Method and device for scheduling distributed flexible job shop with automated guided vehicle

Abstract

The application discloses a scheduling method and device for a distributed flexible job shop with an automatic guided vehicle, and the method comprises the following steps: establishing encoding and decoding rules by taking the maximum completion time and energy consumption of a scheduling sequence as optimization targets; randomly obtaining two solutions from a feature space, performing a crossover-mutation strategy on the obtained two solutions, and generating two new solutions; selecting a solution and using a double deep neural network to decide a local search operator used; executing the selected local search operator, obtaining a new scheduling scheme, and updating the feature space; judging whether a termination condition is met, and if yes, outputting a current best scheduling sequence and two target values; and if not, returning to execute the step of randomly obtaining two solutions from the feature space. The application solves the scheduling problem of the distributed flexible job shop with the automatic guided vehicle, shortens the completion time of the scheduling sequence in workshop production, and reduces the energy consumption in processing.

Description

Technical Field

[0001] The present invention relates to the technical field of flexible job shop scheduling, and in particular to a scheduling method and device for a distributed flexible job shop with an automatic guided vehicle. Background Art

[0002] Manufacturing is the cornerstone of intelligent manufacturing, and its production scheduling needs to be capable of autonomous learning and continuous improvement to meet the rapidly changing production environment and societal demands of today. Distributed flexible job shop scheduling is a common production scheduling problem in the manufacturing industry, with widespread applications in areas such as steelmaking and continuous casting, glassmaking, logistics and transportation, and drone scheduling. Distributed flexible job shop scheduling is a complex, discrete, large-scale combinatorial optimization problem characterized by NP-hardness. When faced with slightly larger-scale problems, it is difficult to solve using precise or traditional mathematical methods. In actual production, it is also necessary to consider that different tasks or jobs need to be transported to other machines for processing after completing a certain process, making the problem even more complex and difficult to solve. When faced with large-scale production tasks, minimizing time costs and energy consumption has become a pressing challenge that companies currently need to address.

[0003] Currently, research on the distributed flexible job shop scheduling problem with automated guided vehicles (AGVs) is insufficient. There are few inventions or relevant literature establishing encoding and decoding rules for this problem, and even fewer algorithms. Therefore, establishing encoding and decoding rules and designing corresponding optimization algorithms for this problem are crucial. Furthermore, task assignment based on human experience is often suboptimal. Irrational scheduling can lead to significant idle time for machines, extending processing time, increasing inefficient energy consumption, and reducing production efficiency. Summary of the Invention

[0004] In order to solve one of the technical problems existing in the prior art to at least a certain extent, the object of the present invention is to provide a scheduling method and device for a distributed flexible job shop with an automated guided vehicle.

[0005] The technical solution adopted in the present invention is:

[0006] A scheduling method for a distributed flexible job shop with an automated guided vehicle comprises the following steps:

[0007] With the optimization objective of minimizing both the maximum completion time and energy consumption of the scheduling sequence, the encoding and decoding rules for the distributed flexible job shop scheduling problem with automated guided vehicles are established.

[0008] Randomly obtain two solutions from the feature space, perform crossover-mutation strategy on the two solutions obtained, and generate two new solutions;

[0009] If two new solutions do not dominate each other, one of them is selected and a dual deep neural network is used to decide the local search operator to use;

[0010] Execute the selected local search operator to obtain a new solution and update the feature space;

[0011] Determine whether the termination condition is met. If so, output the current optimal scheduling sequence and two target values. If not, return to the step of randomly obtaining two solutions from the feature space.

[0012] Furthermore, the objective function of the distributed flexible job shop scheduling problem with automated guided vehicles is:

[0013]

[0014] Among them, obj1 is the first optimization objective, which is to minimize the maximum completion time of the scheduling sequence; is the wth workpiece i i The first process, w i is the last process of workpiece i; obj2 is the second optimization target, namely the energy consumption value; I, J i , and M i，j Represent the workpiece set, process set and process O respectively i，j Optional machine set;TP i，j，k For process O i，j Processing energy consumption on machine k, TI k is the idle energy consumption of machine k; TT i，j，k，k’ For process O i，j Transport energy consumption from machine k to machine k'.

