A Group Control and Scheduling Method for AGVs in Flexible Production Mode

Abstract

This invention discloses an AGV group control and scheduling method under a flexible production mode. The invention includes: acquiring production information from a matrix manufacturing workshop under a flexible production mode; inputting the production information and optimizer parameters into an optimizer, which searches for the optimal solution to the objective function based on constraints and the objective function, and uses the scheduling scheme corresponding to the optimal solution as the final AGV scheduling scheme; processing the final AGV scheduling scheme to output an AGV material scheduling and delivery information table, a graph showing the change in total energy consumption with the number of iterations, and an AGV operation Gantt chart. This invention solves the problem of excessive energy consumption in the scheduling of flexible production modes in related technologies.

Description

Technical Field

[0001] This invention relates to the field of production scheduling technology, and more particularly to an AGV group control and scheduling method under a flexible production mode. Background Technology

[0002] The widespread application of flexible manufacturing has brought about highly flexible and customized production methods, but it has also placed higher demands on AGV scheduling systems. As a key automated logistics tool in modern manufacturing, AGVs need to respond in real time to task adjustments and path changes in complex and ever-changing production environments to ensure timely material delivery and production safety. However, traditional scheduling methods focus primarily on path optimization and task timeliness, neglecting energy consumption and leading to increased operating costs for enterprises. Therefore, it is necessary to design a reasonable scheduling model to balance task timeliness and energy consumption control to achieve efficient and low-consumption material transportation optimization.

[0003] In flexible production models, the AGV scheduling problem involves the rational allocation, sequence planning, and path optimization of multiple AGVs and material transportation tasks. It is a complex NP-hard problem, and the optimal solution cannot be found through exact methods; therefore, intelligent optimization algorithms are often employed. However, many swarm intelligence algorithms use randomly generated initial populations, resulting in significant uncertainty, and their optimization capabilities vary considerably, easily leading to local optima or deviation from the target. The Artificial Bee Colony Algorithm (ABC), as an optimization method simulating bee foraging behavior, effectively balances exploration and exploitation by combining local search by worker bees and global search by scout bees with information sharing and cooperation mechanisms, avoiding local optima problems. Compared to other intelligent algorithms, ABC exhibits stronger global optimization capabilities and stability, demonstrating significant advantages in solving the AGV scheduling problem.

[0004] Therefore, the technical problem of excessive energy consumption in the scheduling of flexible production modes in related technologies has not yet been effectively solved. Summary of the Invention

[0005] The main objective of this application is to provide an AGV group control and scheduling method under a flexible production mode, so as to at least solve the problem of energy waste caused by the lack of consideration of energy consumption in the scheduling of flexible production modes in related technologies.

[0006] To achieve the above objectives, according to one aspect of this application, an AGV group control and scheduling method under a flexible production mode is provided. The method includes: acquiring production information of the matrix manufacturing workshop under a flexible production mode, including the number of AGVs, operating energy consumption information, speed, coordinates of calling units, material demand time of calling units, and material deadline of calling units; inputting the production information and optimizer parameters into the optimizer, which searches for the optimal solution of the objective function based on constraints and objective function, and uses the scheduling scheme corresponding to the optimal solution as the final AGV scheduling scheme, wherein the constraints include: during scheduling, each calling unit can only be delivered by one AGV once; when delivering materials, the AGV must ensure that each path is continuous; within a production cycle, each AGV departs from the warehouse at most once; the number of AGVs called is less than or equal to the total number of AGVs in the matrix manufacturing workshop; the materials required by each calling unit are delivered within the material deadline of the calling unit; when delivering materials, the amount of materials carried by each AGV must be greater than the amount of materials required by the calling unit; the objective function is to minimize the total energy consumption of AGV operation; and processing the data of the final AGV scheduling scheme to output an AGV material scheduling and delivery information table, a graph of total energy consumption changing with the number of iterations, and an AGV operation Gantt chart.

[0007] Optionally, the order of all calling units is randomly shuffled, and the calling units are assigned to different AGVs in sequence to obtain a task allocation scheme population, wherein one task allocation scheme corresponds to one initial solution, and the task allocation scheme population includes at least one initial solution; the objective function is used as the fitness function, and the fitness values ​​of all initial solutions are calculated to obtain feasible solutions of the objective function, and the feasible solutions constitute a feasible solution group; in step S301, a neighborhood operator is used to generate neighborhood solutions based on feasible solutions, the fitness values ​​of neighborhood solutions are calculated, and a new feasible solution group is obtained, wherein one feasible solution corresponds to one neighborhood solution; Step S302: Arbitrarily select a high-quality position in the new feasible solution group, select at least two feasible solutions as a pre-selected solution group, and compare the pre-selected solutions in the pre-selected solution group with the high-quality solutions. The feasible solution corresponding to the high-quality position is the high-quality solution. Step S303: Check the failure count of each feasible solution. When the failure count of a feasible solution is greater than a specified number of times, reset the feasible solution. Repeat the above steps S301-S303 until the preset convergence condition is met, and output the high-quality solution. The scheduling scheme corresponding to the high-quality solution is the final scheduling scheme of the AGV.

