Multi-AGV cooperative scheduling method and device for matrix workshop

Abstract

The invention discloses a multi-AGV cooperative scheduling method and device for a matrix workshop, and the method comprises the steps: obtaining the machine layout information and AGV parameters of a grid matrix workshop, and generating an initial path of an AGV according to an A * algorithm; when it is detected that the current AGV and other AGVs are located at the same position at the same time, the initial path is re-planned with the purpose of minimizing the total workshop energy consumption, and a re-planned path is obtained; wherein the total energy consumption consists of the total energy consumption of the machine and the total energy consumption of the AGV; according to the method, space-time overlapping conflict detection is carried out on the initial path, when the current AGV and other AGVs are located at the same position at the same time, the other AGVs are also used as obstacles, then path re-planning is carried out with the minimum workshop total energy consumption as the constraint, and therefore mutual influence between the AGVs can be avoided, and the path operation efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of AGV path planning, and in particular relates to a multi-AGV collaborative scheduling method and device for a matrix workshop. Background Art

[0002] Automated Guided Vehicles (AGVs) are automated transport vehicles used to autonomously perform material transport tasks within manufacturing workshops or warehouses. They are often used to replace manual handling or transport vehicles to improve efficiency, reduce costs and enhance operational safety.

[0003] Path planning is a key technology for intelligent AGV operation. Common path planning methods include the A* algorithm and the Dijkstra algorithm. In a multi-AGV system, path planning requires coordination and conflict resolution between vehicles. However, traditional algorithms rely heavily on static environments and cannot handle dynamic obstacles (such as the AGVs themselves) in real time, resulting in frequent path failures. Summary of the Invention

[0004] The purpose of the present invention is to provide a multi-AGV collaborative scheduling method and device for a matrix workshop, which generates AGV paths by minimizing workshop energy consumption as a constraint condition, thereby solving the problem of multi-AGV path conflict.

[0005] The present invention adopts the following technical solution: a multi-AGV collaborative scheduling method for a matrix workshop, comprising the following steps:

[0006] Obtain the machine layout information and AGV parameters of the grid matrix workshop, and generate the initial AGV path based on the A* algorithm;

[0007] When it is detected that the current AGV is at the same location as other AGVs at the same time, the initial path is replanned with the goal of minimizing the total energy consumption of the workshop to obtain a replanned path; the total energy consumption is composed of the total energy consumption of the machine and the total energy consumption of the AGV.

[0008] Furthermore, the total energy consumption of the machine includes operating energy consumption, no-load energy consumption, and start-stop energy consumption;

[0009] Operating energy consumption is calculate;

[0010] No-load energy consumption is determined by calculate;

[0011] Start-stop energy consumption is determined by calculate;

[0012] in, represents the operating energy consumption of machine n, represents the operating power of machine n, T kinrepresents the time spent by the i-th process of the k-th AGV on machine n, N represents the number of machines, O represents the number of processes, represents the no-load energy consumption of machine n, represents the no-load power consumption of machine n, represents the idle time of machine n, represents the start-stop energy consumption of machine n, represents the start-stop power of machine n, represents the start and stop time of machine n, Indicates the number of starts and stops of machine n.

[0013] Furthermore, the corresponding machine is selected for each process based on the machine priority;

[0014] The machine priority is calculated as follows:

[0015]

[0016] Among them, Priority(n) represents the priority of machine n, L n Indicates the current number of tasks for machine n, is the waiting time of machine n, is the current time of the kth AGV, d ij represents the distance from the current process i to the next process j, and α, β, and γ are different weight factors.

[0017] Furthermore, during the replanning process, when the AGV is in the i-th process and the battery level is less than the charging threshold, the processing time of the k-th AGV on the machine n of the i-th process is recalculated;

[0018] The recalculation method is:

[0019]

[0020] Among them, T kin,new It represents the processing time of the kth AGV on the machine n of the i-th process after recalculation, represents the start and stop time of the kth AGV, T kch represents the battery replacement time of the kth AGV, d ijk represents the distance of the kth AGV from process i to the next process j, v min represents the minimum speed of the kth AGV.

[0021] Furthermore, the energy consumption of each AGV includes movement energy consumption, no-load energy consumption, and start-stop energy consumption;

[0022] The mobile energy consumption is calculated based on the mobile power. The mobile power calculation method is:

[0023]

[0024] in, represents the moving power of the kth AGV, soc k Represents the battery capacity of the kth AGV.

[0025] Furthermore, the battery power calculation method of AGV is:

[0026]

[0027] Among them, soc k (t+1) represents the battery capacity of the kth AGV at time t+1, soc k (t) represents the battery capacity of the kth AGV at time t, η represents the efficiency factor of the battery, X kin represents the processing status of the kth AGV on the machine n in the i-th process, v(soc k ) indicates that the kth AGV is in a state where the battery power is soc k The moving speed is , and D represents the maximum distance that can be traveled per unit battery power.

