A AGV configuration optimization method based on simulation particle swarm algorithm
By using the simulation particle swarm algorithm in AGV configuration optimization, the simulation model of the actual production environment and the particle swarm algorithm are optimized, the problems of high AGV configuration cost and unstable production system efficiency are solved, and more efficient production system management is achieved.
Patent Information
- Application Number
- CN202210439562.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-04-25
- Publication Date
- 2024-11-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the prior art, the AGV configuration cost is high, making it difficult to effectively simulate the production environment in large random production systems, resulting in unstable production system efficiency.
AGV configuration optimization method based on simulation particle swarm algorithm is adopted, and a mathematical model is established by building a simulation model of the actual production environment, and the improved particle swarm algorithm is used to calculate the particle fitness value, and the AGV configuration is optimized to achieve more scientific analysis and decision making.
This method can truly characterize the AGV production system, simplify the simulation of large-scale random production systems, reduce production logistics costs, and improve the efficiency of the production system.
Smart Images

Figure CN114819339B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of AGV configuration optimization, and more specifically, to an AGV configuration optimization method based on a simulation particle swarm algorithm. Background Art
[0002] In recent years, with the rapid development of technology, the frequency of product updates has accelerated, the product life cycle has shortened, and large-scale customized production has become a common production method for many companies. Personalized customized production is the future development direction of the entire manufacturing industry. Customized equipment production is a resource-oriented engineering-to-order (ETO) production method. Customized production methods generally have significant characteristics such as multiple optional process paths, order delivery time restrictions, and random production environments. They put forward higher requirements for the flexibility of the production system, prompting its production layout to evolve from the traditional cluster layout to a production system composed of multiple flexible manufacturing units. However, the traditional logistics and transportation system is difficult to meet the needs of process resources in the manufacturing process. Therefore, the random production system containing AGV (Automated Material Handling System, AMHS) transportation coupled with multiple manufacturing units has become a typical production system structure for customized equipment.
[0003] As market competition becomes increasingly fierce, enterprises are constantly looking for innovative technologies to realize and improve mass customized production to improve market competitiveness. In the production activities of enterprises, the main waste is material handling. Efficient logistics flow and reliable logistics supply in the workshop are important success factors for mass customization. This has led customized production enterprises to focus on reducing production logistics costs to reduce manufacturing costs. Among them, the use of material handling systems based on automatic guided vehicles (AGVs) in production logistics has become an important method for enterprises to reduce production logistics costs. AGVs transfer different work-in-progress batches to designated locations to ensure the normal production activities of the production system. The number of AGVs directly affects the system capacity and order delivery time. However, this type of AGV configuration cost is high. In order to reduce the production logistics cost of the production system, it is of great significance to optimize the AGV configuration of flexible flow workshops using AGV-based material handling systems.
[0004] In resource allocation optimization, production system modeling methods are divided into analytical methods and simulation methods. Both analytical methods and simulation methods can simulate the actual production environment. Many scholars use analytical methods to calculate the performance indicators of the system under each configuration, but in large-scale random production systems, using analytical methods to calculate the fitness values of different resource configuration systems is complex and difficult to implement.
[0005] The prior art discloses an AGV optimization scheduling method based on simulated annealing particle swarms. The scheme uses a particle swarm algorithm to initialize the population, determine the fitness function, calculate the fitness value of the particles, and combines the simulated annealing algorithm to update the speed and position of the particles, determine the fitness value of the new particles, and finally obtain the optimization result. It can be considered as the prior art closest to the present application. The defect of this scheme is that it does not perform calculations based on the actual production environment, cannot truly characterize the AGV configuration system, and does not solve the problem of unstable efficiency of the production system. Summary of the invention
[0006] In order to overcome the defects of the above-mentioned prior art such as high AGV configuration cost, extremely difficult to simulate a production system using analytical methods in a large random production system, and unstable production system efficiency, the present invention provides an AGV configuration optimization method based on a simulation particle swarm algorithm.
[0007] The primary purpose of the present invention is to solve the above technical problems. The technical solution of the present invention is as follows:
[0008] An AGV configuration optimization method based on a simulation particle swarm algorithm comprises the following steps:
[0009] S1. Build a simulation model of the actual production environment.
[0010] S2. Establish a mathematical model for AGV configuration optimization and set constraint parameter conditions.
[0011] S3. Using the simulation model to improve the particle swarm algorithm, an improved particle swarm algorithm is obtained; and the fitness value of the particles is calculated using the constraint parameter conditions.
[0012] S4. Use the improved particle swarm algorithm to solve the mathematical model of AGV configuration optimization and obtain the optimized AGV configuration.
[0013] Furthermore, the step S1 includes:
[0014] S1-1. Construct a simulation model according to the actual production environment, wherein the simulation model includes an AGV controller, a raw material warehouse, a finished product warehouse, an AGV pool, an AGV track, and multiple processing procedure modules; the processing procedure modules correspond to different processes in the production system, and the processing procedure modules of adjacent processes can be connected through a unidirectional AGV track; the processing procedure module contains multiple manufacturing units, and the manufacturing units are used to execute manufacturing tasks in the process, wherein each manufacturing unit has two front and rear buffer zones.
[0015] S1-2. The AGV controller monitors the simulation model in real time according to the load balancing principle through the job_asign method, selects the buffer zone to be served first, establishes the task, and generates the pickup point and unloading point of the task through the previous and next processes associated with the buffer zone.
[0016] S1-3. Calculate the number of workpieces that the AGV needs to load in the task according to the preset limiting conditions, and then assign the task to the idle AGV closest to the buffer zone that is served first. The idle AGV goes to the corresponding pickup point to pick up the goods according to the task information, and then transports the goods to the corresponding unloading point for unloading.
[0017] S1-4, after the AGV completes the task, if the AGV controller continues to assign tasks to the AGV, it continues to execute the task, otherwise the AGV returns to the AGV pool to wait for tasks.
[0018] Furthermore, the simulation model constructed according to the production environment also includes setting basic variables, and the basic variables include:
[0019] j: raw material number, j = 1, 2, 3, ..., n
[0020] k: processing type, k = 1, 2, 3, ..., n
[0021] λ j : Arrival rate of the jth type of raw materials (pieces / min)
[0022] μ k : Service rate of the kth processing step (pieces / min)
[0023] r: the rth manufacturing unit in the processing block, r = 1, 2, 3, ..., n
[0024] R k : The total number of manufacturing units in the kth processing step, r = 1, 2, 3, ..., n
[0025] When w is P, it means the front buffer of the rth manufacturing unit in the kth processing step; when w is B, it means the back buffer or finished product buffer of the rth manufacturing unit in the kth processing step
[0026] The current quantity of materials
[0027] C: AGV capacity
[0028] N: Buffer capacity
[0029] Furthermore, the preset limiting conditions are specifically:
[0030] (1) In a system with multiple workpiece inputs, the arrival of the workpieces is limited to be independent of each other and to obey the mean λ j Poisson distribution.
