Unmanned ship cluster task allocation method based on improved discrete grey wolf algorithm

Abstract

An unmanned ship cluster task allocation method based on an improved discrete grey wolf algorithm comprises the following steps that 1, information such as map information and unmanned ship cluster information is input in advance, modeling is conducted on all kinds of information, and a cost function for measuring the advantages and disadvantages of tasks is designed; 2, calling an improved discrete grey wolf algorithm to carry out population initialization coding on all task points; step 3, performing updating iteration on the initialized population, and firstly performing population aggregation on scattered task points by adopting an individual and population interaction strategy; step 4, carrying out first wolf guidance strategy updating on the clustered task points; and step 5, performing individual wolf wandering strategy updating on the population. Step 6, if the current iteration number t is greater than the maximum iteration number T, outputting a first wolf population sequence, otherwise, returning to the step 3; and step 7, decoding the first wolf population sequence. The algorithm has higher convergence speed and higher convergence quality.

Description

Technical Field

[0001] The present invention relates to unmanned boat task allocation technology, in particular to an unmanned boat cluster task allocation method based on an improved discrete grey wolf algorithm. Background Art

[0002] With the continuous development of the intelligent field, marine operations are becoming larger and more complex. Single unmanned surface vessels are largely limited by their payload capacity and short endurance. Therefore, swarms of multiple unmanned surface vessels are becoming an effective means of carrying out large-scale, complex marine operations. Because marine operations vary in location and workload, the coordination of these vessels is crucial when collaboratively completing large-scale, complex marine missions. A rational task allocation system is crucial for multi-tasking surface unmanned surface vessel swarms to successfully complete their missions.

[0003] Collaborative task allocation for a swarm of unmanned aerial vehicles requires assigning each unmanned aerial vehicle a corresponding task point and determining the execution order of each task, while satisfying the unmanned aerial vehicle's own performance constraints and task point condition constraints. This ensures the unmanned aerial vehicles achieve maximum efficiency in executing their tasks. Currently, the algorithms used to solve task allocation problems on the market can be roughly divided into two categories: centralized algorithms, typified by dynamic programming and exhaustive methods, and distributed algorithms, typified by swarm intelligence optimization algorithms. While centralized algorithms can produce relatively accurate solutions, they can be very complex when faced with large-scale problems, resulting in extremely slow solutions.

[0004] The Grey Wolf Algorithm (GWO) is a swarm intelligence optimization algorithm that has been very popular in recent years. It has been applied by many scholars to many engineering problems. It shows strong adaptability when facing complex and large-scale tasks, and has the advantages of few parameters, simple settings, and strong scalability. However, it also has problems such as insufficient population diversity and slow convergence speed. Therefore, when solving the unmanned boat task allocation problem, it may lead to insufficient accuracy of the problem solution or no solution to the problem. Summary of the Invention

[0005] In response to the above-mentioned defects, the purpose of the present invention is to solve the problem of collaborative task allocation of unmanned boat clusters. In view of the problems of insufficient population diversity, poor convergence and poor applicability of the traditional gray wolf algorithm to task allocation problems, the present invention proposes corresponding modification strategies.

[0006] To achieve this object, the present invention adopts the following technical solutions:

[0007] A method for allocating unmanned boat swarm tasks based on an improved discrete grey wolf algorithm comprises the following steps:

[0008] Step 1: Input map information, UAV cluster information, mission point information, maximum number of iterations T, and population size N in advance, model various information, and design a cost function to measure the quality of the mission;

[0009] Step 2: Call the improved discrete grey wolf algorithm to perform population initialization encoding on all task points;

[0010] Step 3: Update and iterate the initialized population. First, use the individual and group interaction strategy to cluster the scattered task points so that the task points with similarity are aggregated into one class, which is conducive to the subsequent allocation of task points to the unmanned boat.

