Multi-stroke multi-load AGV cooperative scheduling method in speed-adjustable job shop

Abstract

The invention relates to the technical field of workshop scheduling, and provides a multi-stroke multi-load AGV cooperative scheduling method in a speed-adjustable workshop, and the method comprises the steps: building a cooperative scheduling model, and enabling an objective function to be minimum maximum operation time and total energy consumption of a machine AGV; carrying out coding representation on the solution multilayer vector model of the AGV multi-stroke speed-adjustable cooperative scheduling model to generate a feasible solution, carrying out first decoding on the feasible solution by adopting a single-stroke constraint condition to generate AGV single-stroke initial scheduling, and carrying out second decoding on the AGV single-stroke initial scheduling by adopting a multi-stroke constraint condition to generate AGV multi-stroke initial scheduling; and according to whether the AGV single-stroke initial scheduling meets the AGV multi-stroke load capacity requirement and the AGV multi-stroke material delivery time requirement, solving the AGV multi-stroke initial scheduling or the AGV single-stroke initial scheduling through an adaptive large neighborhood search algorithm to generate a planning optimal solution. According to the invention, precise energy-saving scheduling of the machine, the multi-stroke multi-load AGV and the speed of the AGV is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of job shop scheduling, and in particular to a method for collaborative scheduling of multi-trip and multi-load AGVs in a speed-adjustable job shop. Background Art

[0002] In conventional job shop scheduling technology, production scheduling and logistics scheduling are carried out independently. However, research shows that in some discrete workshops and long-distance layouts, the impact of transportation and conversion time on the total production cycle cannot be ignored, especially in small-batch, multi-variety production modes, where transportation tasks are crucial. There is a close interaction between production operations and transportation operations. For example, production operations affect transportation time, and transportation delays may cause machines to wait, thereby reducing overall production efficiency.

[0003] Therefore, incorporating the operation transportation time into the workshop scheduling model for scheduling optimization is crucial to achieve efficient production. In order to improve the overall efficiency, researchers have gradually focused on combining AGV (automatic guided vehicle) with production task scheduling to form a collaborative scheduling model.

[0004] In recent years, research on energy-saving job scheduling has continued to emerge. However, most of these collaborative scheduling models do not fully consider the scheduling problem of AGVs, or simply assume that the number of AGVs is infinite, which is quite different from the actual situation. In addition, there are collaborative scheduling models for AGVs in a single-trip and single-load mode, that is, AGVs only perform one transportation task at a time. In actual production environments, AGVs usually have multi-load capabilities and can transport multiple materials to different destinations at a time, thereby improving transportation efficiency and equipment utilization. Therefore, it is of great practical significance to study the workshop scheduling problem of integrated multi-trip and multi-load AGVs with a finite number of AGVs.

[0005] The scheduling problem of a limited number of AGVs involves two NP (Non-deterministic Polynomial) problems: machine scheduling and AGV scheduling. Machine scheduling requires determining the order of operations for tasks to be processed on each machine, while AGV scheduling involves the allocation and sorting of transportation tasks to ensure that limited AGV resources are used reasonably.

[0006] Adjustable speed energy-saving job shop scheduling with transportation resource constraints is also an NP-hard problem, because the machine speed and AGV running speed of the processing task must be determined simultaneously to minimize energy consumption and achieve energy-saving job scheduling. Due to the high complexity of such problems, effective solution methods are needed to meet their challenges.

[0007] ALNS (Adaptive Large Neighborhood Search) is a relatively new heuristic method. Unlike most heuristic techniques, ALNS is a large-scale neighborhood improvement method that runs on top of a built-in heuristic, which uses local search to select between different neighborhoods to achieve optimization, and is used to solve the NP-hard problem of the collaborative scheduling model.

[0008] Although ALNS has been successful in solving various single-objective problems, there is still great potential in extending ALNS to address multi-objective optimization challenges. To extend ALNS to the multi-objective domain, it is necessary to develop mechanisms that can effectively balance the trade-offs between different objectives. Such extensions will enable ALNS to be used in more complex and realistic scenarios, such as multi-objective production scheduling, transportation, and resource allocation, where conflicting objectives must be considered simultaneously.

[0009] Therefore, how to construct a collaborative scheduling model that accurately expresses workshop energy saving by applying and expanding ALNS is a technical problem that needs to be solved. Summary of the invention

[0010] To this end, the present invention provides a method for collaborative scheduling of multi-stroke and multi-load AGVs in an adjustable-speed operation workshop, and constructs a collaborative scheduling model with the minimized maximum operation time and the total energy consumption of the machine AGV as the objective function, thereby achieving accurate expression of the energy saving and operation status of the workshop through mathematical modeling, and solving the collaborative scheduling model step by step through multi-layer vector model encoding and decoding and adaptive large neighborhood search algorithm, thereby achieving accurate scheduling of machines, multi-stroke and multi-load AGVs and their speeds.

[0011] To achieve the above object, the present invention proposes a method for coordinated dispatching of multi-trip and multi-load AGVs in a speed-adjustable workshop, comprising:

[0012] Step S1, constructing an AGV multi-stroke adjustable speed collaborative scheduling model based on a linear programming model, wherein the objective function is to minimize the maximum operation time and the total energy consumption of the machine AGV, and the constraints include operation sequence constraints, load material transportation time constraints, multi-stroke constraints, material transportation and machine operation processing start time constraints, and multi-load transportation task constraints, and the decision variables include machine operation variables, transportation task allocation variables, AGV multi-stroke transportation task allocation variables, machine operation speed level variables, AGV transportation task speed level variables, and AGV single-load and multi-load scheduling variables;

[0013] Step S2, encoding the solution of the AGV multi-trip adjustable speed collaborative scheduling model through a multi-layer vector model with an initialization strategy to generate a feasible solution, performing a first decoding on the feasible solution using a single-trip constraint condition to generate an AGV single-trip initial scheduling, and performing a second decoding on the AGV single-trip initial scheduling using a multi-trip constraint condition to generate an AGV multi-trip initial scheduling;

[0014] Step S3: According to whether the AGV single-trip initial scheduling meets the AGV multi-trip load capacity requirement and the AGV multi-trip material delivery time requirement, the AGV multi-trip initial scheduling or the AGV single-trip initial scheduling is respectively solved by an adaptive large neighborhood search algorithm for the AGV multi-trip adjustable speed collaborative scheduling model to generate the optimal solution for the machine AGV planning of the job shop.

