AGV group control scheduling method in flexible production mode

Abstract

The invention discloses an AGV group control scheduling method in a flexible production mode. The method comprises the following steps: acquiring production information of a matrix manufacturing workshop in a flexible production mode; the production information and optimizer parameters are input into an optimizer, the optimizer searches an optimal solution of the target function according to the constraint conditions and the target function, and a scheduling scheme corresponding to the optimal solution serves as a final AGV scheduling scheme; and carrying out data processing on the AGV final scheduling scheme, and outputting an AGV material scheduling distribution information table, a total energy consumption change graph along with the number of iterations and an AGV operation Gantt chart. According to the invention, the problem of excessive energy consumption in the scheduling of the flexible production mode in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of production scheduling. In particular, it relates to a method for AGV group control scheduling under a flexible production mode. Background Art

[0002] The widespread application of flexible production modes has brought about a highly flexible and customized production method, but it has also posed higher requirements for the AGV scheduling system. As a key automated logistics tool in modern manufacturing, AGVs need to respond to task adjustments and path changes in real time in a complex and changeable production environment to ensure the on-time delivery of materials and production safety. However, traditional scheduling methods mostly focus on path optimization and task timeliness, ignoring the issue of energy consumption, resulting in an increase in enterprise operating costs. 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] Under the flexible production mode, the AGV scheduling problem involves the reasonable allocation, sequence planning, and path optimization of multiple AGV vehicles and material transportation tasks, which belongs to a complex NP-hard problem and cannot be solved by an exact method to obtain the optimal solution. Therefore, intelligent optimization algorithms are often used. However, the initial populations of many swarm intelligence algorithms are randomly generated, with great uncertainty, and the optimization capabilities of the algorithms vary significantly, easily falling into local optima or deviating from the target. As an optimization method that simulates the foraging behavior of bees, the Artificial Bee Colony Algorithm (ABC) can effectively balance exploration and exploitation and avoid local optimum problems through the local search of worker bees and the global search of scout bees, combined with an information sharing and cooperation mechanism. Compared with other intelligent algorithms, ABC has stronger global optimization capabilities and stability, showing significant advantages in solving AGV scheduling problems.

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

[0005] The main purpose of this application is to provide a method for AGV group control scheduling under a flexible production mode 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 object, according to one aspect of the present application, a method for AGV group control scheduling in a flexible production mode is provided. The method includes: obtaining the production information of a matrix manufacturing workshop in a flexible production mode, where the production information includes the number of AGVs, running energy consumption information, speed, call unit coordinates, call unit material requirement time, and call unit material deadline; inputting the production information and optimizer parameters into an optimizer, and the optimizer searches for the optimal solution of the objective function according to the constraint conditions and the objective function, and takes the scheduling plan corresponding to the optimal solution as the final AGV scheduling plan, where the constraint conditions include: when scheduling, each call unit can only be delivered by one AGV once; when an AGV distributes materials, it is necessary to ensure that each of its paths is continuous; within one production cycle, each AGV can start 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 required materials of each call unit are delivered within the call unit material deadline; when distributing materials, the quantity of materials carried by each AGV needs to be greater than the call unit material requirement quantity; the objective function is the minimum total running energy consumption of AGVs; performing data processing on the final AGV scheduling plan, and outputting an AGV material scheduling and distribution information table, a graph of the total energy consumption varying with the number of iterations, and an AGV operation Gantt chart.

[0007] Optionally, randomly shuffle the order of all call units, and sequentially assign the call units to different AGVs to obtain a population of task assignment plans, where one task assignment plan corresponds to one initial solution, and the population of task assignment plans includes at least one initial solution; use the objective function as the fitness function, calculate the fitness values of all initial solutions to obtain the feasible solutions of the objective function, and the feasible solutions form a feasible solution group; step S301, use a neighborhood operator to generate neighborhood solutions based on the feasible solutions, calculate the fitness values of the neighborhood solutions, and obtain a new feasible solution group, where 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 in the new feasible solution group as a preselected solution group, and compare the preselected solutions in the preselected solution group with the high-quality solution, where the feasible solution corresponding to the high-quality position is the high-quality solution; step S303, check the failure count of each feasible solution, and when the failure count of the feasible solution is greater than the specified number of times, reset the feasible solution; repeat the above steps S301 - step S303 until the preset convergence condition is met, and then output the high-quality solution, and the scheduling plan corresponding to the high-quality solution is the final AGV scheduling plan.

[0008] Optionally, 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, replace the feasible solution in the feasible solution group with the neighborhood solution to obtain a new feasible solution group; step S3012, when the neighborhood solution is inferior to the feasible solution corresponding to the neighborhood solution, increase the failure count of the feasible solution by one.

