputs the cage car information, which, along with the starting landmark, is input into the task generation model. The task generation model outputs a single task, including information such as the cage car, starting landmark, and ending landmark. The path point list and maximum path speed are then input into the corresponding AGV model via RCS, outputting task execution information and AGV operation information. Next, the travel route and travel time are input into the operation trajectory model, outputting an AGV operation heatmap. Finally, the vehicle number being loaded is input into the loading model, outputting loading information. The AGV operation heatmap uses color intensity to represent the number of times multiple AGVs travel on each passable route.

[0157] The information output by each model can be written into the corresponding table for storage and transmission, and the resulting table is used as the final output.

[0158] According to the technical solution of this embodiment, digital twin technology is used to construct models identical to those of the actual AGV scheduling system, including a loading port model, a discharging port model, multiple AGV models, and a scene model. Then, based on the spatial constraints and process coupling relationships between the models, and considering the physical spatial structure of each model and their correlation in the operational process, a digital twin model of the AGV scheduling system identical to the actual system is constructed. In this way, simulation is performed using the digital twin model, and by observing the digital twin model, the AGV operation data during the simulation process can be monitored to obtain accurate simulation results. Based on the simulation results, the AGV scheduling system can be optimized, resulting in better optimization effects.

[0159] To ensure the reliability of the optimization suggestions, this application provides an optional embodiment that, when the simulation results include the fidelity of the digital twin model, determines the fidelity of the digital twin model based on the difference between the simulation performance indicators and the actual performance indicators. For example... Figure 7 As shown, its specific implementation includes steps 500, 510, and 520:

[0160] Step 500: Determine the actual operating indicators of the AGV scheduling system when executing the target scheduling task based on the operating data.

[0161] The operational data includes AGV call data and AGV status data.

[0162] Furthermore, the actual operating metrics correspond one-to-one with the simulation operating metrics. The difference between the actual and simulation operating metrics lies in that the actual operating metrics are determined based on historical data, i.e., operational data, while the simulation operating metrics are determined based on simulation data obtained from scheduling simulations. Therefore, similar to simulation operating metrics, the actual operating metrics encompass multiple dimensions, including full-process simulation, single-task execution, AGV positioning, and travel routes. For a detailed description of each metric in the actual operating metrics, please refer to the description of the metric items in the simulation operating metrics above; it will not be repeated here.

[0163] Step 510: For each simulation operation index, the difference between the simulation operation index and its corresponding actual operation index is determined as the index difference, and the fidelity of the simulation operation index is determined based on the index difference.

[0164] Step 520: Determine the fidelity of the digital twin model based on the fidelity of all simulation operation indicators and the fidelity weights corresponding to each simulation operation indicator.

[0165] Among them, the fidelity weight is used to characterize the importance of the fidelity of the corresponding simulation operation index.

[0166] Specifically, based on the difference between the indicators and the values ​​of the actual operating indicators, the similarity between the simulated operating indicators and the actual operating indicators is calculated, and the similarity is used as the fidelity of the simulated operating indicators.

[0167] For example, according to Calculate the fidelity of the simulation performance indicators. Among them, Q... i Let z be the fidelity of the i-th simulation performance index. i For the value of the i-th actual operating indicator corresponding to the i-th simulation operating indicator, Let be the value of the i-th simulation operation index. Let be the difference between the i-th simulated performance indicator and the i-th actual performance indicator. i is a positive integer, and the maximum value of i is the total number of simulated performance indicators.

[0168] The simulation performance index has a fidelity range of [0, 1].

[0169] If a simulation operation index has multiple values, and the corresponding actual operation index also has multiple values, and there is a one-to-one correspondence between the multiple values ​​of the simulation operation index and the multiple values ​​of the actual operation index, the average of the differences between the multiple values ​​of the simulation operation index and the corresponding values ​​of the actual operation index is taken as the index difference.

[0170] Among them, the larger the fidelity weight, the higher the importance of the fidelity of the corresponding simulation operation index.

[0171] Specifically, the fidelity of the digital twin model is obtained by averaging the fidelity of all simulation performance metrics, and this average value is used as the fidelity of the digital twin model.

[0172] For example, according to Where S represents the fidelity of the digital twin model, N1 represents the total number of simulation performance indicators / actual performance indicators, and w i Q is the fidelity weight for the i-th simulation performance index. i Let represent the fidelity of the i-th simulation performance index.