[0008] Optionally, in step S3011, when the fitness value of the neighborhood solution is better than the fitness value of the feasible solution corresponding to the neighborhood solution, the feasible solution in the feasible solution group is replaced with the neighborhood solution to obtain a new feasible solution group; in step S3012, when the neighborhood solution is worse than the feasible solution corresponding to the neighborhood solution, the failure count of the feasible solution is increased by one.

[0009] Optionally, in step S3021, when the fitness value of the preselected solution is better than that of the high-quality solution, the high-quality position is replaced with the position of the preselected solution; in step S3022, when the fitness value of the preselected solution is worse than that of the high-quality solution, the failure count of the feasible solution corresponding to the preselected solution is increased.

[0010] Optionally, calculate the task difficulty index of the task unit. ,in, This represents the task difficulty index. For urgency, For distance, For energy consumption, It is an infinitely small number. for In the calling unit The time required to complete material delivery For calling unit Delivery deadline The first weighting coefficient, This is the second weighting coefficient. The third weighting factor; sort all task units in descending order of task difficulty index, and select the top... The most difficult task units are selected as the priority candidate set; load index parameters are calculated for each AGV, and the calculation formula is as follows: ,in, For load index parameters, for Number of tasks executed for Total running distance, for Total energy consumption As a weighting factor for the task, These are distance-weighted coefficients. The energy consumption weighting coefficient is used to transfer or exchange the task units of the AGVs corresponding to the task units in the candidate set with the AGVs with the lowest load index parameters to obtain the neighborhood solution.

[0011] Optionally, when the AGV with the lowest load index parameter can accept the task unit, the task unit is transferred from the AGV corresponding to the task unit to the AGV with the lowest load index parameter; when the AGV with the lowest load index parameter cannot accept the task unit, the task unit is exchanged between the AGV corresponding to the task unit and the AGV with the lowest load index parameter for tasks that consume energy or have delays.

[0012] Optionally, the fitness values ​​of all solutions in the new feasible solution group are calculated, and at least two candidate solutions are randomly selected from the new feasible solution group as candidate solution groups. From the candidate solution groups, the solution with the better fitness is selected based on its fitness. Probabilistic selection of the optimal solution, with Probabilistically select the suboptimal solution until the pre-selected solution group is filled.

[0013] Optionally, arranging AGVs to deliver materials to calling units includes: recording the current status information of the AGVs, where the status information includes: the current location of each AGV, all currently unfinished calling unit tasks, the material requirements of each calling unit, the earliest delivery time, the latest material delivery completion time, and the current load of each AGV; checking whether all calling unit delivery tasks have been assigned: if all calling unit material delivery tasks have been assigned, the scheduling ends; if the calling unit has tasks to be delivered, extracting all currently unfinished material delivery tasks, calculating the AGV's task scheduling capacity, determining whether the AGV can complete the delivery before the task deadline, and calculating the transportation time from the current location of the AGV to the calling unit; sorting all material delivery tasks according to the earliest completion deadline to obtain a preliminary sort of material delivery tasks; based on the preliminary sort, traversing all AGVs for each material delivery task to obtain an AGV group, and selecting the AGV with the lowest cost for performing the material delivery task from the AGV group to assign each material delivery task.

[0014] Optionally, when the AGV travels to the calling unit, the remaining load of the AGV is updated, and the energy consumption of the AGV during the execution of the task is calculated; when the current AGV's task is completed, the position of the AGV and the available time of the AGV are updated.

[0015] Optionally, when there are still unfinished tasks, new tasks are assigned to the AGV; when all tasks are completed, the scheduling process is terminated and the scheduling scheme is output.

[0016] This application employs the following steps: First, production information from the matrix manufacturing workshop under a flexible production model is obtained. This production information includes the number of AGVs, operating energy consumption, speed, call unit coordinates, call unit material demand time, and call unit material deadline. Second, the production information and optimizer parameters are input into the optimizer. The optimizer searches for the optimal solution to the objective function based on constraints and the objective function, and uses the scheduling scheme corresponding to the optimal solution as the final AGV scheduling scheme. The constraints include: during scheduling, each call unit can only be delivered once by one AGV; when delivering materials, each AGV must ensure that its path is continuous; within a production cycle, each... Each AGV departs from the warehouse at most once; the number of AGVs called is less than or equal to the total number of AGVs in the matrix manufacturing workshop; the materials required by each calling unit are delivered within the material deadline of the calling unit; when delivering materials, the amount of materials carried by each AGV must be greater than the amount of materials required by the calling unit; the objective function is to minimize the total energy consumption of AGV operation; the final AGV scheduling scheme is processed to output an AGV material scheduling and delivery information table, a graph of total energy consumption changing with the number of iterations, and an AGV operation Gantt chart, which solves the problem of excessive energy consumption in the scheduling of flexible production modes in related technologies, thereby achieving the effect of controlling energy consumption in the production process under flexible production modes. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a flowchart of an AGV group control and scheduling method under a flexible production mode provided in an embodiment of this application;

[0019] Figure 2 This is a graph showing the total energy consumption as a function of the number of iterations.

