Joint optimization method for AGV guiding path and loading and unloading point

By constructing a mathematical model and combining genetic algorithms and simple variable neighborhood search algorithms, the loading and unloading points and guiding paths of AGV are optimized, and the problem that AGV handling efficiency is affected by loading and unloading points and paths is solved, and the handling efficiency of AGV is improved.

CN119067551AInactive Publication Date: 2024-12-03XIANGNAN UNIV
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202411293043.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-14
Publication Date
2024-12-03
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

In the intelligent material storage and transportation system, the handling efficiency of AGV is affected by the location of loading and unloading points and the direction of the path. It is difficult for the prior art to optimize both at the same time to improve the handling efficiency of AGV.

Method used

A joint optimization method is adopted to optimize the loading and unloading point positions and guidance paths of AGVs by constructing mathematical models and combining improved genetic algorithms and minimalist variable neighborhood search algorithms to minimize the load and total transportation distance of AGVs.

Benefits of technology

The handling efficiency of AGV has been improved, and by simultaneously optimizing the loading and unloading point locations and guidance paths, the handling cost is reduced and the efficiency of the production process is improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119067551A_ABST
    Figure CN119067551A_ABST
Patent Text Reader

Abstract

A joint optimization method for AGV guiding paths and loading and unloading points comprises the following steps: S100, on the premise that facility layout and logistics quantity among units in a system are known, an undirected path network exists, and each working unit has a plurality of corresponding selectable points which can be used as the loading and unloading points; the position of each unit loading and unloading point and the direction of each guide path need to be determined, and a mathematical model for optimizing the guide paths and the loading and unloading point positions is constructed with the purpose of minimizing the total transportation distance formed by the AGV load distance and the no-load distance; s200, solving the mathematical model through an improved genetic algorithm; and S300, substituting the position data of the AGV loading and unloading point obtained in the step S200 into a simple variable neighborhood search algorithm, so as to solve an AGV guiding path in order to achieve the purpose of minimizing the total transportation distance including the AGV load distance and the no-load distance at the same time.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of AGV application, and in particular to a joint optimization method for an AGV guide path and a loading and unloading point. Background Art

[0002] In the intelligent material storage and transportation system, the cost of material handling accounts for a large proportion of the total production cost and affects the efficiency of production processes and facility planning. The use of AGV can greatly reduce the cost of handling and improve handling efficiency. However, the location of the loading and unloading points determines the starting and ending points of AGV handling, and the direction of each path in the path network determines the moving path of each AGV handling. These two factors will determine the handling distance of the AGV, thereby affecting the handling efficiency. Therefore, jointly optimizing the path direction and the location of the loading and unloading points will improve the AGV handling efficiency better than optimizing either one unilaterally. Summary of the invention

[0003] In view of the above-mentioned defects, the purpose of the present invention is to propose a joint optimization method for AGV guide path and loading and unloading points, which will improve the AGV handling efficiency better than optimizing either one unilaterally.

[0004] To achieve this object, the present invention adopts the following technical solutions:

[0005] A joint optimization method for AGV guidance path and loading and unloading points, comprising the following steps:

[0006] S100. Under the premise of a known system layout of facilities and the amount of logistics between units, there is an undirected path network, and each work unit has corresponding multiple optional points that can be used as loading and unloading points; it is necessary to determine the position of each unit loading and unloading point and the direction of each guide path, and to minimize the total transportation distance formed by the AGV loaded distance and empty distance, and to construct a mathematical model for optimizing the guide path and the loading and unloading point position;

[0007] S200, solving the mathematical model by improving the genetic algorithm, so as to obtain the position data of the AGV loading and unloading point for the purpose of minimizing the total transportation distance of the AGV including the AGV loaded distance and the unloaded distance;

[0008] S300, bringing the location data of the AGV loading and unloading points obtained in step S200 into a simplified variable neighborhood search algorithm, so as to obtain the AGV guidance path with the purpose of minimizing the total transportation distance including both the AGV loaded distance and the empty distance.