[0173] The fidelity of the digital twin model ranges from [0, 1].

[0174] According to the technical solution of this embodiment, after determining the simulation operation indicators based on the simulation data, the actual operation indicators of the AGV scheduling system when executing the target scheduling task are determined based on the operation data. These actual operation indicators correspond one-to-one with the simulation operation indicators determined during simulation. For each simulation operation indicator, the difference between the simulation operation indicator and its corresponding actual operation indicator is determined, and the fidelity of the simulation operation indicator is determined based on the difference. Then, based on the fidelity of all simulation operation indicators and the fidelity weights corresponding to each simulation operation indicator that characterize its fidelity, the fidelity of the digital twin model is determined, and the simulation results containing the fidelity of the digital twin model are output. In this way, based on the fidelity of the digital twin model in the simulation results, the reliability of the current simulation and the obtained optimization suggestions can be well understood. Therefore, when the fidelity of the digital twin model is low, the digital twin model can be adjusted in a timely manner, and optimization suggestions can be re-determined to ensure the optimization effect.

[0175] To determine the parameter accuracy of a digital twin model, this application provides an optional embodiment that, when the simulation results include the parameter accuracy of the digital twin model, determines the parameter accuracy of the digital twin model based on the differences between the model parameters in the digital twin model and the corresponding entity parameters. For example... Figure 8 As shown, its specific implementation includes steps 600 and 610:

[0176] Step 600: For any model parameter in the digital twin model, determine the difference between the value of the model parameter and the value of its corresponding entity parameter as the parameter difference, and determine the accuracy of the model parameter based on the parameter difference.

[0177] The model parameters correspond one-to-one with the entity parameters.

[0178] Based on the above introduction to the model parameters of each model in the digital twin model, it can be seen that the model parameters of the digital twin model include system parameters, ontology parameters, and / or learning parameters.

[0179] Among them, entity parameters refer to the parameters of each part of the actual AGV scheduling system, which correspond one-to-one with the model parameters of the digital twin model.

[0180] Specifically, based on the parameter difference and the values ​​of the entity parameters, the similarity between the model parameters and the entity parameters is calculated, and the similarity is used as the accuracy of the model parameters.

[0181] For example, according to Calculate the accuracy of the model parameters. In this context, P... j Let r be the accuracy of the j-th model parameter. j For the value of the i-th entity parameter corresponding to the j-th model parameter, Let j be the value of the j-th model parameter. Let be the parameter difference between the j-th model parameter and the j-th entity parameter, where j is a positive integer and the maximum value of j is the total number of model parameters in the digital twin model. The accuracy of the model parameters ranges from [0, 1].

[0182] Step 610: Determine the parameter accuracy of the digital twin model based on the accuracy of all model parameters in the digital twin model and the accuracy weight corresponding to each model parameter.

[0183] Among them, the accuracy weight is used to characterize the importance of the accuracy of its corresponding model parameter.

[0184] Since the digital twin model of the AGV scheduling system is a model built based on a simulation of the actual AGV scheduling system, its system parameters and ontology parameters are generally consistent with those of the actual AGV scheduling system. However, the learned parameters may differ. Therefore, the accuracy of the parameters of the digital twin model is usually determined by the learned parameters of each model within the digital twin model.

[0185] Of course, based on different actual needs, in order to more accurately determine the parameter accuracy of the digital twin model, all model parameters of the digital twin model are used to determine the parameter accuracy of the digital twin model.

[0186] Among them, the larger the accuracy weight, the more important the accuracy of the corresponding model parameters is.

[0187] Specifically, the average accuracy of all model parameters is taken to obtain the parameter accuracy of the digital twin model, and this average is used as the parameter accuracy of the digital twin model.

[0188] For example, according to Determine the parameter accuracy of the digital twin model. Where R represents the parameter accuracy of the digital twin model, N² is the total number of model parameters / entity parameters, and w... j P represents the accuracy weight of the j-th model parameter. j Let be the accuracy of the j-th model parameter.

[0189] The parameter accuracy of the digital twin model ranges from [0, 1].

[0190] Alternatively, specifically, the accuracy of the model parameters of each model in the digital twin model is averaged to obtain the parameter accuracy of each model in the digital twin model. Then, based on the parameter accuracy of each model in the digital twin model, operations such as averaging are performed to determine the parameter accuracy of the digital twin model.

