Relay core component double-layer production line AGV scheduling strategy considering task completion time and energy consumption optimization

Abstract

The invention provides a relay core component double-layer production line AGV scheduling strategy considering task completion time and energy consumption optimization, and aims to optimize task completion time and AGV energy consumption distribution of a double-layer relay welding production line and improve the operation efficiency of an AGV in a complex production environment. The method comprises the following steps: firstly, constructing an environment model of a double-layer production line, extracting operation constraints and task requirements of an AGV, and planning an initial path by using an A * algorithm; secondly, whether dynamic or static position conflicts exist is judged according to the real-time position of the AGV, obstacle avoidance is performed in combination with an improved BUG2 algorithm and a priority strategy, and an AGV route after obstacle avoidance is updated; and finally, combined optimization is carried out on the path planning and task allocation of the AGV by using mixed integer programming in combination with an adaptive weight adjustment strategy so as to ensure that the AGV can complete the established task under the lowest energy consumption. The method can be widely applied to the fields of intelligent manufacturing, automatic logistics, unmanned warehousing and the like, the production efficiency is improved, and the operation cost is reduced.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent logistics and intelligent manufacturing technology, and specifically relates to an AGV scheduling strategy for a double-layer production line of relay core components that takes into account task completion time and energy consumption optimization, aiming to optimize task execution time and energy consumption management to improve production efficiency and resource utilization. Background Art

[0002] With the rapid development of intelligent manufacturing and automated logistics technologies, automated guided vehicle (AGV) systems have become a core component of modern manufacturing workshops and warehouse logistics. In traditional single-story production environments, AGV scheduling systems have been able to achieve basic material transportation functions through fixed path planning based on magnetic strips or laser navigation, combined with simple time window management strategies. However, when the application scenario is expanded to two-story or multi-story production lines, existing AGV scheduling technology has exposed obvious limitations.

[0003] Under the two-layer production line architecture, the AGV system faces numerous technical challenges: First, the complexity of the spatial layout increases significantly. AGVs not only need to plan their paths within the same plane but also utilize central elevators for cross-layer transport. This three-dimensional motion pattern makes traditional two-dimensional path planning algorithms difficult to apply. Second, the risk of dynamic collisions increases. Critical nodes such as elevator entrances become system bottlenecks, and the intersecting movements of multiple AGVs within a confined space can easily lead to congestion and deadlocks. Third, task scheduling becomes more complex. Due to the strict timing constraints of production processes, AGV transport delays can cause excessive temperature drops between processes, directly impacting product quality. Furthermore, the additional energy consumption associated with cross-layer movement makes AGV power management challenging. Some AGVs run out of power due to frequent cross-layer tasks, while others remain idle, resulting in uneven resource utilization.

[0004] Therefore, it is urgent to develop a new AGV scheduling method that can simultaneously optimize task completion time and system energy consumption, and realize efficient collaborative operation of multiple AGVs in a two-layer production environment. Summary of the Invention

[0005] Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention proposes a dual-layer production line AGV scheduling strategy for relay core components that takes into account task completion time and energy consumption optimization, integrating task completion efficiency and AGV real-time power monitoring into a dual-objective optimization framework. The proposed multi-AGV scheduling optimization strategy adopts a hierarchical progressive architecture, aiming to achieve dual-objective optimization of task completion time and energy consumption. The core concept of the overall strategy can be summarized as a closed-loop process of "initial path planning-conflict detection-dynamic obstacle avoidance-path update". This method can effectively adapt to the production line's personalized requirements for efficiency and energy consumption, and adapt to the complex structure of the double-layer production line.

[0006] Technical Solution: To achieve the above objectives, the present invention proposes an AGV scheduling strategy for a double-layer production line of relay core components that takes into account task completion time and energy consumption optimization, including the following steps:

[0007] S1: Establish a task completion time and AGV energy consumption model based on a double-layer relay welding production line;

[0008] S2: Design AGV static and dynamic conflict priority rules;

[0009] S3: Use the traditional A* algorithm to obtain the initial path planning;

[0010] S4: Update the AGV path based on the dynamic and static conflict avoidance strategy;

[0011] Furthermore, the specific process of establishing the task completion time and AGV energy consumption model based on the double-layer relay welding production line in step S1 is as follows:

[0012] Assume that the relay welding production line can be abstracted as a checkerboard-like grid space, and the size of the assembly environment is p×q (where p and q are arbitrary positive integers greater than 1). Define the number of AGVs as N, with each AGV index denoted as k, satisfying k∈{1,2,...,N}; define the total number of production line tasks as M, with each task index denoted as m, satisfying m∈{1,2,...,M}; define the total number of elevators in the double-layer relay production line as E, with each elevator index denoted as e, satisfying e∈{1,2,...,E}; define the production line space as having L layers, with each layer index denoted as l, satisfying l∈{1,2,...,L}; define the possible locations of AGVs as P, with each node index denoted as i,j, satisfying i,j∈{1,2,...,P}; define the total time it takes for an AGV to complete a task as T, with each time index denoted as t, satisfying t∈{1,2,...,T}. The task completion time function and AGV energy consumption function are constructed respectively. The weight coefficients of the dual objective functions are determined based on the analytic hierarchy process (AHP), and the constraints related to AGV power, path, task, and elevator are determined.

[0013] Furthermore, the construction task completion time function and the AGV energy consumption function can be expressed as:

[0014]

[0015] Among them, T represents the total time to complete all task scheduling cycles, f mk It represents the time for the kth AGV to complete the mth task, E total Indicates the total amount of electricity consumed by AGV to complete all tasks. The amount of electricity consumed by the kth AGV moving from position i to position j on the lth floor, is a 0-1 variable. If the kth AGV moves from position i to position j on the lth floor, then otherwise y mk is a 0-1 variable. If the kth AGV is assigned to perform the mth task, then y mk =1, otherwise y mk =0.

