Wireless charging AGV cluster operation cooperative scheduling method and device considering battery capacity constraint, and medium
By constructing a Markov decision process model and deep reinforcement learning method to optimize the wireless charging AGV cluster operation scheduling, the problem of low charging efficiency of traditional AGVs is solved, and efficient, economical and environmentally friendly AGV cluster operation collaborative scheduling is achieved, thereby improving the production and operation efficiency of smart manufacturing factories.
Patent Information
- Application Number
- CN202510711244.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-29
- Publication Date
- 2025-09-05
AI Technical Summary
Traditional AGVs rely on wired charging or battery replacement, which makes it difficult to meet the efficiency, economy, and environmental protection needs of smart manufacturing factories. They also have low charging efficiency, long downtime, large space occupation, high energy consumption, and safety risks.
A wireless charging AGV cluster operation collaborative scheduling method considering battery capacity constraints is adopted. By constructing a Markov decision process model and a deep reinforcement learning method based on the Actor-Critic architecture, the operation scheduling of the AGV cluster is optimized, and efficient collaborative scheduling of the wireless charging AGV cluster is achieved.
It improves the operating efficiency and flexibility of AGV cluster operations, can efficiently handle complex and dynamically changing operating environments, ensures the efficiency and robustness of solving the optimal solution, and significantly improves the efficiency of power replenishment.
Smart Images
Figure CN120601566A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of wireless charging technology. Background Art
[0002] With the rapid development of digital-driven green intelligent manufacturing, automated guided vehicles (AGVs) play an important role in factory material handling, warehousing and logistics, and effectively improve operational efficiency.
[0003] However, traditional AGVs usually rely on wired charging or battery replacement, and have problems such as low charging efficiency, long downtime, large space occupation, high energy consumption, and charging safety. They are difficult to meet the high efficiency, economy, and environmental protection needs of smart manufacturing factories.
[0004] Application Contents This application aims to solve the problem that traditional AGVs rely on wired charging or battery replacement, which makes it difficult to meet the efficiency, economy and environmental protection needs of smart manufacturing factories. We now provide a wireless charging AGV cluster operation collaborative scheduling method that considers battery capacity constraints.
[0005] In a first aspect, the present application provides a method for collaboratively scheduling wireless charging AGV cluster operations considering battery capacity constraints, including: The state of the wireless charging AGV cluster is input into the wireless charging AGV cluster operation collaborative scheduling model to obtain the action of the wireless charging AGV cluster, and the wireless charging AGV cluster is collaboratively scheduled according to the action.
[0006] In one possible design, the training process of the wireless charging AGV cluster operation collaborative scheduling model includes: Constructing a virtual environment for wireless charging AGV cluster job scheduling, acquiring job demand data and AGV cluster status data in real time from the virtual environment, and then constructing a Markov decision process model. The Markov decision process model includes the state space, action space, and reward function of the wireless charging AGV cluster job scheduling; Extracting the state space and action space of the wireless charging AGV cluster job scheduling in the Markov decision process model as the input and output of the wireless charging AGV cluster job collaborative scheduling model respectively; The wireless charging AGV cluster operation collaborative scheduling model is trained using a deep reinforcement learning method based on the Actor-Critic architecture.
[0007] In one possible design, building a virtual environment for wireless charging AGV cluster job scheduling includes: A wireless charging AGV cluster scheduling mathematical model is constructed, and a virtual environment for wireless charging AGV cluster operation scheduling is constructed using the wireless charging AGV cluster scheduling mathematical model.
[0008] In one possible design, the objective function expression of the wireless charging AGV cluster scheduling mathematical model is: , in, For the goal, is the maximum task completion time of AGV.
[0009] In one possible design, the constraints of the wireless charging AGV cluster scheduling mathematical model include: ; ; ; ; ; Where, Represents a collection of AGVs, is the total number of AGVs, ; Represents a collection of jobs, Indicates the total number of jobs, ; Represents a set of charging times, Indicates the total number of charging times. ; , For homework The duration of the work, The duration of each charge; is a 0-1 decision variable. In the Perform work after charging but ,otherwise ; is a 0-1 decision variable. Carry out the Second charge ,otherwise ; and They are In the Perform work after charging The charge during loading and unloading process, and meet the following requirements: , , , , , , for In the Perform work after charging The previous power, For homework energy consumption, is the battery capacity of the AGV, For wireless charging efficiency, For wireless charging speed, and They are Execute Job Loading and unloading times, It is the minimum operating SOC ratio of AGV.