[0191] For example, according to Determine the parameter accuracy of each model in the digital twin model. Among them, R...m Let N be the parameter accuracy of the m-th model in the digital twin model. 2n w represents the total number of model parameters for the m-th model. p Let be the accuracy weight of the p-th model parameter in the m-th model, and let be the accuracy of the p-th model parameter in the m-th model, where p is a positive integer and the maximum value of p is N. 2m The parameter accuracy of the m-th model ranges from [0, 1].

[0192] For example, according to Determine the parameter accuracy of the digital twin model. Where R represents the parameter accuracy of the digital twin model, M represents the total number of models in the digital twin model, and w... q R represents the accuracy weights of the q-th model. q Let be the parameter accuracy of the q-th model. q is a positive integer, and its maximum value is M.

[0193] According to the technical solution of this embodiment, for any model parameter in the digital twin model, the accuracy of the model parameter is determined based on the parameter difference between the model parameter and the corresponding entity parameter. Furthermore, the accuracy of the digital twin model's parameters is determined based on the accuracy of all model parameters in the digital twin model and the accuracy weights that characterize the importance of each model parameter's accuracy. Thus, by analyzing the parameter accuracy of the digital twin model in the simulation results, the reliability of the current digital twin model and the optimization suggestions obtained from simulations based on it can be well understood. Therefore, when the parameter accuracy of the digital twin model is low, the model parameters of the digital twin model can be adjusted in a timely manner, and optimization suggestions can be redefined to ensure the optimization effect.

[0194] To determine the intuitiveness of a digital twin model, this application provides an optional embodiment that, when the simulation results include the intuitiveness of the digital twin model, determines the intuitiveness of the digital twin model based on the ratio of the number of target key parameters to the total number of key parameters, and the ratio of the number of target key parameter pairs to the total number of key parameter pairs. Figure 9 As shown, its specific implementation includes steps 700, 710, 720, and 730:

[0195] Step 700: Determine the intuitiveness of the key parameters by the ratio of the number of target key parameters to the total number of key parameters in the digital twin model.

[0196] The number of target key parameters refers to the number of key parameters that can be displayed.

[0197] Step 710: Determine the intuitiveness of the key parameter pairs by the ratio of the number of target key parameter pairs to the total number of key parameter pairs in the digital twin model.

[0198] The number of target key parameter pairs is the total number of displayable key parameter pairs, and a key parameter pair is a combination of parameters that are related.

[0199] Step 720: Determine the intuitiveness of the model simulation process by comparing the visible simulation process to the entire simulation process.

[0200] Step 730: Determine the intuitiveness of the digital twin model based on the intuitiveness of the key parameters, the intuitiveness of the key parameter pairs, and the intuitiveness of the model simulation process.

[0201] Since the construction, operation, optimization, and migration of the model may require in-depth participation or understanding from relevant personnel, it may be difficult for humans to quickly understand and analyze the model files. In order to improve the work efficiency and decision-making quality of relevant personnel, the established model needs to have a certain degree of intuitiveness, that is, a high degree of intuitiveness. When evaluating the intuitiveness of a digital twin model, it is necessary to consider the intuitiveness of the model parameters, structure, and operation, so as to accurately quantify the intuitiveness of the digital twin model and achieve an intuitive evaluation of the digital twin model.

[0202] Among them, key parameters are the model parameters that are of high importance in the digital twin model. Depending on the actual needs, they can also refer to all model parameters in the digital twin model.

[0203] Specifically, the intuitiveness of the digital twin model is determined by the sum of the intuitiveness of the key parameters, the intuitiveness of the key parameter pairs, and the intuitiveness of the model simulation process.

[0204] For example, according to Determine the intuitiveness of the key parameters in the digital twin model. Here, C1 represents the intuitiveness of the key parameters, w1 represents the intuitiveness weight of each key parameter, D1 represents the number of key parameters that can be intuitively presented (displayed visually on the web and output as reports), and N represents the number of key parameters. 3,1 This represents the total number of key parameters in a digital twin model.