[0020] Figure 3 Create a Gantt chart for AGV operation. Detailed Implementation

[0021] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0022] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0023] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0024] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0025] Figure 1 This is a flowchart illustrating an AGV group control and scheduling method under a flexible production mode according to an embodiment of this application. Figure 1 As shown, the method includes the following steps:

[0026] Step S101: Obtain production information of the matrix manufacturing workshop under the flexible production mode. The production information includes the number of AGVs, operating energy consumption information, speed, coordinates of the calling unit, material demand time of the calling unit, and material deadline of the calling unit.

[0027] Step S102: Input production information and optimizer parameters into the optimizer. The optimizer searches for the optimal solution to the objective function based on the constraints and objective function, and uses the scheduling scheme corresponding to the optimal solution as the final AGV scheduling scheme. The constraints include: during scheduling, each calling unit can only be delivered by one AGV once; when delivering materials, each AGV must ensure that its path is continuous; within a production cycle, each AGV can depart from the warehouse at most once; the number of AGVs called is less than or equal to the total number of AGVs in the matrix manufacturing workshop; the materials required for each calling unit are delivered within the material deadline of the calling unit; when delivering materials, the amount of materials carried by each AGV must be greater than the amount of materials required by the calling unit; the objective function is to minimize the total energy consumption of AGV operation.

[0028] Specifically, when a production unit in the matrix manufacturing workshop lacks materials and awaits replenishment by AGVs, its status changes to "Calling Unit." At a designated time, the control system dispatches AGVs to transport the materials to the calling unit. The AGVs depart from the warehouse and unload at the designated calling unit. Based on the material demand information of the calling unit, the control system schedules AGVs to transport materials. After the AGVs deliver the materials from the warehouse to the designated calling units, the control system arranges logistics transportation tasks based on the AGV's operational status until the AGVs have delivered to all calling units.

[0029] The objective function is to minimize the total energy consumption of AGV operation. The mathematical model of the objective function is as follows:

[0030] express The Manhattan distance it travels from call unit i to call unit j In the formula ( , )and( , ) represents the matrix manufacturing workshop call unit and The coordinates of the location;

[0031] express Runtime from call unit i to call unit j ;

[0032] express The time it takes for call unit j to complete material delivery , express Just left the warehouse. for The time from departure from the warehouse to completion of material delivery in the first call unit j. express The time from departure from the warehouse to completion of material delivery in call unit j. express The time it takes for material delivery to be completed in call unit i;

[0033] express Load before unloading at call unit j Its calculated value is The weight before unloading at call unit i is reduced by the amount of material q unloaded by call unit i. express Load before unloading at call unit i;

[0034] express The energy consumption for moving from call unit i to call unit j is calculated as energy consumption per unit time multiplied by the running time. Among them, energy consumption per unit time is from The load energy consumption coefficient is related to its load before unloading at call unit j. The product of these factors, plus the idle energy consumption of the AGV, constitutes the total energy consumption. Energy consumption per unit of AGV under no-load conditions.

[0035] The objective function is to minimize the energy consumption of AGVs in delivering materials within a production cycle. Integer decision variables, if Along the edge Move, then =1; otherwise, =0.

[0036] The constraints are as follows:

[0037] and This means that during scheduling, each calling unit can only be delivered once by one AGV. Represents the decision variables from call unit i to call unit j;

[0038] This means that AGVs must ensure that each of their routes is continuous when delivering materials. Represents the decision variable from call unit j to call unit i;

[0039] This means that within a production cycle, each AGV can depart from the warehouse a maximum of once. Represents any Decision variables from warehouse to call unit i;

[0040] This indicates that the number of AGVs called is not greater than the total number of AGVs in the matrix manufacturing workshop;

[0041] This means that, based on the material demand information of each calling unit, the dispatching system must ensure that the materials required by each calling unit can be delivered within its deadline when dispatching AGVs to deliver materials, thus ensuring timely delivery. This indicates the time when calling unit j issues a material request. In the calling unit The time required to complete material delivery Indicates the deadline for delivery of the material required by call unit j;

[0042] This means that when the dispatching system arranges AGVs to deliver materials to the calling unit, the amount of materials carried by each AGV must be greater than the amount of materials required by the calling unit.

[0043] This indicates that a constraint has been imposed on the decision variable, i.e., if Passing by ,but = 1, otherwise 0.

[0044] Parameter description: and The production unit is numbered. If the production unit needs material replenishment, it will be converted into a calling unit. Therefore, if production unit i needs material replenishment, it can also be represented by calling unit i. This refers to the number of production units; For call time (production unit) (Signal transmission time) For calling unit Delivery deadline; For AGV loading and unloading time; The AGV's operating speed; The quantity of goods to be unloaded from the calling unit; The initial load of the AGV; Energy consumption per unit of AGV under no-load conditions; A collection of production units. D represents the warehouse, and D = {0}. for , It is a collection of production units and warehouses; for A collection of AGVs, B is the set of edges. ; The AGV load capacity energy consumption coefficient;

[0045] Step S103: Process the data of the final AGV scheduling scheme and output the AGV material scheduling and delivery information table, the total energy consumption change graph with the number of iterations, and the AGV operation Gantt chart.