[0009] Optionally, in step S3021, when the fitness value of the preselected solution is better than that of the high-quality solution, replace the high-quality position with the position where the preselected solution is located; in 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.

[0010] Optionally, calculate the task difficulty index of the task unit , where is the task difficulty index, is the urgency, is the distance, is the energy consumption, is an infinitesimal number, is the time to complete material distribution in the calling unit , is the delivery deadline of the calling unit , is the first weighting coefficient, is the second weighting coefficient, is the third weighting coefficient; sort all task units in descending order of the task difficulty index, and select the first most difficult task units as the candidate set for priority processing; calculate the load index parameter for each AGV, and the calculation formula is , where is the load index parameter, is the number of tasks executed, is the total running distance of is the total energy consumption of is the task weighting coefficient, is the distance weighting coefficient, is the energy consumption weighting coefficient; transfer or exchange the task units corresponding to the AGVs in the candidate set with the task units of the AGV with the lowest load index parameter to obtain a neighborhood solution.

[0011] Optionally, when the AGV with the lowest load index parameter can accept the task unit, transfer the task unit 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, exchange the tasks with energy consumption or delay between the AGV corresponding to the task unit and the AGV with the lowest load index parameter.

[0012] Optionally, calculate the fitness values of all solutions in the new feasible solution group, randomly select at least two candidate solutions as the candidate solution group in the new feasible solution group, and select the solution with better fitness according to the fitness in the candidate solution group, and select the optimal solution with probability, and with Probabilistically select sub-optimal solutions until the preselected solution group is filled.

[0013] Optionally, arrange for the AGV to perform material distribution for the calling unit, including: recording the status information of the AGV at the current moment, where the status information includes: the current location of each AGV, all current unfinished calling unit tasks, the material requirements of each calling unit, the earliest available delivery time, the latest material delivery completion time, and the current load of each AGV; checking whether all the distribution tasks of all calling units have been assigned: if all the material distribution tasks of all calling units have been assigned, end the scheduling; if there are pending distribution tasks for the calling unit, extract all current unfinished material distribution tasks, calculate the task scheduling ability of the AGV, determine whether the AGV can complete the delivery before the task deadline, and calculate the transportation time from the current location of the AGV to the calling unit; sort all the material distribution tasks in ascending order according to the earliest deadline for completing the material delivery to obtain a preliminary sorting of the material distribution tasks; traverse all AGVs for each material distribution task based on the preliminary sorting to obtain an AGV group, and select the AGV with the lowest cost for executing the material distribution task in the AGV group to assign each material distribution task.

[0014] Optionally, 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; when the current AGV's task is completed, update the location of the AGV and the available time of the AGV.

[0015] Optionally, when there are still unfinished tasks, continue to assign new tasks to the AGV; when all tasks are completed, terminate the scheduling process and output the scheduling plan.

[0016] Through this application, the following steps are adopted: obtaining the production information of the matrix manufacturing workshop in the flexible production mode, where the production information includes the number of AGVs, operating energy consumption information, speed, call unit coordinates, call unit material requirement time, and call unit material deadline; inputting the production information and optimizer parameters into the optimizer, and the optimizer searches for the optimal solution of the objective function according to the constraint conditions and the objective function, and uses the scheduling plan corresponding to the optimal solution as the final AGV scheduling plan, where the constraint conditions include: when scheduling, each call unit can only be delivered by one AGV once; when the AGV distributes materials, it is necessary to ensure that each of its paths is continuous; within a production cycle, each AGV can start 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 required materials for each call unit are delivered within the call unit material deadline; when carrying out material distribution, the amount of materials carried by each AGV needs to be greater than the call unit material requirement quantity; the objective function is the minimum total operating energy consumption of the AGV; performing data processing on the final AGV scheduling plan, and outputting the AGV material scheduling and distribution information table, the graph of the total energy consumption changing with the number of iterations, and the AGV operation Gantt chart, which solves the problem of excessive energy consumption in the scheduling of the flexible production mode in the related art, and thus achieves the effect of controlling the energy consumption in the production process in the flexible production mode. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 is a flowchart of an AGV group control scheduling method in a flexible production mode provided by an embodiment of the present application;

[0019] Figure 2 is a graph of the total energy consumption changing with the number of iterations;

[0020] Figure 3 is an AGV operation Gantt chart. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The following will refer to the drawings and combine the embodiments to detail the present application.

[0022] To enable those skilled in the art to better understand the solution of this application, the technical solution in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the protection scope of this application.