[0016] Furthermore, the time f for the kth AGV to complete the mth task is mk It can be expressed as:

[0017]

[0018] Among them, t mk represents the time when the kth AGV starts to perform the mth task, t m represents the machine processing time of the mth task, t mk,agv It represents the path movement time required for the kth AGV to perform the mth task, t mk,e It represents the running time of elevator e required for the k-th AGV to perform the m-th task.

[0019] The time t when the kth AGV starts to perform the mth task mk , the time t when each processing position initiates a transport request to the upper computer mr The time f that the AGV assigned to this task completed the previous task m'k The larger of the two is determined, namely:

[0020] t mk =max{t mr ,f m'k}

[0021] The subscript m' represents the index number of the previous task performed by the k-th AGV before performing the m-th task.

[0022] The path movement time t required for the kth AGV to perform the mth task mk,agv Including the running time of AGV in each layer, it can be expressed as:

[0023]

[0024] in, represents the movement time of the kth AGV from position i to position j on the lth floor. If the running speed of each AGV in the corresponding floor remains constant, the movement time of the kth AGV from position i to position j on the lth floor is The speed of the kth AGV on the lth floor The following relationship is satisfied:

[0025]

[0026] The k-th AGV needs the elevator e to run for the m-th task. mk,e The calculation of is subject to three assumptions, as follows:

[0027] Assumption 1: An AGV in the waiting state must wait until all AGVs waiting to exit the elevator on the same floor have completed the exit operation before entering the elevator car.

[0028] Assumption 2: When there are multiple AGVs waiting for the same elevator on the same floor, they should enter the elevator in ascending order according to the time sequence of their arrival at the elevator location.

[0029] Assumption 3: If multiple AGVs on the same floor need to perform exit operations, the order in which they exit the cabin should follow the last-in, first-out (LIFO) principle, which is the opposite of the order in which they enter the cabin.

[0030] The k-th AGV needs the elevator e to run for the m-th task. mk,e , including the time W that the kth AGV waits for the eth elevator to arrive ek , and the elevator operation (up / down) time, t mk,e It can be expressed as:

[0031]

[0032] Among them, z ke is a 0-1 variable. If the kth AGV uses the eth elevator, then z ke =1, otherwise z ke =0, R m Is the flag for whether the mth task needs to use the elevator. If the elevator is needed and it is an upward demand, then R m =1, if the elevator is needed and the demand is for downhill, then R m =-1, if no elevator is needed, then R m =0, represents the ascending time of the e-th elevator, Indicates the ascending time of the e-th elevator.

[0033] The time W that the kth AGV waits for the eth elevator to arrive is ek It is obtained by iterative accumulation, which can be expressed as:

[0034]

[0035] in, It represents the elevator waiting time of the AGV before the k-th AGV in the queue waiting for the e-th elevator.

[0036] Furthermore, the amount of electricity consumed by the kth AGV when it moves from position i to position j on the lth floor is It can be expressed as:

[0037]

[0038] Among them, C k It represents the amount of electricity consumed by the k-th AGV per unit distance. It represents the distance from position i to position j of the kth AGV on the lth floor.

[0039] The amount of electricity C consumed by the kth AGV per unit distance k Based on the dynamic model definition of the AGV moving along each segment of the path, since the influence of wind speed in the warehouse environment can be ignored, its energy consumption can be considered to be proportional to the square of the AGV running speed. The calculation formula is:

[0040]

[0041] Where G represents the total gravity of the AGV and the cargo, and μ represents the ground friction factor.

[0042] Furthermore, the process of determining the weight coefficients of the dual-objective function and constructing the multi-objective optimization function based on the analytic hierarchy process (AHP) is as follows:

[0043] The dual objective function can be expressed as:

[0044] Min(α·T+β·E total )

[0045] Among them, α and β are the relative importance weights of task completion time and AGV energy consumption in the current production line, which can be obtained through the hierarchical analysis method. The calculation method of the weight coefficient is as follows:

[0046] First, based on the process requirements of the relay welding production line, the relative importance of task completion time and energy consumption was compared and evaluated, and a judgment matrix was constructed accordingly:

[0047]

[0048] Among them, a TE Indicates the importance of task completion time relative to energy consumption, a ET Indicates the importance of energy consumption relative to task completion time, satisfying a ET ·a TE =1.

[0049] Calculate the geometric mean of the elements in each row of the judgment matrix:

[0050]

[0051] Among them, g T With g E They represent the geometric mean weight values of task completion time and energy consumption respectively. Finally, through normalization, the weight coefficient satisfying a+β=1 is obtained:

[0052]

[0053] Furthermore, the constraints on AGV power, path, task, and elevator are determined as follows:

[0054] The AGV power constraints include: the remaining battery capacity of each AGV must meet the requirements of its task execution and path traversal to prevent task interruption due to insufficient battery power; the AGV battery power will be updated after completing the current task; and the remaining battery capacity of each AGV cannot exceed the maximum allowed battery capacity. The above three constraints can be expressed as:

[0055]

[0056] Among them, b k represents the remaining power of the kth AGV, B k Represents the battery capacity of the k-th AGV.

[0057] The path-related constraints mean that the AGV cannot pass through obstacles or occupy paths already occupied by other AGVs during movement to avoid path conflicts. It can be expressed as:

[0058]

[0059] The task-related constraints include: each task must be completed by an AGV and each AGV can only perform one task at a time; the completion time of each task must be less than the total time of the scheduling period; and the AGV must complete the current task before starting the next task. The above three constraints can be expressed as:

[0060]

[0061] Here, m' represents the index of the next task that the k-th AGV will perform after completing the m-th task.