[0031] (2) Each manufacturing unit can only process one workpiece at a time. Their processing times are independent of each other and are ordered by the processing steps of each material, respectively subject to the rate μ k The exponential distribution of .
[0032] (3) All manufacturing units in the processing procedure module obey the post-blocking mechanism, and the front and rear buffers of each manufacturing unit have limited capacity and the capacity N is the same.
[0033] (4) Considering that there are multiple different AGV services, each AGV can serve any manufacturing unit in the system, and each operation is independent of each other. The AGV selects the delivery task based on the load balancing principle without considering the loading and unloading time of the AGV.
[0034] (5) When multiple AGVs are idle, the AGV controller selects the AGV based on the proximity principle when assigning tasks to a manufacturing unit.
[0035] (6) When the AGV transports the finished product to the finished product buffer, the finished product leaves the workshop immediately.
[0036] The purpose of setting parameters and limiting conditions of the simulation model is to be able to most realistically characterize the AGV system.
[0037] Among them, in the simulation model, tasks are generated at certain time intervals, and the AGV transports materials according to the task information. The AGV only runs along a single-channel unidirectional track.
[0038] Furthermore, in step S1-3, the calculation process of the number of workpieces required to be loaded by the AGV is as follows:
[0039] (1) According to the working objectives of AGV, the tasks in the AGV handling process are divided into two types: adding materials to the front buffer and reducing materials to the back buffer or finished product buffer.
[0040] (2) For the task of adding materials to the front buffer, the AGV controller first obtains the unloading point, that is, the maximum number of materials that the front buffer of the manufacturing unit can accommodate Then determine the pickup point, that is, the number of goods in the back buffer of the previous processing step Finally, combined with the maximum capacity C of the AGV itself, the minimum value of the three is taken to output the number of workpieces that the AGV needs to load, among which the number of goods in the raw material warehouse is infinite by default.
[0041] (3) For the task of reducing materials for the back buffer or finished product buffer, first obtain the pickup point, that is, the number of goods in the back buffer of the current manufacturing unit Then determine the maximum capacity C of the AGV itself, and finally combine it with the unloading point, that is, the number of materials that the front buffer of the next processing step can accommodate Take the minimum value of the three and output the number of workpieces that the AGV needs to load. The number of materials that the finished product warehouse can accommodate is infinite by default.
[0042] Among them, the purpose of taking the minimum value of the three as the number of workpieces required for the AGV to load is to use the minimum value of the three as the loading capacity, so that the AGV does not need to wait for picking up the goods at the pickup point, nor does it need to wait for unloading at the unloading point, which minimizes the waiting time of the AGV, saves time costs, and improves the work efficiency of the production system.
[0043] Furthermore, the specific process of step S2 is as follows:
[0044] S2-1. According to the production system to which the AGV belongs, a mathematical model for AGV configuration optimization is established, and the minimum AGV configuration cost is taken as the optimization goal. The type and quantity of AGV under this optimization goal are defined as X. * ;
[0045] S2-2. Set target production cycle Γ max It is smaller than the actual average production cycle of the production system, constraining the average production cycle of the production system E{Γ(X:ξ)}.
[0046] S2-3. Set the target average output rate θ min Greater than the actual average output rate of the production system, constraining the average output rate of the production system E{θ(X:ξ)}.
[0047] Furthermore, the mathematical model of the AGV configuration optimization has the following specific functions:
[0048]
[0049] This function is used to set the optimization goal.
[0050] The constraints of the mathematical model for AGV configuration optimization include:
[0051] E{Γ(X:ξ)}≤Γ max
[0052] E{θ(X:ξ)}≥θ min
[0053] X∈N +
[0054] Where E{Γ(X:ξ)}≤Γ maxTo ensure that the order is completed within the delivery period, E{θ(X:ξ)}≥θ min To ensure that the production system has sufficient capacity, X∈N + Used to represent the number of AGVs as a non-negative integer.
[0055] Furthermore, the step S3 is specifically:
[0056] S3-1. Initialize the particle swarm. Using the monotonicity, the simulation model is used to obtain the number of each AGV required when only one type of AGV is considered in the task. This number is used as the upper bound of the solution space of this type of AGV. The lower bound of the solution space is set to 0, indicating that this type of AGV is not used. Randomly generate a position and speed within a certain range in the solution space and assign it to each particle to obtain the initial population. The position of the i-th particle in the particle swarm is recorded as:
[0057] X i =(X i1 ,X i2 ,…,X iD ),i=1,2,…,N
[0058] The velocity of the i-th particle is recorded as:
[0059] V i =(V i1 ,V i2 ,…,V iD ),i=1,2,…,N
[0060] S3-2. Calculate the system performance index and fitness value corresponding to each particle, import the position of each particle into the simulation model, obtain the system performance index values Γ(X:ξ) and θ(X:ξ) corresponding to each particle, and judge whether the performance index meets the constraint conditions. If the conditions are met, use the normal objective function to calculate the fitness value of the particle, otherwise use the objective function with the penalty factor introduced to calculate the fitness value of the particle.
[0061] S3-3, calculate the best fitness value of a single particle, compare the current position fitness value of each particle with its historical best fitness value, if the current position fitness value is better, update the historical best fitness value, otherwise do not change the historical best fitness value, where the current best position experienced by the i-th particle is recorded as:
[0062] P besti =(P i1 ,P i2 ,…,P iD ),i=1,2,…,N
[0063] S3-4, calculate the best fitness value of the entire particle swarm, compare the current position fitness values of all particles with the global best fitness value, if the current position fitness value of a particle in the particle swarm is better than the global best fitness value, then update the global best fitness value, otherwise keep the global best fitness value, where the current best position of the i-th particle in the entire particle swarm is recorded as:
[0064] G besti =(G i1 ,G i2 ,…,G iD ),i=1,2,…,N
[0065] S3-5, updating the position and speed of the particles to obtain a new population, updating the speed of the particles, the position of the particles, and performing boundary preprocessing to obtain a new population; the boundary preprocessing directly defines the solutions that exceed the solution space as the boundary values of the solution space.