[0023] It should be noted that the terms "first", "second", etc. in the description and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so as to implement the embodiments of this application described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

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

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

[0026] Step S101, obtain the production information of the matrix manufacturing workshop under the flexible production mode, where the production information includes the number of AGVs, operation energy consumption information, speed, call unit coordinates, call unit material requirement time, and call unit material deadline;

[0027] Step S102, input the production information and optimizer parameters into the optimizer. The optimizer searches for the optimal solution of the objective function according to the constraint conditions and the objective function, and uses the scheduling plan corresponding to the optimal solution as the final AGV scheduling plan. The constraint conditions include: when scheduling, each call unit can only be delivered by one AGV once; when an AGV distributes materials, it is necessary to ensure that each of its paths is continuous; within a production cycle, each AGV can start 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 required materials of each call unit are delivered within the call unit material deadline; when distributing materials, the quantity of materials carried by each AGV needs to be greater than the call unit material requirement quantity; the objective function is the minimum total operation energy consumption of AGVs;

[0028] Specifically, when the production unit in the matrix manufacturing workshop lacks materials and waits for the AGV to replenish materials, its status changes to a calling unit. At the specified time, the control system will dispatch the AGV to transport the materials to the calling unit. The AGV departs from the warehouse and unloads the materials at the specified calling unit. The control system schedules the AGV to carry materials based on the material demand information of the calling unit. After the AGV departs from the warehouse and delivers the materials to the specified calling unit, the control system will arrange the logistics transportation tasks according to the operating conditions of the AGV until the AGV has delivered materials to all calling units.

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

[0030] Denote the Manhattan distance it travels when moving from calling unit i to calling unit j , where ( , ) and ( , ) represent the coordinate positions of calling units and in the matrix manufacturing workshop;

[0031] Denote the running time from calling unit i to calling unit j ;

[0032] Denote the time to complete material distribution at calling unit j , Denote just departing from the warehouse, and is the time from departing from the warehouse to completing material distribution at the first calling unit j. Denote the time from departing from the warehouse to completing material distribution at calling unit j, Denote the time to complete material distribution at calling unit i;

[0033] Denote the load before unloading at calling unit j , and its calculated value is the load before unloading at calling unit i minus the quantity q of materials unloaded at calling unit i, Denote the load before unloading at calling unit i;

[0034] Denote The energy consumption from call unit i to call unit j is calculated by multiplying the energy consumption per unit time by the running time. . Among them, the energy consumption per unit time is The load energy consumption coefficient of multiplied by its load before unloading at call unit j, plus the no-load energy consumption of the AGV.

[0035] It means that the objective function is to minimize the energy consumption of the AGV for material distribution within a production cycle. Integer decision variable. If moves along the edge , then = 1; otherwise, = 0.

[0036] The constraint conditions are as follows:

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

[0038] It means that the AGV needs to ensure that each of its routes is continuous when distributing materials. It represents the decision variable from call unit j to call unit i;

[0039] It means that within a production cycle, each AGV can depart from the warehouse at most once. It represents any The decision variable from the warehouse to call unit i;

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

[0041] It means that according to the material demand information of the call unit, the scheduling system needs to ensure that the materials required by each call unit can be delivered within its deadline when scheduling the AGV to distribute materials, ensuring that the materials are delivered on time. It represents the time when call unit j issues a material demand. At call unit The time when the material distribution is completed. It represents the deadline for delivering the material demand of call unit j;

[0042] It means that when the scheduling system arranges for the AGV to deliver materials to the calling unit, the quantity of materials carried by each AGV needs to be greater than the material requirement quantity of the calling unit;

[0043] It means that a constraint is imposed on the decision variable, that is, if passes through the edge , then = 1, otherwise it is 0.

[0044] Parameter description: and are the production unit numbers. If a production unit needs material replenishment, it will turn into a calling unit. Therefore, if production unit i needs material replenishment, it can also be expressed as calling unit i; is the number of production units; is the calling time (the time when production unit sends a signal); is the delivery deadline of calling unit ; is the loading and unloading time of the AGV; is the running speed of the AGV; is the quantity of goods unloaded by the calling unit; is the initial load of the AGV; is the unit no-load energy consumption of the AGV; is the set of production units, ; D is the warehouse, D = {0}; is , is the set of production units and the warehouse; is the set of m AGVs, ; B is the set of edges, ; is the load energy consumption coefficient of the AGV;

[0045] In step S103, data processing is performed on the final AGV scheduling plan, and an AGV material scheduling and distribution information table, a graph of the change of total energy consumption with the number of iterations, and an AGV operation Gantt chart are output.