[0015] Furthermore, in the decision variables of step S1:

[0016] The machine operation variables include whether the machine processes the operation variable, the machine operation sequence variable, the machine operation sequence variable and the machine processing sequence variable corresponding to the operation;

[0017] The AGV multi-trip transport task allocation variables include whether the AGV executes a multi-load transport task variable, the AGV executes a multi-transport task sequence variable, and the AGV multi-load task transport material sequence variable;

[0018] The AGV single-load and multi-load scheduling variables include whether the AGV executes a multi-load transportation task variable, whether the AGV executes a material transportation variable, and the AGV multi-load task material transportation order variable.

[0019] Further, the operation sequence constraint includes a single machine processing sequence constraint and an inter-machine processing sequence constraint;

[0020] The single machine processing sequence constraint is composed of a variable of whether the machine processes an operation, a variable of the machine operation speed level, a variable of the machine operation sequence, a completion time of a preceding machine operation, a machine operation processing time, a completion time of a succeeding machine operation, and a first set constant;

[0021] The inter-machine processing sequence constraint is composed of whether the machine processes the operation variable, the machine operation speed level variable, the operation corresponding to the machine processing sequence variable, as well as the completion time of the previous machine operation, the machine operation processing time, the completion time of the subsequent machine operation and the second set constant.

[0022] Further, the multi-trip constraint condition includes a first material delivery time constraint and a second material delivery time constraint;

[0023] The first material delivery time constraint is composed of the first material delivery time, the transportation task start time, the transportation time from the first material starting position to the second material starting position, the transportation time from the second material starting position to the first material ending position, the variable of whether the AGV performs a multi-load transportation task, the AGV multi-load task transportation material sequence variable and a third set constant;

[0024] The second material delivery time constraint is composed of the second material delivery time, the first material delivery time, the transportation time from the second material end position to the first material end position, whether the AGV performs a multi-load transportation task variable, the AGV multi-load task transportation material sequence variable and a fourth set constant.

[0025] In the above scheme, since the workshop includes multiple machines and multiple AGVs, each machine includes multiple jobs, each job includes multiple operations, and the machines and AGVs have multiple speed levels, its collaborative scheduling model constitutes an NP-problem that can accurately describe and constrain the workshop scheduling situation. Solving the NP-problem through a multi-layer vector model and an adaptive large neighborhood search algorithm can speed up the solution speed and improve the quality of the solution, thereby achieving better collaborative scheduling of machines and multi-load AGVs in the workshop.

[0026] Furthermore, the multi-layer vector model is a four-layer vector model, and the vector encoding sequence of each layer is a machine operation order sequence, a machine processing speed level sequence, an AGV task allocation sequence and an AGV transportation speed level sequence.

[0027] Furthermore, the multi-layer vector model is a four-layer vector encoding and decoding model based on a genetic algorithm, and its initialization strategy is population initialization based on topological sorting and random arrangement, wherein the population initialization process of the AGV task allocation sequence is:

[0028] One or more AGVs that complete the transportation task earliest are selected, and one of the multiple AGVs is randomly selected.

[0029] Furthermore, the single-trip constraint conditions include the operation sequence constraint, the load material transportation time constraint, and the material transportation and machine operation processing start time constraint.

[0030] In the above scheme, through the global search capability of the genetic algorithm, the approximate global optimal solution of the AGV single-trip and the approximate global optimal solution of the AGV multi-trip of the collaborative scheduling model can be quickly found, and then the adaptive large neighborhood search algorithm based on this can further optimize the quality of the solution, thereby achieving better collaborative scheduling of machines and multi-load AGVs in the workshop.

[0031] Furthermore, the AGV multi-trip load capacity requirement is that the AGV's transport task allocation is less than or equal to a set value;

[0032] The AGV multi-trip material delivery time requirement is that the AGV's material delivery time is less than or equal to the latest delivery time.

[0033] Furthermore, the constraints also include a single constraint on machine operation and a constraint on the number of AGV transport task executions.

[0034] Furthermore, the adaptive large neighborhood search algorithm is embedded in a Pareto-based genetic algorithm framework and adopts the Metropolis criterion to accept inferior solutions.

[0035] In the above scheme, the Pareto-based genetic algorithm framework is embedded in the adaptive large neighborhood search algorithm, so that the search capability and solution quality of the genetic algorithm are improved through the adaptive large neighborhood search algorithm.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. By minimizing the maximum operating time and the total energy consumption of the machine AGV as the objective function, a collaborative scheduling model is constructed, which realizes the accurate expression of the energy saving and operation status of the workshop through mathematical modeling, and the collaborative scheduling model is solved step by step through multi-layer vector model encoding and decoding and adaptive large neighborhood search algorithm, which realizes the accurate scheduling of machines, multi-trip and multi-load AGVs and their speeds.

[0038] 2. Since the workshop includes multiple machines and multiple AGVs, each machine includes multiple jobs, each job includes multiple operations, and the machines and AGVs have multiple speed levels, their collaborative scheduling model constitutes an NP-hard problem that can accurately describe and constrain the workshop scheduling situation. Solving the NP-hard problem through a multi-layer vector model and an adaptive large neighborhood search algorithm can speed up the solution speed and improve the quality of the solution, thereby achieving better collaborative scheduling of the workshop machines and multi-load AGVs.

[0039] 3. Through the global search capability of the genetic algorithm, the approximate global optimal solution of the AGV single-trip and the approximate global optimal solution of the AGV multi-trip of the collaborative scheduling model can be quickly found, and then the adaptive large neighborhood search algorithm based on this can further optimize the quality of the solution, realizing better collaborative scheduling of workshop machines and multi-load AGVs.