[0205] For example, according to Determine the intuitiveness of the key parameter pairs in the digital twin model, or the intuitiveness of the model structure. Here, C2 represents the intuitiveness of the key parameter pairs in the digital twin model, w2 represents the intuitiveness weight of the key parameter pairs, D2 represents the number of key parameter pairs that can be intuitively presented (visually displayed and output as reports on the web), and N... 3,2 This represents the total number of key parameter pairs in the digital twin model.

[0206] For example, the intuitiveness of the simulation process of the digital twin model, or the intuitiveness of the operation process of the digital twin model, is determined according to C3 = w3 × opv × 100%. Here, C3 represents the intuitiveness of the simulation process of the digital twin model, W3 is the intuitiveness weight of the simulation process, and opv characterizes whether the digital twin model can intuitively reflect the operation process of the physical entity, or in other words, opv is the ratio of the visible process to the entire simulation process of the digital twin model.

[0207] The value of opv ranges from [0,1]. Alternatively, the value of opv can be 0 or 1. If the digital twin model can intuitively represent the operation process of the physical entity, then the value of opv is 1; if the digital twin model cannot intuitively represent the operation process of the physical entity, then the value of opv is 0.

[0208] For example, the intuitiveness of the digital twin model is determined according to C = C1 + C2 + C3, where C represents the intuitiveness of the digital twin model.

[0209] Furthermore, the sum of w1, w2, and w3 is 1. For example, the values ​​of w1, w2, and w3 are 0.25, 0.25, and 0.5, respectively.

[0210] According to the technical solution of this embodiment, the ratio of the number of visually displayable target key parameters to the total number of key parameters in the digital twin model is determined as the intuitiveness of the key parameters. The ratio of visually displayable target key parameter pairs to the total number of key parameter pairs in the digital twin model is determined as the intuitiveness of the key parameter pairs, i.e., the model structure. The intuitiveness of the simulation process is determined based on whether the simulation operation of the digital twin model is visually displayed, thus quantifying the intuitiveness of model parameters, model structure, and simulation process. Subsequently, the intuitiveness of the digital twin model is determined based on the intuitiveness of the key parameters, the intuitiveness of the key parameter pairs that characterize the model structure, and the intuitiveness of the model operation (simulation) process. This allows for a relatively comprehensive and accurate quantification of the intuitiveness of the digital twin model, enabling an evaluation of its intuitiveness.

[0211] For example, the entire simulation process includes a learning phase simulation, an evaluation phase simulation, and a What-if phase simulation. The learning phase simulation is used to adjust the digital twin model to obtain a digital twin model of the AGV scheduling system that closely resembles reality; the evaluation phase simulation is used to adjust scheduling strategies and / or operating environment maps, etc., to ensure business efficiency; and the What-if phase simulation is used to adjust parameters in the configuration file to obtain the optimal simulation that meets business indicators, thereby determining accurate optimization suggestions.

[0212] The entire simulation process can be as follows: Figure 10As shown, on the cloud server side, the simulation system is invoked to load the simulation model (i.e., the aforementioned digital twin model) and the simulation configuration file. Then, the data in the static database tables—that is, the data generated by the AGV scheduling system during actual business production (i.e., the aforementioned operational data and operational environment map), such as site layout, number of cage cars, number of storage locations, number of AGV vehicles, AGV operating routes, etc.—and the dynamic data, i.e., the task data obtained from the production system, are cached and matched for simulation data. Discrete event (i.e., target scheduling task) simulation is then performed. The simulation operation data is fed back step-by-step to the RCS, and the simulation results are output after the simulation is completed, including virtual output data generated during the simulation, such as the AGV simulation operating route and simulation handling time. Afterwards, the simulation results and the data generated during actual business production (such as actual operating energy efficiency, actual AGV handling time, actual picking time, etc.) are analyzed and diagnosed autonomously to obtain a research and development simulation report and a business simulation report.

[0213] Afterwards, the model is retrained on the local terminal based on the R&D simulation report to obtain a new model. The new model is then evaluated, and the business simulation report is fed back to RCS offline.

[0214] When diagnosing and analyzing simulation results, the What-if approach is used to determine the optimal simulation that meets business metrics. This involves adjusting a specific metric or parameter, such as adding 10 AGVs, and then performing simulations according to each condition in the if list, resulting in a list of simulation results for each if condition. Analyzing these results under each if condition determines the optimal simulation / simulation under the optimal if condition that meets the business metrics, a business simulation report is generated. This report includes optimization strategies for scheduling and / or the runtime environment map.