[0062] The elevator-related constraints include: each AGV can only select one elevator at a time, the number of AGVs carried by each elevator at the same time does not exceed its capacity limit, and each elevator can only be in one of the states of up or down at the same time. The above three constraints can be expressed as:

[0063]

[0064] Among them, Q e represents the capacity of the e-th elevator, U e (t) is a 0-1 variable. When the e-th elevator is in the upward state at time t, U e (t)=1, otherwise U e (t) = 0, D e (t) is a 0-1 variable. When the e-th elevator is in the downward state at time t, D e (t)=1, otherwise D e (t)=0.

[0065] Furthermore, the AGV static and dynamic conflict priority rules designed in step S2 include:

[0066] The static obstacle avoidance rule uses the optimized BUG2 obstacle avoidance mechanism as its core algorithm: the AGV's movement in an unknown environment is divided into a straight line towards the target point and a detour when encountering an obstacle. In an obstacle-free area, the AGV maintains its original straight-line motion strategy to minimize the number of turns. Once an obstacle is detected ahead, the system dynamically evaluates the time cost of adjacent positions and selects the position with the lowest time cost as the turning target. After executing the initial obstacle avoidance turn, the system continuously monitors the time cost relationship between the current movement direction and adjacent positions, and determines whether to turn based on the following rules:

[0067] Rule 1: If the time cost of the current movement direction is the same as that of the adjacent position, the AGV prioritizes maintaining the current direction to minimize the number of turns.

[0068] Rule 2: If the time cost of the current direction exceeds the time cost of the adjacent position, a turn adjustment is made immediately to switch to a path with lower cost.

[0069] The dynamic obstacle avoidance rules are formulated based on priority, which is related to the order in which tasks are generated and whether the AGV is loaded:

[0070] Rule 1: When a loaded AGV conflicts with an unloaded AGV, the loaded AGV has a higher priority than the unloaded AGV by default.

[0071] Rule 2: When a conflict occurs between load-carrying AGVs: (1) Each load-carrying AGV has a corresponding task generation time. The AGV with the shorter task generation time has a higher priority. (2) When the generation time of the tasks carried by two vehicles is the same, the AGV with the shorter task generation time has a higher priority based on the estimated completion time of the AGV. (3) If the above two situations still cannot be judged, the overall operation objective function is calculated and the obstacle avoidance solution with the optimal objective function is selected.

[0072] Furthermore, the initial path planning is obtained by using the traditional A* algorithm in step S3 as follows:

[0073] Initialize the parameters such as the starting node and target node in the production line environment; create an open list and a closed list, and store the starting node to be searched in the open list. The closed list is empty at the beginning of the operation; by continuously updating the environmental parameters, all the nodes to be searched in the open list are searched until the open list is empty and the A* algorithm search ends; calculate the heuristic function f(n) of all nodes after the search, and use the node corresponding to the minimum value of f(n) as the next node to be searched by the A* algorithm, and store it in the closed list, and remove it from the open list at the same time; determine whether the node expanded by the A* algorithm coincides with the target node. If so, it means that the node is the best point for optimization and the algorithm search is successful; if not, repeat the previous step to calculate until the best path is output.

[0074] Furthermore, in step S4, the AGV path is updated according to the dynamic and static conflict avoidance strategy as follows:

[0075] After the A* algorithm generates an initial optimal path for each AGV, the system monitors each AGV's next position in real time to detect potential static obstacles (such as equipment and shelves) and dynamic conflicts (such as overlapping paths with other AGVs). If a conflict is detected, the dynamic obstacle avoidance mechanism is triggered, adjusting the AGV's path based on priority rules or implementing a waiting strategy. Finally, the AGV's global path plan is updated, and the above process is repeated until the task is completed.

[0076] The present invention firstly targets the three-dimensional layout characteristics of the double-layer welding production line of the core components of relays, establishes a dual-objective scheduling optimization model with the goals of minimizing task completion time and AGV energy consumption, and constructs a grid production environment based on the cross-layer transportation constraints of multiple AGVs; secondly, considering the static obstacles and dynamic conflicts faced by AGVs during operation, a dynamic obstacle avoidance strategy integrating the improved BUG2 algorithm and the priority decision-making mechanism is designed to realize real-time correction and conflict avoidance of AGV paths; finally, an improved A* algorithm is introduced to obtain the driving trajectory with the lowest path cost, and the mixed integer programming and multi-objective weight adjustment mechanism are combined to realize the integrated scheduling optimization of task allocation and path planning.

[0077] The present invention addresses the problems of low scheduling efficiency, uneven energy consumption distribution, and frequent path conflicts in the collaborative operation of multiple AGVs in a double-layer welding production line. This method proposes a systematic intelligent scheduling method, which specifically includes two core parts: one is to construct a scheduling optimization model with time and energy consumption as the goals to improve the global performance of AGV collaborative scheduling; the other is to design a dynamic path planning mechanism that integrates obstacle avoidance strategies to reduce energy waste caused by path conflicts and redundant motion. The first part includes three key steps: constructing a two-layer production environment grid model and establishing a dual-objective optimization function for task completion time and energy consumption; determining the weight coefficients of task time and energy consumption targets based on the analytic hierarchy process (AHP) to quantify scheduling priorities; and formulating a scheduling allocation strategy based on information such as the AGV's initial position, power status, and task timing to achieve optimal task-AGV matching. The second part includes the following seven core mechanisms: using the improved A* algorithm to construct the initial path and generate a heuristic path planning diagram; using the path conflict detection mechanism to monitor the AGV path status in real time; introducing dynamic priority judgment rules based on task generation order and load status to distinguish the AGV passage priority; using the optimized BUG2 algorithm to design detour paths for static obstacles, and dynamically selecting the detour direction based on time cost as the objective function; introducing a "path pre-occupancy mechanism" to mark the next position status of the AGV to prevent multi-vehicle path conflicts; dynamically updating the AGV scheduling status based on the task execution completion time, waiting time and path energy consumption information; optimizing the path planning strategy through memory playback and experience update mechanism to achieve adaptive iterative improvement of system performance.