[0040] 4. By embedding the Pareto-based genetic algorithm framework into the adaptive large neighborhood search algorithm, the search capability and solution quality of the genetic algorithm are improved through the adaptive large neighborhood search algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic diagram of the general flow of a method for collaboratively dispatching multi-trip and multi-load AGVs in an adjustable-speed workshop according to an embodiment of the present invention;

[0042] Figure 2 A schematic diagram of a feasible solution generated by encoding a multi-layer vector model of a method for collaborative scheduling of multi-trip and multi-load AGVs in a speed-adjustable workshop according to an embodiment of the present invention;

[0043] Figure 3 It is a flowchart of a multi-trip AGV scheduling method in a multi-trip and multi-load AGV collaborative scheduling method in a speed-adjustable workshop according to an embodiment of the present invention;

[0044] Figure 4 A schematic flow chart of a genetic algorithm embedded with an adaptive large neighborhood algorithm in a method for collaborative scheduling of multi-trip and multi-load AGVs in an adjustable-speed workshop according to an embodiment of the present invention;

[0045] Figure 5 It is a flow chart of an adaptive large neighborhood algorithm for a speed-adjustable job shop according to an embodiment of the present invention;

[0046] Figure 6 It is a schematic diagram of single-trip and multi-trip scheduling of the collaborative scheduling model of the adjustable-speed job shop according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0048] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0049] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0050] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0051] like Figures 1 to 6As shown, the present invention provides a method for collaborative scheduling of multi-trip and multi-load AGVs in an adjustable-speed operation workshop. A collaborative scheduling model is constructed by minimizing the maximum operation time and the total energy consumption of the machine AGV as the objective function, and the energy saving and operation status of the workshop are accurately expressed through mathematical modeling. The collaborative scheduling model is solved step by step through multi-layer vector model encoding and decoding and an adaptive large neighborhood search algorithm, thereby realizing accurate scheduling of machines, multi-trip and multi-load AGVs and their speeds.

[0052] like Figures 1 to 6 As shown, this embodiment proposes a multi-trip multi-load AGV collaborative scheduling method in an adjustable speed operation workshop, including: step S1, constructing an AGV multi-trip adjustable speed collaborative scheduling model based on a linear programming model, whose objective function is to minimize the maximum operation time and the total energy consumption of the machine AGV, and whose constraints include operation sequence constraints, load material transportation time constraints, multi-trip constraints, material transportation and machine operation processing start time constraints, and multi-load transportation task constraints, and whose decision variables include machine operation variables, transportation task allocation variables, AGV multi-trip transportation task allocation variables, machine operation speed level variables, AGV transportation task speed level variables, and AGV single-load and multi-load scheduling variables;

[0053] Step S2, encoding the solution of the AGV multi-trip adjustable speed collaborative scheduling model through a multi-layer vector model with an initialization strategy to generate a feasible solution, performing a first decoding on the feasible solution using a single-trip constraint condition to generate an AGV single-trip initial scheduling, and performing a second decoding on the AGV single-trip initial scheduling using a multi-trip constraint condition to generate an AGV multi-trip initial scheduling;

[0054] Step S3: According to whether the AGV single-trip initial scheduling meets the AGV multi-trip load capacity requirement and the AGV multi-trip material delivery time requirement, the AGV multi-trip initial scheduling or the AGV single-trip initial scheduling is respectively solved by an adaptive large neighborhood search algorithm for the AGV multi-trip adjustable speed collaborative scheduling model to generate the optimal solution for the machine AGV planning of the job shop.

[0055] It can be understood that the second decoding is a further decoding of the first decoding.

[0056] It can be understood that the basic elements of the linear programming model include objective function, constraints and decision variables. The machines in the workshop need to process different materials. The processing work includes multiple operations. The sequence of work and processing is represented by the process.

[0057] It should be noted that, if the value of the AGV multi-stroke transport task allocation variable in the AGV multi-stroke adjustable speed collaborative scheduling model described in this embodiment is zero, it can be used for single-stroke scheduling of the AGV. Since it is used for the collaborative scheduling of all AGVs in the workshop, the AGV multi-stroke adjustable speed collaborative scheduling model described in this embodiment is specifically a workshop model for the collaborative scheduling of mixed single-stroke and multi-stroke AGVs and machines.

[0058] Specifically, see Figure 6 In the AGV multi-trip adjustable speed collaborative scheduling model described in this embodiment, the modeling description of the machines and AGVs in the workshop is as follows: there are n different jobs in the workshop that need to be processed on m machines, and each job i (i = 1, 2, ..., n) contains j operations, and operation j is processed by the machine in a predetermined order. There are also TA AGVs (automatic guided vehicles) in the workshop, which have two transportation modes: single load and multi-load. Each job has k transportation tasks. In the single-trip mode, Figure 6 As shown in Example 1, AGV can only perform one transportation task at a time. First, it transports the material from the warehouse (WH) to the machine for the first processing process, and then transports the semi-finished product to other machines for subsequent processing. AGV selects a multi-trip transportation mode within an acceptable cost range, such as Figure 2 As shown in example 2, after the AGV obtains the semi-finished product of workpiece 1 at machine m1, it can continue to go to machine m2 to obtain the semi-finished product of workpiece 2, and then transport the two to the corresponding positions for processing in turn.

[0059] It should be noted that the AGV multi-stroke adjustable speed collaborative scheduling model described in this embodiment can be performed as follows: Figure 6 The single-stroke single-load mode, single-stroke multiple-load mode, and multiple-stroke multiple-load mode shown may also perform only a single load in the single-stroke mode.

[0060] Furthermore, in the decision variables of step S1:

[0061] The machine operation variables include whether the machine processes the operation variable, the machine operation sequence variable, the machine operation sequence variable and the machine processing sequence variable corresponding to the operation, wherein the operation and the operation are a pair of index variables;

[0062] The AGV multi-trip transport task allocation variables include whether the AGV executes a multi-load transport task variable, the AGV executes a multi-transport task sequence variable, and the AGV multi-load task transport material sequence variable;

[0063] The AGV single-load and multi-load scheduling variables include whether the AGV executes a multi-load transportation task variable, whether the AGV executes a material transportation variable, and the AGV multi-load task material transportation order variable.