[0011] Step 4: Update the leader guidance strategy for the clustered task points;

[0012] Step 5: Finally, update the individual wolf wandering strategy of the population;

[0013] Step 6: If the current number of iterations t is greater than the maximum number of iterations T, output the alpha wolf population sequence, otherwise return to step 3;

[0014] Step 7: Decode the alpha wolf population sequence and assign the decoded task sequence to each unmanned boat to complete the unmanned boat task allocation.

[0015] Preferably, in step 2:

[0016] Applying the encoding to the algorithm transforms the GWO algorithm into a discrete GWO algorithm to be applicable to the task allocation problem. The operation is as follows:

[0017] During the encoding initialization process, the number of encodings of the improved discrete GWO algorithm is N, where N represents the population size, also called chromosomes. Each chromosome consists of n+m-1 genes, n is the total number of mission points, and m is the number of unmanned boats performing the mission. The genes of the mission points are numbered {1, 2, 3, …, n}, and the genes of the split points are numbered {n+1, n+2, …, n+m-1}.

[0018] The encoding process is the process of randomly combining these n+m-1 genes into a chromosome. In order to make the initial encoding more diverse, this chapter uses the SPM chaotic mapping method for encoding. The specific process is as follows: the n+m-1 genes on each chromosome of N chromosomes are first generated by SPM for random numbers, and then the n+m-1 genes are assigned corresponding values ​​according to the size of the generated random numbers; the formula of SPM chaotic mapping is as follows:

[0019]

[0020] Among them, η is a random number between [0, 0.5]. Decoding is the process of untangling each chromosome to generate m sub-chromosomes. The chromosome can be divided into m groups of sub-chromosomes through m-1 split points, and these m groups of sub-chromosomes are the execution task sets to which each unmanned boat will be assigned.

[0021] Preferably, in step 3, the traditional gray wolf algorithm only has communication between individual wolves and the three alpha wolves, β, and γ, and lacks communication with other wolf packs, which leads to insufficient population diversity. Therefore, this study introduces the idea of ​​K-means clustering algorithm on the basis of discrete GWO algorithm, so that individual wolves can communicate with the group of wolves, so that wolves with similar individuals can be clustered into a wolf pack.

[0022] The K-means algorithm is an unsupervised learning clustering algorithm whose goal is to minimize the sum of the squares of the distances between each task point and the center point of its corresponding cluster. The mathematical expression is as follows:

[0023]

[0024] Among them, n is the number of mission points, m is the number of unmanned boats, and x i represents the position of the i-th task, μ k Indicates the center point position of the kth cluster, 1 c(i)＝k is the indicator function, which means when x i If it belongs to cluster k, its value is 1, otherwise it is 0.

[0025] When the improved discrete grey wolf algorithm uses K-means to perform individual and group communication behavior, it is necessary to first i , i=1,2,…,N is decoded to generate the task sequence L k ,k=1,2,…,m. Then for each task sequence L k Perform cluster midpoint calculation to obtain the cluster center point μ corresponding to the task sequence k ,k=1,2,…,m;

[0026] Then from each task sequence L k A gene sequence is randomly selected from the original sequence to form a new task set S, and then the task set S is deleted from the original sequence, and the cluster k that makes F the smallest is calculated according to formula (2). At the same time, the task set S is redistributed to the task sequence. After the distribution, the new task sequence and the split point gene are updated. Finally, the new task sequence is re-encoded to form a new population X. i ′.

[0027] Preferably, in step 4, the population update formula of the traditional grey wolf algorithm is:

[0028] d α=|C(P α -P)|,d β =|C(P β -P)|,d γ =|C(P γ -P)| Formula (3)

[0029] P1=P α -Ad α ,P1=P β -A2d β ,P3=P γ -Ad γ Formula (4)

[0030]

[0031] In order to make the population update formula of communication between individuals and the alpha wolf applicable to the discrete gray wolf algorithm, and to avoid the traditional gray wolf algorithm from falling into the local optimal situation, d α ,d β ,d γ Modify to:

[0032] d α =0.3+0.2cos(2πt / T)·(2rand-1) Formula (6)

[0033]