[0046] This invention designs an AGV scheduling method that considers the timely delivery of materials, aiming to reduce the energy consumption of AGVs transporting materials to workstations. The efficient scheduling method proposed in this invention can optimize AGV scheduling efficiency, thereby improving the reliability and stability of the AGV scheduling system. By optimizing scheduling, the energy consumption of AGVs is reduced, thus lowering equipment wear and failure rates. Reducing AGV energy consumption not only helps reduce enterprise energy consumption and operating costs, but also optimizes resource utilization and equipment lifespan, reducing maintenance and replacement costs, thereby enhancing the reliability and stability of the entire AGV scheduling system.

[0047] After initializing the parameters, task encoding is performed, and multiple initial scheduling schemes are randomly generated to form an initial population. Based on this, a bee colony algorithm is used for optimization search according to the AGV path planning and energy consumption optimization objectives to effectively narrow the search space. Subsequently, the energy consumption level of each scheduling scheme is calculated, and the scheme with lower energy consumption (better fitness) is determined. Using the scheme with lower energy consumption as the basis, a neighborhood search is performed in the hired bee phase to generate new solutions and evaluate their energy consumption, selecting the scheduling scheme with lower energy consumption. In the observation bee phase, excellent schemes are selected based on fitness probability to further optimize task scheduling. In the scout bee phase, schemes that have not been optimized for a long time are reset to avoid getting trapped in local optima. Finally, it is determined whether the optimization termination condition is met. If it is, the optimal AGV scheduling scheme is output and the operation stops; otherwise, the calculation continues in a loop until the maximum number of iterations is reached or the optimized scheme remains stable. The AGV scheduling task is encoded according to its number to ensure that AGV transportation for the same task follows the optimal path as much as possible and reduces energy consumption, thereby improving overall scheduling efficiency.

[0048] In an optional embodiment, the order of all calling units is randomly shuffled, and the calling units are sequentially assigned to different AGVs to obtain a task allocation scheme population, wherein one task allocation scheme corresponds to one initial solution, and the task allocation scheme population includes at least one initial solution; the objective function is used as the fitness function, and the fitness values ​​of all initial solutions are calculated to obtain feasible solutions of the objective function, which constitute a feasible solution group; in step S301, a neighborhood operator is used to generate neighborhood solutions based on feasible solutions, and the fitness values ​​of the neighborhood solutions are calculated to obtain a new feasible solution group, wherein one feasible solution corresponds to one initial solution. Neighborhood solution; Step S302, arbitrarily select a high-quality position in the new feasible solution group, select at least two feasible solutions in the new feasible solution group as a pre-selected solution group, compare the pre-selected solutions in the pre-selected solution group with the high-quality solutions, wherein the feasible solution corresponding to the high-quality position is the high-quality solution; Step S303, check the failure count of each feasible solution, when the failure count of a feasible solution is greater than a specified number of times, reset the feasible solution; Repeat the above steps S301-S303 until the preset convergence condition is met, output the high-quality solution, and the scheduling scheme corresponding to the high-quality solution is the final scheduling scheme of AGV.

[0049] Specifically, this invention uses the bee colony algorithm for population optimization. Before applying the bee colony algorithm, it is necessary to first clarify data such as the size of the matrix manufacturing workshop, the number of AGVs, the number of calling units, the material calling time of the calling units, and the material delivery time of the material, and determine the objective function and constraints, and encode the initial data.

[0050] Hired Bee Phase: Each hired bee corresponds to a current candidate solution. Using an improved neighborhood operator, a neighborhood solution is generated for the current initial solution. The fitness value of this neighborhood solution is calculated. If it is better than the original solution, the neighborhood solution replaces the current solution; otherwise, the failure count of the current solution is incremented by one.

[0051] Observation Bee Phase: Based on the fitness values ​​of the candidate solutions obtained in the previous phase, a tournament selection strategy is used. That is, in each round, several candidate solutions (e.g., 4-5) are randomly selected from the population, and the one with the better fitness value is chosen as the observation bee's target. In this phase, the selected individual is no longer subjected to neighborhood perturbation; instead, its current fitness is directly compared to the fitness of its original position. If the selected solution is better, it replaces the current solution and the failure count is reset; otherwise, the original solution remains unchanged, and its failure count is incremented by one.

[0052] Scout bee phase: Iterates through the failure counts of all solutions. If the failure count of a solution exceeds a set threshold (such as LIMIT), the solution is determined to be trapped in a local optimum, discarded, and a new solution is randomly generated to replace it. This is intended to enhance population diversity and avoid premature convergence in the search.

[0053] Optimal Solution Recording and Convergence Judgment: In each iteration, the current optimal solution is continuously tracked, and its AGV path scheme and fitness are recorded. If the maximum number of iterations is met, the algorithm terminates, and the current optimal AGV scheduling solution is output as the final result.

[0054] Population initialization is a crucial step in the bee colony algorithm for AGV scheduling optimization. Its core objective is to generate a set of candidate solutions (food sources), each representing an AGV task allocation scheme. First, the order of all calling units is randomly shuffled to increase solution diversity. Then, a round-robin allocation strategy is used to sequentially assign calling units to different AGVs, ensuring a balanced task distribution and avoiding extreme situations where some AGVs are overloaded or have no tasks. Ultimately, n distinct initial solutions are generated.