[0064] Specifically, the machine operation variables in the decision variables include:

[0065] Whether the machine processes the operation variable is:

[0066]

[0067] The machine operation sequence variables are:

[0068]

[0069] The machine processing sequence variables corresponding to the job are:

[0070]

[0071] The machine operation sequence variables are:

[0072]

[0073] Specifically, the transportation task allocation variable in the decision variables is:

[0074]

[0075] When the variable indicating whether the AGV is executing a multi-load transport task is zero, it is used to allocate a single-trip transport task, which may be a single-load transport task.

[0076] The AGV single-load and multi-load scheduling variables include whether the AGV executes a multi-load transportation task variable, whether the AGV executes a material transportation variable, and the AGV multi-load task material transportation order variable.

[0077] Specifically, the AGV single-load and multi-load scheduling variables in the decision variables include:

[0078] The variable of whether AGV performs multi-load transport task is:

[0079]

[0080] Whether the AGV performs the material transportation variable is:

[0081]

[0082] The order variables of AGV multi-load task transport materials are:

[0083]

[0084] Specifically, the machine operation speed level variable in the decision variables is:

[0085]

[0086] Specifically, the AGV transport task speed level variable in the decision variables is:

[0087]

[0088] Specifically, the AGV multi-trip transport task allocation variables in the decision variables include:

[0089] The variable of whether AGV performs multi-load transport task is:

[0090]

[0091] The order variables of AGV multi-load task transport materials are:

[0092]

[0093] It can be understood that the AGV multi-load task material transportation sequence variable corresponds to two transportation tasks and also corresponds to the machine's processing operation on the material.

[0094] In the above scheme, since each decision variable is an integer with a value of 0 or 1, it constitutes a constraint variable of the linear programming model, which is solved through constraint conditions and objective functions.

[0095] It can be understood that since there are multiple machines and multiple AGVs in the workshop, it is reflected in the value range of k and h. Each AGV corresponds to one or more transport tasks V if , each transport task V if Corresponding to a transport material g / l, each material g / l is related to operation O i Or operation O ij There is a corresponding relationship, and each operation O ij There are multiple speed levels r, each transport task V if There are also multiple speed levels v, which results in a large number of variables that the collaborative model actually needs to determine by decision, which can be roughly estimated as the product of the maximum values ​​of k, h, g, i, j, r and v. In addition, due to the mutual influence between machines and AGVs, for example, the operation of the machine needs to start after the AGV delivers the required materials, and the AGV should not let the machine standby for too long to increase energy consumption. As a result, a large number of variables increases the difficulty of decision-making while accurately expressing the situation in the workshop.

[0096] Further, the operation sequence constraint includes a single machine processing sequence constraint and an inter-machine processing sequence constraint; the single machine processing sequence constraint is composed of whether the machine processes an operation variable, the machine operation speed level variable, the machine operation sequence variable, and the completion time of the preceding machine operation, the machine operation processing time, the completion time of the following machine operation and a first set constant;

[0097] The inter-machine processing sequence constraints are composed of whether the machine processes the operation variable, the machine operation speed level variable, the operation corresponding to the machine processing sequence variable, as well as the completion time of the previous machine operation, the machine operation processing time, the completion time of the subsequent machine operation and the second set constant.

[0098] Specifically, the single machine processing sequence constraint of the operation sequence constraint and the single-stroke constraint condition is:

[0099]

[0100] In the formula, C i'j' represents the completion time of the machine's previous job, x ijk is whether the machine processes an operational variable, γ ijkr is the machine operating speed level variable, represents the processing time of the subsequent operation on machine k with processing speed level r, C ij Indicates the completion time of the machine in the following job, y i'j'ijk is the machine operation sequence variable, B is the first set constant, preferably a large positive integer to ensure that y i'j'ijk When the inequality is taken to zero, it holds true when the processing sequence is satisfied. In the above formula, the constraints on the machine processing operation are realized through the transportation time of the AGV.

[0101] Specifically, the inter-machine processing sequence constraints of the operation sequence constraints and the single-stroke constraints are:

[0102]

[0103] In the formula, C i(j-1) represents the completion time of the machine's previous operation, x ijk is whether the machine processes an operational variable, γ ijkr is the machine operating speed level variable, represents the processing time of the subsequent operation on machine k with processing speed level r, C ij Indicates the completion time of the subsequent operation, z ik′k represents the machine processing sequence variable corresponding to the job, and B represents the second setting constant to ensure z ik′kWhen the inequality is zero, it holds when the processing sequence is satisfied. In the above formula, the constraints on the machine processing operation are realized through the transportation time of the AGV.

[0104] Further, the multi-trip constraint condition includes a first material delivery time constraint and a second material delivery time constraint;

[0105] The first material delivery time constraint is composed of the first material delivery time, the transportation task start time, the transportation time from the first material starting position to the second material starting position, the transportation time from the second material starting position to the first material ending position, the variable of whether the AGV performs a multi-load transportation task, the AGV multi-load task transportation material sequence variable and a third set constant;

[0106] The second material delivery time constraint is composed of the second material delivery time, the first material delivery time, the transportation time from the second material end position to the first material end position, whether the AGV performs a multi-load transportation task variable, the AGV multi-load task transportation material sequence variable and a fourth set constant.

[0107] Specifically, the first material delivery time constraint is:

[0108]

[0109] s g ,s l =0,1,...,m;d g ,d l =1,2,...,m; r=1,2,...,R,h=1,2,...,a; g≠l

[0110] In the formula, C g It represents the delivery time of AGV for material g. Represents the AGV transport task V numbered h if The transportation start time, Indicates that the AGV transport speed level is r from position s g To location l The transportation time, Indicates that the AGV transport speed level is r from position s l To position d g The transportation time is s g Indicates the starting position of material g, s l Indicates the starting position of material l, d g represents the delivery location of material g, ρ ifi'f'h Indicates whether the AGV performs a multi-load transport task variable, specifically, whether the AGV numbered h performs a multi-load transport task, μ glhrepresents the transport material order variable, and B is the third set constant.