[0010] In one possible design, the state space includes: Scheduling workload requirements, energy consumption, and time consumption, AGV load capacity, battery capacity, charging efficiency, and power, as well as wireless charging facility charging capabilities.
[0011] In one possible design, the action space includes: AGV operation arrangement decision, wireless charging operation decision and initial operation power.
[0012] In one possible design, the reward function of the Markov decision process model is The expression is as follows: , in, and Execute actions respectively and actions End time.
[0013] The second aspect of the present application provides a wireless charging AGV cluster operation collaborative scheduling device considering battery capacity constraints. The wireless charging AGV cluster operation collaborative scheduling device considering battery capacity constraints includes a processor and a memory. The memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the wireless charging AGV cluster operation collaborative scheduling method considering battery capacity constraints as described above.
[0014] A third aspect of the present application provides a computer storage medium storing at least one instruction, which is loaded and executed by a processor to implement the wireless charging AGV cluster operation collaborative scheduling method considering battery capacity constraints as described above.
[0015] Beneficial effects of this application: 1. Compared with the traditional wired charging or battery replacement AGV operation scheduling method, the wireless charging AGV operation scheduling proposed in this application has higher operating efficiency and flexibility.
[0016] 2. The deep reinforcement learning optimization method proposed in this application for the collaborative scheduling model of wireless charging AGV cluster operations considering battery capacity constraints can efficiently handle complex and dynamically changing operating environments and ensure the efficiency and robustness of solving the optimal solution.
[0017] In summary, this application aims to build an efficient production material scheduling system for green smart manufacturing factories. It proposes a modeling and optimization method for the coordinated scheduling of wireless charging AGV cluster operations, taking into account battery capacity constraints. By introducing wireless charging technology, the efficiency of energy replenishment during the operation process is significantly improved, and efficient coordination of AGV cluster operation task allocation and charging planning under battery capacity constraints is achieved. This application can be widely used in smart manufacturing material transportation, warehousing and logistics, and other fields, helping to improve production and operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] Figure 1 This is a diagram showing the layout of AGV operation scheduling and wireless charging facilities in a smart manufacturing factory; Figure 2 This is a diagram of the deep reinforcement learning optimization algorithm based on the Actor-Critic architecture; Figure 3 is a comparison of the optimal solutions for AGV operation scheduling under wired charging mode and wireless charging mode, where (a) represents the wired charging mode and (b) represents the wireless charging mode; Figure 4 is a comparison of the SOC of the AGV with the optimal scheduling scheme in the wired charging mode and the wireless charging mode, where (a) represents the wired charging mode and (b) represents the wireless charging mode. DETAILED DESCRIPTION
[0019] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. It should be noted that, in the absence of conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.
[0020] Taking the most widely used wired charging AGV as an example, the battery power of the AGV gradually decreases as it completes its task. When it approaches the charging threshold, it needs to stop working and return to a fixed charging station for charging. After charging is completed, it can rejoin the task execution. This charging process makes the AGV unable to operate, the charging process takes time, and the energy consumption for charging is high. Fixed charging locations take up a lot of space, affecting the layout of factory facilities. Since contact charging is required, there is a risk of equipment wear and short circuit. In the battery swap mode, more complex battery swap facilities are usually required, which takes up a certain amount of space. The centralized storage of spare batteries significantly increases the risk of thermal runaway and its destructiveness. In addition, the AGV also needs to return to a fixed battery swap station for battery swapping, which is time-consuming and energy-consuming.
[0021] Furthermore, the interconnected nature of battery capacity constraints, task allocation, and charging planning during AGV cluster scheduling complicates scheduling. Consequently, the existing technology lacks an optimization method that comprehensively considers battery capacity constraints, wireless charging, and AGV cluster collaborative scheduling, making efficient and environmentally friendly AGV cluster scheduling difficult.
[0022] In view of this, the embodiment of the present application provides a wireless charging AGV cluster operation collaborative scheduling method considering battery capacity constraints, in order to solve the above problems. Figure 1 To 4, the scheme of the implementation method of this application is described in detail.
[0023] Specific implementation method 1: refer to Figure 1 and Figure 2 Specifically describing this embodiment, the method for collaborative scheduling of wireless charging AGV cluster operations considering battery capacity constraints described in this embodiment includes: The state of the wireless charging AGV cluster is input into the wireless charging AGV cluster operation collaborative scheduling model to obtain the action of the wireless charging AGV cluster, and the wireless charging AGV cluster is collaboratively scheduled according to the action.