[0009] Preferably, the mathematical model in step S100 is as follows:

[0010] There is no upper limit to AGV transport capacity;

[0011] Each unit has a unique loading and unloading port, and the position is known;

[0012] The transport paths of AGVs are all the shortest paths;

[0013] Regardless of whether the AGV will be blocked;

[0014] Among them, P n is the feeding port, n∈[1,N]; D n is the unloading port, n∈[1,N]; E(i,j) is the lane from vertex i to vertex j; T ij is the lane direction; H ij is the lane length; F bd From the feed port D b To the feeding port P d Number of handling times; G db From the feeding port P d To the feeding port D b No-load times; L ij is the shortest path from vertex i to vertex j; Is lane E(i,j) located at the unloading port D? b To the feeding port P d on the transport route; Is lane E(i,j) located at loading port P? d To the feeding port D b on empty routes; The loading port of the nth unit is on lane E(i,j); The unloading port of the nth unit is on lane E(i,j); q bd is the load flow matrix between unit b and unit d; is the no-load flow matrix between unit b and unit d;

[0015] The mathematical model of this problem is as follows:

[0016]

[0017]

[0018] T ij +T ji =1# Formula (4.12

[0019] ∑ i∈S T ij ≥1# Formula (4.13

[0020] ∑ j∈S T ij ≥1# Formula (4.14

[0021]

[0022] ∑ d∈N F bd =∑ d∈N G db # Formula (4.17

[0023]

[0024] Formula (4.1) is the main goal of this mathematical programming model. The optimization goal is to minimize the total distance of the AGV, which includes the load travel and the empty travel.

[0025] Constraints (4.2) and (4.3) ensure that after the route is selected, the length of the selected route is equal to the empty distance; in other words, whether it is from the loading port P b Departure, passing the corresponding unloading port D d Then return to the feeding port P b Path, or from the feed port D d Departure, passing through the loading port P b Then return to the feeding port D d The paths of are all equal to the length of the selected path;

[0026] Constraints (4.4) and (4.5) ensure that the feeding port P b The number of lanes is only one; similarly, constraints (4.6) and (4.7) are also to ensure that the inlet and outlet D d The number of lanes is only one;

[0027] The purpose of constraints (4.8) and (4.9) is to ensure that the flow of all vertices (except loading and unloading points) is balanced; that is, the number of times the AGV enters and leaves a non-loading and unloading point vertex must be equal;

[0028] Constraints (4.10) and (4.11) ensure that the lanes are not reused during either the transport phase or the empty return phase;

[0029] Constraint (4.12) is to ensure that each lane has only one direction;

[0030] Constraints (4.13) and (4.14) are to ensure that for all vertices in the vertex set, there is a path starting from and ending at the vertex, so as to ensure that the guided path network is strongly connected;

[0031] Constraints (4.15) and (4.16) ensure that the loading and unloading points are on the load route and the empty route;

[0032] Constraints (4.17) and (4.18) are to ensure the logistics balance of the AGV, that is, its number of transports and empty loads should be equal;

[0033] Constraints (4.19) and (4.20) ensure that each unit has only one corresponding loading port and unloading port, respectively;

[0034] Constraint (4.21) specifies the range of all variables, that is, all variables are 0-1 variables and can only take 0 or 1. ij is a known constant, representing the direction of each guided path in the path network.

[0035] Preferably, step S200 is specifically as follows:

[0036] S210, population initialization, using tuple coding with points instead of edges as the coding method, and using an undirected graph-based method to generate an initial solution;

[0037] S220, calculating the fitness of the initial solution, the function calculates the corresponding fitness of the objective function value of the individual through linear scaling transformation, after the linear transformation, the fitness of a few individuals with high original fitness is proportionally reduced, while the fitness of individuals with poor fitness is proportionally increased, which is conducive to the diversity of the population;

[0038] S230, selection operation, the genetic algorithm adopts the elite retention tournament selection, randomly selects from a certain number of individuals, and then selects the individual with the best fitness as the parent;

[0039] S240, crossover operation. In the crossover operation link of the genetic algorithm, a partial matching crossover with correction and deduplication is adopted to avoid duplication of genes in different units of the two chromosomes undergoing the crossover operation;

[0040] S250, mutation operation, uses single gene mutation; first, it will traverse each chromosome and generate a random number each time. If the random number is greater than the mutation probability, no mutation will be performed; otherwise, proceed to the next step: randomly select a gene on the individual for mutation, and select a position that does not exist in the individual from the corresponding unit optional position as a replacement.

[0041] Furthermore, step S300 is specifically as follows: a simplified variable neighborhood search algorithm: for two strongly connected directed graphs D and D', if D and D' have the same base graph G, then by reversing the path and the circle, D can be converted into D'; the algorithm is used to obtain all loops of the directed graph D, generate a loop neighborhood, and search to determine the corresponding path orientation.