[0034] d γ =1-d α -d β Formula (8)

[0035] Where t is the current iteration number of the population, T is the total iteration number, and the update of P1, P2, and P3 is modified as follows:

[0036]

[0037] Among them, cross(X i ,X′ i ) represents the population X i With population X′ i (Chromosome X i Chromosome X′ i ) perform crossover operations between them, and the detailed operation process is as follows: Assume that the population X i The numbering sequence is X i (1,2,3,4,5,6,7,8,9,10), population X′ i The numbering sequence is X′ i (5,7,1,3,4,8,10,9,6,2), now randomly select population X′ iTwo numbered positions r1 and r2 on the y-axis, assuming r1 = 3 and r2 = 6, then X′ i {r1,r2} is (1,3,4,8), then the population X i Up and X′ i The same genes {r1, r2} are deleted and then from the remaining population X i Randomly select a position r3 and place X′ i {r1, r2} is inserted into this position to generate a new population X i (t+1).

[0038] Furthermore, in step 5, in addition to communicating with other individual wolves and the alpha wolf, individual wolves also need to roam and hunt independently, so that the algorithm can better conduct local exploration and accelerate algorithm convergence. This method proposes a task reordering method based on the original algorithm to reorder the decoded unmanned boat task sequence;

[0039] The detailed operation of task rescheduling is as follows: Randomly select task sequence L i Extract a task sequence L0 from the task list and delete L i The same task points in L0; then add the task points in L0 to L i The positions added are those that minimize ΔF. The formula for ΔF is defined as follows:

[0040]

[0041] in, Indicates adding the jth task in L0 to the task sequence L i The kth position in each task sequence L i All the tasks corresponding to L0 are added to L with the minimum increment of ΔF. i Then, we get the task sequence L′ after task rearrangement i , re-encode the m groups of rearranged task sequences to obtain the updated population X′.

[0042] One of the above technical solutions has the following beneficial effects: Compared to the traditional GWO algorithm, this method for allocating unmanned vehicle (UAV) tasks can more effectively improve the efficiency of UAV cluster allocation, significantly shorten overall task execution time, and effectively solve complex large-scale task allocation problems. Furthermore, compared to the traditional GWO algorithm, this algorithm has a faster convergence speed and higher convergence quality, and can successfully avoid the problem of the algorithm falling into local optimality. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is an algorithm flow chart of the overall steps of an embodiment of the present invention;

[0044] Figure 2 Schematic diagram of the encoding-decoding process of one embodiment of the present invention;

[0045] Figure 3 is a pseudo code of the K-means communication strategy of one embodiment of the present invention;

[0046] Figure 4 is a schematic diagram of the cross operation principle of an embodiment of the present invention;

[0047] Figure 5 This is a pseudo code of the task rescheduling method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following describes embodiments of the present invention in detail. Examples of the embodiments 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 intended only to explain the present invention and are not to be construed as limiting the present invention.

[0049] like Figure 1 As shown in FIG, a method for allocating unmanned boat swarm tasks based on an improved discrete grey wolf algorithm includes the following steps:

[0050] Step 1: Input map information, UAV cluster information, mission point information, maximum number of iterations T, and population size N in advance, model various information, and design a cost function to measure the quality of the mission;

[0051] Step 2: Call the improved discrete grey wolf algorithm to perform population initialization encoding on all task points;

[0052] Step 3: Update and iterate the initialized population. First, use the individual and group interaction strategy to cluster the scattered task points so that the task points with similarity are aggregated into one class, which is conducive to the subsequent allocation of task points to the unmanned boat.

[0053] Step 4: Update the leader guidance strategy for the clustered task points;

[0054] Step 5: Finally, update the individual wolf wandering strategy of the population;

[0055] Step 6: If the current number of iterations t is greater than the maximum number of iterations T, output the alpha wolf population sequence, otherwise return to step 3;

[0056] Step 7: Decode the alpha wolf population sequence and assign the decoded task sequence to each unmanned boat to complete the unmanned boat task allocation.