[0024] In one embodiment, the training process of the wireless charging AGV cluster operation collaborative scheduling model includes: Constructing a virtual environment for wireless charging AGV cluster job scheduling, acquiring job demand data and AGV cluster status data in real time from the virtual environment, and then constructing a Markov decision process model. The Markov decision process model includes the state space, action space, and reward function of the wireless charging AGV cluster job scheduling; Extracting the state space and action space of the wireless charging AGV cluster job scheduling in the Markov decision process model as the input and output of the wireless charging AGV cluster job collaborative scheduling model respectively; The wireless charging AGV cluster operation collaborative scheduling model is trained using a deep reinforcement learning method based on the Actor-Critic architecture.
[0025] In one embodiment, the construction of a virtual environment for wireless charging AGV cluster job scheduling includes: A wireless charging AGV cluster scheduling mathematical model is constructed, and a virtual environment for wireless charging AGV cluster operation scheduling is constructed using the wireless charging AGV cluster scheduling mathematical model.
[0026] In one embodiment, the objective function expression of the wireless charging AGV cluster scheduling mathematical model is: , in, For the goal, is the maximum task completion time of AGV.
[0027] In one embodiment, the constraints of the wireless charging AGV cluster scheduling mathematical model include: ; ; ; ; ; Where, Represents a collection of AGVs, is the total number of AGVs, ; Represents a collection of jobs, Indicates the total number of jobs, ; Represents a set of charging times, Indicates the total number of charging times. ; , For homework The duration of the work, The duration of each charge; is a 0-1 decision variable. In the Perform work after charging but ,otherwise ; is a 0-1 decision variable. Carry out the Second charge ,otherwise ; and They are In the Perform work after charging The charge during loading and unloading process, and meet the following requirements: , , , , , , for In the Perform work after charging The previous power, For homework energy consumption, is the battery capacity of the AGV, For wireless charging efficiency, For wireless charging speed, and They are Execute Job Loading and unloading times, It is the minimum operating SOC ratio of AGV.
[0028] In one embodiment, the state space includes: Scheduling workload requirements, energy consumption, and time consumption, AGV load capacity, battery capacity, charging efficiency, and power, as well as wireless charging facility charging capabilities.
[0029] In one embodiment, the action space includes: AGV operation arrangement decision, wireless charging operation decision and initial operation power.
[0030] In one embodiment, the reward function of the Markov decision process model is The expression is as follows: , in, and Execute actions respectively and actions End time.
[0031] To further introduce the solution of the implementation method of this application, each step is described in detail below: Task allocation. Manufacturing plants use AGVs to supply materials needed for production operations to a group of workstations. The need to transport materials to the workstations is called the AGV's task. The goal is to minimize the time required for the AGV to perform the task of supplying materials to all workstations. Production materials are stored in a central warehouse and transported to the workstations by a group of homogeneous AGVs. Generally speaking, the number of AGVs is much smaller than the number of workstations and tasks, so each AGV usually serves multiple workstations. Since each AGV can only carry one package at a time, an AGV serving multiple workstations must perform multiple round trips, including a loaded trip from the warehouse to the workstation and an empty round trip, see Figure 1 Each task has two parameters: time, which is the sum of the round-trip time and the time required for material loading and unloading operations; and energy, which is the energy consumption to perform the task.
[0032] Charging plan. Each AGV has a battery with limited capacity, and its charge level decreases as the task time and energy consumption increase. Therefore, in order to complete the assigned task, the AGV needs to visit a wireless charging station for charging after the battery power drops to a certain threshold. Since this embodiment adopts wireless charging, it is considered that the AGV can be wirelessly charged at the loading position, unloading position and fixed charging position. Figure 1 A set of charging stations are located in the central warehouse and must perform charging operations before the AGV batteries are completely depleted.
[0033] Based on the above analysis of AGV task allocation and charging planning, a collaborative scheduling method for wireless charging AGV cluster operations considering battery capacity constraints is proposed, including: Step 1: Construct a mathematical model for wireless charging AGV cluster scheduling.
[0034] The model aims to minimize the maximum completion time of wireless charging AGV: (1), in, For the goal, is the maximum task completion time of AGV.