[0042] Furthermore, the RVNS algorithm process is:

[0043] Input:(kmax , feasible solution ε);

[0044] Output: local optimal solution β best ;

[0045] 1:ε best :=β,k:=1;

[0046] 2: While k≤k max do;

[0047] 3:if k=1then;

[0048] 4: Obtain all the circles of the directed graph D corresponding to β, forming the neighborhood N1(β best );

[0049] 5: Beta * :=random(N1(β best ))#Randomly select neighborhood structure N1(β best )

[0050] 6: If OFV(β * ) <ODV(β best ), then let β best :=β * , otherwise k:=k+1;

[0051] 7:ifk=2then;

[0052] 8: Use the network flow algorithm to find that β corresponds to the directed graph D with two arcs that do not intersect each other, forming the neighborhood N2(β best );

[0053] 9: Beta * :=random(N2(β best ))#Randomly select neighborhood structure N2(β best )

[0054] 10: If OFV(β * ) <ODV(β best ), then let β best :=β * , otherwise k:=k+1.

[0055] Furthermore, in step S210, the workshop layout is converted into an undirected graph, and the edges of the undirected graph are used to indicate whether the units have common boundaries, and these boundaries can be used as loading and unloading point locations;

[0056] Since the vertices are located at corners or intersections, in order to avoid congestion of AGVs, the vertices are not considered as loading and unloading points. Instead, it is stipulated that the loading and unloading points can only be located at the center of these boundaries. The units are represented by vertices, and 4 auxiliary vertices (N, S, W, E) are added to represent the boundaries of the workshop that are not adjacent to other workshops.

[0057] By randomly generating a row of codes containing unit numbers and directional letters, a feasible initial solution without duplication can be obtained by decoding;

[0058] After the code is generated according to the above method, it is necessary to decode the code from left to right to get the initial solution. The decoding rules are as follows:

[0059] If the current vertex is an auxiliary vertex and the next vertex is also an auxiliary vertex, the position of the unit's loading and unloading point is uncertain;

[0060] If the current vertex is an auxiliary vertex and the next vertex is a unit, it means that the loading and unloading point of the next vertex unit is located on the common boundary of the two vertex units;

[0061] If the current vertex is a unit, and the unit has no assigned loading and unloading point; if the next vertex is an auxiliary vertex, the unit loading and unloading point is located on the boundary corresponding to the auxiliary vertex; if the next vertex is a unit, the current vertex unit loading and unloading point is located on the common boundary of the two units;

[0062] If the current vertex is a unit, and the unit is assigned a loading and unloading point; if the next vertex is an auxiliary vertex, if the next vertex is a unit, then the loading and unloading point of the next vertex unit is located on the common boundary of the two units;

[0063] If the current vertex is a unit and the unit has no assigned loading and unloading point location; and it is also the last vertex of the route, then the first vertex of the segment is taken, and the loading and unloading point is located on the common boundary.

[0064] Furthermore, in step S220, the calculation formula is:

[0065] F'=aF+b# Formula (3.19)

[0066] Where F is a function that directly uses the objective function value as the fitness value, F' is the fitness value function after linear scaling of F, and F' needs to meet the following two conditions:

[0067] (1) To ensure that the expected number of copies of individuals with average fitness in the generated offspring is 1, the average value of the original fitness F avg = the average value of the new fitness F' avg ;

[0068] (2) In order to limit the number of copies of the individual with the highest fitness in the next generation, we set the average value of the original fitness times the specified value as the maximum fitness after transformation, that is, F' max =Smul*F avg , Smul is a constant that determines the upper bound of the linear transformation, and its default value is generally 2.

[0069] Furthermore, elite retention tournament selection combines the elite retention strategy and the tournament selection strategy, and the specific operations are as follows:

[0070] 1. Select the best individuals from the population as elite individuals and directly copy them to the next generation of population to ensure that the best individuals are retained;

[0071] 2. For the remaining selections, the tournament selection method is adopted; among a certain number of randomly selected individuals, the individual with the best fitness is selected as the parent.

[0072] One of the above technical solutions includes the following beneficial effects: In the intelligent material storage and transportation system, the cost of material handling accounts for a large proportion of the total production cost and affects the efficiency of production processes and facility planning. The use of AGV can greatly reduce the cost of handling and improve the handling efficiency. However, the location of the loading and unloading points determines the starting and ending points of the AGV handling, and the direction of each path in the path network determines the moving path of the AGV each time. These two factors will determine the handling distance of the AGV, thereby affecting the handling efficiency. Therefore, the path direction and the location of the loading and unloading points are jointly optimized; the genetic algorithm is an intelligent optimization algorithm obtained by simulating the process of biological genes undergoing natural selection, crossover mutation, and survival of the fittest in Darwin's theory of evolution. Due to its diverse algorithm encoding methods, it is conducive to global optimization; the reduced variable neighborhood search (RVNS) algorithm, as a variant of VNS, removes the perturbation operation in the traditional neighborhood search, generates solutions based only on random neighborhood operations, and compares them with the currently known best solutions to update the local optimal solutions, showing high search efficiency; the genetic algorithm and the reduced variable neighborhood search algorithm are combined to first calculate the AGV loading and unloading points, and then calculate the guide path based on the loading and unloading point data. The entire scheme optimizes the AGV loading and unloading points and the guide path at the same time, and the improvement of AGV handling efficiency will be better than optimizing either one unilaterally. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 It is a schematic diagram of the overall process of an embodiment of the present invention;

