A multi-load AGV path planning method based on hybrid breeding strategy

Abstract

The application belongs to the technical field of AGV path planning in smart factory, and specifically provides a multi-load AGV path planning method based on hybrid breeding strategy, improves the basic ant colony optimization algorithm, designs a collaborative evolution heterogeneous ant colony optimization algorithm to make up for the deficiency of single colony in the diversity of solution, and proposes a strategy based on population similarity to determine whether communication is needed after a fixed number of iterations, maintains a dynamic balance between population diversity and convergence speed, makes the algorithm have a high global convergence speed while obtaining high-quality solutions, and can be used in the technical field of path planning. The application can quickly obtain the optimal path selection of multi-load AGV when facing multiple tasks, improve the path planning speed under the premise of obtaining the optimal path, can meet the demand of multi-load AGV path planning in the field of smart factory, and provides a more effective technical method for improving the efficiency of AGV path planning.

Description

Technical Field

[0001] This invention relates to the field of AGV path planning technology in smart factories, and more specifically, to a multi-load AGV path planning method based on a hybridization breeding strategy. Background Technology

[0002] With the rapid development of advanced technologies such as digital factories, the Industrial Internet of Things (IIoT), and artificial intelligence, higher demands are being placed on production and logistics methods. Automated Guided Vehicles (AGVs) are an important component of automated logistics equipment, offering advantages such as low transportation costs, high transportation efficiency, high transportation accuracy, and ease of standardized management. They play an irreplaceable role in reducing enterprise costs and improving production and transportation efficiency.

[0003] The development of AGVs involves multiple fields, such as computers, automatic control, electronics, information communication, and mechanical design. It represents a high-tech field integrating optics, electromechanical systems, and computer technology. To ensure the safe and reliable operation of AGVs, the AGV path planning problem must first be solved. This problem involves generating a safe movement path using known environmental information and data acquired from real-time sensors. Ant Colony Optimization (ACO) is a swarm intelligence optimization algorithm proposed by simulating the foraging behavior of ants. It has been widely used in discrete combinatorial optimization problems, primarily transforming the combinatorial optimization problem into an optimization problem of ants searching for a path. Multi-load AGV path planning is a typical combinatorial optimization problem, and ACO is highly applicable to handling such problems.

[0004] Compared to traditional optimization algorithms, ant colony optimization (ACO) offers advantages such as robustness, high search efficiency, and a positive feedback mechanism, leading to successful applications in fields like the Traveling Salesman Problem, intelligent transportation, and network optimization. However, ACO also has some drawbacks, such as high time complexity and a tendency to get trapped in local optima. Furthermore, as the number of iterations increases, ACO struggles to maintain a balance between population diversity and local search capability.

[0005] To improve the optimization performance of the basic ACO, inspired by the hybridization breeding mechanism, this invention designs a multi-load AGV path planning method based on the hybridization breeding co-evolutionary ant colony optimization algorithm. This method fully leverages the advantages of different populations, effectively expands the search range, and thus effectively improves the quality of the solution, enabling the algorithm to steadily move towards the optimal solution. Summary of the Invention

[0006] This invention addresses the technical problem in existing technologies where traditional AGV path planning based on ant colony optimization algorithms cannot maintain a balance between population diversity and local search capability.

[0007] This invention provides a multi-load AGV path planning method based on a hybridization breeding strategy, comprising the following steps:

[0008] Step 1: Initialize the AGV's environment map and obtain the destination address information as the raw input set;

[0009] Step 2: Initialize the population and the initialization parameters required by the algorithm;

[0010] Step 3: Divide the population into three seed populations: Elitist Ant System (EAS), Ant Colony System (ACS), and Max-Min Ant System (MMAS).

[0011] Step 4: Different seed groups evolve independently, completing path visits;

[0012] Step 5: Calculate the population similarity of each subgroup and determine whether inter-population communication is needed. If inter-population communication is needed, then perform inter-population communication.