[0046] The present invention designs an AGV scheduling method considering the on-time delivery of materials for the operation energy consumption of AGVs, and this method aims to reduce the energy consumption of AGVs when transporting materials to the work sites. The efficient scheduling method proposed by the present invention can optimize the scheduling efficiency of AGVs to improve the reliability and stability of the AGV scheduling system. By optimizing the scheduling, the running energy consumption of AGVs is reduced, and the wear and failure rate of equipment are decreased. Reducing the AGV energy consumption not only helps to reduce the enterprise's energy consumption, lower the operation cost, optimize the utilization of resources and the service life of equipment, and reduce the maintenance and replacement costs, but also enhances the reliability and stability of the entire AGV scheduling system.

[0047] After initializing the parameters, task encoding is carried out, and multiple initial scheduling schemes are randomly generated to form an initial population. On this basis, according to the path planning and energy consumption optimization objectives of AGVs, a bee colony algorithm is used for optimization search 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. The scheme with lower energy consumption is selected as the basis, and neighborhood search is carried out in the employed bee stage to generate new solutions and evaluate their energy consumption, and the scheduling scheme with lower energy consumption is selected. In the observing bee stage, excellent schemes are selected based on the fitness probability to further optimize the task scheduling. In the scout bee stage, the schemes that have not been optimized for a long time are reset to avoid falling into local optima. Finally, it is judged whether the optimization termination condition is met. If it is met, the optimal AGV scheduling scheme is output and the operation is stopped; otherwise, the loop calculation continues until the maximum number of iterations is reached or the optimization scheme remains stable. Encoding is carried out according to the numbers of AGV scheduling tasks to ensure that the AGV transportation of the same task follows the optimal path as much as possible and reduces the energy consumption, thereby improving the overall scheduling efficiency.

[0048] In an alternative embodiment, the order of all call units is randomly shuffled, and the call units are sequentially assigned to different AGVs to obtain a population of task assignment schemes, where one task assignment scheme corresponds to an initial solution, and the population of task assignment schemes includes at least one initial solution; the objective function is used as the fitness function to calculate the fitness values of all initial solutions to obtain the feasible solutions of the objective function, and the feasible solutions form a feasible solution group; step S301, a neighborhood operator is used to generate neighborhood solutions based on the feasible solutions, calculate the fitness values of the neighborhood solutions, and obtain a new feasible solution group, where one feasible solution corresponds to one neighborhood solution; step S302, a high-quality position is randomly selected in the new feasible solution group, at least two feasible solutions are selected from the new feasible solution group as a preselected solution group, and the preselected solutions in the preselected solution group are compared with the high-quality solution, where the feasible solution corresponding to the high-quality position is the high-quality solution; step S303, check the failure count of each feasible solution, and when the failure count of the feasible solution is greater than the specified number of times, reset the feasible solution; repeat the above steps S301-step S303 until the preset convergence condition is met, and then output the high-quality solution, and the scheduling scheme corresponding to the high-quality solution is the final scheduling scheme of the AGV.

[0049] Specifically, the present invention uses the bee colony algorithm to optimize the population. Before applying the bee colony algorithm, it is first necessary to clarify data such as the size of the matrix manufacturing workshop, the number of AGVs, the number of call units, the material call time and the material delivery time of the call units, and determine the objective function and constraint conditions, and encode the initial data;

[0050] Employed bee stage: Each employed bee corresponds to a current candidate solution. Through an improved neighborhood operator, a neighborhood solution is generated for the current initial solution. Calculate the fitness value of the neighborhood solution. If it is better than the original solution, replace the current solution with the neighborhood solution; otherwise, increment the failure count of the current solution by one.

[0051] Onlooker bee stage: According to the fitness values of the candidate solutions obtained in the previous stage in the population, the tournament selection strategy is adopted for selection, that is, several candidate solutions (such as 4-5) are randomly selected from the population in each round, and the one with the better fitness value is selected as the selection object of the onlooker bee. In this stage, no neighborhood perturbation operation is performed on the selected individual, but directly compare its current fitness with the fitness of its original position. If the selected solution is better, replace the current solution with it and reset the failure count; otherwise, keep the original solution unchanged and increment the failure count of this solution by one.

[0052] Scout bee stage: Traverse the failure counts of all solutions. If the failure count of a certain solution exceeds the set threshold (such as LIMIT), it is determined that the solution has fallen into a local optimum, discard the solution and randomly generate a new solution to replace it. This is aimed at enhancing the population diversity and avoiding premature search.