[0035] Model constraints include: (1) It must be greater than or equal to the sum of the cumulative operating time and the cumulative charging operation time of each AGV, expressed as: (2), in, ; Indicates homework duration of the operation; Indicates the duration of each charge; Represents a collection of AGVs, Indicates the total number of AGVs, ; Represents a collection of jobs, Indicates the total number of jobs, ; Represents a set of charging times, Indicates the total number of charging times. ; is a 0-1 decision variable, if In the Perform work after charging If it is 1, otherwise it is 0, which is expressed as ; is a 0-1 decision variable, if Carry out the If it is charged once, it is 1, otherwise it is 0, which is expressed as .
[0036] (2) Homework It is completed by and only by a certain AGV after a certain charge, expressed as: (3).
[0037] (3) Energy balance constraint, which means the sum of the initial power of the AGV when performing the task, the charge during loading, and the charge during unloading, minus the energy consumption of the task, equals the power of the AGV after performing the task, which is expressed as: (4), in, (5), (6), and Respectively In the After charging, perform the work The amount of charge during loading and unloading, express In the Perform work after charging The previous power, Indicates homework energy consumption, Indicates the battery capacity of the AGV, Indicates the wireless charging efficiency, Indicates wireless charging speed, and Respectively Execute Job Loading and unloading times.
[0038] Equations (7) to (9) describe The battery level before operation 1, the battery level during loading, and the battery level during unloading: (7), (8), (9).
[0039] Formula (10) represents Initially fully charged: (10).
[0040] (4) Symmetry breaking constraints are used to reduce the decision variable space, which can be expressed as: (11).
[0041] (5) The fluctuation range constraint of AGV power is expressed as: (13), in, It is the minimum operating SOC ratio of AGV.
[0042] For the convenience of optimization and solution, equation (5) is linearized into equations (14)-(17): (14), (15), (16), (17), in, is a 0-1 variable. hour, ,when hour, .
[0043] At the same time, equation (6) is linearized into equations (18)-(21): (18), (19), (20), (twenty one), in, is a 0-1 variable. hour, ,when hour, .
[0044] Finally, with formula (1) as the objective function and formulas (2), (3), (4) and (7)-(21) as constraints, a mathematical model for wireless charging AGV cluster scheduling is constructed.
[0045] Step 2: Design a wireless charging AGV cluster scheduling model optimization algorithm based on deep reinforcement learning.
[0046] 1) Convert the wireless charging AGV cluster scheduling mathematical model into an equivalent Markov decision process.
[0047] First, a mathematical model for wireless charging AGV cluster scheduling was used to construct a virtual environment for wireless charging AGV cluster operation scheduling in an intelligent manufacturing factory. The virtual environment incorporated the intelligent manufacturing factory layout, wireless charging AGV characteristics, and the operation demand characteristics of each workstation. The intelligent manufacturing factory layout information included: operating tracks, workstation locations, and wireless charging facility locations. Their distribution information can be predicted based on historical operation demand data. Wireless charging AGV characteristics included: quantity, load capacity, charging efficiency, and battery capacity. Operation demand characteristics included: load demand, energy consumption, and time consumption.
[0048] Then, real-time data on job requirements and the real-time status of the AGV cluster are obtained from the virtual production environment. Combined with the mathematical model for wireless charging AGV cluster scheduling established in step 1, an equivalent Markov decision process model is constructed. The Markov decision process model includes a state space, an action space, a reward function, and a return function.
[0049] 2) Construct state space.
[0050] Extract external environment characteristics based on task characteristics and mathematical model mechanisms, such as scheduling workload requirements , energy consumption and time-consuming AGV load capacity , battery capacity , charging efficiency and power , wireless charging facilities charging capacity , building a state space that can fully and accurately represent the environment .
[0051] 3) Construct action space.
[0052] Combined with the decision-making mechanism of the wireless charging AGV cluster scheduling mathematical model in step 1, an action matrix including AGV operation scheduling and charging planning decision is proposed, such as AGV operation arrangement decision , wireless charging operation decision and initial power of operation , forming the action space of AGV .
[0053] 4) Construct a reward function.
[0054] Combined with the optimization goal of the wireless charging AGV cluster scheduling mathematical model in step 1, the action selected in the previous step The execution end time minus the action selected in the current step The execution end time is the reward function : (twenty two).
[0055] 5) Use deep reinforcement learning based on the Actor-Critic architecture for optimization training to obtain the final wireless charging AGV cluster operation collaborative scheduling model.