[0215] For example, the system architecture of the simulation system can be as follows: Figure 11a As shown, the software architecture of the simulation system can be as follows: Figure 11b As shown.

[0216] Exemplary device

[0217] Accordingly, embodiments of this application also provide an apparatus, such as Figure 12 As shown, the AGV scheduling system optimization device 800 provided in this embodiment may include: an acquisition module 810, a simulation module 820, and an optimization module 830.

[0218] The system includes an acquisition module 810 for acquiring operational data of the AGV scheduling system. A simulation module 820, using a digital twin model of the AGV scheduling system and based on the operational data and a dynamic scheduling map, simulates AGV scheduling to obtain simulation results. The dynamic scheduling map includes the AGV scheduling system's scheduling strategy and operational environment map. The simulation results include optimization suggestions for the AGV scheduling system, including optimization suggestions for the scheduling strategy and / or the operational environment map. An optimization module 830 optimizes the scheduling strategy and / or the operational environment map based on the optimization suggestions from the AGV scheduling system.

[0219] In some exemplary embodiments, the operational data includes a target task instruction, which instructs multiple AGVs in the AGV scheduling system to execute a target scheduling task. Specifically, the acquisition module 810 can be used to acquire historical task instructions from when the AGV scheduling system executes the target scheduling task, and determine the historical task instructions as the target task instruction; or, it can acquire task data for the target scheduling task from the production system, and generate the target task instruction based on the scheduling strategy of the AGV scheduling system, the operational environment map, and the task data, wherein the scheduling strategy is used for task allocation, path planning, and traffic control for each AGV.

[0220] In some exemplary embodiments, the simulation module 820 is further configured to, before obtaining simulation results by simulating AGV scheduling through the digital twin model of the AGV scheduling system based on the operating data and dynamic scheduling map, construct a loading port model, a discharging port model, multiple AGV models, and a scene model using digital twin technology, and construct a digital twin model of the AGV scheduling system based on the spatial constraint relationship and process coupling relationship between the loading port model, the discharging port model, the multiple AGV models, and the scene model. The spatial constraint relationship refers to the spatial positional relationship between the models in the operating environment map, and the process coupling relationship refers to the association relationship between the models in the work process of executing the scheduling task.

[0221] In some exemplary embodiments, the simulation module 820 is specifically used to simulate the AGV scheduling system performing AGV scheduling in the operating environment map according to the scheduling strategy using the digital twin model of the AGV scheduling system and the operating data, thereby obtaining simulation data; determine simulation operation indicators based on the simulation data; and analyze the scheduling strategy and the operating environment map based on the simulation operation indicators to obtain the optimization suggestions, wherein the optimization suggestions include optimization suggestions for the operating environment map and / or optimization suggestions for the scheduling strategy of the AGV scheduling system.

[0222] In some exemplary embodiments, the simulation results also include the fidelity of the digital twin model. Specifically, the simulation module 820 is further configured to, after determining the simulation operation indicators based on the simulation data, determine the actual operation indicators when the AGV scheduling system executes the target scheduling task based on the operation data, wherein the actual operation indicators correspond one-to-one with the simulation operation indicators; for each simulation operation indicator, determine the difference between the simulation operation indicator and its corresponding actual operation indicator as an indicator difference, and determine the fidelity of the simulation operation indicator based on the indicator difference; determine the fidelity of the digital twin model based on the fidelity of all simulation operation indicators and the fidelity weight corresponding to each simulation operation indicator, wherein the fidelity weight is used to characterize the importance of the fidelity of its corresponding simulation operation indicator.

[0223] In some exemplary embodiments, the simulation results also include the parameter accuracy of the digital twin model. Specifically, the simulation module 820 is further configured to, for any model parameter in the digital twin model, determine the difference between the value of the model parameter and the value of its corresponding entity parameter as a parameter difference, and determine the accuracy of the model parameter based on the parameter difference, wherein the model parameter and the entity parameter correspond one-to-one; and determine the parameter accuracy of the digital twin model based on the accuracy of all model parameters in the digital twin model and the accuracy weight corresponding to each model parameter, wherein the accuracy weight is used to characterize the importance of the accuracy of its corresponding model parameter.