[0028] Furthermore, the kth AGV is at a battery level of soc k The calculation method of moving speed is:

[0029]

[0030] in, Indicates that the kth AGV is in a state where the battery power is 0.20≤soc k Movement speed when ≤0.35, v max represents the maximum speed of the kth AGV.

[0031] Another technical solution of the present invention: a multi-AGV collaborative scheduling method for a matrix workshop, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.

[0032] The beneficial effect of the present invention is that the present invention performs spatiotemporal overlap conflict detection on the initial path. When the current AGV and other AGVs are at the same position at the same time, the other AGVs are also regarded as obstacles, and then the path is replanned with minimizing the total energy consumption of the workshop as the constraint, thereby avoiding mutual influence between AGVs and improving path operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a flow chart of a multi-AGV collaborative scheduling method for a matrix workshop according to an embodiment of the present invention;

[0034] Figure 2 This is a schematic diagram of process connection in an embodiment of the present invention;

[0035] Figure 3 This is a schematic plan view of a matrix workshop in an embodiment of the present invention;

[0036] Figure 4 A schematic diagram comparing energy consumption between the method of the present invention and the existing method in the embodiment for verifying the present invention;

[0037] Figure 5 This is a schematic diagram comparing energy consumption when verifying whether a retry mechanism is included in an embodiment of the present invention;

[0038] Figure 6 This is a schematic diagram of energy consumption comparison for verifying whether the embodiment of the present invention includes a dynamic speed strategy. DETAILED DESCRIPTION

[0039] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0040] In a multi-AGV environment, traditional algorithms are prone to falling into local optima, where multiple AGVs repeatedly select paths, leading to congestion and waiting. Furthermore, when multiple AGVs share limited resources, spatiotemporal overlap frequently occurs, necessitating real-time detection and dynamic adjustment. Furthermore, existing research generally focuses on optimizing a single objective (path length or task completion time), ignoring the synergistic effects of energy consumption and machine load. Therefore, dynamic scheduling and multi-objective optimization are essential challenges for matrix workshops.

[0041] Traditional AGV path planning methods have the following defects:

[0042] 1) Insufficient dynamic adaptability: Unable to handle dynamic obstacles caused by AGV movement in real time. Manual intervention is required after path failure, resulting in low production efficiency.

[0043] 2) Causes local optimality: Using a single heuristic rule (distance priority) causes multiple AGVs to focus on neighboring machines, leading to unbalanced machine loads and energy consumption during idle time, resulting in increased energy consumption and costs.

[0044] 3) Lack of energy consumption optimization: AGVs use fixed power consumption and speed parameters, and their battery replacement strategy is simple, which increases AGV energy consumption.

[0045] 4) Single optimization objective: only the path distance or task completion time is optimized.

[0046] Therefore, the present invention first defines the initial machine layout and AGV parameters, introduces relevant information into the path planning method (a scheduling optimization algorithm that integrates a dynamic retry mechanism and multi-factor rules), and uses the retry mechanism to replan the path in the event of a deadlock position.

[0047] Specifically, whenever a path deadlock is detected, the system randomly deletes the most recently occupied nodes through a retry mechanism and re-plans the path. Furthermore, for node updates, the present invention utilizes a spatiotemporal overlap conflict detection mechanism. This mechanism detects the path planning tables of other AGVs in real time, determines the current position of each AGV at the target time, and dynamically updates the most recently occupied nodes. This process requires the combined effect of a retry mechanism and a conflict detection mechanism to achieve efficient and accurate path planning, effectively reducing path planning time and avoiding path deadlock in dynamic environments.

[0048] The present invention then dynamically adjusts the AGV's speed and energy consumption to minimize the frequency of AGV visits to battery swap stations, thereby reducing unnecessary energy waste. Finally, the present invention comprehensively considers three factors to select the appropriate machine for each scenario, rather than simply pursuing a local optimal solution. This prevents multiple AGVs from waiting at the same machine, thereby avoiding the additional energy consumption caused by idle workshops.

[0049] Therefore, the present invention discloses a multi-AGV collaborative scheduling method for a matrix workshop, such as Figure 1 As shown, the method includes the following steps: obtaining the machine layout information and AGV parameters of the grid matrix workshop, and generating the initial path of the AGV according to the A* algorithm; when it is detected that the current AGV and other AGVs are at the same position at the same time, the initial path is replanned with the goal of minimizing the total energy consumption of the workshop to obtain a replanned path; wherein the total energy consumption is composed of the total energy consumption of the machine and the total energy consumption of the AGV.