[0055] To prevent the algorithm from getting stuck in local optima, avoid search stagnation, and enhance search diversity, a new solution is randomly generated again. Fitness calculation, neighborhood solution generation, and solution selection are repeated for the new generation population until the pre-set convergence condition is met. The optimal solution is then output as the final AGV scheduling decision.

[0056] In an optional embodiment, in step S3011, when the fitness value of the neighborhood solution is better than the fitness value of the feasible solution corresponding to the neighborhood solution, the feasible solution in the feasible solution group is replaced with the neighborhood solution to obtain a new feasible solution group; in step S3012, when the neighborhood solution is worse than the feasible solution corresponding to the neighborhood solution, the failure count of the feasible solution is increased by one.

[0057] Specifically, for each high-difficulty task in the candidate task set: find its current AGV route (denoted as...). ) and the AGV with the lowest load index (denoted as Perform a task transfer (or exchange) if If the task can be accepted (without violating capacity / time window constraints), then the task will be moved from... Transferred to If direct transfer is not feasible, then in and When swapping tasks, prioritize task pairs that significantly improve energy consumption or latency. After this priority transfer / swap operation, a new solution is obtained as a neighborhood solution and returned for use in the improved bee colony algorithm.

[0058] In an optional embodiment, step S3021, when the fitness value of the preselected solution is better than that of the high-quality solution, the high-quality position is replaced with the position of the preselected solution; step S3022, when the fitness value of the preselected solution is worse than that of the high-quality solution, the failure count of the feasible solution corresponding to the preselected solution is increased.

[0059] In one alternative embodiment, a task difficulty index for the task unit is calculated. ,in, This represents the task difficulty index. For urgency, For distance, For energy consumption, It is an infinitely small number. for In the calling unit The time required to complete material delivery For calling unit Delivery deadline The first weighting coefficient, This is the second weighting coefficient. The third weighting factor; sort all task units in descending order of task difficulty index, and select the top... The most difficult task units are selected as the priority candidate set; load index parameters are calculated for each AGV, and the calculation formula is as follows: ,in, For load index parameters, for Number of tasks executed for Total running distance, for Total energy consumption As a weighting factor for the task, These are distance-weighted coefficients. The energy consumption weighting coefficient is used to transfer or exchange the task units of the AGVs corresponding to the task units in the candidate set with the AGVs with the lowest load index parameters to obtain the neighborhood solution.

[0060] In one optional embodiment, when the AGV with the lowest load index parameter can accept the task unit, the task unit is transferred from the AGV corresponding to the task unit to the AGV with the lowest load index parameter; when the AGV with the lowest load index parameter cannot accept the task unit, the task unit is exchanged between the AGV corresponding to the task unit and the AGV with the lowest load index parameter for tasks that consume energy or have delays.

[0061] In one optional embodiment, the fitness values ​​of all solutions in the new feasible solution group are calculated, at least two candidate solutions are randomly selected from the new feasible solution group as candidate solution groups, and the solution with better fitness is selected from the candidate solution groups based on its fitness. Probabilistic selection of the optimal solution, with Probabilistically select the suboptimal solution until the pre-selected solution group is filled.

[0062] Specifically, the fitness of each solution is calculated. Since the optimization objective is to minimize the total energy consumption, the smaller the fitness value, the higher the quality of the solution (i.e., the lower the energy consumption).

[0063] Calculate the selection probability of a solution, randomly select T candidate solutions from the population, and define the better individual based on fitness. Human definition ,by Probabilistic selection of the optimal solution, with Probabilistic selection of the suboptimal solution.

[0064] Directly selecting the optimal value can lead to excessively rapid convergence and loss of population diversity; probability These are parameters that are manually adjusted; the goal of the observation bee stage is not to find the global optimal solution, but to select a "potential" solution to replace the current bee position. The role of the suboptimal solution is to preserve diversity, activate potential solutions, increase the search width, and help to escape local optima.

[0065] In one optional embodiment, arranging AGVs to deliver materials to calling units includes: recording the current status information of the AGVs, wherein the status information includes: the current location of each AGV, all currently unfinished calling unit tasks, the material requirements of each calling unit, the earliest delivery time, the latest material delivery completion time, and the current load of each AGV; checking whether all calling unit delivery tasks have been assigned: if all calling unit material delivery tasks have been assigned, the scheduling ends; if the calling unit has tasks to be delivered, extracting all currently unfinished material delivery tasks, calculating the task scheduling capacity of the AGVs, determining whether the AGVs can complete the delivery before the task deadline, and calculating the transportation time from the current location of the AGV to the calling unit; sorting all material delivery tasks according to the earliest completion deadline to obtain a preliminary sort of material delivery tasks; traversing all AGVs for each material delivery task according to the preliminary sort to obtain an AGV group, and selecting the AGV with the lowest cost for performing the material delivery task from the AGV group to assign each material delivery task.

[0066] In one alternative embodiment, the AGV travels to the calling unit, updates the AGV's remaining load, and calculates the AGV's energy consumption during task execution; when the current AGV's task is completed, the AGV's position and available time are updated.

[0067] In one alternative embodiment, when there are still unfinished tasks, new tasks are continuously assigned to the AGV; when all tasks are completed, the scheduling process is terminated and the scheduling scheme is output.