[0057] Compared to the traditional GWO algorithm, this method for unmanned aerial vehicle (UAV) task allocation can more effectively improve the efficiency of UAV swarm allocation, significantly shorten the overall task execution time, and effectively solve complex large-scale task allocation problems. Furthermore, compared to the traditional GWO algorithm, this algorithm has a faster convergence speed and higher convergence quality, successfully avoiding the problem of the algorithm being trapped in local optimality, effectively improving the rationality of task scheduling and resource utilization, and significantly shortening the overall task execution time. Secondly, the algorithm exhibits greater robustness and scalability in complex and large-scale task environments, and can adapt to the collaborative allocation requirements under multi-objective and multi-constraint conditions. This enables it to achieve faster convergence speed and higher convergence accuracy, more stably find the global optimal solution, and successfully avoid being trapped in local optimality. Experimental results show that this method not only enhances the algorithm's global search capability but also effectively balances the relationship between exploration and exploitation, significantly improving the overall execution efficiency and task completion rate of UAV swarms in dynamic and complex task scenarios.

[0058] like Figure 2 As shown, in step 2:

[0059] Applying the encoding to the algorithm transforms the GWO algorithm into a discrete GWO algorithm to be applicable to the task allocation problem. The operation is as follows:

[0060] During the encoding initialization process, the number of encodings of the improved discrete GWO algorithm is N, where N represents the population size, also called chromosomes. Each chromosome consists of n+m-1 genes, n is the total number of mission points, and m is the number of unmanned boats performing the mission. The genes of the mission points are numbered {1, 2, 3, …, n}, and the genes of the split points are numbered {n+1, n+2, …, n+m-1}.

[0061] The encoding process is the process of randomly combining these n+m-1 genes into a chromosome. In order to make the initial encoding more diverse, this chapter uses the SPM chaotic mapping method for encoding. The specific process is as follows: the n+m-1 genes on each chromosome of N chromosomes are first generated by SPM for random numbers, and then the n+m-1 genes are assigned corresponding values ​​according to the size of the generated random numbers; the formula of SPM chaotic mapping is as follows:

[0062]

[0063] Among them, η is a random number between [0, 0.5]. Decoding is the process of untangling each chromosome to generate m sub-chromosomes. The chromosome can be divided into m groups of sub-chromosomes through m-1 split points, and these m groups of sub-chromosomes are the execution task sets to which each unmanned boat will be assigned.

[0064] What are the good technical effects of doing this? The core benefit of discretizing GWO into DGWO and applying it to task allocation problems is that it closely fits the discrete nature of the problem, which can achieve more efficient search and better solution quality; DGWO effectively avoids the decoding bias and local optimal traps that are prone to occur in continuous algorithms in discrete problems, and is particularly suitable for the efficient allocation of large-scale and complex tasks, and has stronger robustness and practical value.

[0065] like Figure 3 As shown in Figure 3, in step 3, the traditional gray wolf algorithm only has communication between individual wolves and the three alpha wolves, β, and γ, and lacks communication with other wolf packs, which leads to insufficient population diversity. Therefore, this study introduces the idea of ​​K-means clustering algorithm on the basis of discrete GWO algorithm, so that individual wolves can communicate with the group of wolves, so that wolves with similar individuals can be clustered into a wolf pack.

[0066] The K-means algorithm is an unsupervised learning clustering algorithm whose goal is to minimize the sum of the squares of the distances between each task point and the center point of its corresponding cluster. The mathematical expression is as follows:

[0067]

[0068] Among them, n is the number of mission points, m is the number of unmanned boats, and x i represents the position of the i-th task, μ k Indicates the center point position of the kth cluster, 1 c(i)＝k is the indicator function, which means when x i If it belongs to cluster k, its value is 1, otherwise it is 0.