[0050] The present invention performs spatiotemporal overlap conflict detection on the initial path. When the current AGV and other AGVs are at the same location at the same time, the other AGVs are also considered as obstacles. Then, the path is replanned with minimizing the total energy consumption of the workshop as the constraint, thereby avoiding mutual influence between AGVs and improving path operation efficiency.

[0051] In one embodiment, it is assumed that there are 5 AGVs, numbered 1-5, and the AGV driving speed v is designed to be 2m / s. The AGV is only allowed to move in four directions: front, back, left, and right, that is, it can only move through the edges in the matrix workshop. The start and stop time is 1s, the start and stop energy consumption is approximately 0.2% of the total power, and the operating energy consumption is 0.1% of the total power consumption per unit time.

[0052] like Figure 2As shown in the figure, the AGV takes 5 seconds to pick up copper wires from the machine that places the copper sheets and the machine that places the copper sheets. The copper sheets and wires are picked up randomly. Starting from 0 seconds, a total of 9 processes are executed (10 processes if battery replacement is required).

[0053] The first process has two machines, two machines for placing copper wire and two machines for placing copper sheets. The start and stop time of both machines is 4s, and the start and stop energy consumption is 1.5 times the operating energy consumption. The operating energy consumption is 0.9kw, and the placement time is 4s; the second process (energy consumption is the same as the machine in the first step) goes to the empty box machine to place the empty box. The machine start and stop time is 4s, the placement time is 5s, and there are two empty box machines; the third process (energy consumption is the same as the machine in the first step) is that the AGV goes to the copper sheet machine to pick up the copper sheets, with a total of two start and stop times of 4s and a picking time of 6s; the fourth process (energy consumption is the same as the machine in the first step) goes to the copper wire machine to pick up the copper wire, with a total of two start and stop times The stop time is 4s, and the picking time is 5s; the fifth process goes to the welding machine for welding, a total of four machines have a start-stop time of 2s, the start-stop energy consumption is 1.5 times the operating energy consumption, the operating energy consumption is 2kw, and the welding time is 3s; the sixth process (energy consumption is the same as the machine in the first step) to inspect the welded finished products, a total of two inspection machines have a start-stop time of 2s, and the inspection time is 6s; the seventh process puts one into the finished product warehouse, there is no machine start-stop time, only the AGV start-stop is involved, and the placement time is 4s; the eighth process returns to the empty box area to pick up the empty box, which is the empty box machine in the second process. The start-stop time is the same as the second step, and the picking time is 5s; the ninth process returns to the starting warehouse.

[0054] In addition, there is an additional process (the tenth process), which can occur between any two processes. When the AGV battery is less than 10%, it goes to the exchange station to exchange the battery. Except for the starting warehouse, which is numbered 0, other machines are randomly assigned numbers from 1 to 17. Figure 3 As shown, at this time, each machine is placed in a 5×5 matrix, excluding the finished product warehouse and the initial warehouse. The initial warehouse is fixed at the coordinate of (0, 0), and the finished product warehouse is fixed at the coordinate of (25, 25). The two battery swap stations are placed at the (0, 25) and (25, 0) positions of the matrix respectively. What is important is that the machine for taking copper wire must be generated at the adjacent points on the four sides of the machine for placing copper wire (one to one). The same is true for copper sheets. There are 2 machines for placing copper wire, 2 machines for placing copper sheets, 2 machines for placing empty boxes, 2 machines for taking copper wire, 2 machines for taking copper sheets, 4 machines for welding, and 2 machines for testing. These machines need to be randomly generated at the intersection of the middle 4×4 matrix, and the interval between each point in the matrix is ​​5m.

[0055] As a more specific embodiment, the layout coordinates of each machine are shown in Table 1.

[0056] Table 1

[0057]

[0058] First, initialize the machine, establish a two-dimensional plane area Z, and set the set of machines to set N. The attributes of each machine n∈N are defined as, where W and H are set to 25m:

[0059] Z=[0,W]×[0,H] (1)

[0060] n=(machine id ,type,(x n ,y n )) (2)

[0061] Among them, machine id Indicates the machine number, type∈{testing, welding, empty_box, ...} indicates its type, and the items in the brackets represent the testing process, welding process, and empty box removal process, respectively. (x n ,y n ) represent the horizontal and vertical coordinates of machine n, which are distributed in a 25m×25m matrix workshop at intervals of 5m, and the main process machines are allocated at the central 20m×20m matrix vertex.

[0062] The two-dimensional plane area Z of the matrix workshop is divided into grids with a grid side length of Δ=5m, where the grid node coordinates are: q a，b =(a·Δ,b·Δ), a, b∈(0, 1…, 5). Finally, the actual coordinates are converted into grid coordinates

[0063] Next, we design the AGV cluster. First, we define the state of each AGV. Let the set of AGVs be K, and the state of each AGV be:

[0064] k=(AGV_id,position(t),soc k (t),path_schedule(t)) (3)

[0065] Among them, position(t) is the coordinate of the kth AGV at time t, soc k (t)∈[0, 1] is the power of AGV at time t, and path_schedule(t) is the combined information including time nodes and path nodes.