[0015] Furthermore, the constraints of the distributed flexible job shop scheduling problem with automated guided vehicles include:

[0016] 1) A workpiece can only be assigned to one factory for processing;

[0017] 2) Each process has one and only one direct predecessor and successor in the scheduling sequence. If the process is the first or last in the sequence, its predecessor and successor are both empty;

[0018] 3) One machine can only process one process at a time, and one process can only be processed by one machine at a time;

[0019] 4) A certain process of a workpiece can only be transported by one automated guided vehicle, and one automated guided vehicle can only transport a certain process of a workpiece at a time;

[0020] 5) The AGV cannot be preempted or interrupted during transportation;

[0021] 6) Once a process starts, it cannot be preempted or interrupted by other processes;

[0022] 7) The time for setting up, adjusting and testing the machine is included in the processing time;

[0023] 8) At time zero, all machines and automated guided vehicles are available;

[0024] 9) The length of the process processing time is greater than 0.

[0025] Furthermore, the encoding and decoding rules include:

[0026] Generate four integer vectors: factory assignment FA, process sorting OS, machine selection MS, and automatic guided vehicle selection AS;

[0027] Traverse the data in the four vectors and extract the numbers of the factory, machine and AGV assigned to each process;

[0028] Decoding is performed based on the allocation of the code: After determining the process and the workpiece corresponding to the process, the workpiece is first allocated to the corresponding factory, and then the machine to be processed is selected;

[0029] If it is the first process, the AGV goes to the warehouse to obtain the workpiece, and then transports the workpiece to the preset machine for processing; otherwise, the AGV goes to the machine processed in the previous process to obtain the workpiece, and then transports the workpiece to the machine determined by the process for processing;

[0030] Calculate the completion time of the entire sequence and obtain the target value, characteristics and corresponding solutions.

[0031] Furthermore, the scheduling method further includes the following steps:

[0032] Generate an initial solution under the framework of the quality-diversity optimization algorithm and store the solution in the feature space;

[0033] If the corresponding cell in the feature space is empty, the solution is directly stored;

[0034] If the corresponding cell in the feature space is not empty, the newly generated solution is compared with the original solution in the cell; if the new solution dominates the original solution, the new solution replaces the original solution; if the new solution and the original solution are not mutually dominant, the number of Pareto solutions in the cell is determined; if the number of Pareto solutions in the cell exceeds the preset upper limit, a solution is randomly selected, otherwise, it is directly stored; in other cases, the new solution is discarded.

[0035] Further, the initial solution generated under the quality-diversity optimization algorithm framework includes:

[0036] According to the existence of the corresponding feature cell in the action space and the size of the target value, the solution in the grid is updated, and whether the filling condition is met is determined.

[0037] If there is no solution in the grid, the current solution is saved in the grid.

[0038] If there is a solution in the grid, it is determined whether the target value of the solution in the grid is less than the target value of the current solution; if it is less, the newly generated solution replaces the solution in the grid; otherwise, the solution in the grid remains unchanged.

[0039] Further, the crossover-mutation strategy includes a crossover operation and a mutation operation, and the crossover operation adopts a two-point crossover operation.

[0040] The mutation operation includes:

[0041] Randomly exchange any two positions of one of the four integer vectors of factory allocation FA, process sequencing OS, machine selection MS and automated guided vehicle selection AS to generate two new solutions.

[0042] Further, if the two new solutions are not mutually dominant, one of the solutions is selected, and a double deep neural network is used to determine the local search algorithm to be used, including:

[0043] From the two new solutions, a solution with better performance is found; if the two new solutions are not mutually dominant, a solution is randomly selected, the solution is re-recorded as x, and the solution x is stored in the feature space;

[0044] The solution x containing the four integer vectors of factory allocation FA, process sequencing OS, machine selection MS and automated guided vehicle selection AS is input into the double deep neural network.

[0045] The double deep neural network outputs an execution action as a local search operator.

[0046] Furthermore, the local search operators include four types: a first local search operator based on process sequencing design, a second local search operator based on machine allocation design, a third local search operator based on factory allocation design, and a fourth local search operator based on automatic guided vehicle allocation design;

[0047] The operation steps of the first local search operator based on process sorting design include:

[0048] Find the OS vector of the process order in the current solution;

[0049] Randomly select two processes O that do not belong to the same workpiece i from the process sorting OS vector i，j and process O i’，j ;

[0050] Exchange process O i，j and process O i’，j Position in the process sorting OS vector;

[0051] Update the information in the process sorting OS vector to obtain a new solution;

[0052] The second local search operator based on machine allocation design has the following steps:

[0053] Find the OS vector of the process order in the current solution;

[0054] Randomly select a process O from the process sorting OS vector i，j ;

[0055] Find process O i，j Optional processing machine set M i，j ;

[0056] If there are other i，j If the machine is different from the optional machine, then replace process O i，j processing machines;