[0066] S3-6. Set the maximum number of iterations and determine whether to continue the next iteration. If the number of iterations reaches the maximum number of iterations, the optimization ends; otherwise, return to step S3-2 for the next iteration.
[0067] Wherein, the step S3-2 is used to optimize the performance of the AGV configuration production system and establish an optimization model with dual constraints of the production system output rate and the production cycle.
[0068] Among them, the purpose of boundary preprocessing after updating the particle's velocity and position is to deal with solutions that are beyond the feasibility range due to algorithm randomness problems.
[0069] Furthermore, in step S3-5, the d-dimensional velocity update formula of the ith particle is:
[0070]
[0071] The update formula for the d-dimensional position of the i-th particle is:
[0072]
[0073] Where w is a non-negative inertia factor used to adjust the convergence of the algorithm; c1 and c2 are acceleration constants; r1 and r2 are randomly generated numbers in the range of (0,1); k is the number of iterations; P iD Represents the optimal fitness value of a single particle; P gD Represents the optimal fitness value of the particle swarm.
[0074] The second aspect of the present invention provides an AGV configuration optimization system based on a simulation particle swarm algorithm, the system comprising: a memory, a processor, the memory comprising an AGV configuration optimization method program based on a simulation particle swarm algorithm, the AGV configuration optimization method program based on a simulation particle swarm algorithm when executed by the processor implements the following steps:
[0075] S1. Build a simulation model of the actual production environment;
[0076] S2, establish a mathematical model for AGV configuration optimization and set constraint parameter conditions;
[0077] S3, using the simulation model to improve the particle swarm algorithm to obtain an improved particle swarm algorithm; using the constraint parameter conditions to calculate the fitness value of the particles;
[0078] S4. Use the improved particle swarm algorithm to solve the mathematical model of AGV configuration optimization and obtain the optimized AGV configuration.
[0079] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:
[0080] The present invention provides an AGV configuration optimization method based on a simulation particle swarm algorithm. A simulation model is constructed according to an actual production environment. It is relatively simple to simulate a production system in a large random production system using a simulation method, and the AGV production system can be most realistically characterized. The simulation model is used to improve the particle swarm algorithm, which provides a more scientific analysis method and decision-making basis for AGV configuration optimization. The improved particle swarm algorithm is used to solve the mathematical model of AGV configuration optimization, which can reduce production logistics costs, reduce blockages in the production system, and improve production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0081] Figure 1 The present invention is a flow chart of an AGV configuration optimization method based on a simulation particle swarm algorithm.
[0082] Figure 2 It is a schematic diagram of a simulation model of an AGV configuration optimization method based on a simulation particle swarm algorithm of the present invention.
[0083] Figure 3 This is a flow chart of optimizing the particle swarm algorithm based on the simulation model in the present invention.
[0084] Figure 4 A schematic diagram of an AGV configuration optimization system based on a simulation particle swarm algorithm according to another embodiment of the present invention DETAILED DESCRIPTION
[0085] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.
[0086] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.
[0087] Example 1
[0088] like Figure 1-Figure 3 As shown, the present invention provides an AGV configuration optimization method based on a simulation particle swarm algorithm, specifically, comprising the following steps:
[0089] S1. Build a simulation model of the actual production environment.
[0090] S2. Establish a mathematical model for AGV configuration optimization and set constraint parameter conditions.
[0091] S3. Using the simulation model to improve the particle swarm algorithm, an improved particle swarm algorithm is obtained; and the fitness value of the particles is calculated using the constraint parameter conditions.
[0092] S4. Use the improved particle swarm algorithm to solve the mathematical model of AGV configuration optimization and obtain the optimized AGV configuration.
[0093] Firstly, a simulation model is constructed to simulate the production process of the production system. A mathematical model for AGV configuration optimization is established and constraint parameters are set. The performance indicators of the production system are solved through the simulation model. Then, the simulation model is embedded in the improved particle swarm algorithm to establish an optimization model with dual constraints of production system output rate and production cycle. The particle swarm algorithm is used to solve the optimization scheme of AGV quantity configuration.
[0094] Furthermore, the step S1 includes:
[0095] S1-1. Construct a simulation model according to the actual production environment, wherein the simulation model includes an AGV controller, a raw material warehouse, a finished product warehouse, an AGV pool, an AGV track, and multiple processing procedure modules; the processing procedure modules correspond to different processes in the production system, and the processing procedure modules of adjacent processes can be connected through a unidirectional AGV track; the processing procedure module contains multiple manufacturing units, and the manufacturing units are used to execute manufacturing tasks in the process, wherein each manufacturing unit has two front and rear buffer zones.
[0096] S1-2. The AGV controller monitors the simulation model in real time according to the load balancing principle through the job_asign method, selects the buffer zone to be served first, establishes the task, and generates the pickup point and unloading point of the task through the previous and next processes associated with the buffer zone.
[0097] S1-3. Calculate the number of workpieces that the AGV needs to load in the task according to the preset limiting conditions, and then assign the task to the idle AGV closest to the buffer zone that is served first. The idle AGV goes to the corresponding pickup point to pick up the goods according to the task information, and then transports the goods to the corresponding unloading point for unloading.
[0098] S1-4, after the AGV completes the task, if the AGV controller continues to assign tasks to the AGV, it continues to execute the task, otherwise the AGV returns to the AGV pool to wait for tasks.
[0099] like Figure 2 As shown, the naming rule of the manufacturing unit is S+the number of the processing procedure in the production system+the number of the manufacturing unit in the processing procedure, such as manufacturing unit S12 represents the second manufacturing unit of the first manufacturing procedure.
[0100] Each manufacturing unit has two front and rear buffers with the same capacity. The buffers are named B+process number+buffer category+to which the manufacturing unit belongs. Buffer category 1 represents the front buffer, and 2 represents the back buffer. In particular, in the last process, each manufacturing unit has two front buffers, and the corresponding names are AB+to which the manufacturing unit belongs+the number of the front buffer of the manufacturing unit, such as AB11 and AB12 represent the two front buffers of manufacturing unit S41. Correspondingly, the finished product buffer is named AB+manufacturing unit number.
[0101] The arrow direction on the AGV track is the AGV running direction, and each track is named in the form of T + number. In a specific embodiment, the constructed simulation model has multiple processing module, and the processing modules corresponding to adjacent processing steps can be connected through a unidirectional AGV track.