[0066] After obtaining the above information, the path generation method is used to generate the initial path, and then the time-space overlap conflict detection is performed.

[0067] In an embodiment of the present invention, AGV path conflict detection uses spatiotemporal overlap detection, which determines the spatiotemporal overlap of the AGV at the current time node based on the initial path. For example, each time the AGV to be detected plans a path (i.e., at each moment), a conflict detection is performed to determine whether the current AGV and other AGVs are at the same location at the same time. If so, a conflict is determined.

[0068] As shown in Table 2, which is an embodiment of a conflict, it can be seen from the table that AGV1 and AGV2 are both located at the position (5, 5) at t=5s.

[0069] Table 2

[0070] AGV Time window Path node sequence Conflict Point AGV1 [0,10]s (0,5)-(5,5)-(5,10) t=5s is located at (5,5) AGV2 [0,10]s (10,5)-(5,5)-(5,0) t=5s is located at (5,5)

[0071] Then, it can be expressed as follows using a mathematical model:

[0072]

[0073] That is, there is a node q=(5,5), at t=5s and Indicates the state of AGV1 at node (5,5) at t=5s. If yes, it is equal to 1, otherwise it is equal to 0. Indicates the state of AGV2 at node (5,5) at t=5s. If yes, it is equal to 1, otherwise it is equal to 0.

[0074] However, the probability of this conflict point occurring will be greatly reduced under the constraints of different AGV power levels and speeds. The mathematical process used in this part is as follows, and the time-space occupied path segment is represented as follows:

[0075]

[0076] Among them, P k represents the planned path of the kth AGV. In this embodiment, the planned path is divided into the combination of e nodes, q e Represents path P k The e-th node, Indicates the starting time of the planned path, Indicates the end time of the planned path, d ijk represents the distance of the kth AGV from process i to the next process j, v(soc k ) indicates that k AGVs have a battery capacity of soc k The speed of time.

[0077] For the path P of the kth AGV kFor each node in the AGV, calculate its spatiotemporal occupancy information and check whether there are other AGVs at the same time t e Occupy the node q e Through this part, it is possible to dynamically update the AGV occupied nodes and judge in real time whether there will be conflicts between AGVs.

[0078] When a conflict is detected, path replanning is required. In the present invention, path replanning is performed with minimizing the total energy consumption of the workshop as a constraint.

[0079] Specifically, the total energy consumption of the machine includes operating energy consumption, no-load energy consumption and start-stop energy consumption; the operating energy consumption is composed of Calculation; no-load energy consumption is calculated by Calculation; start-stop energy consumption is calculated by Calculate; where, represents the operating energy consumption of machine n, represents the operating power of machine n, T kin represents the time spent by the i-th process of the k-th AGV on machine n, N represents the number of machines, O represents the number of processes, represents the no-load energy consumption of machine n, represents the no-load power consumption of machine n, represents the idle time of machine n, represents the start-stop energy consumption of machine n, represents the start-stop power of machine n, represents the start and stop time of machine n, Indicates the number of starts and stops of machine n.

[0080] Similarly, the energy consumption of each AGV includes moving energy consumption, no-load energy consumption, and start-stop energy consumption. In order to further reduce energy consumption, the present invention reconstructs the AGV speed model. The change in the AGV's power level plays an important role in the AGV's decision-making. First, the priority constraint is that the AGV's current battery power must not exceed the AGV's maximum battery power:

[0081] soc k ≤soc kmax (7)

[0082] Among them, soc k Indicates the battery capacity of the kth AGV, soc kmax represents the maximum battery capacity of the k-th AGV.

[0083] The battery power of the kth AGV reaches the speed change threshold range (i.e. 0.20≤soc k ≤0.35), the AGV will slow down as the battery capacity decreases and reach a lower speed limit to ensure subsequent operation. kis the current battery capacity of the k-th AGV, is the moving speed corresponding to when the AGV power level is in the speed change threshold range, and the kth AGV is at a battery power level of soc k The calculation method of moving speed is:

[0084]

[0085] in, Indicates that the kth AGV is in a state where the battery power is 0.20≤soc k Movement speed when ≤0.35, v max represents the maximum speed of the kth AGV.

[0086] In addition, there are:

[0087]

[0088] In this formula, when the battery charge of the kth AGV is less than or equal to 0.2 (assuming it is 1 when the battery is fully charged), the speed is 1 (i.e., the 2 m / s mentioned above); when the battery charge of the kth AGV is greater than or equal to 0.35, the speed is 2 (i.e., 2 m / s × 2 = 4 m / s).