[0057] Update the information in the machine selection MS vector to obtain a new solution;

[0058] The operation steps of the third local search operator based on factory allocation design include:

[0059] Find the critical factory f_critical of the current solution and randomly select a job cri_job from the critical factory f_critical;

[0060] Randomly select a non-critical plant f_rand from the solution;

[0061] Randomly select a job rnd_job from a non-critical factory f_rand;

[0062] Exchange the factories assigned to the artifacts rnd_job and cri_job;

[0063] Update the workpiece information in the factory allocation FA vector and get a new solution;

[0064] The operation steps of the fourth local search operator based on AGV allocation design include:

[0065] Find the OS vector of the process order in the current solution;

[0066] Randomly select a process O from the process sorting OS vector i，j ;

[0067] Change process O i，j assigned automated guided vehicles;

[0068] Update the information in the AGV selection AS vector and get a new solution.

[0069] Another technical solution adopted in the present invention is:

[0070] A scheduling device for a distributed flexible job shop with an automated guided vehicle, comprising:

[0071] at least one processor;

[0072] at least one memory for storing at least one program;

[0073] When the at least one program is executed by the at least one processor, the at least one processor implements the above method.

[0074] Another technical solution adopted in the present invention is:

[0075] A computer-readable storage medium stores a program executable by a processor, wherein the program executable by the processor is used to perform the method described above when executed by the processor.

[0076] The beneficial effects of the present invention include:

[0077] Method level:

[0078] (1) This invention proposes a quality-diversity algorithm that can store Pareto solution sets, based on which the diversity of solutions is greatly improved;

[0079] (2) A four-layer encoding and decoding rule is proposed. According to the four vectors of factory allocation FA, machine selection MS, process sorting OS, and automatic guided vehicle selection AS, appropriate factories, machines, and automatic guided vehicles are assigned to the processes in the scheduling sequence. The decoding rule is then used to obtain the two target values ​​of maximum completion time and energy consumption.

[0080] (3) The combination of dual-depth neural networks and quality-diversity algorithms can intelligently help the algorithm make decisions more effectively, thereby avoiding the short-term nature of random selection. The intelligent selection of operators by the dual-depth neural network exceeds the random selection method, promoting the development of understanding in the optimal direction.

[0081] (4) Four local search operators are designed to take automated guided vehicles into account and work in synergy with the local search operators designed for processes, machines, and factories to comprehensively improve the quality of the solution.

[0082] Application level:

[0083] (1) Compared with the traditional distributed flexible job shop scheduling method, the optimization algorithm provided by the present invention studies the problem from the perspective of feature space, which is convenient for users to design and use;

[0084] (2) Ability to provide more high-quality solutions and be able to learn and improve intelligently;

[0085] (3) It can effectively optimize the scheduling sequence according to user needs, improve the efficiency of the use of machines and automated guided vehicles, and improve the overall productivity of the enterprise;

[0086] (4) By effectively ensuring load balancing between multiple factories, uneven resource allocation during the processing process can be avoided, thereby maximizing machine utilization. Compared with the original solution, the optimized solution significantly shortens the machine's idle time and completion time, effectively reduces resource waste, and further reduces the company's production costs. The present invention can not only improve production efficiency, but also optimize the factory's production layout and resource allocation, thereby bringing greater competitive advantages to the company.

[0087] In general, the present invention provides a solution for a distributed flexible job shop with automated guided vehicles, which provides a good solution for distributed flexible job shop scheduling, can ensure the resource allocation of the factory, improve the utilization efficiency of machines and automated guided vehicles, and reduce the completion time and energy consumption of the entire solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following introduction is made to the drawings of the embodiments of the present invention or the related technical solutions in the prior art. It should be understood that the drawings introduced below are only for the convenience of clearly describing some embodiments of the technical solutions of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative work.

[0089] Figure 1 This is a flow chart of a method for scheduling a distributed flexible job shop with an automated guided vehicle in an embodiment of the present invention;

[0090] Figure 2 This is a performance comparison chart of the present invention and other algorithms;

[0091] Figure 3 This is a comparison chart of the convergence curves of the present invention and other algorithms. DETAILED DESCRIPTION

[0092] The embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar components or components having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention and are not to be construed as limiting the present invention. For the step numbers in the following embodiments, they are provided only for the convenience of explanation and do not limit the order between the steps. The execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0093] In the description of the present invention, it should be understood that descriptions involving orientations, such as up, down, front, back, left, right, etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or component referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, they cannot be understood as limitations on the present invention.