[0053] Optimal solution record and convergence judgment: Continuously track the current optimal solution in each iteration, and record its AGV path scheme and fitness. If the maximum number of iterations is met, terminate the algorithm and output the current optimal AGV scheduling solution 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), where each solution represents an AGV task allocation scheme. First, randomly shuffle the order of all call units to increase the diversity of solutions. Then, adopt a round-robin allocation strategy to sequentially allocate call units to different AGVs, ensuring an even distribution of tasks and avoiding extreme situations where some AGVs are overloaded or have no tasks. Finally, generate n different initial solutions.

[0055] Randomly generate new solutions again to prevent the algorithm from falling into local optima, avoid search stagnation, and enhance search diversity; repeat the operations of fitness calculation, generating neighborhood solutions, and screening solutions for the new generation of the population until the pre-set convergence conditions are met, and output the scheduling scheme of the optimal solution as the final AGV scheduling decision output.

[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, replace the feasible solution in the feasible solution group with the neighborhood solution to obtain a new feasible solution group; in step S3012, when the neighborhood solution is inferior to the feasible solution corresponding to the neighborhood solution, increase the failure count of the feasible solution by one.

[0057] Specifically, for each difficult 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 swap). If can accept the task (without violating the capacity / time window constraint), then transfer the task from to . If the direct transfer is not feasible, then swap a task between and , and preferentially select a task pair that can significantly improve energy consumption or delay. After the above preferential transfer / swap operations, obtain a new solution as the neighborhood solution and return it for use in the improved bee colony algorithm.

[0058] In an optional embodiment, in step S3021, when the fitness value of the preselected solution is better than that of the high-quality solution, replace the position of the high-quality solution with the position where the preselected solution is located; in step S3022, when the fitness value of the preselected solution is inferior to that of the high-quality solution, increase the failure count of the feasible solution corresponding to the preselected solution.

[0059] In an alternative embodiment, the task difficulty index of the computing task unit , where is the task difficulty index, is the urgency, is the distance, is the energy consumption, is an infinitesimal number, is the time when the calling unit completes the material delivery, is the delivery deadline of the calling unit , is the first weighting coefficient, is the second weighting coefficient, is the third weighting coefficient; all task units are sorted in descending order of the task difficulty index, and the first most difficult task units are selected as the candidate set for priority processing; for each AGV, the load index parameter is calculated, and the calculation formula is , where is the load index parameter, is the number of tasks executed, is the total running distance of is the total energy consumption of is the task weighting coefficient, is the distance weighting coefficient, is the energy consumption weighting coefficient; the task units corresponding to the AGVs in the candidate set are transferred or exchanged with the task units of the AGV with the lowest load index parameter to obtain a neighborhood solution.

[0060] In an alternative 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 tasks with energy consumption or delay between the AGV corresponding to the task unit and the AGV with the lowest load index parameter are exchanged.

[0061] In an alternative 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 the candidate solution group, and the solution with better fitness is selected according to the fitness in the candidate solution group. The optimal solution is selected with a probability, and the sub-optimal solution is selected with a probability until the preselected solution group is filled.

[0062] Specifically, calculate the fitness of each solution. Since the optimization goal 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 the solution. Randomly select T candidate solutions from the population, and define better individuals according to the fitness. Artificially define , and probability to select the optimal solution, and probability to select the sub-optimal solution.

[0064] Selecting the optimal directly will lead to too fast convergence and the loss of population diversity; the probability is an artificially adjusted parameter; the goal of the observing bee stage is not to find the global optimal solution, but to select a "promising" solution to replace the current bee position. The role of the sub-optimal solution is to retain diversity, activate the potential solution, increase the search width, and help jump out of the local optimum.

[0065] In an alternative embodiment, arrange for the AGV to perform material distribution for the calling unit, including: recording the status information of the AGV at the current moment, where the status information includes: the current location of each AGV, all current unfinished calling unit tasks, the material requirements of each calling unit, the earliest deliverable time, the latest material delivery completion time, and the current load of each AGV; check whether all the distribution tasks of all calling units have been completed: if all the material distribution tasks of all calling units have been completed, end the scheduling; if there are pending distribution tasks for the calling unit, extract all current unfinished material distribution tasks, calculate the task scheduling ability of the AGV, judge whether the AGV can complete the delivery before the task deadline, and calculate the transportation time from the current location of the AGV to the calling unit; sort all the material distribution tasks in ascending order according to the earliest completion time of the material delivery to obtain the preliminary sorting of the material distribution tasks; traverse all AGVs for each material distribution task according to the preliminary sorting to obtain the AGV group, and select the AGV with the lowest cost for executing the material distribution task in the AGV group to allocate each material distribution task.

[0066] In an alternative embodiment, when the AGV travels to the calling unit, update the remaining load of the AGV and calculate the energy consumption of the AGV during the task execution; when the current task of the AGV is completed, update the location of the AGV and the available time of the AGV.