[0078] Beneficial effects: Compared with the existing technology, the present invention is suitable for the collaborative scheduling of AGVs in double-layer production lines at different structural levels. By constructing a joint optimization model of task completion time and energy consumption, the overall scheduling efficiency of the multi-AGV system can be improved. At the same time, based on the optimal path planning results, a cross-layer movement and obstacle avoidance strategy is designed, and the AGV task execution order is adaptively adjusted through a dynamic priority judgment mechanism, which can greatly reduce the path conflict rate, improve the production line logistics flow efficiency, and reduce the system energy consumption and scheduling costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 Schematic diagram of the framework of the method of the present invention;

[0080] Figure 2 This is a schematic diagram of the AGV static obstacle avoidance that improves BUG2 in the present invention;

[0081] Figure 3 This is a priority flow chart of the AGV dynamic obstacle avoidance of the present invention;

[0082] Figure 4 A spatial layout diagram of a relay production line according to the method of the present invention;

[0083] Figure 5A rasterized environment map for the relay production line simulation of the present invention;

[0084] Figure 6 Comparison chart of AGV energy consumption using A* algorithm, Dijkstra algorithm, and genetic algorithm path planning;

[0085] Figure 7 Gantt chart for path planning using A* algorithm, Dijkstra algorithm, and genetic algorithm. DETAILED DESCRIPTION

[0086] The present invention is further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, modifications of various equivalent forms of the present invention made by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0087] like Figure 1 As shown, the present invention proposes an AGV scheduling strategy for a double-layer production line of relay core components that takes into account task completion time and energy consumption optimization, including the following steps:

[0088] S1: Establish a task completion time and AGV energy consumption model based on a double-layer relay welding production line;

[0089] S2: Design AGV static and dynamic conflict priority rules;

[0090] S3: Use the traditional A* algorithm to obtain the initial path planning;

[0091] S4: Update the AGV path based on the dynamic and static conflict avoidance strategy;

[0092] Based on the above solution, this embodiment applies the above-mentioned relay core component double-layer production line AGV scheduling strategy considering task completion time and energy consumption optimization as an example, as follows:

[0093] Figure 2 The figure shows the AGV static obstacle avoidance method using the improved BUG2 in this embodiment, where B represents an obstacle, T represents the target position, and the red dotted line represents the moving path of the AGV. The numbers in the grid represent the time cost required to reach the target from that position. According to the improved BUG2 algorithm, the movement of the AGV in an unknown environment is divided into straight-line driving facing the target point and detours when encountering obstacles. In obstacle-free areas, the AGV maintains its original straight-line motion strategy to minimize the number of turns. Once an obstacle is detected ahead, the system will dynamically evaluate the time cost of adjacent positions and select the position with the lowest time cost as the turning target. After executing the initial obstacle avoidance turn, the system will continue to monitor the time cost relationship between the current direction of movement and adjacent positions, and determine whether to turn according to the following rules:

[0094] Rule 1: If the time cost of the current movement direction is the same as that of the adjacent position, the AGV prioritizes maintaining the current direction to minimize the number of turns.

[0095] Rule 2: If the time cost of the current direction exceeds the time cost of the adjacent position, a turn adjustment is made immediately to switch to a path with lower cost.

[0096] Figure 3 The embodiment shown is based on the priority of the AGV dynamic obstacle avoidance rules. The priority is related to the time when the task is generated and whether the AGV is loaded:

[0097] Rule 1: When a loaded AGV conflicts with an unloaded AGV, the loaded AGV has a higher priority than the unloaded AGV by default.

[0098] Rule 2: When a conflict occurs between load-carrying AGVs: (1) Each load-carrying AGV has a corresponding task generation time. The AGV with the shorter task generation time has a higher priority. (2) When the generation time of the tasks carried by two vehicles is the same, the AGV with the shorter task generation time has a higher priority based on the estimated completion time of the AGV. (3) If the above two situations still cannot be judged, the overall operation objective function is calculated and the obstacle avoidance solution with the optimal objective function is selected.

[0099] like Figure 4 As shown, the relay welding production line in this embodiment is a double-layer space architecture, which can be abstracted as a flat grid environment, such as Figure 5 As shown in the figure. The arrow on the AGV indicates the direction of the vehicle's head; the positions of the AGV's starting and destination points can be modified arbitrarily; the task nodes for raw material transportation, iron sheet transportation, copper wire transportation, and finished product transportation are marked; the gray strip area represents the elevator location, with the left side of the strip as the dividing line, the upper space, and the right side as the lower space; the white grid represents the area that the AGV can occupy; the grid sequence formed by a series of adjacent white grids between any two grids represents the AGV's travel path; the number of grids the AGV passes through represents its travel distance. Based on the gridded environmental map, a task completion time and AGV energy consumption model for a double-layer relay welding production line can be constructed.

[0100] Assume the dimensions of the final assembly environment are 18×26. Define the number of AGVs as N, with each AGV indexed as k, satisfying k∈{1,2,...,N}; define the total number of production line tasks as M, with each task indexed as m, satisfying m∈{1,2,...,M}; define the total number of elevators in the double-layer relay production line as E, with each elevator indexed as e, satisfying e∈{1,2,...,E}; define the total number of production line floors as L, with each floor indexed as l, satisfying l∈{1,2,...,L}; define the possible locations of AGVs as P, with each node indexed as i,j, satisfying i,j∈{1,2,...,P}; define the total time it takes for an AGV to complete a task as T, with each moment indexed as t, satisfying t∈{1,2,...,T}. A task completion time function and an AGV energy consumption function are constructed, respectively. The weight coefficients of the dual-objective functions are determined using the Analytic Hierarchy Process (AHP), and the constraints related to AGV power, routing, tasks, and elevators are determined.