[0111] Specifically, the second material delivery time constraint is:

[0112]

[0113] d g ,d l =1,2,...,m; r=1,2,...,R,h=1,2,...,a,g≠l

[0114] In the formula, C l represents the delivery time of AGV for material l, C g It represents the delivery time of AGV for material g. Indicates that the AGV transport speed level is r from position d g To position d l The transportation time is d g Indicates the delivery location of material g, d l represents the delivery location of material l, ρ ifi'f'h Indicates whether the AGV performs a multi-load transport task variable, specifically, whether the AGV numbered h performs a multi-load transport task, μ glh represents the transport material order variable, and B is the third set constant.

[0115] More specifically, the multi-trip constraint condition also includes:

[0116] μ glh +μ lgh =1

[0117] In the formula, μ glh , μ lgh are both material transport sequence variables, which respectively indicate that the AGV numbered h transports material g first and then material l, or transports material l first and then material g.

[0118] The above formula realizes the arrival time of each material when the AGV performs multi-load transportation tasks.

[0119] Furthermore, the single-trip constraint conditions include the operation sequence constraint, the load material transportation time constraint, and the material transportation and machine operation processing start time constraint.

[0120] Specifically, the load material transportation time constraint of the single-trip constraint condition is:

[0121]

[0122] In the formula, C g It represents the delivery time of AGV for material g. Represents the AGV transport task V numbered h if The transportation start time, Indicates that the AGV transport speed level is r from position s g To position d g The transportation time is s g Indicates the starting position of material g, d g Indicates the delivery location of material g. In the above formula, the constraint that the arrival time of material g (the time of loaded transportation) must be greater than the start time of the transportation task plus the transportation time is realized.

[0123]

[0124] In the formula, Indicates that the AGV numbered h performs the following transport task V if The transportation start time, β i′f′ifh Indicates whether the AGV executes multiple load transport task variables, specifically whether to execute the transport task first V i′f′ Execute the transport task V again if , Indicates that the AGV numbered h performs the previous transport task V i′f′ The transportation end time of . In the above formula, the constraint that the next transportation task can only be started after the previous transportation task is completed is realized.

[0125] Specifically, the material transportation and machine operation processing start time constraints of the single-trip constraint condition are:

[0126]

[0127] In the formula, S g Indicates the start time of transportation of material g, Indicates that the AGV numbered h performs the transport task V if The transportation start time, represents the transportation time from position e to e′ of AGV with transportation speed level r, C i(j-1) Indicates the completion time of the previous operation of the machine. In the above formula, it is realized that the material g can be transported to the next machine for processing only after the current operation is completed and the AGV arrives.

[0128] S ij ≥max(C g ,C i′j′ )

[0129]

[0130] In the formula, S ij Indicates operation o ij The start time of Cg represents the delivery time of AGV for material g, C i'j' Indicates the completion time of the previous operation of the machine. In the above formula, it is realized that the material g reaches the corresponding machine position and the machine can start processing the current operation only after completing the previous operation.

[0131] In the above scheme, since the workshop includes multiple machines and multiple AGVs, each machine includes multiple jobs, each job includes multiple operations, and the machines and AGVs have multiple speed levels, its collaborative scheduling model constitutes an NP-problem that can accurately describe and constrain the workshop scheduling situation. Solving the NP-problem through a multi-layer vector model and an adaptive large neighborhood search algorithm can speed up the solution speed and improve the quality of the solution, thereby achieving better collaborative scheduling of machines and multi-load AGVs in the workshop.

[0132] Specifically, the multi-load transport task constraints of the constraint conditions include the AGV multi-trip load capacity requirement and the AGV multi-trip material delivery time requirement.

[0133] The AGV multi-trip load capacity requirement is that the AGV's transport task allocation is less than or equal to the set value, specifically:

[0134]

[0135] In the formula, α ifh Assign variables to the transport tasks, Q represents the maximum loading capacity of the AGV, and preferably takes 2 task quantities. Therefore, the transport capacity constraint on the AGV is achieved.

[0136] The AGV multi-trip material delivery time requirement is that the AGV's material delivery time is less than or equal to the latest delivery time, specifically:

[0137] C g ≤D g

[0138] In the formula, C g represents the delivery time of AGV for material g, D g represents the latest delivery time of material g under multiple trips. Furthermore, when the AGV performs the single trip initial scheduling, D g represents the start processing time of material g. The above formula ensures that the material is delivered before its deadline. Specifically, the latest delivery time D g Defined as the machine can operate O ij The processing start time is set so that the AGV can deliver the material g first, and then the machine starts processing and allows the machine to operate O ij Other operations are being carried out when the materials arrive.

[0139] Specifically, the constraints also include a single machine operation constraint and an AGV transport task execution quantity constraint, specifically:

[0140]

[0141] In the formula, x ijk Whether the machine processes the operation variable. It ensures that each operation can only be processed on one machine at a time.

[0142]

[0143] In the formula, x ijk Whether the machine processes the operation variable. It ensures that each machine can only process one operation at a time.

[0144]

[0145] Where η i′ik is the machine job order variable. It ensures that a job executed by a machine must follow exactly one predecessor job, unless it is the first job of the machine.

[0146]

[0147] Where η i′ik is the machine job sequence variable. It ensures that when a job is processed on a machine, only a different job can be selected for subsequent processing, unless the job is the last job of the machine.

[0148]

[0149] In the formula, α i f h Assign variables to transport tasks. This ensures that each transport task can only be performed by one AGV at a time.

[0150]

[0151] In the formula, α if h Assign variables to transport tasks, which means each AGV can perform at most two transport tasks at a time.

[0152]

[0153] In the formula, γ ijkr It is the machine operating speed level variable. It ensures that only one machine speed can be selected for each operation.