[0067] In an alternative embodiment, when there are still unfinished tasks, continue to allocate new tasks to the AGV; when all tasks are completed, terminate the scheduling process and output the scheduling plan.

[0068] The present invention has the following technical effects: The present invention optimizes the AGV task scheduling based on the bee colony algorithm, processes the allocation results of AGV tasks, integrates the demand of call units to be delivered into input data, including task time, material demand, and AGV operation parameters, and inputs them into the scheduling optimization module. In the optimization process, according to the preset energy consumption constraint, path planning goal, and time window limit, the bee colony algorithm is used to search for the optimal solution and output the optimal AGV delivery plan. The results after scheduling optimization are processed to form the final AGV operation path; through the global optimization characteristics of the bee colony algorithm, the AGV task sequence is dynamically adjusted to make the scheduling of AGV tasks on different paths more reasonable, reduce the AGV no-load operation time, and optimize the path selection to minimize the overall AGV operation energy consumption. Finally, users can formulate a more efficient AGV operation plan based on the optimized AGV scheduling results, improve the logistics distribution efficiency, and shorten the production logistics cycle.

[0069] In another embodiment:

[0070] The algorithm parameters are set as follows: The population size is 50, representing the number of initial solutions (food sources) in the bee colony algorithm; the maximum number of iterations is 100, which controls the number of running rounds of the bee colony algorithm and affects the optimization degree of the final solution; the failure limit number is 100, which sets the upper limit of the failure times for scout bees to trigger new solutions in the bee colony algorithm to avoid falling into local optima; the ratio of the three types of bees in the bee colony algorithm is set to 1, which controls the optimization capabilities of employed bees, observing bees, and scout bees, making the generation of new solutions more stable and more exploratory; ; The flexible production workshop is defined as a 5*5 matrix manufacturing workshop, with 3 AGVs. Within 10 - 40 unit times, 15 production units are randomly transformed into call units. For each call unit, the material call time and the material delivery deadline are randomly generated. The deadline is randomly increased by 5 - 10 seconds based on the call time; The following are the settings for each AGV operation parameter: The AGV running speed v is 1 unit / unit time, the AGV loading and unloading time is 1 unit time, the AGV unit no-load energy consumption is 1, the AGV load energy consumption coefficient is 1, the maximum load Q of the AGV is 100, and the material quantity delivered by the AGV each time is 10.

[0071] Results and Analysis: After running the bee colony algorithm, an AGV scheduling plan is generated according to the optimization results, and a Gantt chart of AGV operation is drawn. Through this optimization process, the execution order, completion time, and energy consumption cost of each AGV task can be clearly understood. The optimized scheduling plan mainly improves the task allocation order, making the path of AGV more reasonable in material distribution, optimizing the running time of AGV, and reducing the task waiting time. In addition, the task load of AGV is reasonably adjusted to avoid overloading or idling of some AGVs, and the overall scheduling efficiency is improved. Based on the optimization results, users can formulate a more reasonable AGV running path, optimize the workshop logistics, improve the material distribution efficiency, and shorten the task execution cycle. The present invention can provide certain reference value for the problem of energy consumption optimization in multi-AGV collaborative scheduling.

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

[0073] Running the algorithm with the above data gives the graph of the total energy consumption TEC varying with the number of iterations, and the results are as Figure 2 shown. The horizontal axis is the number of iterations, the vertical axis is the total energy consumption, and the blue broken line marks the TEC values of each iteration point with dots. It can be seen from the graph that the total energy consumption shows a gradually decreasing trend with the increase of the number of iterations, indicating that the algorithm is continuously optimizing the AGV scheduling plan and reducing the total energy consumption of AGV operation. In the early stage (0 - 20 generations): the energy consumption drops the fastest, indicating that in the initial search stage, the algorithm quickly finds a better solution; in the middle stage (20 - 60 generations): the decreasing trend becomes slow, and local optimization is carried out step by step, and the quality of the solution is gradually improved; in the later stage (60 - 100 generations): the energy consumption tends to be stable, indicating that the algorithm is close to convergence and the space for further optimization is limited. The overall optimization trend is fast optimization speed in the early stage, gradual optimization in the middle stage, and stability in the later stage, which conforms to the typical convergence characteristics of intelligent optimization algorithms, indicating that this optimization process successfully reduces the energy consumption of AGV operation and improves the logistics distribution efficiency.