[0101] The construction task completion time function and AGV energy consumption function can be expressed as:

[0102]

[0103] Among them, T represents the total time to complete all task scheduling cycles, f mk It represents the time for the kth AGV to complete the mth task, E total Indicates the total amount of electricity consumed by AGV to complete all tasks. The amount of electricity consumed by the kth AGV moving from position i to position j on the lth floor, is a 0-1 variable. If the kth AGV moves from position i to position j on the lth floor, then otherwise y mk is a 0-1 variable. If the kth AGV is assigned to perform the mth task, then y mk =1, otherwise y mk =0.

[0104] The time f for the kth AGV to complete the mth task mk It can be expressed as:

[0105]

[0106] Among them, t mk represents the time when the kth AGV starts to perform the mth task, t m represents the machine processing time of the mth task, t mk,agv It represents the path movement time required for the kth AGV to perform the mth task, t mk,e It represents the running time of elevator e required for the k-th AGV to perform the m-th task.

[0107] The time t when the kth AGV starts to perform the mth task mk , the time t when each processing position initiates a transport request to the upper computer mr The time f that the AGV assigned to this task completed the previous task m'k The larger of the two is determined, namely:

[0108] t mk =max{t mr ,f m'k}

[0109] The subscript m' represents the index number of the previous task performed by the k-th AGV before performing the m-th task.

[0110] The path movement time t required for the kth AGV to perform the mth task mk,agv Including the running time of AGV in each layer, it can be expressed as:

[0111]

[0112] in, represents the movement time of the kth AGV from position i to position j on the lth floor. If the running speed of each AGV in the corresponding floor remains constant, the movement time of the kth AGV from position i to position j on the lth floor is The speed of the kth AGV on the lth floor The following relationship is satisfied:

[0113]

[0114] The elevator e running time t required for the kth AGV to perform the mth task mk,e The calculation of is subject to three assumptions, as follows:

[0115] Assumption 1: An AGV in the waiting state must wait until all AGVs waiting to exit the elevator on the same floor have completed the exit operation before entering the elevator car.

[0116] Assumption 2: When there are multiple AGVs waiting for the same elevator on the same floor, they should enter the elevator in ascending order according to the time sequence of their arrival at the elevator location.

[0117] Assumption 3: If multiple AGVs on the same floor need to perform exit operations, the order in which they exit the cabin should follow the last-in, first-out (LIFO) principle, which is the opposite of the order in which they enter the cabin.

[0118] The elevator e running time t required for the kth AGV to perform the mth task mk,e , including the time W that the kth AGV waits for the eth elevator to arrive ek, and the elevator operation (up / down) time, t mk,e It can be expressed as:

[0119]

[0120] Among them, z ke is a 0-1 variable. If the kth AGV uses the eth elevator, then z ke =1, otherwise z ke =0, R m Is the flag for whether the mth task needs to use the elevator. If the elevator is needed and it is an upward demand, then R m =1, if the elevator is needed and the demand is for downhill, then R m =-1, if no elevator is needed, then R m =0, represents the ascending time of the e-th elevator, Indicates the ascending time of the e-th elevator.

[0121] The time W that the kth AGV waits for the arrival of the eth elevator ek It is obtained by iterative accumulation, which can be expressed as:

[0122]

[0123] in, It represents the elevator waiting time of the AGV before the k-th AGV in the queue waiting for the e-th elevator.

[0124] The amount of electricity consumed by the kth AGV moving from position i to position j on the lth floor It can be expressed as:

[0125]

[0126] Among them, C k It represents the amount of electricity consumed by the k-th AGV per unit distance. It represents the distance from position i to position j of the kth AGV on the lth floor.

[0127] The amount of electricity C consumed by the kth AGV per unit distance k Based on the dynamic model definition of the AGV moving along each segment of the path, since the influence of wind speed in the warehouse environment can be ignored, its energy consumption can be considered to be proportional to the square of the AGV running speed. The calculation formula is:

[0128]

[0129] Where G represents the total gravity of the AGV and the cargo, and μ represents the ground friction factor.

[0130] The process of determining the weight coefficients of the dual-objective function and constructing the multi-objective optimization function based on the analytic hierarchy process (AHP) is as follows:

[0131] The dual objective function can be expressed as:

[0132] Min(α·T+β·E total )

[0133] Among them, α and β are the relative importance weights of task completion time and AGV energy consumption in the current production line, which can be obtained through the hierarchical analysis method. The calculation method of the weight coefficient is as follows:

[0134] First, based on the process requirements of the relay welding production line, the relative importance of task completion time and energy consumption was compared and evaluated, and a judgment matrix was constructed accordingly:

[0135]

[0136] Among them, a TE Indicates the importance of task completion time relative to energy consumption, a ET Indicates the importance of energy consumption relative to task completion time, satisfying a ET ·a TE =1.

[0137] Calculate the geometric mean of the elements in each row of the judgment matrix:

[0138]

[0139] Among them, g T With g E They represent the geometric mean weight values of task completion time and energy consumption respectively. Finally, through normalization, the weight coefficient satisfying a+β=1 is obtained:

[0140]

[0141] Determine the constraints related to AGV power, path, task, and elevator, including:

[0142] AGV power constraints include: the remaining battery capacity of each AGV must meet the requirements of its task execution and path traversal to prevent task interruption due to insufficient battery power; the battery power of the AGV will be updated after completing the current task; and the remaining battery capacity of each AGV cannot exceed the maximum allowed battery capacity. The above three constraints can be expressed as:

[0143]

[0144] Among them, b k represents the remaining power of the kth AGV, Bk Represents the battery capacity of the k-th AGV.

[0145] Path-related constraints mean that the AGV cannot pass through obstacles or occupy paths already occupied by other AGVs during movement to avoid path conflicts. It can be expressed as:

[0146]

[0147] Task-related constraints include: each task must be completed by an AGV and each AGV can only perform one task at a time; the completion time of each task must be less than the total time of the scheduling period; and the AGV must complete the current task before starting the next task. The above three constraints can be expressed as:

[0148]

[0149] Here, m' represents the index of the next task that the k-th AGV will perform after completing the m-th task.