[0074] Figure 2 is a schematic diagram of the principle of a four-unit diagram of an embodiment of the present invention;

[0075] Figure 3 is a schematic diagram of the encoding principle of an embodiment of the present invention;

[0076] Figure 4 is a schematic diagram of the principle of generating an initial solution according to an embodiment of the present invention;

[0077] Figure 5 is a schematic diagram of a partially matched crossover according to an embodiment of the present invention;

[0078] Figure 6 It is a schematic diagram of a variation operation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0079] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0080] like Figure 1-2 As shown, a joint optimization method for AGV guidance path and loading and unloading points includes the following steps:

[0081] S100. Under the premise of a known system layout of facilities and the amount of logistics between units, there is an undirected path network, and each work unit has corresponding multiple optional points that can be used as loading and unloading points; it is necessary to determine the position of each unit loading and unloading point and the direction of each guide path, and to minimize the total transportation distance formed by the AGV loaded distance and empty distance, and to construct a mathematical model for optimizing the guide path and the loading and unloading point position;

[0082] S200, solving the mathematical model by improving the genetic algorithm, so as to obtain the position data of the AGV loading and unloading point for the purpose of minimizing the total transportation distance of the AGV including the AGV loaded distance and the unloaded distance;

[0083] S300, bringing the location data of the AGV loading and unloading points obtained in step S200 into a simplified variable neighborhood search algorithm, so as to obtain the AGV guidance path with the purpose of minimizing the total transportation distance including both the AGV loaded distance and the empty distance.

[0084] In intelligent material storage and transportation systems, the cost of material handling accounts for a large proportion of the total production cost and affects the efficiency of production processes and facility planning. The use of AGV can greatly reduce the cost of handling and improve handling efficiency. However, the location of the loading and unloading points determines the starting and ending points of AGV handling, and the direction of each path in the path network determines the moving path of each AGV handling. These two factors will determine the handling distance of the AGV, thereby affecting the handling efficiency. Therefore, the path direction and the location of the loading and unloading points are jointly optimized; the genetic algorithm is an intelligent optimization algorithm obtained by Professor J. Holland in 1975 by simulating the process of natural selection, crossover mutation and survival of the fittest of biological genes in Darwin's theory of evolution. Due to its diverse algorithm encoding methods, it is conducive to global optimization; the reduced variable neighborhood search (RVNS) algorithm, as a variant of VNS, removes the perturbation operation in the traditional neighborhood search, generates solutions based only on random neighborhood operations, and compares them with the currently known best solutions to update the local optimal solutions, showing high search efficiency; the genetic algorithm and the reduced variable neighborhood search algorithm are combined to first calculate the AGV loading and unloading points, and then calculate the guidance path based on the loading and unloading point data. The entire scheme optimizes the AGV loading and unloading points and the guidance path at the same time, and the improvement of AGV handling efficiency will be better than optimizing either one unilaterally.

[0085] The mathematical model in step S100 is as follows:

[0086] There is no upper limit to AGV transport capacity;

[0087] Each unit has a unique loading and unloading port, and the position is known;

[0088] The transport paths of AGVs are all the shortest paths;

[0089] It does not consider whether the AGV will be blocked;

[0090] Among them, P n is the feeding port, n∈[1,N]; D n is the unloading port, n∈[1,N]; E(i,j) is the lane from vertex i to vertex j; T ij is the lane direction; H ij is the lane length; F bd From the feed port D b To the feeding port P d Number of transports; G db From the feeding port P d To the feeding port D b No-load times; L ij is the shortest path from vertex i to vertex j; Is lane E(i,j) located at the unloading port D?b To the feeding port P d on the transport route; Is lane E(i,j) located at loading port P? d To the feeding port D b on empty routes; The loading port of the nth unit is on lane E(i,j); The unloading port of the nth unit is on lane E(i,j); q bd is the load flow matrix between unit b and unit d; is the no-load flow matrix between unit b and unit d;

[0091] The mathematical model of this problem is as follows:

[0092]

[0093]

[0094] T ij +T ji =1# Formula (4.12

[0095] Σ i∈S T ij ≥1# Formula (4.13

[0096] Σ j∈S T ij ≥1# Formula (4.14

[0097]

[0098] Σ d∈N F bd =∑ d∈N G db # Formula (4.17

[0099]

[0100] Formula (4.1) is the main goal of this mathematical programming model. The optimization goal is to minimize the total distance of the AGV, which includes the load travel and the empty travel.