[0074] Run the algorithm by inputting the above data, and make the AGV material distribution scheduling arrangement according to the optimized results. The results are shown in Table 1. In the table, the AGV number column lists the AGV vehicles assigned to execute tasks, the call unit (coordinate) column lists the call unit numbers and target positions (two-dimensional coordinates) of material distribution, the AGV task start time column lists the time when the AGV starts to execute the task from the initial position, the AGV arrival time column lists the time when the AGV actually arrives at the call unit, the material delivery time column lists the time when the material distribution is completed, the call time column lists the time when the material distribution task is triggered, that is, the time when the production unit requests materials, and the deadline column lists the latest completion time of the material distribution task. Taking AGV1 being dispatched to call unit 11 (2, 4) to execute the material distribution task as an example, call unit 11 (2, 4) requests material distribution at the 16th moment. The task scheduling system receives the request and selects AGV1 for scheduling according to the AGV availability and path conditions. AGV1 receives the assignment of this task at the 13th moment. Although the request time of the call unit is the 16th moment, the AGV may plan the path in advance to improve the overall scheduling efficiency. AGV 1 departs from the starting position (1, 3), travels a certain distance, and finally arrives at call unit 11 (2, 4) at the 15th moment. After arriving at the 15th moment, AGV1 starts to unload materials and completes the material distribution at the 16th moment. The call time of this task is 16, and the AGV completes the distribution at the 16th moment, that is, the task is completed on time without any delay. The deadline is 22, indicating that this task is successfully completed within the acceptable time window. The subsequent AGV scheduling arrangements are similar. From this, it can be clearly understood the call units responsible for material distribution by each AGV, the distribution order of each material distribution task, and the time when the materials arrive at the call unit.

[0075] Table 1 is as follows:

[0076]

[0077] The Gantt chart of production scheduling obtained according to the above scheduling arrangement is as Figure 3 shown. The horizontal axis is the time span of task execution, the vertical axis is the AGV equipment number, and the colored rectangular blocks are the tasks executed by the AGV in a certain time period. Different colors distinguish the tasks of different call units (CUs). It can be seen more simply from the graph the processing order of each material distribution task and the arrangement of each AGV. For example, the call units served by AGV1 are 2, 11, 14, 3, 1 respectively, and the running time of AGV1 is between the 5th and 30th moments. The scheduling arrangements of the call units responsible for material distribution by other AGVs are similar. From this, it can be clearly understood the distribution order of each material distribution task and the time when the materials arrive at the call unit.

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

[0079] Table 2 Comparison of experimental results:

[0080]

[0081] As can be seen from Table 2, in the solution of the 20 extended test cases by the improved ABC algorithm proposed in the present invention:

[0082] 1. Obtained 20 optimal values, which are significantly better than other comparison algorithms.

[0083] 2. The on-time delivery rate of the materials of the improved ABC algorithm is significantly higher than that of other algorithms.

[0084] In summary, the present invention proposes an improved artificial bee colony algorithm (ABC), which realizes the global optimization of AGV scheduling by introducing a multi-agent cooperation mechanism. In the employed bee stage, an adaptive neighborhood transfer operator based on the task execution difficulty and the current load of the AGV is designed, which can dynamically generate high-quality local solutions; in the observing bee stage, an improved selection mechanism is introduced to retain excellent solutions while improving the overall search efficiency and algorithm stability of the population. The experimental results show that compared with the traditional algorithm, the scheduling method proposed in the present invention has obvious advantages in reducing the total system energy consumption, improving the on-time delivery rate of materials, reducing no-load and path repetition, and can realize the efficient cooperation of multiple AGVs and improve the overall operation efficiency of the workshop logistics scheduling system under the 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, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions 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, Including: Obtain the production information of the matrix manufacturing workshop under the flexible production mode, where the production information includes the number of AGVs, running energy consumption information, speed, call unit coordinates, call unit material requirement time, and call unit material deadline; Input the production information and optimizer parameters into the optimizer. The optimizer searches for the optimal solution of the objective function according to the constraint conditions and the objective function, and takes the scheduling plan corresponding to the optimal solution as the final AGV scheduling plan. Among them, the constraint conditions include: when scheduling, each call unit can only be delivered by one AGV once; when the AGV distributes materials, it is necessary to ensure that each of its paths 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 required materials for each call unit are delivered within the call unit material deadline; when distributing materials, the quantity of materials carried by each AGV needs to be greater than the call unit material requirement quantity; the objective function is the minimum total running energy consumption of the AGV; Perform data processing on the final AGV scheduling plan, and output the AGV material scheduling and distribution information table, the graph of the total energy consumption varying with the number of iterations, and the AGV operation Gantt chart.