[0102] Furthermore, the simulation model constructed according to the production environment also includes setting basic variables, and the basic variables include:
[0103] j: raw material number, j = 1, 2, 3, ..., n
[0104] k: processing type, k = 1, 2, 3, ..., n
[0105] λ j : Arrival rate of the jth type of raw materials (pieces / min)
[0106] μk : Service rate of the kth processing step (pieces / min)
[0107] r: the rth manufacturing unit in the processing block, r = 1, 2, 3, ..., n
[0108] R k : The total number of manufacturing units in the kth processing step, r = 1, 2, 3, ..., n
[0109] When w is P, it means the front buffer of the rth manufacturing unit in the kth processing step; when w is B, it means the back buffer or finished product buffer of the rth manufacturing unit in the kth processing step
[0110] The current quantity of materials
[0111] C: AGV capacity
[0112] N: Buffer capacity
[0113] Furthermore, the preset limiting conditions are specifically:
[0114] (1) In a system with multiple workpiece inputs, the arrival of the workpieces is limited to be independent of each other and to obey the mean λ j Poisson distribution.
[0115] (2) Each manufacturing unit can only process one workpiece at a time. Their processing times are independent of each other and are ordered by the processing steps of each material, respectively subject to the rate μ k The exponential distribution of .
[0116] (3) All manufacturing units in the processing procedure module obey the post-blocking mechanism, and the front and rear buffers of each manufacturing unit have limited capacity and the capacity N is the same.
[0117] (4) Considering that there are multiple different AGV services, each AGV can serve any manufacturing unit in the system, and each operation is independent of each other. The AGV selects the delivery task based on the load balancing principle without considering the loading and unloading time of the AGV.
[0118] (5) When multiple AGVs are idle, the AGV controller selects the AGV based on the proximity principle when assigning tasks to a manufacturing unit.
[0119] (6) When the AGV transports the finished product to the finished product buffer, the finished product leaves the workshop immediately.
[0120] The purpose of setting parameters and limiting conditions of the simulation model is to be able to most realistically characterize the AGV system.
[0121] Among them, in the simulation model, tasks are generated at certain time intervals, and the AGV transports materials according to the task information. The AGV only runs along a single-channel unidirectional track.
[0122] In this simulation model, considering that the batch transportation characteristics of AGV can make the production system operate more efficiently, and the transportation batch is random, it is limited that the workpieces arrive at the buffer before the first processing step are independent of each other, and the probability of each manufacturing unit receiving the workpiece obeys the mean value λ j Poisson distribution, and each manufacturing unit can only process one workpiece at a time, and the processing time is consistent with the rate μ k The exponential distribution of .
[0123] The quantity that AGV carries each time is affected by the service object and task type. Therefore, when each manufacturing unit follows the "first come, first served" service rule, it can be assumed that the workpieces processed by the manufacturing unit will be accumulated in the back buffer at a predictable rate according to the back-blocking mechanism.
[0124] Furthermore, in step S1-3, the calculation process of the number of workpieces required to be loaded by the AGV is as follows:
[0125] (1) According to the working objectives of AGV, the tasks in the AGV handling process are divided into two types: adding materials to the front buffer and reducing materials to the back buffer or finished product buffer.
[0126] (2) For the task of adding materials to the front buffer, the AGV controller first obtains the unloading point, that is, the maximum number of materials that the front buffer of the manufacturing unit can accommodate Then determine the pickup point, that is, the number of goods in the back buffer of the previous processing step Finally, combined with the maximum capacity C of the AGV itself, the minimum value of the three is taken to output the number of workpieces that the AGV needs to load, among which the number of goods in the raw material warehouse is infinite by default.
[0127] (3) For the task of reducing materials for the back buffer or finished product buffer, first obtain the pickup point, that is, the number of goods in the back buffer of the current manufacturing unit Then determine the maximum capacity C of the AGV itself, and finally combine it with the unloading point, that is, the number of materials that the front buffer of the next processing step can accommodate Take the minimum value of the three and output the number of workpieces that the AGV needs to load. The number of materials that the finished product warehouse can accommodate is infinite by default.
[0128] Among them, the purpose of taking the minimum value of the three as the number of workpieces required for the AGV to load is to use the minimum value of the three as the loading capacity, so that the AGV does not need to wait for picking up the goods at the pickup point, nor does it need to wait for unloading at the unloading point, which minimizes the waiting time of the AGV, saves time costs, and improves the work efficiency of the production system.
[0129] In a specific embodiment, the AGV controller monitors that the capacity of the back buffer B122 of manufacturing unit S12 is almost full, which is the buffer that needs to be served first, and generates a task to reduce materials for B122. By comparing the remaining capacity of the front buffer of the three manufacturing units S21, S22, and S23 in the next process, it is found that the remaining capacity of the front buffer B213 is the smallest, so B122 is set as the pickup point and B213 is set as the unloading point. After calculation, the number of goods currently existing at the pickup point is Greater than the AGV capacity C and greater than the amount of material that can be accommodated at the unloading point Take the minimum value of the three and get the number of workpieces that the AGV needs to carry: After searching, the idle AGV (hereinafter referred to as AGV1) closest to B122 is located on track T12, directly below manufacturing unit S23. The above task is assigned to AGV1; AGV1 passes through tracks T12 and T13 and then turns to T11 to pick up the goods at B122. The quantity of goods picked up is Then AGV1 goes up along T11, turns to T9, then turns to T10, and unloads at B213 At this time, the AGV controller does not assign AGV1 another task, so AGV1 returns to the AGV pool along T10, T13, and T11 to wait.
[0130] Furthermore, the specific process of step S2 is:
[0131] S2-1. According to the production system to which the AGV belongs, a mathematical model for AGV configuration optimization is established, and the minimum AGV configuration cost is taken as the optimization goal. The type and quantity of AGV under this optimization goal are defined as X. * .
[0132] S2-2. Set target production cycle Γ max It is smaller than the actual average production cycle of the production system, constraining the average production cycle of the production system E{Γ(X:ξ)}.
[0133] S2-3. Set the target average output rate θ min Greater than the actual average output rate of the production system, constraining the average output rate of the production system E{θ(X:ξ)}.
[0134] Furthermore, the mathematical model of the AGV configuration optimization has the following specific functions:
[0135]
[0136] This function is used to set the optimization goal.