[0224] In some exemplary embodiments, the simulation results also include the intuitiveness of the digital twin model. Specifically, the simulation module 820 is further configured to determine the intuitiveness of the key parameters as the ratio of the number of target key parameters to the total number of key parameters in the digital twin model, where the number of target key parameters is the number of displayable key parameters; to determine the intuitiveness of key parameter pairs as the ratio of the number of target key parameter pairs to the total number of key parameter pairs in the digital twin model, where the number of target key parameter pairs is the total number of displayable key parameter pairs, and key parameter pairs are parameter combinations with a correlation; to determine the intuitiveness of the model simulation process as the ratio of the visible simulation process to the entire simulation process; and to determine the intuitiveness of the digital twin model based on the intuitiveness of the key parameters, the intuitiveness of the key parameter pairs, and the intuitiveness of the model simulation process.

[0225] The AGV scheduling system optimization device provided in this embodiment belongs to the same application concept as the AGV scheduling system optimization method provided in the above embodiments of this application. It can execute the AGV scheduling system optimization method provided in any of the above embodiments of this application and has the corresponding functional modules and beneficial effects of executing the AGV scheduling system optimization method. Technical details not described in detail in this embodiment can be found in the specific processing content of the AGV scheduling system optimization method provided in the above embodiments of this application, and will not be repeated here.

[0226] It should be understood that the devices in the above apparatus can be implemented in the form of a processor calling software. For example, the apparatus includes a processor connected to a memory containing instructions. The processor calls the instructions stored in the memory to implement any of the above methods or to implement the functions of each unit of the cargo transportation system. The processor can be a general-purpose processor, such as a CPU or microprocessor, and the memory can be internal or external to the apparatus. Alternatively, the units in the apparatus can be implemented in the form of hardware circuits. By designing the hardware circuits, some or all of the unit functions can be implemented. The hardware circuits can be understood as one or more processors. For example, in one implementation, the hardware circuit is an ASIC, and the functions of some or all of the above units are implemented by designing the logical relationships between the components within the circuit. In another implementation, the hardware circuit can be implemented using a PLD, such as an FPGA, which can include a large number of logic gates. The connection relationships between the logic gates are configured through a configuration file, thereby implementing the functions of some or all of the above units. All units of the above cargo transportation system can be implemented entirely through processor calling software, entirely through hardware circuits, or partially through processor calling software with the remaining parts implemented through hardware circuits.

[0227] In this application embodiment, a processor is a circuit with signal processing capabilities. In one implementation, the processor can be a circuit with instruction reading and execution capabilities, such as a CPU, microprocessor, GPU, or DSP. In another implementation, the processor can implement certain functions through the logical relationships of hardware circuits. These logical relationships are fixed or reconfigurable. For example, the processor may be a hardware circuit implemented as an ASIC or PLD, such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document and configuring the hardware circuit can be understood as the processor loading instructions to implement the functions of some or all of the above units. Furthermore, it can also be a hardware circuit designed for artificial intelligence, which can be understood as an ASIC, such as an NPU, TPU, or DPU.

[0228] As can be seen, each unit in the above cargo transportation system can be one or more processors (or processing circuits) configured to implement the above methods, such as: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor types.

[0229] Furthermore, the units in the above devices can be integrated in whole or in part, or they can be implemented independently. In one implementation, these units are integrated together and implemented in the form of a System-on-Chip (SoC). The SoC may include at least one processor for implementing any of the above methods or for implementing the functions of the units of the cargo transportation system. The at least one processor may be of different types, such as CPU and FPGA, CPU and artificial intelligence processor, CPU and GPU, etc.

[0230] Exemplary electronic devices

[0231] This application provides an electronic device, see [link to relevant documentation] Figure 13 As shown, the electronic device includes a memory 1300 and a processor 1310 connected to the memory 1300.

[0232] The memory 1300 is used to store programs.

[0233] Processor 1310 is configured to execute an optimization method for any of the AGV scheduling systems described in any of the above embodiments.

[0234] For details on the specific processing procedure of the processor 1310 described above, please refer to the description of the above method embodiments. For details on the specific implementation of the processor 1310, please refer to the description of the above embodiments.

[0235] Specifically, the aforementioned electronic device may also include: a bus, a communication interface 1320, an input device 1330, and an output device 1340.