2. The method according to claim 1, wherein Input the production information and optimizer parameters into the optimizer. The optimizer searches for the optimal solution of the objective function according to the constraint conditions and the objective function, and takes the scheduling plan corresponding to the optimal solution as the final AGV scheduling plan, including: Randomly shuffle the order of all call units, and sequentially assign the call units to different AGVs to obtain a population of task assignment plans. Among them, one task assignment plan corresponds to an initial solution, and the population of task assignment plans includes at least one of the initial solutions; Take the objective function as the fitness function, calculate the fitness values of all the initial solutions, obtain the feasible solutions of the objective function, and the feasible solutions form a feasible solution group; Step S301, use the neighborhood operator to generate neighborhood solutions based on the feasible solutions, calculate the fitness values of the neighborhood solutions, and obtain a new feasible solution group. Among them, 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 in the new feasible solution group as the preselected solution group, and compare the preselected solutions in the preselected solution group with the high-quality solution. Among them, 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 the feasible solution is greater than the specified number of times, reset the feasible solution; Repeat the above steps S301 - S303 until the preset convergence condition is met, and then output the high-quality solution. The scheduling plan corresponding to the high-quality solution is the final AGV scheduling plan.

3. The method according to claim 2, 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, replace the feasible solution in the feasible solution group with the neighborhood solution to obtain the new feasible solution group; Step S3012: When the neighborhood solution is inferior to the feasible solution corresponding to the neighborhood solution, increment the failure count of the feasible solution by one.

4. The method according to claim 2, 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, replace the position of the high-quality solution with the position where the preselected solution is located; Step S3022: When the fitness value of the preselected solution is inferior to that of the high-quality solution, increment the failure count of the feasible solution corresponding to the preselected solution.

5. The method according to claim 2, characterized in that, The said Step S301 includes: Calculate the task difficulty index of the computing task unit, , where, is the task difficulty index, is the urgency, is the distance, is the energy consumption, is an infinitesimal number, is the time when the calling unit completes the material delivery, is the delivery deadline of the calling unit , is the first weighting coefficient, is the second weighting coefficient, is the third weighting coefficient; Sort all the said task units in descending order according to the said task difficulty index, and select the first most difficult said task units as the candidate set to be processed preferentially; Calculate the load index parameter for each of the AGVs. The calculation formula is , where is the load index parameter, is the number of tasks executed, is the total running distance of is the total energy consumption of is the task weighting factor, is the distance weighting factor, is the energy consumption weighting factor; Transfer or exchange the task units corresponding to the AGVs in the candidate set with the task units of the AGV with the lowest load index parameter to obtain the neighborhood solution.

6. The method according to claim 5, characterized in that, Transfer or exchange the task units corresponding to the AGVs in the candidate set with the task units of the AGV with the lowest load index parameter to obtain the neighborhood solution, including: When the AGV with the lowest load index parameter can accept the task unit, transfer the task unit 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, exchange the tasks with energy consumption or delay between the AGV corresponding to the task unit and the AGV with the lowest load index parameter.

7. The method according to claim 2, characterized in that, The said 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 a candidate solution group, and select a solution with better fitness from the candidate solution group according to the fitness. Probability selects the optimal solution, Suboptimal solutions are probabilistically selected until the preselected solution group is filled.

8. The method according to claim 2, wherein Take the objective function as the fitness function, calculate the fitness values of all the initial solutions, obtain the feasible solutions of the objective function, and the feasible solutions form a feasible solution group, including: Arrange the AGV to perform material distribution for the call unit, including: Record the status information of the AGV at the current moment, where the status information includes: the current position of each AGV, all the current unfinished call unit tasks, the material requirements of each call unit, the earliest available delivery time, the latest material delivery completion time, and the current load of each AGV; Check whether the distribution tasks of all the call units have been completed: If the material distribution tasks of all the call units have been completed, end the scheduling; If there are pending delivery tasks for the call unit, extract all the current unfinished material distribution tasks, calculate the task scheduling ability 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 call unit; Sort all the material distribution tasks according to the earliest completion time of the material delivery to obtain the preliminary sorting of the material distribution tasks; Traverse all the AGVs for each material distribution task according to the preliminary sorting to obtain an AGV group, and select the AGV with the lowest cost for executing the material distribution task in the AGV group to assign each material distribution task.

9. The method according to claim 8, wherein Arrange the AGV to perform material distribution for the call unit, including: The AGV travels to the call unit, updates the remaining load of the AGV, and calculates the energy consumption of the AGV during the task execution; When the task of the current AGV is completed, update the position of the AGV and the available time of the AGV.

10. The method according to claim 9, wherein When the current task of the AGV is completed, update the location of the AGV and the available time of the AGV, including: When there are still unfinished tasks, continue to assign new tasks to the AGV; When all tasks are completed, terminate the scheduling process and output the scheduling plan.

Images