[0089] This part can effectively reduce the energy consumption of AGV, and ensure that there is enough power to go to the remaining machines with unfinished processes before the power is exhausted, so as to prevent long-term full-speed operation from accelerating battery consumption and causing unnecessary energy waste by multiple trips to battery swap stations.

[0090] Regarding the dynamic energy consumption of AGV, the battery loss of AGV itself is calculated first. AGV itself needs to calculate the battery consumption according to the gradually changing battery power and update the SOC of the next time. k , where η is the efficiency factor of the battery and D is the maximum distance that can be traveled per unit battery power. The battery power calculation method for the AGV is:

[0091]

[0092] Among them, soc k (t+1) represents the battery capacity of the kth AGV at time t+1, soc k (t) represents the battery capacity of the kth AGV at time t, X kin represents the processing status of the kth AGV on the machine n in the i-th process, v(soc k ) indicates that the kth AGV is in a state where the battery power is soc k The moving speed when .

[0093] Since AGV uses different energy consumption under different battery power conditions, (unit: watt) is the mobile energy consumption power of the kth AGV under the current power. The mobile energy consumption is calculated based on the mobile power. The mobile power calculation method is:

[0094]

[0095] in, represents the moving power of the kth AGV, soc k Represents the battery capacity of the kth AGV.

[0096] Finally, the energy consumption parameters of AGV are defined. is the moving energy consumption of the kth AGV, is the start-stop energy consumption of AGVK, and the energy consumption model is:

[0097]

[0098] is the no-load energy consumption of the k-th AGV, The k-th AGV no-load power consumption is:

[0099]

[0100] Among them, K is the number of AGVs, O is the number of processes, N is the number of machines, T kin represents the time that the kth AGV spends on machine n in process i.

[0101] The number of times the kth AGV triggers the task is accumulated by judging whether the AGV reaches machine n. is the start and stop time of the kth AGV.

[0102]

[0103] X kin Represents the state of the kth AGV, then:

[0104]

[0105] It indicates the number of times the k-th AGV triggered a task last time. That is, the number of times the task was triggered increases by 1 each time a task is triggered.

[0106] is the start and stop time of the kth AGV, is the start-stop energy consumption of the kth AGV:

[0107]

[0108] Therefore, the total energy consumption of the kth AGV is for:

[0109]

[0110] In addition, when the AGV enters the charging state, it cannot continue to perform other tasks. During the replanning process, when the AGV is in the i-th process and the battery power is less than the charging threshold, the processing time of the k-th AGV on the machine n in the i-th process is recalculated. The recalculation method is:

[0111]

[0112] Among them, T kin，new It represents the processing time of the kth AGV on the machine n of the i-th process after recalculation, represents the start and stop time of the kth AGV, T kch represents the battery replacement time of the kth AGV, d ijk represents the distance of the kth AGV from process i to the next process j, v min represents the minimum speed of the kth AGV.

[0113] The charging process affects the start time of subsequent tasks, and the updated completion time incorporates the charging process time into the process time without the need for rescheduling:

[0114] S ki，new =S ki +Y kin T kin，new ≤P k(i+1) (19)

[0115] Among them, S ki，new S represents the processing completion time of the kth AGV after adding the battery replacement process in the i-th process, ki represents the processing completion time of the kth AGV in the i-th process (excluding the battery replacement process), Y kin Indicates the processing status of the k-th AGV on the machine n of the i-th process, which is 1 if it is processing, otherwise 0.

[0116] In this case, if the battery power of the AGV reaches the charging threshold during the route, it will go to the battery swap station first, and then go to the target location of the original route after the battery is replaced.

[0117] In one embodiment, a corresponding machine is selected for each process according to machine priority;

[0118] The machine priority is calculated as follows:

[0119]

[0120] Among them, Priority(n) represents the priority of machine n, L n Indicates the current number of tasks for machine n, is the waiting time for machine n to complete, is the current time of the kth AGV, d ij Represents the distance from the current process i to the next process j, α, β and γ are different weight factors. n The machine load factor is used to penalize machines with high loads and give priority to machines with low loads. γ·d is the machine availability factor, which is used to penalize machines that require a long waiting time and give priority to machines with a long available time. ij is the distance factor, which is used to penalize machines that are farther away and give priority to machines that are closer.

[0121] According to the characteristics of this embodiment and method, the weight of the load factor is emphasized to prevent excessive pursuit of the selection of the nearest machine. The objective function is to give priority to the machine with the minimum comprehensive value:

[0122]

[0123] On this basis, it is necessary to constrain the process time and task sequence. During the processing, it is necessary to ensure that one processing equipment can only process a certain process of the AGV:

[0124]

[0125]

[0126] X kjn Represents the state of the k-th AGV on machine n' in the j-th process.