[0094] In the description of the present invention, the meaning of "several" is one or more, the meaning of "many" is two or more, and "greater than", "less than", "exceed", etc. are understood to exclude the number itself, and "above", "below", "within", etc. are understood to include the number itself. If there is a description of the first and the second, it is only for the purpose of distinguishing the technical features, and cannot be understood as indicating or implying the relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features. In addition, "and / or" describes the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship.

[0095] In the description of the present invention, unless otherwise clearly defined, terms such as setting, installing, and connecting should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in the present invention based on the specific content of the technical solution.

[0096] In order to efficiently solve the scheduling problem of distributed flexible job shops with automated guided vehicles, the present invention proposes an optimization algorithm with self-learning and adaptive technology, with the goal of simultaneously optimizing the maximum completion time and energy consumption, establishing reasonable and effective encoding and decoding rules and a quality-diversity algorithm enhanced by dual deep neural networks. Compared with traditional distributed flexible job shop scheduling methods, the optimization algorithm provided by the present invention searches for solutions from the behavioral space from a completely new perspective, requires fewer algorithm parameters, can be designed according to the specific needs of users, and finds higher-quality solutions within a limited time. The optimization algorithm of the present invention can automatically allocate all tasks, effectively balance the load of machines and automated guided vehicles between multiple factories, avoid unreasonable allocation caused by human factors, improve the stability and reliability of the production line, reduce time and energy waste, and thus reduce overall production costs. Using the optimization algorithm of the present invention, the completion time and energy consumption of enterprise production tasks can be significantly reduced, the effective utilization rate of machines and automated guided vehicles in the factory can be significantly improved, and ultimately the productivity level of the enterprise can be improved.

[0097] like Figure 1 As shown, this embodiment provides a scheduling method for a distributed flexible job shop with an automated guided vehicle, comprising the following steps:

[0098] Step 1: With the optimization objective of simultaneously minimizing the maximum completion time and energy consumption of the scheduling sequence, encoding and decoding rules for the distributed flexible job shop scheduling problem with automated guided vehicles (AGVs) are established. Based on the four vectors of plant allocation (FA), machine selection (MS), process sequencing (OS), and AGV selection (AS) in the four-layer encoding rule, appropriate plants, machines, and AGVs are assigned to the processes in the scheduling sequence. The decoding rule is then used to initialize the scheduling sequence and calculate the corresponding target values. The objective function of the distributed flexible job shop scheduling problem with AGVs is:

[0099]

[0100] Among them, obj1 is the first optimization objective, which is to minimize the maximum completion time of the scheduling sequence; is the wth workpiece i i The first process, w i is the last process of workpiece i; obj2 is the second optimization target, namely the energy consumption value; I, J i , and Mi，j Represent the workpiece set, process set and process O respectively i，j Optional machine set;TP i，j，k For process O i，j Processing energy consumption on machine k, TI k is the idle energy consumption of machine k; TT i，j，k，k’ For process O i，j Transport energy consumption from machine k to machine k'.

[0101] Step 2: Generate an initial solution under the framework of the quality-diversity optimization algorithm and store it in the feature space; if the corresponding cell in the feature space is empty, store it directly; otherwise, compare the newly generated solution with the original solution in the cell: if the new solution dominates the original solution, replace the original solution with the new solution; if the new solution and the original solution do not dominate each other, determine the number of Pareto solutions in the cell: if the number of Pareto solutions in the cell exceeds the upper limit, randomly select a solution, otherwise, store it directly; in other cases, the new solution is discarded.

[0102] Step 3: Randomly select two solutions x and y from the feature space (one of which comes from the global Pareto solution set) and perform a crossover-mutation strategy on them. The crossover uses the traditional two-point crossover operation, and the mutation randomly changes the order of one of the four vectors FA, MS, OS, and AS. Then, two new solutions x′ and y′ are generated.

[0103] Step 4: Select the solution with better performance from the two solutions x′ and y′ obtained above, denoted as solution z. If the two solutions do not dominate each other, one of the solutions is randomly selected, and then a dual deep neural network is used to decide the local search operator to be used.

[0104] Step 5: Execute the selected local search operator and update the solution saved in the feature space. The method of updating the feature space is consistent with that described in step 2.

[0105] Step 6: Determine whether the termination condition is met. If so, the evaluation ends and the current optimal scheduling sequence and two target values ​​obj1 and obj2 are output. Otherwise, continue to step 3.