[0236] The processor 1310, memory 1300, communication interface 1320, input device 1330, and output device 1340 are interconnected via a bus. Among them:

[0237] A bus can include a pathway for transmitting information between various components of a computer system.

[0238] The processor 1310 can be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present invention. It can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0239] The processor 1310 may include a main processor, as well as a baseband chip, modem, etc.

[0240] The memory 1300 stores a program that executes the technical solution of the present invention, and may also store an operating system and other key business functions. Specifically, the program may include program code, which includes computer operation instructions. More specifically, the memory 1300 may include read-only memory (ROM), other types of static storage devices capable of storing static information and instructions, random access memory (RAM), other types of dynamic storage devices capable of storing information and instructions, disk storage, flash memory, etc.

[0241] Input device 1330 may include a device for receiving user input data and information, such as an error microphone, keyboard, mouse, camera, scanner, light pen, voice input device, touch screen, pedometer, or gravity sensor.

[0242] Output device 1340 may include devices that allow information to be output to a user, such as a speaker, display screen, printer, etc.

[0243] The communication interface 1320 may include a device that uses any transceiver to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0244] The processor 1310 executes the program stored in the memory 1300 and calls other devices, which can be used to implement the various steps of any of the AGV scheduling system optimization methods provided in the above embodiments of this application.

[0245] This application also proposes a chip, which includes a processor and a data interface. The processor reads and runs a program stored in the memory through the data interface to execute the optimization method of the AGV scheduling system described in any of the above embodiments. For the specific processing procedure and its beneficial effects, please refer to the embodiments of the optimization method of the AGV scheduling system described above.

[0246] Exemplary computer program products and storage media

[0247] In addition to the methods and devices described above, embodiments of this application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps in the optimization method of the AGV scheduling system according to various embodiments of this application as described in any of the above embodiments of this specification.

[0248] Computer program products can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0249] Furthermore, embodiments of this application may also be storage media storing a computer program, which is executed by a processor in the steps of the optimization method of the AGV scheduling system according to various embodiments of this application as described in any of the above embodiments of this specification.

[0250] For the foregoing method embodiments, in order to simplify the description, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0251] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For apparatus embodiments, since they are basically similar to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0252] The steps in the methods of the various embodiments of this application can be adjusted, merged, or deleted in order according to actual needs, and the technical features described in each embodiment can be replaced or combined.

[0253] The modules and sub-modules in the various embodiments of the present application's devices and terminals can be merged, divided, and deleted according to actual needs.

[0254] It should be understood that the disclosed terminals, devices, and methods can be implemented in other ways, given the several embodiments provided in this application. For example, the terminal embodiments described above are merely illustrative. For instance, the division of modules or sub-modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple sub-modules or modules may be combined or integrated into another module, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or modules, and may be electrical, mechanical, or other forms.

[0255] The modules or submodules described as separate components may or may not be physically separate. The components that constitute a module or submodule may or may not be physical modules or submodules; that is, they may be located in one place or distributed across multiple network modules or submodules. Some or all of the modules or submodules can be selected to achieve the purpose of this embodiment's solution, depending on actual needs.

[0256] Furthermore, the functional modules or sub-modules in the various embodiments of this application can be integrated into one processing module, or each module or sub-module can exist physically separately, or two or more modules or sub-modules can be integrated into one module. The integrated modules or sub-modules described above can be implemented in hardware or in the form of software functional modules or sub-modules.

[0257] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0258] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software unit executed by a processor, or a combination of both. The software unit can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0259] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0260] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An optimization method for an AGV scheduling system, characterized in that, The method includes: Obtain the operating data of the AGV scheduling system; Using the digital twin model of the AGV scheduling system, simulation results are obtained by simulating AGV scheduling based on the operating data and dynamic scheduling map. The dynamic scheduling map includes the scheduling strategy and operating environment map of the AGV scheduling system. The simulation results include optimization suggestions for the AGV scheduling system, which include optimization suggestions for the scheduling strategy and / or the operating environment map. Based on the optimization suggestions of the AGV scheduling system, optimize the scheduling strategy and / or the operating environment map.