[0127] The AGV can only process on this device and leave after completing the task:

[0128]

[0129] Ensure that the AGV must leave the warehouse to transport the goods and eventually return to the warehouse:

[0130]

[0131] And ensure that the time for the current workpiece processing shall not exceed the time limit, and must be carried out in accordance with the order of the processes, and cannot be directly processed to the next step. ki is the starting time of the i-th process of the k-th AGV, S ki The start and end time of the i-th process of the k-th AGV:

[0132] Iki +T kin *X kin ≤S ki (25)

[0133] S ki ≤I k(i+1) (26)

[0134] T kin It is the time spent by the k-th AGV on the machine n to perform the i-th process.

[0135] The time required to complete all workpiece processing cannot exceed the maximum processing time, and the processing time and completion time must be greater than or equal to 0:

[0136] I ki ≥0 (27)

[0137] S ki ≥0 (28)

[0138] It is determined that a machine can only process one process at a time, and the machine will not process other workpieces except the workpiece on the current AGV. H is a sufficiently large integer:

[0139] I ki +T kin ≤I cd +H(1-Y kidcn ) (29)

[0140] S ki ≤I k(i+1) +H(1-Y k(i+1)dcn ) (30)

[0141] Among them, when X kin When Y is equal to 1, kidcn =1.

[0142] The workpiece is processed from the beginning to the end without interruption, that is, S ki -P ki =T kin The following represents the distance the AGV travels from process i to process j, and the time it takes to travel that distance:

[0143] d ijk =|x in +x jn` |+|y in +y jn` | (31)

[0144]

[0145] If the kth AGV starts from the warehouse and goes to the first process i, the time required to reach the task is this formula, where V = {0, 1, 2, ..., n}, 0 represents the warehouse:

[0146]

[0147] represents the arrival time from the warehouse to the first process, t 0i Indicates the time used in the process, and T0 indicates the departure time.

[0148] If the kth AGV starts from the i-th process and goes to the j-th process, the time required to reach the task is:

[0149]

[0150] Among them, t ij It represents the time it takes for the AGV to move from the current process i to the next process j.

[0151] The total distance of the path traveled by the kth AGV can be expressed as:

[0152]

[0153] In order to verify the effectiveness of the present invention in reducing workshop energy consumption, this case conducted multiple task experiments and analyzed the experimental data. The total energy consumption data are shown in Tables 3, 4 and 5.

[0154] In this experiment, five AGVs were used to perform five tasks simultaneously and continuously. Each group used its own algorithm to schedule the matrix workshop and collect energy data of each AGV and machine, and calculated the final total energy consumption, which was then compared with the method of the present invention (DREAMS).

[0155] According to Table 3 and Figure 4 (a) Data Figure 4 (The horizontal axis 1-5 represents the five AGVs in Task 1, 6-10 represent the five AGVs in Task 2, and so on.) During the execution of five consecutive tasks, the DREAMS algorithm consumed a total energy of 1004.3 kWh and covered a total distance of 4655.19 meters. This compares to the traditional A algorithm, which consumed 1105.6 kWh and covered a total distance of 4798.02 meters. This demonstrates that the DREAMS algorithm outperforms the traditional A algorithm in both energy consumption and distance.

[0156] From Table 3 and Figure 4The data in (b) show that in 5 consecutive tasks, the total energy consumption of the DREAMS algorithm is 1004.3 kwh, and the total energy consumption of the Dijkstra algorithm is 1133.13 kwh. The energy consumption of the DREAMS algorithm is reduced by 11.4%.

[0157] In Task 1, the total energy consumption of the workshop using the DREAMS algorithm was 181.86 kWh, while that of the Dijkstra algorithm was 181.96 kWh. In Task 3, the DREAMS algorithm consumed 206.72 kWh, while that of the Dijkstra algorithm was 244.49 kWh. In Task 5, the DREAMS algorithm consumed 45.77 kWh, while that of the Dijkstra algorithm was 89.86 kWh. These data demonstrate that the DREAMS algorithm performs well in energy control.

[0158] Table 3

[0159]

[0160] In addition, to verify the effectiveness of the present invention, by removing the optimization algorithms and workshop rules mentioned in the present invention one by one, 5 AGVs and 5 consecutive tasks were also used to collect data on the energy consumption of each task in each rule and compare their energy consumption data.

[0161] Figure 5 (a) represents the energy consumption comparison of each task of the present invention and the one without the retry mechanism in the present invention, (b) represents the energy consumption comparison of each task of the present invention and the one without the spatiotemporal overlap conflict detection mechanism in the present invention, Figure 6 (a) Comparison of energy consumption per task between the present invention and the heuristic rule without the machine selection part of the present invention, and (b) Comparison of energy consumption per task between the present invention and the dynamic speed and dynamic energy consumption model without the AGV.