[0106] As an optional implementation, in step 1, the encoding and decoding rules used include the following steps:

[0107] First, four integer vectors are generated: factory assignment FA, process sorting OS, machine selection MS, and automated guided vehicle selection AS;

[0108] Secondly, traverse the data in the four vectors, extract the number of the factory, machine and automated guided vehicle assigned to each process;

[0109] Next, decode according to the assignment of encoding: after determining the process and its corresponding workpiece, first assign it to the corresponding factory, and then select the machine to be processed;

[0110] Next, if it is the first process, the automated guided vehicle goes to the warehouse to pick up the workpiece, and then transports it to the machine determined in advance for processing; otherwise, the automated guided vehicle goes to the machine processed by the previous process to pick up the workpiece, and then transports it to the machine determined for this process for processing.

[0111] Finally, calculate the completion time of the entire sequence to obtain the target value, characteristics and corresponding solution.

[0112] The constraint conditions of the distributed flexible job shop scheduling problem with automated guided vehicles include:

[0113] 1) Ensure that a workpiece can only be assigned to one factory for processing;

[0114] 2) Ensure that each process has and only has one direct predecessor and successor in the scheduling sequence, and if the position of the process is at the first or last of the sequence, its predecessor and successor are empty;

[0115] 3) Ensure that a machine can only process one process at a time, and a process can only be processed by one machine at a time;

[0116] 4) Ensure that a workpiece can only be transported by one automated guided vehicle for a process, and one automated guided vehicle can only transport one workpiece for a process at a time;

[0117] 5) Ensure that the automated guided vehicle cannot be preempted or interrupted during transportation;

[0118] 6) Ensure that once a process starts, it cannot be preempted or interrupted by other processes;

[0119] 7) Ensure that the setup, adjustment and detection time of the machine is included in the processing time;

[0120] 8) Ensure that at time zero, all machines and automated guided vehicles are available;

[0121] 9) Ensure that the process processing time is greater than 0.

[0122] As an optional implementation, in step 2, the initial solution generated under the quality-diversity optimization algorithm framework includes the following steps:

[0123] First, the solution in the grid is updated based on the presence or absence of the corresponding feature unit in the behavior space and the size of the target value to determine whether the filling condition is met.

[0124] Second, if no solution exists in the grid, the current solution is saved in the grid;

[0125] Finally, if there is a solution in the grid, it is determined whether the target value of the solution in the grid is less than the target value of the current solution. If so, the newly generated solution replaces the solution in the grid; otherwise, the solution in the grid remains unchanged.

[0126] As an optional implementation, in step 3, the crossover-mutation strategy step includes:

[0127] First, the crossover operation uses the traditional two-point crossover operation;

[0128] Secondly, the mutation operation randomly swaps any two positions of one of the four vectors FA, MS, OS, and AS;

[0129] Finally, two new solutions are generated and recorded as: x′ and y′.

[0130] As an optional implementation, in step 4, after performing the crossover-mutation operation, using a dual deep neural network to decide the local search operator to use includes the following steps:

[0131] First, find the solution with better performance from the two new solutions x′ and y′. If the two solutions do not dominate each other, randomly select a solution, re-record it as solution z, and store it in the feature space;

[0132] Secondly, the solution z containing four vectors FA, MS, OS and AS is passed as input to the dual deep neural network;

[0133] Finally, the neural network outputs an execution action, which is the local search operator to be executed.

[0134] There are four local search operators to choose from. The first local search operator based on process sorting design has the following steps:

[0135] First, find the OS vector in the current solution p;

[0136] Secondly, randomly select two processes O from the OS vector that do not belong to the same workpiece i i，j and O i’，j ;

[0137] Then, exchange O i，j and O i’，jPosition in the OS vector;

[0138] Finally, the information in the machine sort vector OS is updated and a new solution p′ is obtained.

[0139] The second local search operator based on machine allocation design has the following steps:

[0140] First, find the OS vector in the current solution p;

[0141] Next, randomly select a process O from the OS vector i，j ;

[0142] Then, find process O i，j Optional processing machine set M i，j ;

[0143] Then, if there are other i，j If the machine is different from the optional machine, replace O i，j processing machines;

[0144] Finally, the information in the machine sort vector MS is updated and a new solution p′ is obtained.

[0145] The operation steps of the third local search operator based on factory allocation design include:

[0146] First, find the critical factory f_critical of the current solution x and randomly select a job cri_job from f_critical;

[0147] Secondly, randomly select a non-critical factory f_rand from the solution p;

[0148] Then, a job rnd_job is randomly selected from f_rand;

[0149] Next, swap the factories assigned to rnd_job and cri_job;

[0150] Finally, the workpiece information in the factory allocation vector FA is updated and a new solution p′ is obtained.