2. The method according to claim 1, characterized in that, The operational data includes target task instructions, which are used to instruct multiple AGVs in the AGV scheduling system to perform target scheduling tasks. The process of obtaining the operational data of the AGV scheduling system includes: Obtain the historical task instructions from when the AGV scheduling system executes the target scheduling task, and determine the historical task instructions as the target task instructions; or, The system obtains the task data of the target scheduling task from the production system, and generates the target task instruction based on the scheduling strategy of the AGV scheduling system, the operating environment map, and the task data. The scheduling strategy is used for task allocation, path planning, and traffic control for each AGV.

3. The method according to claim 1, characterized in that, Before obtaining simulation results by simulating AGV scheduling using the digital twin model of the AGV scheduling system based on the operating data and dynamic scheduling map, the method further includes: Based on digital twin technology, a loading port model, a discharging port model, multiple AGV models, and a scene model are constructed. Based on the spatial constraints and process coupling relationships between the loading port model, the discharging port model, the multiple AGV models, and the scene model, a digital twin model of the AGV scheduling system is constructed. The spatial constraints are the spatial positional relationships between the models in the operating environment map, and the process coupling relationships are the association relationships between the models in the work process of executing the scheduling task.

4. The optimization method for the AGV scheduling system according to claim 1, characterized in that, The simulation results obtained by using the digital twin model of the AGV scheduling system, based on the operating data and dynamic scheduling map, to simulate AGV scheduling by the AGV scheduling system include: Using the digital twin model of the AGV scheduling system, and based on the operational data, the simulation data is obtained by simulating the AGV scheduling system performing AGV scheduling in the operational environment map according to the scheduling strategy. The simulation performance indicators are determined based on the simulation data; The optimization suggestions are obtained by analyzing the scheduling strategy and the operating environment map based on the simulation operation indicators.

5. The method according to claim 4, characterized in that, The simulation results also include the realism of the digital twin model; After determining the simulation performance indicators based on the simulation data, the method further includes: Based on the operational data, the actual operational indicators of the AGV scheduling system when executing the target scheduling task are determined, and the actual operational indicators correspond one-to-one with the simulation operational indicators. For each of the simulation operation indicators, the difference between the simulation operation indicator and its corresponding actual operation indicator is determined as the indicator difference, and the fidelity of the simulation operation indicator is determined based on the indicator difference. The fidelity of the digital twin model is determined based on the fidelity of all the simulation operation indicators and the fidelity weights corresponding to each simulation operation indicator. The fidelity weights are used to characterize the importance of the fidelity of the corresponding simulation operation indicators.

6. The method according to claim 4, characterized in that, The simulation results also include the parameter accuracy of the digital twin model; The method further includes: For any model parameter in the digital twin model, the difference between the value of the model parameter and the value of its corresponding entity parameter is determined as the parameter difference, and the accuracy of the model parameter is determined based on the parameter difference. The model parameter and the entity parameter correspond one-to-one. The parameter accuracy of the digital twin model is determined based on the accuracy of all model parameters in the digital twin model and the accuracy weight corresponding to each model parameter. The accuracy weight is used to characterize the importance of the accuracy of the corresponding model parameter.

7. The method according to claim 4, characterized in that, The simulation results also include the intuitiveness of the digital twin model; The method further includes: The ratio of the number of target key parameters to the total number of key parameters in the digital twin model is determined as the intuitiveness of the key parameters, where the number of target key parameters refers to the number of key parameters that can be displayed. The ratio of the number of target key parameter pairs to the total number of key parameter pairs in the digital twin model is determined as the intuitiveness of the key parameter pairs. The number of target key parameter pairs is the total number of displayable key parameter pairs, and the key parameter pairs are parameter combinations with correlation relationships. The ratio of the visible simulation process to the entire simulation process in the simulation of the digital twin model is determined as the intuitiveness of the model simulation process. The intuitiveness of the digital twin model is determined based on the intuitiveness of the key parameters, the intuitiveness of the key parameter pairs, and the intuitiveness of the model simulation process.

8. An electronic device, characterized in that, Including memory and processor; The memory is connected to the processor and is used to store programs; The processor is used to implement the optimization method of the AGV scheduling system as described in any one of claims 1 to 7 by running the program in the memory.

9. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the optimization method for the AGV scheduling system as described in any one of claims 1-7.

10. A computer program product, characterized in that, The computer program product includes computer program instructions, which, when executed by a processor, cause the processor to perform the optimization method for the AGV scheduling system as described in any one of claims 1-7.

Images