[0013] Step 6: Complete pheromone update;

[0014] Step 7: Determine if the termination condition is met. The termination condition is set to reach the maximum number of iterations. If not, return to step 4; if yes, output the globally optimal path.

[0015] Preferably, the initialization parameters include pheromone factor α, heuristic factor β, pheromone decay coefficient ρ, pheromone quality coefficient Q, and maximum number of iterations Nc. max Population size m.

[0016] Preferably, step 4 specifically includes:

[0017] When the subpopulation is EAS, the optimal path when the population completes path visits is represented as T. bs Its length is L bs The pheromone concentration remaining on path (i,j) is τ. ij The pheromone levels are calculated as shown in formulas (1) and (2):

[0018]

[0019]

[0020]

[0021]

[0022] in, This represents the increase in pheromone along path (i,j) caused by elite ants after T time steps of path visits at time t. σ represents the number of elite ants, Q represents the total amount of pheromone released by an ant completing one cycle, and L represents the total amount of pheromone released by an ant completing one cycle. k This represents the length of the path ant k traversed in this iteration. This represents the increase in pheromone level of elite ant k on path (i,j) after T time steps of path visits at time t. k This indicates the path length of elite ant k in this path visit.

[0023] Preferably, step 4 specifically includes:

[0024] When the subpopulation is ACS, the pseudo-random proportional rule followed by individual ants when choosing the next task point after the population completes path visits is as shown in formula (5):

[0025]

[0026] Among them, s k Let q be the next city chosen for ant k to move to, q be a random number in the range [0,1], q0 be the threshold set initially, and η be the city to move to. ij Let τ be the visibility of path (i,j), i.e., the heuristic function, reflecting the degree of heuristic effect of moving from city i to city j. ij Let pheromone be the amount of ant k left on path (i,j) in this iteration.

[0027] Preferably, step 4 specifically includes: before selecting the next city, the ant first conducts a random experiment to obtain a random number q. If q ≤ q0, then it selects a city from all feasible cities. The largest city is the next city to be selected; otherwise, the next city is selected according to formula (6). Let city j be the probability that ant k chooses city j as its next city to move to in city i.

[0028]

[0029] Preferably, step 4 specifically includes: after one loop is completed, only the ants that have found the optimal path are updated globally for pheromones, and the pheromone update rules are as shown in formulas (7) and (8):

[0030] τ ij (t+T)=(1-ρ)τ ij (t)+ρΔτ ij (t,t+T) (7)

[0031]

[0032] Where ρ is a constant in the interval [0,1]; L gb This represents the optimal path length obtained by the ant colony in this iteration.

[0033] Preferably, step 4 specifically includes: when the subpopulation is MMAS, the concentration range of pheromones along the path is limited to the interval [τ]. min ,τ max Within this range, when the pheromone concentration on the path exceeds this limit, it will be forcibly set to τ. min or τ max Meanwhile, the initial value of the pheromone is set to the maximum value τ. max .

[0034] Preferably, step 5 specifically includes:

[0035] Calculate the similarity coefficient;

[0036] Calculate the similarity within the subpopulation. The similarity within the ant colony represents the degree of overlap between the paths traversed by all ants in the subpopulation after several iterations and the current best traversal path.

[0037] Calculate the similarity between subpopulations. The similarity between ant colonies represents the similarity between subpopulations after several iterations, indicating the difference in the evolutionary direction of the two subpopulations during the evolutionary process.

[0038] Calculate the frequency of inter-population communication;

[0039] When selecting a communication partner between populations, assuming that subpopulation colony(i) chooses subpopulation colon(y)j as its communication partner, the selection of the communication partner is determined by the following formula:

[0040]

[0041] Where r represents the number of subpopulations.