[0151] The operation steps of the fourth local search operator based on AGV allocation design include:

[0152] First, find the OS vector in the current solution p;

[0153] Next, randomly select a process O from the OS vector i，j ;

[0154] Then, change process O i，jAn automated guided vehicle is assigned;

[0155] Finally, the information in the automatic guided vehicle assignment vector AS is updated, and a new solution p' is obtained.

[0156] The embodiment improves the global search capability of the algorithm and increases the diversity of solutions by using the quality-diversity framework. The local search capability of the algorithm is further improved and the quality of solutions is improved by using a three-layer local search strategy and a machine idle time saving method.

[0157] The embodiment overcomes the short-sightedness of human factors by combining a double deep neural network and a quality-diversity algorithm and intelligently making strategic decisions for the algorithm by using deep reinforcement learning technology, and balances the search and development capabilities. The application will be further described and explained below through specific embodiments:

[0158] The simulation experiment uses 20 standard examples, wherein the number of workpieces j = {10, 20, 30, 40, 50, 100}, the number of machines m = 5, the number of automated guided vehicles a = 2, the number of factories f = {2, 3, 4, 5, 6, 7}, the workpiece processing time is uniformly distributed in the range of [5, 20], and the transportation time is uniformly distributed in the range of [2, 12]. In the specific examples of the application, detailed description and experimental verification will be performed.

[0159] The quality-diversity algorithm based on the double deep neural network proposed in the application is referred to as DQN-QD, which is compared with the original multi-objective quality-diversity algorithm QD (MOME), NSGA-II and MOEA / D. In addition, the most effective solving algorithm SPAMA and DQCE are used to verify the effectiveness of the application. The above-mentioned comparison algorithms are high-performance optimization algorithms proposed in recent years. To reduce errors, each example is repeated 20 times. The hyper volume HV is used as an evaluation index of the performance of the algorithm. HV can evaluate the convergence and diversity of the solution at the same time, and the larger the value, the better the comprehensive performance of the algorithm.

[0160] Figure 2 is a visual representation of the Pareto solution set obtained by all comparison algorithms. Specifically, from Figure 2 It can be known from the HV value corresponding to the distribution range of the solution that DQN-QD is significantly better than NSGA-II, MOEA / D, MOME, SPAMA and DQCE algorithms. In addition, Figure 3 The distribution of the Pareto solution set of the six algorithms in one run of the 100 workpiece 7 factory example is given. From Figure 3 It can be known that the quality of the Pareto solution set obtained by the DQN-QD algorithm of the application is far superior to other comparison algorithms.

[0161] This embodiment further provides a scheduling device for a distributed flexible job shop with an automated guided vehicle, comprising:

[0162] at least one processor;

[0163] at least one memory for storing at least one program;

[0164] When the at least one program is executed by the at least one processor, the at least one processor implements Figure 1 The method shown.

[0165] The scheduling device of a distributed flexible job shop with an automatic guided vehicle in this embodiment can execute a scheduling method of a distributed flexible job shop with an automatic guided vehicle provided by the method embodiment of the present invention, can execute any combination of implementation steps of the method embodiment, and has the corresponding functions and beneficial effects of the method.

[0166] The present application also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs Figure 1 The method shown.

[0167] This embodiment also provides a storage medium storing instructions or programs that can execute a scheduling method for a distributed flexible job shop with an automatic guided vehicle provided by an embodiment of the method of the present invention. When the instructions or program are run, any combination of implementation steps of the method embodiment can be executed, and the corresponding functions and beneficial effects of the method can be obtained.

[0168] In some optional embodiments, the function / operation mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the function / operation involved, the two boxes shown in succession can actually be executed substantially simultaneously or the boxes can sometimes be executed in reverse order. In addition, the embodiment presented and described in the flow chart of the present invention is provided in an exemplary manner for the purpose of providing a more comprehensive understanding of the technology. The disclosed method is not limited to the operation and logic flow presented herein. Optional embodiments are contemplated in which the order of the various operations is changed and the sub-operations described as a part of a larger operation are performed independently.

[0169] Furthermore, although the present application is described in the context of functional modules, it is to be understood that one or more of the described functions and / or features can be integrated in a single physical device and / or software module, or one or more functions and / or features can be implemented in separate physical devices or software modules. It will also be appreciated that detailed discussion of the actual implementation of each module is not necessary to an understanding of the application. Rather, the actual implementation of the modules, in combination with their attributes, functions, and internal relationships, are to be understood within the context of the devices disclosed herein. Thus, those skilled in the art with access to patents, scientific journals, and other public sources known by those skilled in the art will be able to practice the application as set forth in the claims without undue experimentation, using ordinary skill in the art along with the present disclosure. It is also to be understood that the specific concepts disclosed are merely illustrative and that the scope of the present application is to be determined by the entire scope of claims appended hereto along with all equivalents of the claims.