[0162] According to Table 4 and Figure 5 The data in (a) shows that the total energy consumption of the DREAMS algorithm is 1004.3 kwh, a 5.1% reduction compared to 1058.39 kwh for the NrDREAMS algorithm, which does not use a retry mechanism. In Task 5, AGV5 consumes 45.77 kwh with the DREAMS algorithm and 46.2 kwh with the NrDREAMS algorithm, a 0.9% reduction. The AGV consumes 43.77 kwh with the DREAMS algorithm and 58.36 kwh with the NrDREAMS algorithm, a 25% reduction. The standard deviation of the DREAMS algorithm's energy consumption per task is ±5.8 kwh, lower than the NrDREAMS algorithm's ±7.3 kwh.

[0163] According to Table 4 and Figure 5As shown in data (b), the DREAMS algorithm's total energy consumption is 5.2% lower than the 1059.37 kWh of the NcDREAMS algorithm, which lacks spatiotemporal conflict detection. In Task 3, the DREAMS algorithm consumed 206.72 kWh, while the NcDREAMS algorithm consumed 239.6 kWh, a 13.7% reduction. In Task 5, the DREAMS algorithm consumed 43.77 kWh, while the NcDREAMS algorithm consumed 56.66 kWh. The standard deviation of the DREAMS algorithm's energy consumption is ±5.8 kWh, while the NcDREAMS algorithm's energy consumption is ±8.9 kWh.

[0164] As well as Table 5 and Figure 6 (a) Compared with the data of the algorithm without the other two rules, the total energy consumption of the DREAMS algorithm in 5 consecutive tasks is 9% lower than that of the NhDREAMS algorithm without heuristic rules, which is 1103.69 kwh. In Task 3, the DREAMS algorithm consumed 206.72 kwh, while the NhDREAMS algorithm consumed 244.49 kwh, a decrease of 15.5%. In Task 5, the energy consumption of AGV5 was 5.77 kwh under the DREAMS algorithm and 58.3 kwh under the NhDREAMS algorithm, a decrease of 21.5%. The energy consumption of the DREAMS algorithm dropped from 56.06 kwh to 45.77 kwh, a decrease of 18.4%. The standard deviation of the energy consumption of the DREAMS algorithm is ±5.8 kwh, while that of the NhDREAMS algorithm is ±9.1 kwh.

[0165] According to Table 5 and Figure 6 The data in (b) shows that the total energy consumption of the DREAMS algorithm over five consecutive tasks was 1004.3 kWh, a 2.5% reduction compared to 1030.14 kWh for the NtDREAMS algorithm, which does not employ a dynamic speed strategy. In Task 5, the total energy consumption of the AGV workshop was 45.77 kWh under the DREAMS algorithm, compared to 48.94 kWh under the NtDREAMS algorithm, a 6.5% reduction. In Task 3, AGV3 consumed 35.48 kWh under the DREAMS algorithm, compared to 44.6 kWh under the NtDREAMS algorithm, a 20.5% reduction. The runtime was reduced from 219.08 seconds to 196.58 seconds, a 10.3% decrease. In Task 4, AGV4 consumed 43.77 kWh under the DREAMS algorithm, compared to 53.59 kWh under the NtDREAMS algorithm, a 22.4% increase. In Task 4, the energy consumption of AGV4 using the DREAMS algorithm was 43.77 kWh, while that using the NtDREAMS algorithm was 53.59 kWh, a 22.4% increase. This shows that the present invention can rationally plan AGV path allocation and machine allocation in complex environments, effectively reducing workshop energy consumption while improving production efficiency.

[0166] Table 4

[0167]

[0168]

[0169] Table 5

[0170]

[0171]

[0172] In summary, compared to existing technologies, this invention is suitable for AGV scheduling in matrix manufacturing workshops of varying sizes and machine layouts. By designing dynamic path planning, it enables the coordinated scheduling of multiple AGVs to improve workshop production efficiency. It also utilizes AGVs to dynamically adjust their parameters to reduce both AGV and workshop energy consumption. Finally, it optimizes machine selection to reduce idle energy consumption. These rules enable the coordinated scheduling of multiple AGVs in a dynamic environment while minimizing workshop energy consumption.

[0173] This invention addresses the issues of inefficient AGV scheduling and low machine utilization, which lead to idle energy consumption in matrix workshops. It optimizes workshop energy consumption through the dynamic path planning capabilities of the DREAMS (Dynamic Retry and Energy-Aware Multi-factor Scheduling) algorithm. This functionality is divided into three main components: a dynamic retry mechanism, AGV energy optimization, and heuristic rules. The first component consists of two sub-components: the retry mechanism and spatiotemporal overlap conflict detection. By updating obstacles in real time, it enables the coordinated scheduling of multiple AGVs in a dynamic environment. The second component consists of three sub-components: a dynamic power rate, a dynamic energy consumption model, and a battery swap strategy. This strategy reduces the number of visits to battery swap stations and eliminates unnecessary energy loss. The third component uses heuristic rules to balance machine selection. Using load factors, machine availability time factors, and machine distance factors, AGVs optimally select machines. This prevents excessive AGV concentration on a single machine, which increases total operating time, increases the number of AGV starts and stops, and leaves other machines idle, thereby reducing the energy consumption of other AGVs. This reduces overall workshop energy consumption.