[0137] The constraints of the mathematical model for AGV configuration optimization include:
[0138] E{Γ(X:ξ)}≤Γ max
[0139] E{θ(X:ξ)}≥θ min
[0140] X∈N +
[0141] Where E{Γ(X:ξ)}≤Γ max To ensure that the order is completed within the delivery period, E{θ(X:ξ)}≥θ min To ensure that the production system has sufficient capacity, X∈N + Used to represent the number of AGVs as a non-negative integer.
[0142] Among them, the target production cycle Γ is set based on the actual production system max and the target output rate θ min The role is to ensure that the optimal AGV configuration obtained by the solution will not only consider the lowest AGV configuration cost, but also consider that the production capacity under actual production conditions is better than the original production system, as well as improve production efficiency.
[0143] Since there is no way to solve the system performance parameters of the optimization model using the traditional nonlinear integer programming method, and it is also impossible to express E{Γ(X:ξ)} and E{θ(X:ξ)} in the above mathematical model using the closed form of the number of AGVs, this application proposes to use a particle swarm algorithm embedded in a simulation model to solve the above AGV optimization configuration mathematical model.
[0144] In resource allocation optimization, production system modeling methods are divided into analytical methods and simulation methods. Both analytical methods and simulation methods can simulate the actual production environment. In large-scale random production systems, it is very complicated and difficult to calculate the fitness values of different resource configuration systems using analytical methods. Therefore, this application uses simulation methods to calculate system performance indicators.
[0145] Further, such as Figure 3 As shown, the step S3 is specifically as follows:
[0146] S3-1. Initialize the particle swarm. Using the monotonicity, the simulation model is used to obtain the number of each AGV required when only one type of AGV is considered in the task. This number is used as the upper bound of the solution space of this type of AGV. The lower bound of the solution space is set to 0, indicating that this type of AGV is not used. Randomly generate a position and speed within a certain range in the solution space and assign it to each particle to obtain the initial population. The position of the i-th particle in the particle swarm is recorded as:
[0147] X i =(X i1 ,X i2 ,…,X iD ),i=1,2,…,N
[0148] The velocity of the i-th particle is recorded as:
[0149] V i =(V i1 ,V i2 ,…,V iD ),i=1,2,…,N
[0150] S3-2. Calculate the system performance index and fitness value corresponding to each particle, import the position of each particle into the simulation model, obtain the system performance index value corresponding to each particle, and judge whether the performance index meets the constraint conditions. If the conditions are met, use the normal objective function to calculate the fitness value of the particle, otherwise use the objective function with the penalty factor introduced to calculate the fitness value of the particle.
[0151] S3-3, calculate the best fitness value of a single particle, compare the current position fitness value of each particle with its historical best fitness value, if the current position fitness value is better, update the historical best fitness value, otherwise do not change the historical best fitness value, where the current best position experienced by the i-th particle is recorded as:
[0152] P besti =(P i1 ,P i2 ,…,P iD ),i=1,2,…,N
[0153] In the optimization process, when calculating the fitness value of each particle, the performance index of the system is obtained by establishing data communication with the simulation, and then the fitness value of each particle is calculated using the obtained performance index.
[0154] S3-4, calculate the best fitness value of the entire particle swarm, compare the current position fitness values of all particles with the global best fitness value, if the current position fitness value of a particle in the particle swarm is better than the global best fitness value, then update the global best fitness value, otherwise keep the global best fitness value, where the current best position of the i-th particle in the entire particle swarm is recorded as:
[0155] G besti =(G i1 ,G i2 ,…,G iD ),i=1,2,…,N
[0156] S3-5, updating the position and speed of the particles to obtain a new population, updating the speed of the particles, the position of the particles, and performing boundary preprocessing to obtain a new population; the boundary preprocessing directly defines the solutions that exceed the solution space as the boundary values of the solution space.
[0157] S3-6. Set the maximum number of iterations and determine whether to continue the next iteration. If the number of iterations reaches the maximum number of iterations, the optimization ends; otherwise, return to step S3-2 for the next iteration.
[0158] Furthermore, in step S3-5, the d-dimensional velocity update formula of the ith particle is:
[0159]
[0160] The update formula for the d-dimensional position of the i-th particle is:
[0161]
[0162] Where w is a non-negative inertia factor used to adjust the convergence of the algorithm; c1 and c2 are acceleration constants; r1 and r2 are randomly generated numbers in the range of (0,1); k is the number of iterations; P iD Represents the optimal fitness value of a single particle; P gD Represents the optimal fitness value of the particle swarm.
[0163] Furthermore, the mathematical model described in step S2 is solved by using the iterative particle swarm algorithm to obtain the optimized AGV configuration.
[0164] Example 2
[0165] Based on the above-mentioned embodiment 1, this embodiment elaborates on the specific process of improving the particle swarm algorithm by using the simulation model.
[0166] Firstly, the monotonicity is used to obtain the number of each AGV required when only one type of AGV is considered through simulation, and the obtained number is defined as the upper bound of the solution space of this type of AGV.
[0167] Furthermore, the particle swarm is initialized, and the data exchange between Plant simulation and pycharm is realized according to the TCP / IP protocol. Then the upper bound obtained above is input into the particle swarm algorithm as the upper bound of the solution space, and the lower bound is set to 0, that is, this type of AGV is not used. Next, a position and speed within a certain range are generated in the solution space and assigned to each particle to obtain the initial population, and the position and speed of the i-th particle in the particle swarm are recorded.
[0168] Furthermore, the position of each particle is imported into the simulation model to obtain the system performance index value corresponding to each particle. The system performance index is judged to be qualified according to the constraint parameter conditions in the optimization model. If qualified, the normal objective function is used to calculate the fitness value; otherwise, the objective function with the penalty factor introduced is used to calculate its fitness value.
[0169] The reason for importing particles into the simulation model to calculate the system performance index is that it becomes extremely difficult to calculate using traditional analytical methods in large-scale production systems. This step provides a method for calculating the performance index of large-scale random production systems.
[0170] In a large-scale random production and manufacturing system, it is extremely difficult to calculate the system performance index directly through an analytical mathematical expression, but the system performance index can be calculated by using a simulation to simulate the production process of the system and statistically analyzing the data in the production process. This means that the fitness value in the particle swarm algorithm is not obtained by analytical calculation, but by establishing data communication between the algorithm and the simulation software, so that the algorithm obtains the system performance index value. Therefore, the present invention also provides a method for solving the calculation of performance indicators of large-scale random manufacturing systems.