[0056] The training method uses the Proximal Policy Optimization (PPO) method under the Actor-Critic architecture, namely ACPPO. Specifically: definition Indicates Actor network strategy function with parameters; Indicates Critic network value function with parameters; express The value function under the policy; represents the advantage function; is the ratio of the probability of the new and old strategies; represents the strategy parameters before updating; represents the parameters that characterize the decision-making process network; Represents the parameters of the estimated reward RNN (Recurrent Neural Network); Indicates that Restricted to The function of fluctuations between is a parameter between [0,1]; and Represent the environment state and action respectively; Represents computational expectation.
[0057] In order to avoid sudden changes in the strategy loss function update, this embodiment uses the clipping loss function to update the Actor network strategy. Restricted to a certain range, the calculation method is shown in Equations (23) and (24). To minimize the error between the target state value function and its predicted value As the goal, the critic network is trained to estimate the state value function and effectively evaluate the expected cumulative return. The calculation method is defined in Equation (25).
[0058] (twenty three), (twenty four), (25).
[0059] Based on the above analysis, the deep reinforcement learning optimization process is formed: S1: Initialize the parameters of each neural network, including: Actor network strategy function with parameters ,by Critic network value function with parameter ,by Characterize the network for its parameter decision process and RNN intrinsic reward network with parameters .
[0060] S2: Embed the decision flow (action space) to obtain the AGV cluster job scheduling solution based on Strategy (with The Actor network strategy function with parameters is used to sample actions and calculate the reward network respectively. , the ratio of the probability of the new and old strategies And the advantage function , and save the job scheduling collaborative trajectory to form a job scheduling trajectory experience pool.
[0061] S3: Sampling from the experience pool for training parameters, and updating the Actor network strategy function in sequence , Critic network value function , Decision Flow Representation Network and intrinsic reward networks .
[0062] S4: Update strategy parameters , and clear the experience pool.
[0063] S5: After repeated iterative training of the above steps, until the maximum number of iterations is reached, the optimal decision-making scheme for the job allocation and charging planning of the wireless charging AGV cluster is obtained. The process is shown in Figure 2 .
[0064] Referring to the battery-constrained AGV operation scheduling experiment designed by Boccia et al. (2023), a wireless charging AGV operation scheduling test experiment was designed. In the experiment, five homogeneous wireless charging AGVs were set, each with a battery capacity of 10. Fifty operations were set, and the time and energy consumption of each operation were randomly generated by a normal distribution function, where the mean of the normal distribution function was 10 and the average energy consumption was 2, and the standard deviation was half of the average time and average energy consumption, respectively. In addition, the wireless charging efficiency was set to 92% and the wireless charging power was set to 0.5. The loading and unloading times of each operation were randomly generated by a normal distribution function, where the mean of the normal distribution function was 1 and the average unloading time was 2, respectively, and the standard deviation was half of the average loading time and average unloading time, respectively.
[0065] Figure 3 compares the optimal makespan for AGV scheduling under a wired charging scheme (Boccia et al., 2023) and the wireless charging scheme of this application. Due to the need for wired charging at a fixed charging station (shown as the yellow charging operation C1 in Figure 3(a)), the scheduling efficiency of AGVs under the wired charging scheme is significantly lower than that of the wireless charging scheme. The results show that the optimal makespan for AGV scheduling under the wired charging scheme is 168.5, while the optimal makespan for AGV scheduling under the wireless charging scheme is 108.5, which is 35.6% lower than the former.
[0066] Figure 4 shows the changes in the state of charge (SOC) of each AGV under the optimal AGV operation scheduling scheme in wired charging mode and wireless charging mode. It can be seen that due to the ability to dynamically charge during operation, the AGV in wireless charging mode can always maintain a high SOC level. However, in wired charging mode, the SOC of the AGV gradually decreases as the operation progresses, and eventually has to return to a fixed charging station for charging, which reduces operation efficiency.
[0067] Specific embodiment 2: The wireless charging AGV cluster operation collaborative scheduling device considering battery capacity constraints described in this embodiment includes a processor and a memory, and the memory stores at least one instruction. The at least one instruction is loaded and executed by the processor to implement the wireless charging AGV cluster operation collaborative scheduling method considering battery capacity constraints as described in specific embodiment 1.
[0068] Specific embodiment three: A computer storage medium stores at least one instruction, and the at least one instruction is loaded and executed by a processor to implement the wireless charging AGV cluster operation collaborative scheduling method considering battery capacity constraints as described in specific embodiment one.