[0042] Preferably, step 6 specifically includes:

[0043] When the subpopulation colony(i) that needs to communicate is ACS, and the population that is the communication partner is colony(j), then the ACS pheromone update satisfies the following formula:

[0044] τ ij (t+T)=(1-ρ)τ ij (t)+ρΔτ ij (t, t+T)+ET

[0045] Where Δτ ij (t, t+T) is updated as follows:

[0046]

[0047] Where ET is the pheromone contribution of subpopulation colour(j) to subpopulation colour(i), i.e., reinforcement information, L pg N represents the optimal path for the subpopulation colour(j) in this iteration. c This represents the current iteration number.

[0048] When the subpopulation colony(i) that needs to communicate is EAS, and the population that is the communication partner is colony(j), then the pheromone update of colony(i) satisfies the following formula:

[0049]

[0050]

[0051]

[0052] When the subpopulation colony(i) that needs to communicate is MMAS, and the population that is the communication partner is colony(j), then the MMAS pheromone update method is the same as ACS.

[0053] Beneficial Effects: This invention provides a multi-load AGV path planning method based on a hybridization breeding strategy. By improving the basic ant colony optimization algorithm, a co-evolutionary heterogeneous ant colony optimization algorithm is designed to compensate for the lack of solution diversity in a single population. Furthermore, a strategy based on population similarity is proposed to determine whether communication is needed after a fixed number of iterations, maintaining a dynamic balance between population diversity and convergence speed. This allows the algorithm to achieve high global convergence speed while yielding high-quality solutions, making it applicable to path planning related technical fields. This invention can quickly obtain the optimal path selection for multi-load AGVs facing multiple tasks. While ensuring the acquisition of the optimal path, it improves the path planning speed, meeting the needs of multi-load AGV path planning in fields such as smart factories, and providing a more effective technical method for improving AGV path planning efficiency. Attached Figure Description

[0054] Figure 1 The present invention provides a flowchart of a multi-load AGV path planning method based on a hybridization breeding strategy. Detailed Implementation

[0055] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0056] Figure 1The present invention provides a multi-load AGV path planning method based on a hybridization breeding strategy, comprising the following steps:

[0057] Step 1: Initialize the AGV's environment map and obtain the destination address information as the raw input set;

[0058] Step 2: Initialize the population and algorithm parameters, including pheromone factor α, heuristic factor β, pheromone decay coefficient ρ, pheromone quality coefficient Q, and maximum number of iterations Nc. max Population size mA ;

[0059] Step 3: Divide the population. Inspired by the hybridization breeding strategy, the population is divided into three seed groups: Elastic Ant System (EAS), Ant Colony System (ACS), and Max-Min Ant System (MMAS).

[0060] Step 4: Different subpopulations evolve independently, completing path visits;

[0061] Step 5: Calculate the population similarity of each subpopulation and determine whether inter-population communication is needed. If inter-population communication is needed, then perform inter-population communication.

[0062] Step 6: Complete pheromone update;

[0063] Step 7: Determine if the termination condition is met. The termination condition is set to reach the maximum number of iterations. If not, return to step 4; if yes, output the globally optimal path.

[0064] In the multi-load AGV path planning method based on the co-evolutionary ant colony optimization algorithm of the hybridization breeding strategy described above, step 4 specifically includes the following sub-steps:

[0065] Step 4.1: When the subpopulation is EAS, the optimal path is represented as T when the population completes path visits. bs Its length is L bs The pheromone concentration remaining on path (i,j) is τ. ij The pheromone is calculated as shown in formulas (1) and (2):

[0066]

[0067]

[0068] in,

[0069]

[0070]

[0071] in, This represents the increase in pheromone along path (i,j) caused by elite ants after T time steps of path visits at time t. σ represents the number of elite ants, Q represents the total amount of pheromone released by an ant completing one cycle, and L represents the total amount of pheromone released by an ant completing one cycle. k This represents the length of the path ant k traversed in this iteration. This represents the increase in pheromone level of elite ant k on path (i,j) after T time steps of path visits at time t. k This indicates the path length of elite ant k in this path visit.