[0150] Elevator-related constraints include: each AGV can only select one elevator at a time, the number of AGVs carried by each elevator at the same time does not exceed its capacity limit, and each elevator can only be in one of the states of upward or downward movement at the same time. The above three constraints can be expressed as:

[0151]

[0152] Among them, Q e represents the capacity of the e-th elevator, U e (t) is a 0-1 variable. When the e-th elevator is in the upward state at time t, U e (t)=1, otherwise U e (t) = 0, D e (t) is a 0-1 variable. When the e-th elevator is in the downward state at time t, D e (t)=1, otherwise D e (t)=0.

[0153] To verify the effectiveness of the present invention in optimizing the production efficiency and AGV energy consumption of a double-layer relay welding production line, this example uses the Dijkstra algorithm and the genetic algorithm as comparisons. The AGV motion in the production line must meet the following assumptions:

[0154] (1) The distance between two adjacent work points is equal, and the map is bidirectional; (2) One AGV can only perform one task at a time, and the same task can only be performed by one AGV; (3) The AGV's driving speed is constant on each floor; (4) The size of each work point is the same; (5) The number of AGVs is constant, and the sudden increase or decrease of AGVs is not considered; (6) The time for AGV loading and unloading and the probability of failure are ignored. When a task is issued in the transportation environment, the system needs to select an AGV from the idle AGVs online that can reach the task starting point the fastest to perform the task and complete the AGV scheduling.

[0155] Initial parameter settings for the simulation experiment: The task generation time is set to satisfy the Erlang distribution. A total of 15 tasks are allocated, and the task list is shown in Table 1. The initial state of charge (SOC) of the AGVs is set to 100%, and the battery capacity of each AGV is assumed to be 100Wh. When the AGV battery is less than 5%, task allocation to the AGV will be terminated. The map grid corresponds to 2m. It is assumed that the four AGVs have the same driving speed, with the AGV driving at 2m / s on the upper floor and 1m / s on the lower floor. The elevator takes 5s for one cross-floor transport (up / down). The simulation experiment uses α = 0.8 and β = 0.2.

[0156] Table 1 Task list

[0157]

[0158] Figure 6 The energy consumption comparison of three different path planning algorithms is shown. Figure 6 Figures (a), (b), and (c) show the SOC changes over time for four AGVs completing 15 tasks using the A*-based path planning strategy of the present invention, the Dijkstra path planning algorithm, and the genetic algorithm path planning method. The total energy consumption of the AGVs using the present invention's path planning strategy, the Dijkstra path planning algorithm, and the genetic algorithm path planning method is 183.50%, 188.25%, and 180.25%, respectively. This shows that under the same task sequence, the AGV using the genetic algorithm has the lowest energy consumption.

[0159] The energy consumption of an AGV is proportional to the square of its speed. When moving on the lower level, energy consumption is also lower due to the reduced speed. In the Dijkstra path planning results, the AGV prefers to return to the upper level to avoid obstacles, while the A* algorithm chooses to detour on the lower level to avoid obstacles. Therefore, the total energy consumption of the Dijkstra algorithm is higher than that of the A* algorithm. In addition, both the A* algorithm and the Dijkstra algorithm require AGV3 to return to the upper level after completing the task on the lower level, while the random initialization of the genetic algorithm population causes AGV3 to spend a lot of time in the elevator. Therefore, when the next task arrives, it cannot be assigned to AGV3 like the A* algorithm, but is assigned to AGV2 instead. Therefore, AGV3 does not need to return to the upper level, which reduces energy consumption but increases time costs.

[0160] Figure 7 The Gantt charts corresponding to the three path planning algorithms are shown. Figure 7 (a), (b), and (c) show Gantt charts for completing 15 tasks using the A*-based path planning strategy of the present invention, the Dijkstra path planning algorithm, and the genetic algorithm path planning method. The total time consumed by the path planning strategy of the present invention, the Dijkstra path planning algorithm, and the genetic algorithm path planning method is 51.29 seconds, 50.29 seconds, and 52.34 seconds, respectively. This shows that under the same task sequence, the genetic algorithm takes the longest to complete the task, while the Dijkstra algorithm takes the shortest.

[0161] The Dijkstra path planning algorithm executed 1.99% less time than the A* algorithm. This is because the AGVs travel at a lower speed on the lower level, and the Dijkstra path is shorter there, thus reducing the overall execution time. In contrast, the genetic algorithm executed 4.08% longer. Due to the random initialization of the population and the higher relative weight of time in the fitness function, and the lower speed of the AGVs on the lower level, the genetic algorithm favored paths leading to the upper level. This caused AGV3 to change direction after reaching the lower level and take the elevator again, further extending the execution time.

[0162] Through calculation, it can be seen that the objective function values of the path planning algorithm based on the A* algorithm, the Dijkstra algorithm, and the genetic algorithm of the present invention are 77.732, 77.882, and 77.922, respectively. Although the A* algorithm does not show a significant advantage in optimizing task completion time or AGV energy consumption alone, it can effectively improve the AGV scheduling results of the production line after reasonably allocating the importance weights of time and energy consumption according to production line requirements. Compared with the Dijkstra algorithm, the objective function of the A* algorithm is improved by 0.193%, and compared with the genetic algorithm, it is improved by 0.244%. These results show that the improved BUG2 using the A* path planning algorithm can effectively optimize the task completion time and AGV energy consumption in the production line scenario.