[0069] Although the present application is described herein with reference to specific embodiments, it should be understood that these embodiments are merely illustrative of the principles and applications of the present application. It should therefore be understood that many modifications may be made to the illustrative embodiments, and that other arrangements may be devised, without departing from the spirit and scope of the present application as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in ways other than those described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in other described embodiments.
Claims
1. A wireless charging AGV cluster operation collaborative scheduling method considering battery capacity constraints is characterized by: include: The state of the wireless charging AGV cluster is input into the wireless charging AGV cluster operation collaborative scheduling model to obtain the action of the wireless charging AGV cluster, and the wireless charging AGV cluster is collaboratively scheduled according to the action.
2. The method for coordinated scheduling of wireless charging AGV cluster operations considering battery capacity constraints according to claim 1 is characterized in that: The training process of the wireless charging AGV cluster operation collaborative scheduling model includes: Constructing a virtual environment for wireless charging AGV cluster job scheduling, acquiring job demand data and AGV cluster status data in real time from the virtual environment, and then constructing a Markov decision process model. The Markov decision process model includes the state space, action space, and reward function of the wireless charging AGV cluster job scheduling; Extracting the state space and action space of the wireless charging AGV cluster job scheduling in the Markov decision process model as the input and output of the wireless charging AGV cluster job collaborative scheduling model respectively; The wireless charging AGV cluster operation collaborative scheduling model is trained using a deep reinforcement learning method based on the Actor-Critic architecture.
3. The method for coordinated scheduling of wireless charging AGV cluster operations considering battery capacity constraints according to claim 2 is characterized in that: The virtual environment for wireless charging AGV cluster job scheduling is constructed, including: A wireless charging AGV cluster scheduling mathematical model is constructed, and a virtual environment for wireless charging AGV cluster operation scheduling is constructed using the wireless charging AGV cluster scheduling mathematical model.
4. The method for coordinated scheduling of wireless charging AGV cluster operations considering battery capacity constraints according to claim 3 is characterized in that: The objective function expression of the wireless charging AGV cluster scheduling mathematical model is: ， in, For the goal, is the maximum task completion time of AGV.
5. The method for coordinated scheduling of wireless charging AGV cluster operations considering battery capacity constraints according to claim 4 is characterized in that: The constraints of the wireless charging AGV cluster scheduling mathematical model include: ； ； ； ； ； Where, Represents a collection of AGVs, is the total number of AGVs, ; Represents a collection of jobs, Indicates the total number of jobs, ; Represents a set of charging times, Indicates the total number of charging times. ; , For homework The duration of the work, The duration of each charge; is a 0-1 decision variable. In the Perform work after charging but ,otherwise ; is a 0-1 decision variable. Carry out the Second charge ,otherwise ; and They are In the Perform work after charging The charge during loading and unloading process, and meet the following requirements: ， ， ， ， ， ， for In the Perform work after charging The previous power, For homework energy consumption, is the battery capacity of the AGV, For wireless charging efficiency, For wireless charging speed, and They are Execute Job Loading and unloading times, It is the minimum operating SOC ratio of AGV.
6. The method for coordinated scheduling of wireless charging AGV cluster operations considering battery capacity constraints according to claim 2 is characterized in that: The state space includes: Scheduling workload requirements, energy consumption, and time consumption, AGV load capacity, battery capacity, charging efficiency, and power, as well as wireless charging facility charging capabilities.
7. The method for coordinated scheduling of wireless charging AGV cluster operations considering battery capacity constraints according to claim 6 is characterized in that: The action space includes: AGV operation arrangement decision, wireless charging operation decision and initial operation power.
8. The method for coordinated scheduling of wireless charging AGV cluster operations considering battery capacity constraints according to claim 7 is characterized in that: The reward function of the Markov decision process model The expression is as follows: ， in, and Execute actions respectively and actions End time.
9. A wireless charging AGV cluster operation collaborative scheduling device considering battery capacity constraints, characterized by: The wireless charging AGV cluster operation collaborative scheduling device considering battery capacity constraints includes a processor and a memory, wherein the memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the wireless charging AGV cluster operation collaborative scheduling method considering battery capacity constraints as described in one of claims 1 to 8.
10. A computer storage medium, characterized in that The computer storage medium stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the wireless charging AGV cluster operation collaborative scheduling method considering battery capacity constraints as described in any one of claims 1 to 8.