[0072] Step 4.2: When the subpopulation is ACS, the pseudo-random proportional rule followed by individual ants when selecting the next task point after the population completes path visits is as shown in formula (5):

[0073]

[0074] Among them, s k Let q be the next city chosen for ant k to move to, q be a random number in the range [0,1], q0 be the threshold set initially, and η be the city to move to. ij Let τ be the visibility of path (i,j), i.e., the heuristic function, reflecting the degree of heuristic effect of moving from city i to city j. ij Let q be the pheromone level of ant k on path (i,j) in this iteration. Before choosing the next city, the ant first conducts a random experiment to obtain a random number q. If q ≤ q0, then it chooses a city from all feasible cities. The largest city is the next city to be selected; otherwise, the next city is selected according to formula (6). Let ant k be the probability of choosing city j as the next city to move to in city i, and let allow. k ={1,2,...,n} represents the cities that ant k can choose to move next.

[0075]

[0076] After one loop, only the ants that have found the optimal path are updated globally for pheromones. The pheromone update rules are as shown in formulas (7) and (8):

[0077] τ ij (t+T)=(1-ρ)τ ij (t)+ρΔτ ij (t,t+T) (7)

[0078]

[0079] Where ρ is a constant in the interval [0,1]; L gb This represents the optimal path length obtained by the ant colony in this iteration.

[0080] In ACS, ants perform local updates while constructing their paths. The pheromone local update rule is shown in formula (9):

[0081] τ ij (t+T)=(1-ρ)τ ij (t)+ρτ ij (t,t+T) (9)

[0082] Where ρ is a constant in the interval [0,1], and τ ij (t, t+T) is usually taken as the initial value of the pheromone on the path.

[0083] Step 4.3: When the subpopulation is MMAS, the concentration range of pheromones along the path is limited to the interval [τ]. min ,τ max Within this range, when the pheromone concentration on the path exceeds this limit, it will be forcibly set to τ. min or τ max Meanwhile, the initial value of the pheromone is set to the maximum value τ. max .

[0084] In the multi-load AGV path planning method based on the co-evolutionary ant colony optimization algorithm of the hybridization breeding strategy mentioned above, step 5 specifically includes the following sub-steps:

[0085] Step 5.1: Calculate the similarity coefficient. The similarity coefficient is used to describe the degree of overlap between the paths obtained by the ants during traversal. The calculation method can be expressed by formulas (10) and (11):

[0086]

[0087]

[0088] Among them, S ij For ant a i With ant a j The similarity of the paths obtained through traversal Ant a i The k-th node in the traversed path is related to ant a in the same iteration. j Check if the traversal paths overlap; if they do, the value is 1, otherwise the value is 0.

[0089] Step 5.2: Calculate the similarity within the subpopulation. The similarity within the ant colony represents the degree of overlap between the paths traversed by all ants in the subpopulation after several iterations and the current best traversal path. The calculation is shown in formulas (12) and (13):

[0090]

[0091]

[0092] Among them, S colony(i) This indicates the similarity between the path traversed by each ant in the subpopulation colony(i) and the current best traversal path. This indicates that in the subpopulation, each ant a i Does the k-th node of the optimal traversal path obtained in this iteration coincide with ant a in the subpopulation? best If the optimal traversal path overlaps with another path, its value is 1; otherwise, its value is 0.

[0093] Step 5.3: Calculate the similarity between subpopulations. The similarity between ant colonies represents the similarity between subpopulations after several iterations, indicating the difference in evolutionary direction between the two subpopulations during the evolutionary process. The calculation is as shown in formula (14):

[0094]

[0095] Among them, S colony(i,j) This indicates the degree of similarity between the optimal paths obtained by traversing subpopulation colour(i) and subpopulation colour(j). This indicates whether the k-th node of the best path obtained by the subpopulation colony(i) in the current iteration overlaps with the best path obtained by the subpopulation colony(j) in the current iteration. If they overlap, the value is 1, otherwise it is 0.