[0171] Furthermore, the current position fitness values of all particles are compared with the global optimal fitness value. If the current position fitness value of a particle in the particle swarm is better than the global optimal fitness value, the current position fitness value of the particle is updated to the global optimal fitness value, otherwise the original global optimal fitness value remains unchanged.
[0172] Furthermore, the particle speed and particle position are updated, and the solutions exceeding the solution space are subjected to boundary preprocessing to obtain a new population.
[0173] Furthermore, under the premise of a preset maximum number of iterations, it is determined whether the global optimal position of the new population meets the error condition. If not, it returns to the step of calculating the particle fitness value and repeats the above operation to continue iterating. If the error condition is met, it is determined whether the number of iterations reaches the preset maximum number of iterations. If not, it also returns to the step of calculating the particle fitness value and repeats the above operation to continue iterating. If the condition is met, the optimization ends and a particle swarm algorithm embedded in the simulation model is obtained.
[0174] Example 3
[0175] like Figure 4 As shown, this embodiment provides an AGV configuration optimization system based on a simulation particle swarm algorithm, characterized in that the system includes: a memory, a processor, the memory includes an AGV configuration optimization method program based on a simulation particle swarm algorithm, and the AGV configuration optimization method program based on a simulation particle swarm algorithm is executed by the processor to implement the following steps:
[0176] S1. Build a simulation model of the actual production environment;
[0177] S2, establish a mathematical model for AGV configuration optimization and set constraint parameter conditions;
[0178] S3, using the simulation model to improve the particle swarm algorithm to obtain an improved particle swarm algorithm; using the constraint parameter conditions to calculate the fitness value of the particles;
[0179] S4. Use the improved particle swarm algorithm to solve the mathematical model of AGV configuration optimization and obtain the optimized AGV configuration.
[0180] Furthermore, the step S1 is specifically as follows:
[0181] S1-1. Construct a simulation model according to the actual production environment, wherein the simulation model includes an AGV controller, a raw material warehouse, a finished product warehouse, an AGV pool, an AGV track, and multiple processing procedure modules; the processing procedure modules correspond to different processes in the production system, and the processing procedure modules of adjacent processes can be connected through a unidirectional AGV track; the processing procedure module contains multiple manufacturing units, and the manufacturing units are used to execute manufacturing tasks in the process, wherein each manufacturing unit has two front and rear buffer zones.
[0182] S1-2. The AGV controller monitors the model in real time according to the load balancing principle through the job_asign method, selects the buffer zone to be served first, establishes the task, and generates the pickup point and unloading point of the task through the previous and next processes associated with the buffer zone.
[0183] S1-3. Calculate the number of workpieces that the AGV needs to load in the task according to the preset limiting conditions, and then assign the task to the idle AGV closest to the buffer zone that is served first. The idle AGV goes to the corresponding pickup point to pick up the goods according to the task information, and then transports the goods to the corresponding unloading point for unloading.
[0184] S1-4, after the AGV completes the task, if the AGV controller continues to assign tasks to the AGV, it continues to execute the task, otherwise the AGV returns to the AGV pool to wait for tasks.
[0185] Furthermore, the simulation model constructed according to the production environment also includes setting basic variables, and the basic variables include:
[0186] j: raw material number, j = 1, 2, 3, ..., n
[0187] k: processing type, k = 1, 2, 3, ..., n
[0188] λ j : Arrival rate of the jth type of raw materials (pieces / min)
[0189] μ k : Service rate of the kth processing step (pieces / min)
[0190] r: the rth manufacturing unit in the processing block, r = 1, 2, 3, ..., n
[0191] R k : The total number of manufacturing units in the kth processing step, r = 1, 2, 3, ..., n
[0192] When w is P, it means the front buffer of the rth manufacturing unit in the kth processing step; when w is B, it means the back buffer or finished product buffer of the rth manufacturing unit in the kth processing step
[0193] The current quantity of materials
[0194] C: AGV capacity
[0195] N: Buffer capacity
[0196] Furthermore, the preset limiting conditions are specifically:
[0197] (1) In a system with multiple workpiece inputs, the arrival of the workpieces is limited to be independent of each other and to obey the mean λ j Poisson distribution.
[0198] (2) Each manufacturing unit can only process one workpiece at a time. Their processing times are independent of each other and are ordered by the processing steps of each material, respectively subject to the rate μ k The exponential distribution of .
[0199] (3) All manufacturing units in the processing procedure module obey the post-blocking mechanism, and the front and rear buffers of each manufacturing unit have limited capacity and the capacity N is the same.
[0200] (4) Considering that there are multiple different AGV services, each AGV can serve any manufacturing unit in the system, and each operation is independent of each other. The AGV selects the delivery task based on the load balancing principle without considering the loading and unloading time of the AGV.
[0201] (5) When multiple AGVs are idle, the AGV controller selects the AGV based on the proximity principle when assigning tasks to a manufacturing unit.
[0202] (6) When the AGV transports the finished product to the finished product buffer, the finished product leaves the workshop immediately.
[0203] Furthermore, in step S1-3, the calculation process of the number of workpieces required to be loaded by the AGV is as follows:
[0204] (1) According to the working objectives of AGV, the tasks in the AGV handling process are divided into two types: adding materials to the front buffer and reducing materials to the back buffer or finished product buffer.
[0205] (2) For the task of adding materials to the front buffer, the AGV controller first obtains the unloading point, that is, the maximum number of materials that the front buffer of the manufacturing unit can accommodate Then determine the pickup point, that is, the number of goods in the back buffer of the previous processing step Finally, combined with the maximum capacity C of the AGV itself, the minimum value of the three is taken to output the number of workpieces that the AGV needs to load, among which the number of goods in the raw material warehouse is infinite by default.
[0206] (3) For the task of reducing materials for the back buffer or finished product buffer, first obtain the pickup point, that is, the number of goods in the back buffer of the current manufacturing unit Then determine the maximum capacity C of the AGV itself, and finally combine it with the unloading point, that is, the number of materials that the front buffer of the next processing step can accommodate Take the minimum value of the three and output the number of workpieces that the AGV needs to load. The number of materials that the finished product warehouse can accommodate is infinite by default.
[0207] Furthermore, the specific process of step S2 is as follows:
[0208] S2-1. According to the production system to which the AGV belongs, a mathematical model for AGV configuration optimization is established, and the minimum AGV configuration cost is taken as the optimization goal. The type and quantity of AGV under this optimization goal are defined as X. * ;
[0209] S2-2. Set target production cycle Γ max It is smaller than the actual average production cycle of the production system, constraining the average production cycle of the production system E{Γ(X:ξ)}.