[0101] Constraints (4.2) and (4.3) ensure that after the route is selected, the length of the selected route is equal to the empty distance; in other words, whether it is from the loading port P b Departure, passing through the corresponding unloading port D d Then return to the feeding port P b Path, or from the feed port D d Departure, passing through the loading port P b Then return to the feeding port D dThe paths of are all equal to the length of the selected path;

[0102] Constraints (4.4) and (4.5) ensure that the feeding port P b The number of lanes is only one; similarly, constraints (4.6) and (4.7) are also to ensure that the inlet and outlet D d The number of lanes is only one;

[0103] The purpose of constraints (4.8) and (4.9) is to ensure that the flow of all vertices (except loading and unloading points) is balanced; that is, the number of times the AGV enters and leaves a non-loading and unloading point vertex must be equal;

[0104] Constraints (4.10) and (4.11) ensure that the lanes are not reused during either the transport phase or the empty return phase;

[0105] Constraint (4.12) is to ensure that each lane has only one direction;

[0106] Constraints (4.13) and (4.14) are to ensure that for all vertices in the vertex set, there is a path starting from and ending at the vertex, so as to ensure that the guided path network is strongly connected;

[0107] Constraints (4.15) and (4.16) ensure that the loading and unloading points are on the load route and the empty route;

[0108] Constraints (4.17) and (4.18) are to ensure the logistics balance of the AGV, that is, its number of transports and empty loads should be equal;

[0109] Constraints (4.19) and (4.20) ensure that each unit has only one corresponding loading port and unloading port, respectively;

[0110] Constraint (4.21) specifies the range of all variables, that is, all variables are 0-1 variables and can only take 0 or 1. ij is a known constant, representing the direction of each guided path in the path network.

[0111] Wherein, step S200 is specifically as follows:

[0112] S210, population initialization, using tuple coding with points instead of edges as the coding method, and using an undirected graph-based method to generate an initial solution;

[0113] S220, calculating the fitness of the initial solution, the function calculates the corresponding fitness of the objective function value of the individual through linear scaling transformation, after the linear transformation, the fitness of a few individuals with high original fitness is proportionally reduced, while the fitness of individuals with poor fitness is proportionally increased, which is conducive to the diversity of the population;

[0114] S230, selection operation, the genetic algorithm adopts the elite retention tournament selection, randomly selects from a certain number of individuals, and then selects the individual with the best fitness as the parent;

[0115] S240, crossover operation, such as Figure 5 As shown, in the crossover operation of the genetic algorithm, a partial matching crossover with modified deduplication is used to avoid duplication of genes in different units of the two chromosomes undergoing the crossover operation;

[0116] S250, mutation operation, such as Figure 6 As shown, a single gene mutation is used; first, each chromosome is traversed, and a random number is generated each time. If the random number is greater than the mutation probability, no mutation will be performed; otherwise, the next step is carried out: a gene on the individual is randomly selected for mutation, and a position that does not exist in the individual is selected from the corresponding unit optional positions as a replacement.

[0117] Based on the characteristics of the facility layout problem, taking into account the superior global search ability of the genetic algorithm and the particularity of the loading and unloading point problem, the genetic algorithm encoding and decoding, the generation of initial solutions, the crossover operation and the mutation operation are adjusted accordingly to adapt to the solution of the loading and unloading point problem, which can ensure the validity of the generated solutions and the uniqueness of the genes in each solution. By comparing the performance of different algorithms in solving examples of different scales, the characteristics of the algorithm for solving the loading and unloading point problem are obtained.

[0118] Among them, step S300 is specifically as follows: a simplified variable neighborhood search algorithm: for two strongly connected directed graphs D and D', if D and D' have the same base graph G, then by reversing the path and the circle, D can be converted into D'; the algorithm is used to obtain all loops of the directed graph D, generate a loop neighborhood, and search to determine the corresponding path orientation.

[0119] The RVNS algorithm process is as follows:

[0120] Input:(k max , feasible solution ε);

[0121] Output: local optimal solution β best ;

[0122] 1:ε best :=β,k:=1;

[0123] 2: While k≤k max do;

[0124] 3:if k=1then;

[0125] 4: Obtain all the circles of the directed graph D corresponding to β, forming the neighborhood N1(β best );

[0126] 5: Beta * :=random(N1(β best ))#Randomly select neighborhood structure N1(β best )

[0127] 6: If OFV(β * ) <ODV(β best ), then let β best :=β * , otherwise k:=k+1;

[0128] 7:ifk=2then;

[0129] 8: Use the network flow algorithm to find that β corresponds to the directed graph D with two arcs that do not intersect each other, forming the neighborhood N2(β best );

[0130] 9: Beta * :=random(N2(β best ))#Randomly select neighborhood structure N2(β best )

[0131] 10: If OFV(β * ) <ODV(β best ), then let β best :=β * , otherwise k:=k+1.