[0096] Step 5.4: Calculate the frequency of interpopulation communication. Suppose that the subpopulation colony(i) has undergone [a certain number of steps] after [a certain number of steps]. After the iteration, its exchange index is P. interval(i) This value determines whether or not the species will interact with other subpopulations; the exchange probability P. interval(i) The calculation is as shown in formula (15):

[0097] P interval(i) =1-S colony(i) (15)

[0098] When the similarity S corresponding to the subpopulation colony(i) colony(i) When P is large, interval(i)A small similarity value, less than 0.5, indicates that the solutions of the subpopulation colony(i) are relatively concentrated, reducing population diversity and hindering further exploration. Therefore, it is necessary to increase communication between the subpopulation colony(i) and other external subpopulations to improve the evolutionary environment of the subpopulation. Conversely, a large similarity value indicates a high degree of similarity. colony(i) When P is small interval(i) When the value is large, when it is greater than 0.5, the solutions of the subpopulation colony(i) are relatively dispersed and have good diversity. Therefore, there is no need for the subpopulation colony(i) to communicate with other external subpopulations to prevent disruption of the exploration within the population. Therefore, the communication frequency of the population is calculated as shown in formula (16):

[0099]

[0100] Step 5.5: Select the inter-population information exchange target. Assuming that subpopulation colony(i) selects subpopulation colony(j) as the information exchange target, the selection of the inter-population information exchange target is determined by formula (17):

[0101]

[0102] Where r represents the number of subpopulations.

[0103] The multi-load AGV path planning method based on the co-evolutionary ant colony optimization algorithm of the above-mentioned hybridization breeding strategy includes the following sub-steps in step 6:

[0104] Step 6.1: Pheromones Update. During pheromone updates, in order to preserve the contribution of the subpopulation that serves as the communication partner to the current population, the optimal solution of the other group and the similarity between the two groups are also considered as reinforcement information and are included as part of the pheromone update for this population. When the subpopulation colony(i) that needs to communicate is ACS, and the population that serves as the communication partner is colony(j), then the ACS pheromone update satisfies formula (18):

[0105] τ ij (t+T)=(1-ρ)τ ij (t)+ρΔτ ij (t,t+T)+ET (18)

[0106] Where Δτ ij (t, t+T) is updated according to formula (19):

[0107]

[0108] Where ET is the pheromone contribution of subpopulation colour(j) to subpopulation colour(i), i.e., reinforcement information, L pg N represents the optimal path for the subpopulation colour(j) in this iteration. c This represents the current iteration number.

[0109] When the subpopulation colony(i) that needs to communicate is EAS, and the population that is the communication partner is colony(j), then the pheromone update of colony(i) satisfies formula (20):

[0110]

[0111] Where τ ij (t, t+T) is updated according to formula (21). Update according to formula (22), and ET is updated according to formula (19).

[0112]

[0113]

[0114] When the subpopulation colony(i) that needs to communicate is MMAS, and the population that is the communication partner is colony(j), then the MMAS pheromone update method is the same as ACS.