[0170] If the functions are implemented in software, the functions can be stored in or implemented as one or more computer program products, which can be incorporated into a computer-readable medium for use by or in connection with an apparatus, method, or system as described herein. The computer-readable medium can be a machine-readable storage device, a machine-readable storage substrate, an article of manufacture, or a computer program product. The computer-readable medium can be, for example, but is not limited to, volatile or non-volatile memory, a floppy diskette, a compact disk, a tape, a hard disk drive, a solid state memory drive, a memory array, or a computer program product. The computer program product can be tangibly embodied in an information carrier, such as in a non-transitory computer-readable medium or memory, including a non-transitory machine-readable storage medium. The computer program product can also contain instructions that, when executed by a processor, can cause the processor to carry out a method, such as the method(s) for providing a user interface as described herein. The computer program product can be a software program, software, an application including one or more software modules, a computer program, a procedure, a function, an application program interface (API), an app, an applet, an app or applet extension, or the like.

[0171] The logic and / or steps represented in the flowcharts and / or otherwise described herein, for example, can be embodied in computer-readable instructions, which can be used to cause a general purpose computing device, such as a personal computer, server, or network device, or other computing / processing device to perform functions described by the various embodiments. Such computer-readable instructions can be stored in one or more computer-readable media, such as a floppy disk, hard disk, universal serial bus (USB) drive, memory card, read only memory (ROM), random access memory (RAM), or other computer-readable media. The computer-readable media, having such computer-readable instructions stored thereon, can be caused to be executed by a processor such that the instructions result in performance of the functions represented therein.

[0172] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0173] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0174] In the above description of this specification, reference to the terms "one embodiment / example," "another embodiment / example," or "certain embodiments / examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0175] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

[0176] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A method for scheduling a distributed flexible job shop with an automated guided vehicle, characterized in that: The following steps are involved: With the optimization objective of minimizing the maximum completion time and energy consumption of the scheduling sequence, the encoding and decoding rules for the distributed flexible job shop scheduling problem with automated guided vehicles are established. Randomly obtain two solutions from the feature space, perform crossover-mutation strategy on the two solutions obtained, and generate two new solutions; If two new solutions do not dominate each other, one of them is selected and a dual deep neural network is used to decide the local search operator to use; Execute the selected local search operator to obtain a new solution and update the feature space; Determine whether the termination condition is met. If so, output the current optimal scheduling sequence and two target values. If not, return to the step of randomly obtaining two solutions from the feature space. The objective function of the distributed flexible job shop scheduling problem with automated guided vehicles is: Among them, obj1 is the first optimization objective, which is to minimize the maximum completion time of the scheduling sequence; is the wth workpiece i i The first process, w i is the last process of workpiece i; obj2 is the second optimization target, namely the energy consumption value; I, J i , and M i，j Represent the workpiece set, process set and process O respectively i，j Optional machine set;TP i,j,k For process O i，j Processing energy consumption on machine k, TI k is the idle energy consumption of machine k; TT i，j，k，k’ For process O i，j Transport energy consumption from machine k to machine k'; The constraints of the distributed flexible job shop scheduling problem with automated guided vehicles include: 1) A workpiece can only be assigned to one factory for processing; 2) Each process has one and only one direct predecessor and successor in the scheduling sequence. If the process is located at the first or last position in the sequence, the predecessor and successor of the process are both empty; 3) One machine can only process one process at a time, and one process can only be processed by one machine at a time; 4) A certain process of a workpiece can only be transported by one automated guided vehicle, and one automated guided vehicle can only transport a certain process of a workpiece at a time; 5) The AGV cannot be preempted or interrupted during transportation; 6) Once a process starts, it cannot be preempted or interrupted by other processes; 7) The time for setting up, adjusting and testing the machine is included in the processing time; 8) At time zero, all machines and automated guided vehicles are available; 9) The length of the process processing time is greater than 0; The encoding and decoding rules include: Generate four integer vectors: factory assignment FA, process sorting OS, machine selection MS, and automatic guided vehicle selection AS; Traverse the data in the four vectors and extract the numbers of the factory, machine and AGV assigned to each process; Decoding is performed based on the allocation of the code: After determining the process and the workpiece corresponding to the process, the workpiece is first allocated to the corresponding factory, and then the machine to be processed is selected; If it is the first process, the AGV goes to the warehouse to obtain the workpiece, and then transports the workpiece to the preset machine for processing; otherwise, the AGV goes to the machine processed in the previous process to obtain the workpiece, and then transports the workpiece to the machine determined by the process for processing; Calculate the completion time of the entire sequence and obtain the target value, characteristics and corresponding solutions.