[0068] This invention offers the following technical advantages: It optimizes AGV task scheduling based on a bee colony algorithm. The allocation results of AGV tasks are processed, integrating the demands of the calling units to be delivered into input data, including task time, material requirements, and AGV operating parameters, which are then input into the scheduling optimization module. The optimization process uses a bee colony algorithm to search for the optimal solution based on preset energy consumption constraints, path planning objectives, and time window limitations, outputting the optimal AGV delivery plan. The optimized scheduling results are processed to form the final AGV operating path. Through the global optimization characteristics of the bee colony algorithm, the order of AGV tasks is dynamically adjusted, making the scheduling of AGV tasks on different paths more reasonable, reducing AGV idle running time, and optimizing path selection to minimize overall AGV operating energy consumption. Ultimately, users can formulate more efficient AGV operation plans based on the optimized AGV scheduling results, improving logistics and delivery efficiency and shortening production logistics cycles.

[0069] In another embodiment:

[0070] The algorithm parameters are set as follows: population size is 50, representing the number of initial solutions (food sources) in the bee colony algorithm; maximum number of iterations is 100, controlling the number of rounds the bee colony algorithm runs and affecting the optimization degree of the final solution; failure limit is 100, setting the upper limit on the number of times a scout bee fails to trigger a new solution in the bee colony algorithm, avoiding getting trapped in local optima; the ratio of the three types of bees in the bee colony algorithm is also specified. Setting it to 1 controls the optimization capabilities of the mercenary bees, observer bees, and scout bees, making the generation of new solutions more stable and more exploratory; The flexible production workshop is defined as a 5x5 matrix manufacturing workshop with 3 AGVs. Within a 10-40 unit time period, 15 production units are randomly converted into calling units. Each calling unit randomly generates a material calling time and a material delivery deadline, with the deadline being the calling time plus 5-10 seconds randomly. The AGV operating parameters are set as follows: AGV running speed v is 1 unit / unit time, and AGV loading / unloading time... Energy consumption of an AGV per unit of time when it is not in use The AGV load energy consumption coefficient is 1, the maximum load Q of the AGV is 100, and the amount of material delivered by the AGV each time is 10.

[0071] Results and Analysis: After running the bee colony algorithm, an AGV scheduling scheme was generated based on the optimization results, and an AGV operation Gantt chart was plotted. This optimization process clearly shows the execution order, completion time, and energy cost of each AGV task. The optimized scheduling scheme mainly improves the task allocation order, making the AGV paths more reasonable in material delivery, and optimizes AGV running time, reducing task waiting time. Furthermore, it reasonably adjusts the AGV task load to avoid overloading or idling some AGVs, improving overall scheduling efficiency. Users can use the optimization results to formulate more reasonable AGV running paths, optimize workshop logistics, improve material delivery efficiency, and shorten task execution cycles. This invention can provide some reference value for the energy consumption optimization problem of multi-AGV collaborative scheduling.

[0072] Based on the above scheduling optimization conditions, 15 call units requiring material delivery are generated within a production cycle. The call unit numbers involved in the material delivery tasks are (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15). The scheduling system arranges for these tasks to be completed jointly by 3 AGVs, which need to travel to the 15 different call units to deliver materials. During task execution, the AGVs run on the set grid paths, and the AGVs responsible for delivery are numbered AGV1, AGV2, and AGV3 in sequence. The scheduling scheme ensures that each AGV completes all tasks within a limited time window, minimizes resource conflicts and energy consumption, and improves overall logistics efficiency.

[0073] The algorithm was run using the above data to obtain a graph showing the total energy consumption (TEC) as a function of the number of iterations. The results are as follows: Figure 2 As shown in the figure, the horizontal axis represents the number of iterations, the vertical axis represents the total energy consumption, and the blue broken line represents the TEC value of each iteration point marked with dots. The graph shows that the total energy consumption gradually decreases with the increase of the number of iterations, indicating that the algorithm continuously optimizes the AGV scheduling scheme, reducing the total energy consumption of AGV operation. Early stage (generations 0-20): Energy consumption decreases the fastest, indicating that the algorithm quickly finds a relatively optimal solution in the initial search phase; Mid-stage (generations 20-60): The decreasing trend slows down, local optimization proceeds gradually, and the quality of the solution gradually improves; Late stage (generations 60-100): Energy consumption tends to stabilize, indicating that the algorithm is close to convergence, and the space for further optimization is limited. The overall optimization trend is that the optimization speed is fast in the early stage, gradually optimizes in the mid-stage, and tends to stabilize in the late stage, which is consistent with the typical convergence characteristics of intelligent optimization algorithms, indicating that the optimization process successfully reduces the energy consumption of AGV operation and improves logistics and distribution efficiency.