[0132] As a variant of VNS, the Reduced Variable Neighborhood Search (RVNS) algorithm removes the perturbation operation in the traditional neighborhood search, generates solutions based only on random neighborhood operations, and compares them with the currently known best solution to update the local optimal solution, showing high search efficiency.

[0133] like Figure 3-4 As shown, in step S210, the workshop layout is converted into an undirected graph, and the edges of the undirected graph are used to indicate whether the units have common boundaries, and these boundaries can be used as loading and unloading point locations;

[0134] Since the vertices are located at corners or intersections, in order to avoid congestion of AGVs, the vertices are not considered as loading and unloading points. Instead, it is stipulated that the loading and unloading points can only be located at the center of these boundaries. The units are represented by vertices, and 4 auxiliary vertices (N, S, W, E) are added to represent the boundaries of the workshop that are not adjacent to other workshops.

[0135] By randomly generating a row of codes containing unit numbers and directional letters, a feasible initial solution without duplication can be obtained by decoding;

[0136] After the code is generated according to the above method, it is necessary to decode the code from left to right to get the initial solution. The decoding rules are as follows:

[0137] If the current vertex is an auxiliary vertex and the next vertex is also an auxiliary vertex, the position of the unit's loading and unloading point is uncertain;

[0138] If the current vertex is an auxiliary vertex and the next vertex is a unit, it means that the loading and unloading point of the next vertex unit is located on the common boundary of the two vertex units;

[0139] If the current vertex is a unit, and the unit has no assigned loading and unloading point; if the next vertex is an auxiliary vertex, the unit loading and unloading point is located on the boundary corresponding to the auxiliary vertex; if the next vertex is a unit, the current vertex unit loading and unloading point is located on the common boundary of the two units;

[0140] If the current vertex is a unit, and the unit is assigned a loading and unloading point; if the next vertex is an auxiliary vertex, if the next vertex is a unit, then the loading and unloading point of the next vertex unit is located on the common boundary of the two units;

[0141] If the current vertex is a unit and the unit has no assigned loading and unloading point location; and it is also the last vertex of the route, then the first vertex of the segment is taken, and the loading and unloading point is located on the common boundary.

[0142] Among them, in step S220, the calculation formula is:

[0143] F'=aF+b# Formula (3.19)

[0144] Where F is a function that directly uses the objective function value as the fitness value, F' is the fitness value function after linear scaling of F, and F' needs to meet the following two conditions:

[0145] (1) To ensure that the expected number of copies of individuals with average fitness in the generated offspring is 1, the average value of the original fitness F avg = the average value of the new fitness F' avg ;

[0146] (2) In order to limit the number of copies of the individual with the highest fitness in the next generation, we set the average value of the original fitness times the specified value as the maximum fitness after transformation, that is, F' max =Smul*F avg , Smul is a constant that determines the upper bound of the linear transformation, and its default value is generally 2.

[0147] Among them, the elite retention tournament selection combines the elite retention strategy and the tournament selection strategy. The specific operations are as follows:

[0148] 1. Select the best individuals from the population as elite individuals and directly copy them to the next generation of population to ensure that the best individuals are retained;

[0149] 2. For the remaining selections, the tournament selection method is adopted; among a certain number of randomly selected individuals, the individual with the best fitness is selected as the parent.

[0150] The technical principle of the present invention is described above in conjunction with specific embodiments. These descriptions are only for explaining the principle of the present invention and cannot be interpreted as limiting the scope of protection of the present invention in any way. Based on the explanations herein, those skilled in the art can associate other specific implementations of the present invention without paying creative labor, and these methods will fall within the scope of protection of the present invention.

Claims

1. A joint optimization method for AGV guidance path and loading and unloading points, characterized in that: The steps include: S100. Under the premise of a known system layout of facilities and the amount of logistics between units, there is an undirected path network, and each work unit has corresponding multiple optional points that can be used as loading and unloading points; it is necessary to determine the position of each unit loading and unloading point and the direction of each guide path, and to minimize the total transportation distance formed by the AGV loaded distance and empty distance, and to construct a mathematical model for optimizing the guide path and the loading and unloading point position; S200, solving the mathematical model by improving the genetic algorithm, so as to obtain the position data of the AGV loading and unloading point for the purpose of minimizing the total transportation distance of the AGV including the AGV loaded distance and the unloaded distance; S300, bringing the location data of the AGV loading and unloading points obtained in step S200 into a simplified variable neighborhood search algorithm, so as to obtain the AGV guidance path with the purpose of minimizing the total transportation distance including both the AGV loaded distance and the empty distance.