[0210] S2-3. Set the target average output rate θ min Greater than the actual average output rate of the production system, constraining the average output rate of the production system E{θ(X:ξ)}.
[0211] Furthermore, the mathematical model of the AGV configuration optimization has the following specific functions:
[0212]
[0213] This function is used to set the optimization goal.
[0214] The constraints of the mathematical model for AGV configuration optimization include:
[0215] E{Γ(X:ξ)}≤Γ max
[0216] E{θ(X:ξ)}≥θ min
[0217] X∈N +
[0218] Where E{Γ(X:ξ)}≤Γ max To ensure that the order is completed within the delivery period, E{θ(X:ξ)}≥θ min To ensure that the production system has sufficient capacity, X∈N + Used to represent the number of AGVs as a non-negative integer.
[0219] Furthermore, the step S3 is specifically as follows:
[0220] S3-1. Initialize the particle swarm. Using the monotonicity, the simulation model is used to obtain the number of each AGV required when only one type of AGV is considered in the task. This number is used as the upper bound of the solution space of this type of AGV. The lower bound of the solution space is set to 0, indicating that this type of AGV is not used. Randomly generate a position and speed within a certain range in the solution space and assign it to each particle to obtain the initial population. The position of the i-th particle in the particle swarm is recorded as:
[0221] X i =(X i1 ,X i2 ,…,X iD ),i=1,2,…,N
[0222] The velocity of the i-th particle is recorded as:
[0223] V i =(V i1 ,V i2 ,…,V iD ),i=1,2,…,N
[0224] S3-2. Calculate the system performance index and fitness value corresponding to each particle, import the position of each particle into the simulation model, obtain the system performance index values Γ(X:ξ) and θ(X:ξ) corresponding to each particle, and judge whether the performance index meets the constraint conditions. If the conditions are met, use the normal objective function to calculate the fitness value of the particle, otherwise use the objective function with the penalty factor introduced to calculate the fitness value of the particle.
[0225] S3-3, calculate the best fitness value of a single particle, compare the current position fitness value of each particle with its historical best fitness value, if the current position fitness value is better, update the historical best fitness value, otherwise do not change the historical best fitness value, where the current best position experienced by the i-th particle is recorded as:
[0226] P besti =(P i1 ,P i2 ,…,P iD ),i=1,2,…,N
[0227] S3-4, calculate the best fitness value of the entire particle swarm, compare the current position fitness values of all particles with the global best fitness value, if the current position fitness value of a particle in the particle swarm is better than the global best fitness value, then update the global best fitness value, otherwise keep the global best fitness value, where the current best position of the i-th particle in the entire particle swarm is recorded as:
[0228] G besti =(G i1 ,G i2 ,…,G iD ),i=1,2,…,N
[0229] S3-5, updating the position and speed of the particles to obtain a new population, updating the speed of the particles, the position of the particles, and performing boundary preprocessing to obtain a new population; the boundary preprocessing directly defines the solutions that exceed the solution space as the boundary values of the solution space.
[0230] S3-6. Set the maximum number of iterations and determine whether to continue the next iteration. If the number of iterations reaches the maximum number of iterations, the optimization ends; otherwise, return to step S3-2 for the next iteration.
[0231] The icons describing the structural positional relationships in the accompanying drawings are for illustrative purposes only and should not be construed as limitations on this patent.
[0232] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.
Claims
1. An AGV configuration optimization method based on simulation particle swarm algorithm, characterized in that: The following steps are involved: S1. Build a simulation model of the actual production environment; S1-1. Construct a simulation model according to the actual production environment, wherein the simulation model includes an AGV controller, a raw material warehouse, a finished product warehouse, an AGV pool, an AGV track, and multiple processing procedure modules; the processing procedure modules correspond to different processes in the production system, and the processing procedure modules of adjacent processes can be connected through a unidirectional AGV track; the processing procedure module includes multiple manufacturing units, and the manufacturing units are used to perform manufacturing tasks in the process, wherein each manufacturing unit has two front and rear buffer zones; S1-2, the AGV controller monitors the simulation model in real time according to the load balancing principle through the job_asign method, selects the buffer zone to be served first, establishes the task, and generates the pickup point and unloading point of the task through the previous and next processes associated with the buffer zone; S1-3, calculate the number of workpieces that the AGV needs to load in the task according to the preset limiting conditions, and then assign the task to the idle AGV closest to the buffer zone served first. The idle AGV goes to the corresponding pickup point to pick up the goods according to the task information, and then transports the goods to the corresponding unloading point for unloading; S1-4, after the AGV completes the task, if the AGV controller continues to assign tasks to the AGV, it continues to execute the task, otherwise the AGV returns to the AGV pool to wait for tasks; S2, establish a mathematical model for AGV configuration optimization and set constraint parameter conditions; S3, using the simulation model to improve the particle swarm algorithm to obtain an improved particle swarm algorithm; using the constraint parameter conditions to calculate the fitness value of the particles; S3-1. Initialize the particle swarm. Using the monotonicity, the simulation model is used to obtain the number of each AGV required when only one type of AGV is considered in the task. This number is used as the upper bound of the solution space of this type of AGV. The lower bound of the solution space is set to 0, indicating that this type of AGV is not used. Randomly generate a position and speed within a certain range in the solution space and assign it to each particle to obtain the initial population. The position of the i-th particle in the particle swarm is recorded as: X i =(X i1 ,X i2 ,…,X iD ),i=1,2,…,N The velocity of the i-th particle is recorded as: V i =(V i1 ,V i2 ,…,V iD ),i=1,2,…,N; S3-2, calculate the system performance index and fitness value corresponding to each particle, import the position of each particle into the simulation model, obtain the system performance index value Γ(X:ξ) and θ(X:ξ) corresponding to each particle, and judge whether the performance index meets the constraint conditions. If the conditions are met, the fitness value of the particle is calculated using the normal objective function, otherwise, the fitness value of the particle is calculated using the objective function with the penalty factor introduced; S3-3, calculate the best fitness value of a single particle, compare the current position fitness value of each particle with its historical best fitness value, if the current position fitness value is better, update the historical best fitness value, otherwise do not change the historical best fitness value, where the current best position experienced by the i-th particle is recorded as: P besti =(P i1 ,P i2 ,...,P iD ),i=1,2,...,N; S3-4, calculate the best fitness value of the entire particle swarm, compare the current position fitness values of all particles with the global best fitness value, if the current position fitness value of a particle in the particle swarm is better than the global best fitness value, then update the global best fitness value, otherwise keep the global best fitness value, where the current best position of the i-th particle in the entire particle swarm is recorded as: G besti =(G i1 ,G i2 ,...,G iD ),i=1,2,...,N; S3-5, updating the position and speed of the particles to obtain a new population, updating the speed and position of the particles, and performing boundary preprocessing to obtain a new population; the boundary preprocessing directly defines the solutions exceeding the solution space as the boundary values of the solution space; S3-6, set the maximum number of iterations, and determine whether to continue the next iteration. If the number of iterations reaches the maximum number of iterations, the optimization ends, otherwise return to step S3-2 for the next iteration; S4. Use the improved particle swarm algorithm to solve the mathematical model of AGV configuration optimization and obtain the optimized AGV configuration.