2. The method for scheduling a distributed flexible job shop with an automated guided vehicle according to claim 1, characterized in that: The scheduling method further comprises the following steps: Generate an initial solution under the framework of the quality-diversity optimization algorithm and store the solution in the feature space; If the corresponding cell in the feature space is empty, the solution is directly stored; If the corresponding cell in the feature space is not empty, the newly generated solution is compared with the original solution in the cell; if the new solution dominates the original solution, the new solution replaces the original solution; if the new solution and the original solution do not dominate each other, the number of Pareto solutions in the cell is determined; if the number of Pareto solutions in the cell exceeds the preset upper limit, a solution is randomly selected, otherwise it is directly stored; in other cases, the new solution is discarded.

3. The method for scheduling a distributed flexible job shop with an automated guided vehicle according to claim 2, characterized in that: Generating an initial solution under the quality-diversity optimization algorithm framework includes: Update the solution in the grid based on the presence or absence of corresponding feature units in the behavior space and the size of the target value to determine whether the filling conditions are met; If there is no solution in the grid, the current solution is saved in the grid; If there is a solution in the grid, determine whether the target value of the solution in the grid is less than the target value of the current solution; if so, the newly generated solution replaces the solution in the grid; otherwise, the solution in the grid remains unchanged.

4. The method for scheduling a distributed flexible job shop with an automated guided vehicle according to claim 1, characterized in that: The crossover-mutation strategy includes a crossover operation and a mutation operation, wherein the crossover operation adopts a two-point crossover operation; Mutation operations include: Randomly swap any two positions of one of the four integer vectors: factory assignment FA, operation sorting OS, machine selection MS, and automatic guided vehicle selection AS to generate two new solutions.

5. The method for scheduling a distributed flexible job shop with an automated guided vehicle according to claim 1, characterized in that: If two new solutions do not dominate each other, one of the solutions is selected and a dual deep neural network is used to decide the local search operator to use, including: If the two new solutions do not dominate each other, randomly select a solution, rename it as x, and store the solution x in the feature space; The solution x, which contains four integer vectors of factory assignment FA, process sorting OS, machine selection MS, and automated guided vehicle selection AS, is passed as input to the dual deep neural network; The dual deep neural network outputs an execution action as a local search operator.

6. The method for scheduling a distributed flexible job shop with an automated guided vehicle according to claim 1, characterized in that: The local search operators include four types: a first local search operator based on process sequencing design, a second local search operator based on machine allocation design, a third local search operator based on factory allocation design, and a fourth local search operator based on automatic guided vehicle allocation design; The operation steps of the first local search operator based on process sorting design include: Find the OS vector of the process order in the current solution; Randomly select two processes O that do not belong to the same workpiece i from the process sorting OS vector i，j and process O i’，j ; Exchange process O i，j and process O i’，j Position in the process sorting OS vector; Update the information in the process sorting OS vector to obtain a new solution; The second local search operator based on machine allocation design has the following steps: Find the OS vector of the process order in the current solution; Randomly select a process O from the process sorting OS vector i，j ; Find process O i，j Optional processing machine set M i，j ; If there are other i,j If the machine is different from the optional machine, then replace process O i,j processing machines; Update the information in the machine selection MS vector to obtain a new solution; The operation steps of the third local search operator based on factory allocation design include: Find the critical factory f_critical of the current solution and randomly select a job cri_job from the critical factory f_critical; Randomly select a non-critical plant f_rand from the solution; Randomly select a job rnd_job from a non-critical factory f_rand; Exchange the factories assigned to the artifacts rnd_job and cri_job; Update the workpiece information in the factory allocation FA vector and get a new solution; The operation steps of the fourth local search operator based on AGV allocation design include: Find the OS vector of the process order in the current solution; Randomly select a process O from the process sorting OS vector i，j ; Change process O i，j assigned automated guided vehicles; Update the information in the AGV selection AS vector and get a new solution.

7. A dispatching device for a distributed flexible workshop with an automated guided vehicle, characterized in that: include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the method according to any one of claims 1 to 6.

Images