[0115] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0116] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A multi-load AGV path planning method based on a hybridization breeding strategy, characterized in that, Includes the following steps: Step 1: Initialize the AGV's environment map and obtain the destination address information as the raw input set; Step 2: Initialize the population and the initialization parameters required by the algorithm; Step 3: Divide the population into three seed populations: Elitist Ant System (EAS), Ant Colony System (ACS), and Max-Min Ant System (MMAS). Step 4: Different seed groups evolve independently, completing path visits; Step 5: Calculate the population similarity of each subpopulation and determine whether inter-population communication is needed. If inter-population communication is needed, then perform inter-population communication; specifically including: Calculate the similarity coefficient; Calculate the similarity within the subpopulation. The similarity within the ant colony represents the degree of overlap between the paths traversed by all ants in the subpopulation after several iterations and the current best traversal path. Calculate the similarity between subpopulations. The similarity between ant colonies represents the similarity between subpopulations after several iterations, indicating the difference in the evolutionary direction of the two subpopulations during the evolutionary process. Calculate the frequency of inter-population communication; Step 6: Complete pheromone update; specifically includes: When subgroups need to communicate When it is ACS, the population that acts as the communication partner is Then the ACS pheromone update satisfies the following equation: ； ； in, This represents the optimal path length obtained by the ant colony in this iteration. The pheromone attenuation coefficient; in Update as follows: ； in, For subpopulation Pair population The contribution of pheromones, i.e., reinforcing information. Subpopulation The optimal path in this iteration This represents the current iteration number. For subpopulation With subpopulation The degree of path similarity between the optimal paths obtained through traversal; When subgroups need to communicate When it is EAS, the population that acts as the communication partner is ,So Pheromones are updated according to the following formula: ； ； ； When subgroups need to communicate When it is MMAS, the population that acts as the communication partner is Therefore, the MMAS pheromone update method is the same as that of ACS; Step 7: Determine if the termination condition is met. The termination condition is set to reach the maximum number of iterations. If not, return to step 4; if yes, output the globally optimal path.

2. The multi-load AGV path planning method based on hybridization breeding strategy according to claim 1, characterized in that, The initialization parameters include pheromone factors. Heuristic factors pheromone attenuation coefficient pheromone quality coefficient Maximum number of iterations Population size .

3. The multi-load AGV path planning method based on hybridization breeding strategy according to claim 1, characterized in that, Step 4 specifically includes: When the subpopulation is EAS, the optimal path is represented as follows when the population completes path visits: Its length is ,path The concentration of pheromones remaining on the surface is The pheromone levels are calculated as shown in formulas (1) and (2): ； ； ； ； in, Indicates at time go through After the path visit at a certain moment, the path caused by the elite ant The increase of pheromones on the body This indicates the number of elite ants. This represents the total amount of pheromones released by an ant during one cycle. Ants The length of the path traversed in this loop. Indicates at time go through After the path visit at a certain moment, the elite ant In the path The increase of pheromones on the body Elite ants The path length in this path access.

4. The multi-load AGV path planning method based on hybridization breeding strategy according to claim 1, characterized in that, Step 4 specifically includes: When the subpopulation is ACS, the pseudo-random proportional rule followed by individual ants when choosing the next task point after the population completes path visits is as shown in formula (5): ； in, For ants The next city to move to It is a random number in the range [0,1]. The threshold set initially. For path Visibility, or heuristic function, reflects the visibility of the city. Move to the city The degree of inspiration, For ants Stay on the path in this iteration Pheromones.

5. The multi-load AGV path planning method based on hybridization breeding strategy according to claim 4, characterized in that, Step 4 specifically includes: before choosing the next city, the ant first conducts a random experiment to obtain a random number. ,like Then find among all feasible cities The largest city is the next city to be selected; otherwise, the next city is selected according to formula (6). For ants In the city Select City As a measure of the probability of relocation to the next city; ； in, For ants In the city Select City As a measure of the probability of moving to the next city. Ants The next city to choose from.

6. The multi-load AGV path planning method based on hybridization breeding strategy according to claim 5, characterized in that, Step 4 specifically includes: after one loop is completed, only the ants that have found the optimal path are updated globally for pheromones. The pheromone update rules are as shown in formulas (7) and (8): ； ； in, It is a range The constants between; This represents the optimal path length obtained by the ant colony in this iteration.

7. The multi-load AGV path planning method based on hybridization breeding strategy according to claim 1, characterized in that, Step 4 specifically includes: when the subpopulation is MMAS, the concentration range of pheromones along the path is limited to the interval [range missing]. Within this range, when the pheromone concentration on the path exceeds this limit, it will be forcibly set to [a certain value]. or Meanwhile, the initial value of pheromones is set to the maximum value. .

Images