[0174] The present invention also discloses a multi-AGV collaborative scheduling method for a matrix workshop, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The method is implemented when the processor executes the computer program.

[0175] The present invention further discloses an embodiment that provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned various method embodiments are implemented.

[0176] The present invention also provides a computer program product, which, when executed on a data storage device, enables the data storage device to implement the steps in the above-mentioned method embodiments.

[0177] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0178] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0179] In the embodiments provided herein, it should be understood that the disclosed devices and methods may be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of modules or units is merely a logical functional division. In actual implementation, other division methods may be used, such as combining or integrating multiple modules or components into another system, or omitting or not implementing certain features.

[0180] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed across multiple network elements. Some or all of the modules may be selected to achieve the purpose of the solution of this embodiment according to actual needs.

Claims

1. A multi-AGV collaborative scheduling method for a matrix workshop, characterized in that: The following steps are involved: Obtain the machine layout information and AGV parameters of the grid matrix workshop, and generate the initial AGV path based on the A* algorithm; When it is detected that the current AGV is at the same location as other AGVs at the same time, the initial path is replanned with the goal of minimizing the total energy consumption of the workshop to obtain a replanned path; wherein the total energy consumption is composed of the total energy consumption of the machine and the total energy consumption of the AGV.

2. A multi-AGV collaborative scheduling method for a matrix workshop according to claim 1, characterized in that: The total energy consumption of the machine includes operating energy consumption, no-load energy consumption and start-stop energy consumption; The operating energy consumption is given by calculate; The no-load energy consumption is given by calculate; The start-stop energy consumption is given by calculate; in, represents the operating energy consumption of machine n, represents the operating power of machine n, T kin represents the time spent by the i-th process of the k-th AGV on machine n, N represents the number of machines, O represents the number of processes, represents the no-load energy consumption of machine n, represents the no-load power consumption of machine n, represents the idle time of machine n, represents the start-stop energy consumption of machine n, represents the start-stop power of machine n, represents the start and stop time of machine n, Indicates the number of starts and stops of machine n.

3. A multi-AGV collaborative scheduling method for a matrix workshop according to claim 2, characterized in that: During the re-planning process, a corresponding machine is selected for each process according to the machine priority; The machine priority calculation method is: Among them, Priority(n) represents the priority of machine n, L n Indicates the current number of tasks for machine n, is the waiting time of machine n, is the current time of the kth AGV, d ij represents the distance from the current process i to the next process j, and α, β, and γ are different weight factors.

4. A multi-AGV collaborative scheduling method for a matrix workshop according to claim 2 or 3, characterized in that: During the replanning process, when the AGV is in the i-th process and the battery power is less than the charging threshold, the processing time of the k-th AGV on the machine n in the i-th process is recalculated; The recalculation method is: Among them, T kin，new It represents the processing time of the kth AGV on the machine n of the i-th process after recalculation, represents the start and stop time of the kth AGV, T kch represents the battery replacement time of the kth AGV, d ijk represents the distance of the kth AGV from process i to the next process j, v min represents the minimum speed of the kth AGV.

5. The multi-AGV collaborative scheduling method for a matrix workshop according to claim 4, characterized in that: The energy consumption of each AGV includes moving energy consumption, no-load energy consumption, and start-stop energy consumption; The movement energy consumption is calculated based on the movement power, and the movement power calculation method is: in, represents the moving power of the kth AGV, sco k Represents the battery capacity of the kth AGV.

6. A multi-AGV collaborative scheduling method for a matrix workshop according to claim 5, characterized in that: The calculation method of AGV battery power is: Among them, soc k (t+1) represents the battery capacity of the kth AGV at time t+1, sco k (t) represents the battery capacity of the kth AGV at time t, η represents the efficiency factor of the battery, X kin represents the processing status of the kth AGV on the machine n in the i-th process, v(soc k ) indicates that the kth AGV is in a state where the battery power is soc k The moving speed is , and D represents the maximum distance that can be traveled per unit battery power.

7. A multi-AGV collaborative scheduling method for a matrix workshop according to claim 6, characterized in that: The kth AGV has a battery charge of soc k The calculation method of moving speed is: in, Indicates that the kth AGV is in a state where the battery power is 0.20≤soc k Movement speed when ≤0.35, v max represents the maximum speed of the kth AGV.

8. A multi-AGV collaborative scheduling method for a matrix workshop, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

Images