[0074] The above data is input, the algorithm is run, and the AGV material delivery scheduling is arranged according to the optimized results. The results are shown in Table 1. In the table, the AGV number column is the AGV vehicle assigned to perform the task, the calling unit (coordinate) column is the calling unit number and target location (two-dimensional coordinates) of the material delivery, the AGV task start time column is the time when the AGV starts executing the task from the initial position, the AGV arrival time column is the time when the AGV actually arrives at the calling unit, the material delivery time column is the time when the material delivery is completed, the calling time column is the time when the material delivery task is triggered, that is, the time when the production unit requests materials, and the deadline column is the latest completion time of the material delivery task. Taking AGV1 being dispatched to calling unit 11 (2,4) to perform the material delivery task as an example, calling unit 11 (2,4) requests material delivery at 16 o'clock. The task scheduling system receives the request and selects AGV1 for scheduling based on AGV availability and path conditions. AGV1 receives the task assignment at 13 o'clock. Although the calling unit's request time is 16 o'clock, the AGV may plan the path in advance to improve the overall scheduling efficiency. AGV 1 starts from its initial position (1, 3), travels a certain distance, and finally arrives at calling unit 11 (2, 4) at time 15. After arriving at time 15, AGV 1 begins unloading materials and completes the material delivery at time 16. The calling time for this task is 16, and the AGV completes the delivery at time 16, meaning the task is completed on time without any delay. The deadline is 22, indicating that the task was successfully completed within an acceptable time window. Subsequent AGV scheduling follows the same pattern, thus clearly showing which calling unit each AGV is responsible for delivering materials to, the delivery sequence of each material delivery task, and the time when materials arrive at the calling unit.

[0075] Table 1 shows:

[0076]

[0077] The production scheduling Gantt chart obtained based on the above scheduling arrangement is as follows: Figure 3 As shown, the horizontal axis represents the time span of task execution, the vertical axis represents the AGV device number, and the colored rectangles represent the tasks performed by the AGV within a certain time period. The colors distinguish different calling unit (CU) tasks. The diagram clearly illustrates the processing sequence of each material delivery task and the arrangement of each AGV. For example, AGV1 delivers calling units 2, 11, 14, 3, and 1, with AGV1's running time ranging from 5 to 30 seconds. The scheduling of calling units for other AGVs responsible for material delivery follows the same logic. This clearly demonstrates the delivery sequence of each material delivery task and the time it takes for materials to arrive at the calling units.

[0078] Table 2 presents the experimental comparison of the algorithms for 20 cases. The first column of the table gives the name of the case, the second column gives the best value obtained by the algorithm, the next five columns show the best target value obtained by the five algorithms for each case, and the last five columns give the on-time delivery rate of materials in each case.

[0079] Table 2 Comparison of Experimental Results:

[0080]

[0081] As shown in Table 2, the improved ABC algorithm proposed in this invention performs well in solving the 20 extended examples:

[0082] 1. It obtained 20 optimal values, which is significantly better than other comparison algorithms.

[0083] 2. The improved ABC algorithm has a significantly higher on-time delivery rate for materials than other algorithms.

[0084] In summary, this invention proposes an improved Artificial Bee Colony (ABC) algorithm, which achieves global optimization of AGV scheduling by introducing a multi-agent cooperative mechanism. In the hired bee phase, an adaptive neighborhood transfer operator based on task execution difficulty and the current load of the AGV is designed, which can dynamically generate high-quality local solutions. In the observation bee phase, an improved selection mechanism is introduced, which retains excellent solutions while improving the overall search efficiency and algorithm stability of the population. Experimental results show that, compared with traditional algorithms, the scheduling method proposed in this invention has significant advantages in reducing total system energy consumption, improving on-time material delivery rate, and reducing empty loads and path duplication. It can achieve efficient collaborative operation of multiple AGVs and improve the overall operating efficiency of the workshop logistics scheduling system under flexible production mode.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An AGV group control and scheduling method under a flexible production mode, characterized in that, include: Obtain production information of the matrix manufacturing workshop under the flexible production mode, wherein the production information includes the number of AGVs, operating energy consumption information, speed, coordinates of the calling unit, material demand time of the calling unit, and material deadline of the calling unit; The production information and optimizer parameters are input into the optimizer. The optimizer searches for the optimal solution to the objective function based on constraints and the objective function, and uses the scheduling scheme corresponding to the optimal solution as the final AGV scheduling scheme. The constraints include: during scheduling, each calling unit can only be delivered by one AGV once; when delivering materials, each AGV must ensure that its path is continuous; within a production cycle, each AGV departs from the warehouse at most once; the number of AGVs called is less than or equal to the total number of AGVs in the matrix manufacturing workshop; the materials required for each calling unit must be delivered within the material deadline of the calling unit; when delivering materials, the amount of materials carried by each AGV must be greater than the amount of materials required by the calling unit; and the objective function is to minimize the total energy consumption of AGV operation. The production information and optimizer parameters are input into the optimizer. The optimizer searches for the optimal solution of the objective function based on the constraints and objective function, and uses the scheduling scheme corresponding to the optimal solution as the final AGV scheduling scheme. This includes: randomly shuffling the order of all the calling units and sequentially assigning the calling units to different AGVs to obtain a task allocation scheme population, wherein one task allocation scheme corresponds to one initial solution, and the task allocation scheme population includes at least one initial solution; using the objective function as the fitness function, calculating the fitness values ​​of all the initial solutions to obtain feasible solutions of the objective function, and the feasible solutions constitute a feasible solution group; step S301, using a neighborhood operator to generate neighborhood solutions based on the feasible solutions, and calculating the... Step S302: The fitness value of the neighborhood solutions is used to obtain a new feasible solution group, wherein one feasible solution corresponds to one neighborhood solution; Step S303: A high-quality position is arbitrarily selected in the new feasible solution group, and at least two feasible solutions are selected as pre-selected solutions. The pre-selected solutions in the pre-selected solution group are compared with the high-quality solutions, wherein the feasible solution corresponding to the high-quality position is the high-quality solution; Step S304: The failure count of each feasible solution is checked. When the failure count of the feasible solution is greater than a predetermined number, the feasible solution is reset; Steps S305-S306 are repeated until a preset convergence condition is met, and the high-quality solution is output. The scheduling scheme corresponding to the high-quality solution is the final scheduling scheme of the AGV. Step S301 includes: calculating the task difficulty index of the task unit. ,in, This represents the difficulty index of the task. For urgency, For distance, For energy consumption, It is an infinitely small number. for In the calling unit The time required to complete material delivery For calling unit Delivery deadline The first weighting coefficient, This is the second weighting coefficient. The third weighting coefficient is used; all task units are sorted in descending order according to the task difficulty index, and the top ones are selected. The most difficult task units are selected as the priority candidate set; for each AGV, the load index parameters are calculated using the following formula: ,in, The load index parameters, for Number of tasks executed for Total running distance, for Total energy consumption As a weighting factor for the task, These are distance-weighted coefficients. The energy consumption weighting coefficient is used to transfer or exchange the task unit corresponding to the task unit in the candidate set with the task unit of the AGV with the lowest load index parameter to obtain the neighborhood solution. The final AGV scheduling scheme is processed to output an AGV material scheduling and delivery information table, a graph showing the change in total energy consumption with the number of iterations, and an AGV operation Gantt chart.