[0154]

[0155] In the formula, σ ifhvis the AGV transport task speed level variable, which ensures that only one AGV speed can be selected for each transport task.

[0156] Specifically, the objective function of the AGV multi-stroke adjustable speed collaborative scheduling model is:

[0157] min(C max ,TEC)

[0158] In the formula, C max represents the maximum completion time, and TEC represents the total energy consumption of all machines and AGVs in the workshop.

[0159] The maximum completion time is defined as:

[0160] C max =max{C i},i∈1,2,...,n

[0161] In the formula, C i represents the completion time of job i.

[0162] The total energy consumption is defined as:

[0163] TEC=TEM+TEA

[0164] In the formula, TEC represents the total energy consumption of all machines and AGVs in the workshop, TEM represents the energy consumption generated by all machines, and TEA represents the total energy consumption generated by AGVs. The total energy consumption is further split and calculated.

[0165] Among them, the energy consumption generated by all machines is defined as:

[0166] TEM=TEM p +TEM s

[0167] In the formula, TEM represents the energy consumption generated by all machines, TEM p Represents the energy consumption generated by all machine processing, TEM S It represents the energy consumption generated by all machines in standby mode. It further defines the energy consumption generated by the machine as the sum of the machine processing energy consumption and the standby energy consumption, where the processing energy consumption and the standby energy consumption are obtained by multiplying the power by time.

[0168] Among them, the total energy consumption generated by AGV is defined as:

[0169] TEA=TEA s +TEA p +TEA L

[0170] In the formula, TEA represents the total energy consumption of AGV, TEA SIndicates the energy consumption generated by AGV standby, TEA p TEA represents the energy consumption of AGV when performing empty transport tasks. L It represents the energy consumption generated by AGV when performing load transportation tasks. It further defines the total energy consumption of AGV as the sum of standby energy consumption, energy consumption of empty transportation tasks, and energy consumption of load transportation tasks.

[0171] Among them, the energy consumption generated by machine processing is defined as:

[0172]

[0173] Where, TEM p represents the energy consumption of all machine processing, x ijk is whether the machine processes the operation variable, γ ijkr is the machine operating speed level variable, is the processing time of the operation on machine k with processing speed class r, It represents the processing power of machine k when the processing speed is set to r, which can be the average processing power of the machine to perform the operation. Therefore, it is feasible to perform a split calculation of the energy consumption generated by the machine processing in combination with the decision variables.

[0174] Among them, the energy consumption generated by all machines in standby mode is defined as:

[0175]

[0176] Where, TEM S Indicates the energy consumption of all machines in standby mode. represents the standby power of machine k, C k represents the completion time of machine k, x ijk is whether the machine processes the operation variable, γ ijkr is the machine operating speed level variable, is the processing time of the machine k with the processing speed level r. Therefore, it is feasible to perform a split calculation of the energy consumption generated by the machine standby in combination with the decision variables.

[0177] Among them, the energy consumption generated by AGV standby is defined as:

[0178]

[0179] In the formula, TEA S Indicates the energy consumption generated by AGV standby, It represents the standby power of AGV numbered h. Indicates the end time of the transport task of AGV numbered h. Indicates that the AGV transport speed level is r from position sg To location l The transportation time, Indicates that the AGV transport speed level is r from position s l To position d g The transportation time, Indicates that the AGV transport speed level is r from position d g To position d l The transportation time is s g Indicates the starting position of material g, s l Indicates the starting position of material l, d l Indicates the delivery location of material l, d g Indicates the delivery location of material g, U gh Indicates whether the AGV executes the material transportation variable, μ glh represents the transport material order variable, α ifh Assign variables to the transport task, σ ifhv is the speed level variable of the AGV transport task.

[0180] Among them, the energy consumption generated by AGV performing empty-load transportation tasks is defined as:

[0181]

[0182] In the formula, TEA p It represents the energy consumption of AGV when performing empty transport tasks. represents the transportation time from position e to e′ of AGV with transportation speed level r, α ifh Assign variables to the transport task, σ ifhv is the AGV transport task speed level variable, It represents the transport power of AGV numbered h when performing a single load task and the transport speed is set to v, which can be the average operating power of AGV when it is in standby mode. Therefore, it is feasible to perform a split calculation of the energy consumption generated by AGV standby mode in combination with decision variables.

[0183] Among them, the energy consumption generated by AGV performing load transportation tasks is defined as:

[0184]

[0185] In the formula, TEA L It represents the energy consumption of AGV in performing load transportation tasks. Indicates that the AGV transport speed level is r from position s g To position d g The transportation time, Indicates that the AGV transport speed level is r from position s l To location gThe transportation time, Indicates that the AGV transport speed level is r from position d g To position d l The transportation time, Indicates that the AGV transport speed level is r from position s l To position d g The transportation time is s g Indicates the starting position of material g, s l Indicates the starting position of material l, d l Indicates the delivery location of material l, d g Indicates the delivery location of material g, U gh Indicates whether the AGV executes the material transportation variable, μ glh represents the transport material order variable, α ifh Assign variables to the transport task, σ ifhv is the AGV transport task speed level variable, It represents the transport power of AGV numbered h when performing a single load task and the transport speed is set to v. It represents the transport power of AGV numbered h when performing multi-load task and the transport speed is set to r. The first part of the above formula is the transport energy consumption of single load (1 unit) in the multi-load transport task of AGV, and the second part is the transport energy consumption of multiple loads (2 units).

[0186] Furthermore, the multi-layer vector model is a four-layer vector model, and the vector encoding sequence of each layer is a machine operation order sequence, a machine processing speed level sequence, an AGV task allocation sequence and an AGV transportation speed level sequence.

[0187] Furthermore, the multi-layer vector model is a four-layer vector encoding and decoding model based on a genetic algorithm, and its initialization strategy is a population initialization based on topological sorting and random arrangement, wherein the process of population initialization of the AGV task assignment sequence is: selecting one or more AGVs that complete the transportation task earliest, and randomly selecting one of the multiple AGVs.