[0163] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will readily appreciate that other variations or modifications based on the above descriptions are possible. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A double-layer production line AGV scheduling strategy for relay core components that takes into account task completion time and energy consumption optimization, characterized by: The steps include: S1: Establish a task completion time and AGV energy consumption model based on a double-layer relay welding production line; S2: Design AGV static and dynamic conflict priority rules; S3: Use the traditional A* algorithm to obtain the initial path planning; S4: Update the AGV path according to the dynamic and static conflict avoidance strategy.

2. The AGV scheduling strategy for a double-layer production line of a relay core component considering task completion time and energy consumption optimization according to claim 1 is characterized in that: The specific process of establishing the task completion time and AGV energy consumption model based on the double-layer relay welding production line in step S1 is as follows: Assume that the relay welding production line can be abstracted as a chessboard-like grid space, the size of the assembly environment is p×q (where p and q are any positive integers greater than 1), define the number of AGVs as N, and the index of each AGV is k, satisfying k∈{1,2,...,N}; define the total number of production line tasks as M, and the index of each task is m, satisfying m∈{1,2,...,M}; define the total number of elevators in the double-layer relay production line as E, and the index of each elevator is e, satisfying e∈{1,2,...,E}; define the production line space to have a total of L layers, The index of each layer is represented as l, satisfying l∈{1,2,...,L}; the possible location nodes of AGV are defined as P, and the index of each node is represented as i,j, satisfying i,j∈{1,2,...,P}; the total time for AGV task completion is defined as T, and the index of each moment is represented as t, satisfying t∈{1,2,...,T}. The task completion time function and AGV energy consumption function are constructed respectively, and the weight coefficients of the dual objective functions are determined based on the analytic hierarchy process (AHP), and the constraints related to AGV power, path, task, and elevator are determined.

3. The AGV scheduling strategy for a double-layer production line of a relay core component considering task completion time and energy consumption optimization according to claim 2 is characterized in that: The construction task completion time function and AGV energy consumption function can be expressed as: Among them, T represents the total time to complete all task scheduling cycles, f mk It represents the time for the kth AGV to complete the mth task, E total Indicates the total amount of electricity consumed by AGV to complete all tasks. The amount of electricity consumed by the kth AGV moving from position i to position j on the lth floor, is a 0-1 variable. If the kth AGV moves from position i to position j on the lth floor, then otherwise y mk is a 0-1 variable. If the kth AGV is assigned to perform the mth task, then y mk =1, otherwise y mk =0.

4. The AGV scheduling strategy for a double-layer production line of a relay core component considering task completion time and energy consumption optimization according to claim 3 is characterized in that: The time f that the kth AGV completes the mth task mk It can be expressed as: Among them, t mk represents the time when the kth AGV starts to perform the mth task, t m represents the machine processing time of the mth task, t mk,agv It represents the path movement time required for the kth AGV to perform the mth task, t mk,e represents the running time of elevator e required for the k-th AGV to perform the m-th task; The time t when the kth AGV starts to perform the mth task mk , the time t when each processing position initiates a transport request to the upper computer mr The time f that the AGV assigned to this task completed the previous task m'k The larger of the two is determined, namely: t mk =max{t mr ,f m'k } Wherein, the subscript m' represents the index number of the previous task performed by the k-th AGV before performing the m-th task; The path movement time t required for the kth AGV to perform the mth task mk,agv Including the running time of AGV in each layer, it can be expressed as: in, represents the movement time of the kth AGV from position i to position j on the lth floor. If the running speed of each AGV in the corresponding floor remains constant, the movement time of the kth AGV from position i to position j on the lth floor is The speed of the kth AGV on the lth floor The following relationship is satisfied: The k-th AGV needs the elevator e to run for the m-th task. mk,e The calculation of is subject to three assumptions, as follows: Assumption 1: An AGV in the waiting state must wait until all other AGVs waiting to exit the elevator on the same floor have completed the exit operation before entering the elevator car; Assumption 2: When there are multiple AGVs waiting for the same elevator on the same floor, they should enter the elevator in ascending order according to the time sequence of their arrival at the elevator location; Assumption 3: If multiple AGVs on the same floor need to exit the cabin, their exit order should follow the last-in, first-out (LIFO) principle, which is the opposite of their entry order. The k-th AGV needs the elevator e to run for the m-th task. mk,e , including the time W that the kth AGV waits for the eth elevator to arrive ek , and the elevator operation (up / down) time, t mk,e It can be expressed as: Among them, z ke is a 0-1 variable. If the kth AGV uses the eth elevator, then z ke =1, otherwise z ke =0, R m Is the flag for whether the mth task needs to use the elevator. If the elevator is needed and it is an upward demand, then R m =1, if the elevator is needed and the demand is for downhill, then R m =-1, if no elevator is needed, then R m =0, represents the ascending time of the e-th elevator, Indicates the ascending time of the e-th elevator; The kth AGV waits for the arrival time of the eth elevator W ek It is obtained by iterative accumulation, which can be expressed as: in, It represents the elevator waiting time of the AGV before the k-th AGV in the queue waiting for the e-th elevator.

5. The AGV scheduling strategy for a double-layer production line of a relay core component considering task completion time and energy consumption optimization according to claim 3 is characterized in that: The amount of electricity consumed by the kth AGV moving from position i to position j on the lth floor It can be expressed as: Among them, C k It represents the amount of electricity consumed by the k-th AGV per unit distance. represents the distance from position i to position j of the kth AGV on the lth floor; The amount of electricity C consumed by the kth AGV per unit distance k Based on the dynamic model definition of the AGV moving along each segment of the path, since the influence of wind speed in the warehouse environment can be ignored, its energy consumption can be considered to be proportional to the square of the AGV running speed. The calculation formula is: Where G represents the total gravity of the AGV and the cargo, and μ represents the ground friction factor.