2. The method according to claim 1, characterized in that, Step S301 includes: Step S3011: When the fitness value of the neighborhood solution is better than the fitness value of the feasible solution corresponding to the neighborhood solution, the feasible solution in the feasible solution group is replaced with the neighborhood solution to obtain the new feasible solution group; Step S3012: When the neighborhood solution is worse than the feasible solution corresponding to the neighborhood solution, increment the failure count of the feasible solution by one.

3. The method according to claim 1, characterized in that, Step S302 includes: Step S3021: When the fitness value of the preselected solution is better than that of the high-quality solution, the high-quality position is replaced with the position of the preselected solution. Step S3022: When the fitness value of the preselected solution is worse than that of the high-quality solution, increase the failure count of the feasible solution corresponding to the preselected solution.

4. The method according to claim 1, characterized in that, The neighborhood solution is obtained by transferring or exchanging the AGV corresponding to the task unit in the candidate set with the AGV with the lowest load index parameter, including: When the AGV with the lowest load index parameter can accept the task unit, the task unit is transferred from the AGV corresponding to the task unit to the AGV with the lowest load index parameter. When the AGV with the lowest load index parameter cannot accept the task unit, the AGV corresponding to the task unit is swapped with the AGV with the lowest load index parameter for tasks that have energy consumption or delay.

5. The method according to claim 1, characterized in that, Step S302 includes: Calculate the fitness values ​​of all solutions in the new feasible solution group, randomly select at least two candidate solutions from the new feasible solution group as candidate solution groups, and select the solution with better fitness from the candidate solution groups. Probabilistic selection of the optimal solution, with The suboptimal solution is selected probabilistically until the pre-selected solution group is filled.

6. The method according to claim 1, characterized in that, Using the objective function as the fitness function, the fitness values ​​of all the initial solutions are calculated to obtain feasible solutions to the objective function. These feasible solutions constitute a feasible solution group, including: Arranging AGVs to deliver materials to the calling unit includes: Record the status information of the AGV at the current moment, wherein the status information includes: the current location of each AGV, all unfinished call unit tasks, the material requirements of each call unit, the earliest delivery time, the latest material delivery completion time, and the current load of each AGV; Check whether all delivery tasks of the calling units have been assigned: if all material delivery tasks of the calling units have been assigned, end the scheduling; if the calling unit has a delivery task to be delivered, extract all the currently unfinished material delivery tasks, calculate the task scheduling capacity of the AGV, determine whether the AGV can complete the delivery before the task deadline, and calculate the transportation time from the current position of the AGV to the calling unit. Obtain a preliminary sort of the material delivery tasks by sorting all the material delivery tasks according to the earliest deadline for completion of material delivery; Based on the preliminary sorting, all AGVs are traversed for each material delivery task to obtain an AGV group. In the AGV group, the AGV with the lowest cost for performing the material delivery task is selected and assigned to each material delivery task.

7. The method according to claim 6, characterized in that, AGVs will be arranged to deliver materials to the calling unit, including: The AGV travels to the calling unit, updates the remaining load of the AGV, and calculates the energy consumption of the AGV during the task execution process; When the current AGV completes its task, update the AGV's position and its available time.

8. The method according to claim 7, characterized in that, When the current AGV completes its task, update the AGV's position and its available time, including: When there are still unfinished tasks, new tasks will be assigned to the AGV. Once all tasks are completed, the scheduling process terminates and the scheduling scheme is output.

Images