[0188] It can be understood that the execution process of the genetic algorithm is: initializing the population to generate an initial population, mapping the chromosomes of the population, i.e., encoding. This process can be analogized to binary encoding, and in this embodiment, the process is as follows: Figure 2 The encoding process of the multi-layer vector model shown.

[0189] After the initial population is generated, according to the principle of survival of the fittest, the evolution of each generation produces better and better approximate solutions. In each generation, individuals are selected according to the fitness of individuals in the problem domain, and combined crossover and mutation are performed with the help of genetic operators of natural genetics to generate a population representing a new aggregation. In this process, this embodiment embeds an adaptive large neighborhood search algorithm to improve the quality of the solution.

[0190] Specifically, the decision variables are divided into four sub-problems: the operation sequence of the job, AGV allocation, machine processing speed and AGV transportation speed, so that they correspond to the four-layer vector model, and each layer of vectors encodes the expression of each solution, and the length of each layer is equal to the total number of operations. It can be understood that the four-layer vector model encodes the solution of the AGV multi-trip adjustable speed collaborative scheduling model, which can be simply understood as generating a feasible solution that conforms to the encoding format, and this feasible solution does not need to meet the constraints.

[0191] More specifically, if Figure 2 As shown, the first layer of the four-layer vector model is the operation sequence (OS), each element represents the job number, and the j-th appearance represents the j-th operation of the job; the second layer is the machine processing speed (MS), which represents the speed selection of the corresponding operation; the third layer is the AGV allocation (AS), which is used to allocate the AGV of the corresponding transportation task; the fourth layer is the AGV transportation speed (ASS), which represents the transportation speed selection. Figure 2 An example of a feasible solution with 4 jobs (9 processes), 3 speed levels, and 2 AGVs is shown.

[0192] An effective initialization strategy is used to generate high-quality feasible solutions. Specifically, the initial population of OS is generated by topological sorting and random arrangement of the order of all operations. The AGV allocation sequence is generated by using the selection based on scheduling rules and random selection allocation strategies. In the selection of scheduling rules, the allocation rule is to select the AGV that leaves the previous transportation task the earliest. If there is more than one such AGV, one is randomly selected.

[0193] The process of decoding the feasible solution of the four-layer vector model is as follows Figure 3 The generation of a single-trip initial schedule is shown before the hybrid metaheuristic algorithm solves it.

[0194] In the above scheme, through the global search capability of the genetic algorithm, the approximate global optimal solution of the AGV single-trip and the approximate global optimal solution of the AGV multi-trip of the collaborative scheduling model can be quickly found, and then the adaptive large neighborhood search algorithm based on this can further optimize the quality of the solution, thereby achieving better collaborative scheduling of machines and multi-load AGVs in the workshop.

[0195] Furthermore, the adaptive large neighborhood search algorithm is embedded in a Pareto-based genetic algorithm framework and adopts the Metropolis criterion to accept inferior solutions.

[0196] Specifically, the adaptive large neighborhood search algorithm is embedded in the Pareto genetic algorithm framework as follows: Figure 4 shown.

[0197] The Adaptive Large Neighborhood Search (ALNS) algorithm is an efficient and flexible optimization algorithm that is applicable to a variety of complex combinatorial optimization problems. By adaptively adjusting the weights of the destruction and repair operators and using the Metropolis criterion to accept inferior solutions, the algorithm's search capability and solution quality are improved. The process of the adaptive large neighborhood search algorithm is as follows: Figure 5 shown.

[0198] Specifically, the population size of the genetic algorithm embedded in the adaptive large neighborhood search algorithm is 100, the termination condition is iterated 100 times, and the fitness function is the maximum completion time of the objective function and the total energy consumption of all machines and AGVs in the workshop. The smaller the two objectives, the higher the fitness. The initial temperature setting needs to be obtained based on preliminary experiments. The temperature drop strategy is: the temperature of the next iteration = current temperature * annealing rate coefficient. Here, the annealing rate coefficient is 0.95. The stop condition can be modified to: total number of processes * 0.5ms. You can stop based on the calculation time instead of the number of iterations.

[0199] Specifically, the probability that the Metropolis criterion accepts an inferior solution is:

[0200]

[0201] The initial temperature is determined through preliminary experiments. The annealing rate is an important module of SAA and directly affects the accuracy of the algorithm. The annealing rate is closely related to the probability of accepting inferior solutions. As the iteration proceeds, the temperature gradually decreases, the probability of accepting inferior solutions decreases, and the algorithm gradually focuses on local area search.

[0202] The annealing rate function is:

[0203] T k+1 =T k *θ where θ∈[0,1] is the annealing rate coefficient. It usually ranges from 0.75 to 0.95; T k+1and T k are the current annealing temperature and the previous annealing temperature respectively; k is the number of iterations.

[0204] In the above scheme, the Pareto-based genetic algorithm framework is embedded in the adaptive large neighborhood search algorithm, so that the search capability and solution quality of the genetic algorithm are improved through the adaptive large neighborhood search algorithm.

[0205] It can be understood that this embodiment constructs a collaborative scheduling model by minimizing the maximum operation time and the total energy consumption of the machine AGV as the objective function, realizes the accurate expression of the energy saving and operation status of the workshop through mathematical modeling, and solves the collaborative scheduling model step by step through the multi-layer vector model encoding and decoding and the adaptive large neighborhood search algorithm, and realizes the accurate scheduling of the machine, multi-trip multi-load AGV and its speed. Since the workshop includes multiple machines and multiple AGVs, each machine includes multiple jobs, each job includes multiple operations, and the machine and AGV have multiple speed levels, its collaborative scheduling model constitutes an NP problem that can accurately describe and constrain the workshop scheduling situation. Solving the NP problem through the multi-layer vector model and the adaptive large neighborhood search algorithm can speed up the solution speed and improve the quality of the solution, and realize a better collaborative scheduling of the workshop machines and multi-load AGVs. Through the global search capability of the genetic algorithm, the AGV single-trip approximate global optimal solution and the AGV multi-trip approximate global optimal solution of the collaborative scheduling model can be quickly found, so that the adaptive large neighborhood search algorithm based on this can further optimize the quality of the solution, and realize a better collaborative scheduling of the workshop machines and multi-load AGVs. By embedding the adaptive large neighborhood search algorithm into the Pareto-based genetic algorithm framework, the search capability and solution quality of the genetic algorithm are improved through the adaptive large neighborhood search algorithm.