6. The AGV scheduling strategy for a double-layer production line of a relay core component considering task completion time and energy consumption optimization according to claim 2 is characterized in that: The process of determining the weight coefficients of the dual-objective function and constructing the multi-objective optimization function based on the analytic hierarchy process (AHP) is as follows: The dual objective function can be expressed as: Min(α·T+β·E total ) Among them, α and β are the relative importance weights of task completion time and AGV energy consumption in the current production line, which can be obtained through the hierarchical analysis method. The calculation method of the weight coefficient is as follows: First, based on the process requirements of the relay welding production line, the relative importance of task completion time and energy consumption was compared and evaluated, and a judgment matrix was constructed accordingly: Among them, a TE Indicates the importance of task completion time relative to energy consumption, a ET Indicates the importance of energy consumption relative to task completion time, satisfying a ET ·a TE =1; Calculate the geometric mean of the elements in each row of the judgment matrix: Among them, g T With g E They represent the geometric mean weight values of task completion time and energy consumption respectively. Finally, through normalization, the weight coefficient satisfying a+β=1 is obtained:

7. The AGV scheduling strategy for a double-layer production line of a relay core component considering task completion time and energy consumption optimization according to claim 2 is characterized in that: Determine the constraints related to AGV power, path, task, and elevator, including: The AGV power constraints include: the remaining battery capacity of each AGV must meet the requirements of its task execution and path traversal to prevent task interruption due to insufficient battery power; the battery power of the AGV will be updated after completing the current task; the remaining battery capacity of each AGV cannot exceed the maximum allowed battery capacity. The above three constraints can be expressed as follows: Among them, b k represents the remaining power of the kth AGV, B k represents the battery capacity of the kth AGV; The path-related constraints mean that the AGV cannot pass through obstacles or occupy paths already occupied by other AGVs during movement to avoid path conflicts, which can be expressed as: The task-related constraints include: each task must be completed by an AGV and each AGV can only perform one task at a time; the completion time of each task must be less than the total time of the scheduling period; and the AGV must complete the current task before starting the next task. The above three constraints can be expressed as follows: Where m' represents the index of the next task that the k-th AGV will perform after completing the m-th task; The elevator-related constraints include: each AGV can only select one elevator at a time, the number of AGVs carried by each elevator at the same time does not exceed its capacity limit, and each elevator can only be in one of the states of upward or downward movement at the same time. The above three constraints can be expressed as follows: Among them, Q e represents the capacity of the e-th elevator, U e (t) is a 0-1 variable. When the e-th elevator is in the upward state at time t, U e (t)=1, otherwise U e (t) = 0, D e (t) is a 0-1 variable. When the e-th elevator is in the downward state at time t, D e (t)=1, otherwise D e (t)=0.

8. The AGV scheduling strategy for a double-layer production line of a relay core component considering task completion time and energy consumption optimization according to claim 1 is characterized in that: The AGV static and dynamic conflict priority rules designed in step S2 include: The static obstacle avoidance rule uses the optimized BUG2 obstacle avoidance mechanism as the core algorithm: the AGV's movement in an unknown environment is divided into straight-line driving towards the target point and detours when encountering obstacles. In obstacle-free areas, the AGV maintains its original straight-line motion strategy to minimize the number of turns. Once an obstacle is detected ahead, the system dynamically evaluates the time cost of adjacent positions and selects the position with the lowest time cost as the turning target. After executing the initial obstacle avoidance turn, the system continuously monitors the time cost relationship between the current movement direction and adjacent positions, and determines whether to turn based on the following rules: Rule 1: If the time cost of the current movement direction is the same as that of the adjacent position, the AGV will prioritize maintaining the current direction to minimize the number of turns. Rule 2: If the time cost of the current direction exceeds the time cost of the adjacent position, immediately make a turn adjustment and switch to the path with lower cost; The dynamic obstacle avoidance rules are formulated based on priority, which is related to the order in which tasks are generated and whether the AGV is loaded: Rule 1: When a loaded AGV conflicts with an unloaded AGV, the loaded AGV has a higher priority than the unloaded AGV by default; Rule 2: When a conflict occurs between load-carrying AGVs: (1) Each load-carrying AGV has a corresponding task generation time. The AGV with the shorter task generation time has a higher priority. (2) When the generation time of the tasks carried by two vehicles is the same, the AGV with the shorter task generation time has a higher priority based on the estimated completion time of the AGV. (3) If the above two situations still cannot be judged, the overall operation objective function is calculated and the obstacle avoidance solution with the optimal objective function is selected.

9. The AGV scheduling strategy for a double-layer production line of a relay core component considering task completion time and energy consumption optimization according to claim 1 is characterized in that: In step S3, the initial path planning is obtained by using the traditional A* algorithm as follows: Initialize the parameters such as the starting node and target node in the production line environment; create an open list and a closed list, and store the starting node to be searched in the open list. The closed list is empty at the beginning of the operation; by continuously updating the environmental parameters, all the nodes to be searched in the open list are searched until the open list is empty and the A* algorithm search ends; calculate the heuristic function f(n) of all nodes after the search, and use the node corresponding to the minimum value of f(n) as the next node to be searched by the A* algorithm, and store it in the closed list, and remove it from the open list at the same time; determine whether the node expanded by the A* algorithm coincides with the target node. If so, it means that the node is the best point for optimization and the algorithm search is successful; if not, repeat the previous step to calculate until the best path is output.

10. The AGV scheduling strategy for a double-layer production line of relay core components considering task completion time and energy consumption optimization according to claim 1 is characterized in that: In step S4, the AGV path is updated according to the dynamic and static conflict avoidance strategy as follows: After the system generates an initial optimal path for each AGV using the A* algorithm, it monitors the next position of each AGV in real time to detect potential static obstacles (such as equipment and shelves) and dynamic conflicts (such as overlapping paths with other AGVs). If a conflict is detected, the dynamic obstacle avoidance mechanism is triggered, and the AGV's driving path is adjusted or a waiting strategy is implemented according to priority rules. Finally, the AGV's global path planning is updated, and the above process is repeated until the task is completed.

Images