2. The combined optimization method for AGV guidance path and loading and unloading points according to claim 1, characterized in that: The mathematical model in step S100 is as follows: There is no upper limit to AGV transport capacity; Each unit has a unique loading and unloading port, and the position is known; The transport paths of AGVs are all the shortest paths; Regardless of whether the AGV will be blocked; Among them, P n is the feeding port, n∈[1,N]; D n is the unloading port, n∈[1,N]; E(i,j) is the lane from vertex i to vertex j; T ij is the lane direction; H ij is the lane length; F bd From the feed port D b To the feeding port P d Number of handling times; G db From the feeding port P d To the feeding port D b No-load times; L ij is the shortest path from vertex i to vertex j; Is lane E(i,j) located at the unloading port D? b To the feeding port P d on the transport route; Is lane E(i,j) located at loading port P? d To the feeding port D b on empty routes; The loading port of the nth unit is on lane E(i,j); The unloading port of the nth unit is on lane E(i,j); q bd is the load flow matrix between unit b and unit d; is the no-load flow matrix between unit b and unit d; The mathematical model of this problem is as follows: T ij +T ji =1# Formula (4.12) ∑ i∈S T ij ≥1# Formula (4.13) ∑ j∈S T ij ≥1# Formula (4.14) ∑ d∈N F bd =∑ d∈N G db # Formula (4.17) Formula (4.1) is the main goal of this mathematical programming model. The optimization goal is to minimize the total distance of the AGV, which includes the load travel and the empty travel. Constraints (4.2) and (4.3) ensure that after the route is selected, the length of the selected route is equal to the empty distance; in other words, whether it is from the loading port P b Departure, passing the corresponding unloading port D d Then return to the feeding port P b Path, or from the feed port D d Departure, passing through the loading port P b Then return to the feeding port D d The paths of are all equal to the length of the selected path; Constraints (4.4) and (4.5) ensure that the feeding port P d The number of lanes is only one; similarly, constraints (4.6) and (4.7) are also to ensure that the inlet and outlet D d The number of lanes is only one; The purpose of constraints (4.8) and (4.9) is to ensure that the flow of all vertices (except loading and unloading points) is balanced; that is, the number of times the AGV enters and leaves a non-loading and unloading point vertex must be equal; Constraints (4.10) and (4.11) ensure that the lanes are not reused during either the transport phase or the empty return phase; Constraint (4.12) is to ensure that each lane has only one direction; Constraints (4.13) and (4.14) are to ensure that for all vertices in the vertex set, there is a path starting from and ending at the vertex, so as to ensure that the guided path network is strongly connected; Constraints (4.15) and (4.16) ensure that the loading and unloading points are on the load route and the empty route; Constraints (4.17) and (4.18) are to ensure the logistics balance of the AGV, that is, its number of transports and empty loads should be equal; Constraints (4.19) and (4.20) ensure that each unit has only one corresponding loading port and unloading port, respectively; Constraint (4.21) specifies the range of all variables, that is, all variables are 0-1 variables and can only take 0 or 1. ij is a known constant, representing the direction of each guided path in the path network.

3. The joint optimization method for AGV guidance path and loading and unloading points according to claim 1 is characterized in that: Step S200 is specifically as follows: S210, population initialization, using tuple coding with points instead of edges as the coding method, and using an undirected graph-based method to generate an initial solution; S220, calculating the fitness of the initial solution, the function calculates the corresponding fitness of the objective function value of the individual through linear scaling transformation, after the linear transformation, the fitness of a few individuals with high original fitness is proportionally reduced, while the fitness of individuals with poor fitness is proportionally increased, which is conducive to the diversity of the population; S230, selection operation, the genetic algorithm adopts the elite retention tournament selection, randomly selects from a certain number of individuals, and then selects the individual with the best fitness as the parent; S240, crossover operation. In the crossover operation link of the genetic algorithm, a partial matching crossover with correction and deduplication is adopted to avoid duplication of genes in different units of the two chromosomes undergoing the crossover operation; S250, mutation operation, uses single gene mutation; first, it traverses each chromosome and generates a random number each time. If the random number is greater than the mutation probability, no mutation will be performed; Otherwise, proceed to the next step: randomly select a gene on the individual for mutation, and select a position that does not exist in the individual from the corresponding unit optional positions as a replacement.