[0206] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0207] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for coordinating multi-travel and multi-load AGVs in a speed-adjustable workshop, characterized in that: include: Step S1, constructing an AGV multi-stroke adjustable speed collaborative scheduling model based on a linear programming model, wherein the objective function is to minimize the maximum operation time and the total energy consumption of the machine AGV, and the constraints include operation sequence constraints, load material transportation time constraints, multi-stroke constraints, material transportation and machine operation processing start time constraints, and multi-load transportation task constraints, and the decision variables include machine operation variables, transportation task allocation variables, AGV multi-stroke transportation task allocation variables, machine operation speed level variables, AGV transportation task speed level variables, and AGV single-load and multi-load scheduling variables; Step S2, encoding the solution of the AGV multi-trip adjustable speed collaborative scheduling model through a multi-layer vector model with an initialization strategy to generate a feasible solution, performing a first decoding on the feasible solution using a single-trip constraint condition to generate an AGV single-trip initial scheduling, and performing a second decoding on the AGV single-trip initial scheduling using a multi-trip constraint condition to generate an AGV multi-trip initial scheduling; Step S3: According to whether the AGV single-trip initial scheduling meets the AGV multi-trip load capacity requirement and the AGV multi-trip material delivery time requirement, the AGV multi-trip initial scheduling or the AGV single-trip initial scheduling is respectively solved by an adaptive large neighborhood search algorithm for the AGV multi-trip adjustable speed collaborative scheduling model to generate the optimal solution for the machine AGV planning of the job shop.

2. The method for coordinated dispatching of multi-travel and multi-load AGVs in a speed-adjustable workshop according to claim 1, characterized in that: In the decision variables of step S1: The machine operation variables include whether the machine processes the operation variable, the machine operation sequence variable, the machine operation sequence variable and the machine processing sequence variable corresponding to the operation; The AGV multi-trip transport task allocation variables include whether the AGV executes a multi-load transport task variable, the AGV executes a multi-transport task sequence variable, and the AGV multi-load task transport material sequence variable; The AGV single-load and multi-load scheduling variables include whether the AGV executes a multi-load transportation task variable, whether the AGV executes a material transportation variable, and the AGV multi-load task material transportation order variable.

3. The method for coordinated dispatching of multi-travel and multi-load AGVs in a speed-adjustable workshop according to claim 2, characterized in that: The operation sequence constraints include single machine processing sequence constraints and inter-machine processing sequence constraints; The single machine processing sequence constraint is composed of a variable of whether the machine processes an operation, a variable of the machine operation speed level, a variable of the machine operation sequence, a completion time of a preceding machine operation, a machine operation processing time, a completion time of a succeeding machine operation, and a first set constant; The inter-machine processing sequence constraint is composed of whether the machine processes the operation variable, the machine operation speed level variable, the operation corresponding to the machine processing sequence variable, as well as the completion time of the previous machine operation, the machine operation processing time, the completion time of the subsequent machine operation and the second set constant.

4. The method for coordinated dispatching of multi-travel and multi-load AGVs in a speed-adjustable workshop according to claim 2, characterized in that: The multi-trip constraints include a first material delivery time constraint and a second material delivery time constraint; The first material delivery time constraint is composed of the first material delivery time, the transportation task start time, the transportation time from the first material starting position to the second material starting position, the transportation time from the second material starting position to the first material ending position, the variable of whether the AGV performs a multi-load transportation task, the AGV multi-load task transportation material sequence variable and a third set constant; The second material delivery time constraint is composed of the second material delivery time, the first material delivery time, the transportation time from the second material end position to the first material end position, whether the AGV performs a multi-load transportation task variable, the AGV multi-load task transportation material sequence variable and a fourth set constant.

5. The method for coordinated dispatching of multi-travel and multi-load AGVs in a speed-adjustable workshop according to claim 1, characterized in that: The multi-layer vector model is a four-layer vector model, and the vector encoding sequence of each layer is a machine operation order sequence, a machine processing speed level sequence, an AGV task allocation sequence and an AGV transportation speed level sequence.

6. The method for coordinated dispatching of multi-travel and multi-load AGVs in a speed-adjustable workshop according to claim 5, characterized in that: The multi-layer vector model is a four-layer vector encoding and decoding model based on a genetic algorithm, and its initialization strategy is population initialization based on topological sorting and random arrangement, wherein the population initialization process of the AGV task allocation sequence is: One or more AGVs that complete the transportation task earliest are selected, and one of the multiple AGVs is randomly selected.

7. The method for coordinated dispatching of multi-travel and multi-load AGVs in a speed-adjustable workshop according to any one of claims 1 to 6, characterized in that: The single-trip constraint conditions include the operation sequence constraint, the load material transportation time constraint, and the material transportation and machine operation processing start time constraint.

8. The method for coordinated dispatching of multi-travel and multi-load AGVs in a speed-adjustable workshop according to any one of claims 1 to 6, characterized in that: The AGV multi-trip load capacity requirement is that the AGV's transport task allocation is less than or equal to a set value; The AGV multi-trip material delivery time requirement is that the AGV's material delivery time is less than or equal to the latest delivery time.

9. The method for coordinated dispatching of multi-travel and multi-load AGVs in a speed-adjustable workshop according to any one of claims 1 to 6, characterized in that: The constraints also include a single machine operation constraint and an AGV transport task execution quantity constraint.

10. The method for coordinated dispatching of multi-travel and multi-load AGVs in a speed-adjustable workshop according to claim 1, characterized in that: The adaptive large neighborhood search algorithm is embedded in a Pareto-based genetic algorithm framework and adopts the Metropolis criterion to accept inferior solutions.

Images