2. The AGV configuration optimization method based on simulation particle swarm algorithm according to claim 1 is characterized in that: The simulation model constructed according to the production environment also includes setting basic variables, and the basic variables include: j: raw material number, j = 1, 2, 3, ..., n; k: processing operation type, k = 1, 2, 3, ..., n; λ j : Arrival rate of the jth type of raw materials (pieces / min); μ k : The service rate of the kth processing step (pieces / min); r: the rth manufacturing unit in the processing block, r = 1, 2, 3, ..., n; R k : The total number of manufacturing units in the kth processing step, r = 1, 2, 3, ..., n; When w is P, it indicates the front buffer of the rth manufacturing unit in the kth processing step; when w is B, it indicates the back buffer or finished product buffer of the rth manufacturing unit in the kth processing step; The quantity of materials on hand; C: AGV capacity; N: Buffer capacity.
3. The AGV configuration optimization method based on simulation particle swarm algorithm according to claim 2 is characterized in that: The preset limiting conditions are specifically: (1) In a system with multiple workpiece inputs, the arrival of the workpieces is limited to be independent of each other and to obey the mean λ j Poisson distribution of (2) Each manufacturing unit can only process one workpiece at a time. Their processing times are independent of each other and are ordered by the processing steps of each material, respectively subject to the rate μ k exponential distribution of (3) All manufacturing units in the processing module obey the back-blocking mechanism, and the front and back buffers of each manufacturing unit have limited capacity and the capacity N is the same; (4) Considering that there are multiple different AGV services, each AGV can serve any manufacturing unit in the system, and each operation is independent of each other. The AGV selects the delivery task based on the load balancing principle without considering the loading and unloading time of the AGV; (5) When multiple AGVs are idle, the AGV controller selects the AGV based on the proximity principle when assigning tasks to a manufacturing unit; (6) When the AGV transports the finished product to the finished product buffer, the finished product leaves the workshop immediately.
4. The AGV configuration optimization method based on simulation particle swarm algorithm according to claim 1 is characterized in that: In step S1-3, the calculation process of the number of workpieces required to be loaded by the AGV is as follows: (1) According to the working objectives of AGV, the tasks in the AGV handling process are divided into two types: adding materials to the front buffer and reducing materials to the back buffer or finished product buffer; (2) For the task of adding materials to the front buffer, the AGV controller first obtains the unloading point, that is, the maximum number of materials that the front buffer of the manufacturing unit can accommodate Then determine the pickup point, that is, the number of goods in the back buffer of the previous processing step Finally, combined with the maximum capacity C of the AGV itself, the minimum value of the three is taken to output the number of workpieces that the AGV needs to load, where the number of goods in the raw material warehouse is infinite by default; (3) For the task of reducing materials for the back buffer or finished product buffer, first obtain the pickup point, that is, the number of goods in the back buffer of the current manufacturing unit. Then determine the maximum capacity C of the AGV itself, and finally combine it with the unloading point, that is, the number of materials that the front buffer of the next processing step can accommodate Take the minimum value of the three and output the number of workpieces that the AGV needs to load. The number of materials that the finished product warehouse can accommodate is infinite by default.
5. The AGV configuration optimization method based on simulation particle swarm algorithm according to claim 1 is characterized in that: The specific process of step S2 is: S2-1. According to the production system to which the AGV belongs, a mathematical model for AGV configuration optimization is established, and the minimum AGV configuration cost is taken as the optimization goal. The type and quantity of AGV under this optimization goal are defined as X. * ; S2-2. Set target production cycle Γ max Less than the actual average production cycle of the production system, constraining the average production cycle of the production system E{Γ(X:ξ)}; S2-3. Set the target average output rate θ min Greater than the actual average output rate of the production system, constraining the average output rate of the production system E{θ(X:ξ)}.
6. The AGV configuration optimization method based on simulation particle swarm algorithm according to claim 5 is characterized in that: The mathematical model of AGV configuration optimization, the specific function is as follows: This function is used to set the optimization goal; The constraints of the mathematical model for AGV configuration optimization include: E{Γ(X:ξ)}≤Γ max E{θ(X:ξ)}≥θ min X∈N + Among them, E{Γ(X:ξ)} represents the average production cycle of the production system, E{θ(X:ξ)} represents the average output rate of the production system, and Γ max Target production cycle and target output rate θ min It means that E{Γ(X:ξ)}≤Γ max To ensure that the order is completed within the delivery period, E{θ(X:ξ)}≥θ min To ensure that the production system has sufficient capacity, X∈N + Used to represent the number of AGVs as a non-negative integer.
7. An AGV configuration optimization method based on simulation particle swarm algorithm according to claim 6, characterized in that: In step S3-5, the formula for updating the d-dimensional velocity of the ith particle is: The update formula for the d-dimensional position of the i-th particle is: Where w is a non-negative inertia factor used to adjust the convergence of the algorithm; c1 and c2 are acceleration constants; r1 and r2 are randomly generated numbers in the range of (0, 1); k is the number of iterations; P iD Represents the optimal fitness value of a single particle; P gD Represents the optimal fitness value of the particle swarm.
8. An AGV configuration optimization system based on simulation particle swarm algorithm, characterized in that: The system includes: a memory and a processor, wherein the memory includes an AGV configuration optimization method program based on a simulation particle swarm algorithm, and when the AGV configuration optimization method program based on a simulation particle swarm algorithm is executed by the processor, the steps of an AGV configuration optimization method based on a simulation particle swarm algorithm as described in any one of claims 1-7 are implemented.