4. The combined optimization method for AGV guide path and loading and unloading points according to claim 1, characterized in that: Step S300 is specifically as follows: a simplified variable neighborhood search algorithm: for two strongly connected directed graphs D and D', if D and D' have the same base graph G, then D can be converted into D' by reversing the path and the circle; the algorithm is used to obtain all loops of the directed graph D, generate a loop neighborhood, and search to determine the corresponding path orientation.

5. The combined optimization method for AGV guidance path and loading and unloading points according to claim 4 is characterized in that: The RVNS algorithm process is as follows: Input:(k max , feasible solution ε); Output: local optimal solution β best ; 1:e best :=β,k:=1; 2:While k≤k max do; 3:if k=1 then; 4: Obtain all the circles of the directed graph D corresponding to β, forming the neighborhood N1(β best ); 5: Beta * :=random(N1(β best ))#Randomly select neighborhood structure N1(β best ) 6: If OFV(β * ) <ODV(β best ), then let β best :=β * , otherwise k:=k+1; 7:ifk=2then; 8: Use the network flow algorithm to find that β corresponds to the directed graph D with two arcs that do not intersect each other, forming the neighborhood N2(β best ); 9: Beta * :=random(N2(β best ))#Randomly select neighborhood structure N2(β best ) 10: If OFV(β * ) <ODV(β best ), then let β best :=β * , otherwise k:=k+1.

6. The combined optimization method for AGV guide path and loading and unloading points according to claim 3 is characterized in that: In step S210, the workshop layout is converted into an undirected graph, and the edges of the undirected graph are used to indicate whether the units have common boundaries, and these boundaries can be used as loading and unloading point locations; Since the vertices are located at corners or intersections, in order to avoid congestion of AGVs, the vertices are not considered as loading and unloading points. Instead, it is stipulated that the loading and unloading points can only be located at the center of these boundaries. The units are represented by vertices, and 4 auxiliary vertices (N, S, W, E) are added to represent the boundaries of the workshop that are not adjacent to other workshops. By randomly generating a row of codes containing unit numbers and directional letters, a feasible initial solution without duplication can be obtained by decoding; After the code is generated according to the above method, it is necessary to decode the code from left to right to get the initial solution. The decoding rules are as follows: If the current vertex is an auxiliary vertex and the next vertex is also an auxiliary vertex, the position of the unit's loading and unloading point is uncertain; If the current vertex is an auxiliary vertex and the next vertex is a unit, it means that the loading and unloading point of the next vertex unit is located on the common boundary of the two vertex units; If the current vertex is a unit, and the unit has no assigned loading and unloading point; if the next vertex is an auxiliary vertex, the unit loading and unloading point is located on the boundary corresponding to the auxiliary vertex; if the next vertex is a unit, the current vertex unit loading and unloading point is located on the common boundary of the two units; If the current vertex is a unit, and the unit is assigned a loading and unloading point position; If the next vertex is an auxiliary vertex, and if the next vertex is a unit, the next vertex unit loading and unloading point is located on the common boundary of the two units; If the current vertex is a unit and the unit has no assigned loading and unloading point location; and it is also the last vertex of the route, then the first vertex of the segment is taken, and the loading and unloading point is located on the common boundary.

7. The combined optimization method for AGV guide path and loading and unloading points according to claim 3 is characterized in that: In step S220, the calculation formula is: F'=aF+b# Formula (3.19) Where F is a function that directly uses the objective function value as the fitness value, F' is the fitness value function after linear scaling of F, and F' needs to meet the following two conditions: (1) To ensure that the expected number of copies of individuals with average fitness in the generated offspring is 1, the average value of the original fitness F avg = the average value of the new fitness F' avg ; (2) In order to limit the number of copies of the individual with the highest fitness in the next generation, we set the average value of the original fitness times the specified value as the maximum fitness after transformation, that is, F' max =Smul*F avg , Smul is a constant that determines the upper bound of the linear transformation, and its default value is generally 2.

8. The combined optimization method for AGV guide path and loading and unloading points according to claim 3 is characterized in that: Elite retention tournament selection combines the elite retention strategy with the tournament selection strategy. The specific operation is as follows:

1. Select the best individuals from the population as elite individuals and directly copy them to the next generation of population to ensure that the best individuals are retained; 2. For the remaining selections, the tournament selection method is adopted; among a certain number of randomly selected individuals, the individual with the best fitness is selected as the parent.

Citation Information

Patent Citations

  • Job shop motion orbit path guiding optimization method based on hybrid genetic algorithm

    CN114444239A