[0069] When the improved discrete grey wolf algorithm uses K-means to perform individual and group communication behavior, it is necessary to first i , i=1,2,…,N is decoded to generate the task sequence L k ,k=1,2,…,m. Then for each task sequence L k Perform cluster midpoint calculation to obtain the cluster center point μ corresponding to the task sequence k ,k=1,2,…,m;

[0070] Then from each task sequence L k A gene sequence is randomly selected from the original sequence to form a new task set S, and then the task set S is deleted from the original sequence, and the cluster k that makes F the smallest is calculated according to formula (2). At the same time, the task set S is redistributed to the task sequence. After the distribution, the new task sequence and the split point gene are updated. Finally, the new task sequence is re-encoded to form a new population X. i ′.

[0071] What are the good technical effects of doing this? This strategy is conducive to making the task points in the same unmanned boat task sequence as close as possible, thereby avoiding the situation where the task execution cost increases due to the task points being too far apart. This strategy helps to optimize the spatial distribution of each task point in the task sequence executed by the same unmanned boat, so that the task points are as close as possible, thereby effectively reducing the path cost and execution cost caused by the large distance between task points. By improving the coherence and compactness of the task path, it can not only significantly shorten the execution time of a single boat and reduce energy consumption, but also improve the rationality of the overall task scheduling and the efficiency of cluster execution. In addition, this strategy can also reduce the possibility of conflicts and resource competition between tasks in a multi-task collaborative environment, and enhance the stability and execution robustness of the system.

[0072] like Figure 4 As shown, in step 4, the population update formula of the traditional gray wolf algorithm is:

[0073] d α =|C(P α -P)|,d β =|C(P β -P)|,d γ =|C(P γ -P)| Formula (3)

[0074] P1=P α -Ad α ,P1=P β -A2d β ,P3=P γ -Ad γ Formula (4)

[0075]

[0076] In order to make the population update formula of communication between individuals and the alpha wolf applicable to the discrete gray wolf algorithm, and to avoid the traditional gray wolf algorithm from falling into the local optimal situation, d α ,d β ,d γ Modify to:

[0077] d α =0.3+0.2cos(2πt / T)·(2rand-1) Formula (6)

[0078]

[0079] d γ =1-d α -d β Formula (8)

[0080] Where t is the current iteration number of the population, T is the total iteration number, and the update of P1, P2, and P3 is modified as follows:

[0081]

[0082] Among them, cross(X i ,X′ i ) represents the population X i With population X′ i (Chromosome X i Chromosome X′ i ) perform crossover operations between them, and the detailed operation process is as follows: Assume that the population X i The numbering sequence is X i (1,2,3,4,5,6,7,8,9,10), population X′ i The numbering sequence is X′ i (5,7,1,3,4,8,10,9,6,2), now randomly select population X′ i Two numbered positions r1 and r2 on the y-axis, assuming r1 = 3 and r2 = 6, then X′ i {r1,r2} is (1,3,4,8), then the population X i Up and X′ i The same genes {r1, r2} are deleted and then from the remaining population X i Randomly select a position r3 and place X′ i {r1, r2} is inserted into this position to generate a new population X i (t+1).

[0083] In discrete optimization problems such as task allocation, crossover algorithms improve solution quality and accelerate convergence. Introducing crossover operations is an effective strategy for increasing the diversity and quality of solutions to these problems. By reorganizing the task allocation sequence among multiple candidate solutions, crossover algorithms can explore new regions of the solution space while maintaining the characteristics of high-quality solutions, thereby accelerating the algorithm's convergence. This strategy not only helps escape local optimality but also enhances the global search capabilities of the population, improving the stability of the final solution and the overall optimization quality. It is particularly effective when dealing with large-scale, highly complex task allocation problems.

[0084] like Figure 5 As shown in Figure 5, in addition to communicating with other individual wolves and the leader wolf, individual wolves also need to roam and hunt independently, so that the algorithm can better perform local exploration and accelerate algorithm convergence. This method proposes a task reordering method based on the original algorithm to reorder the decoded unmanned boat task sequence;

[0085] The detailed operation of task rescheduling is as follows: Randomly select task sequence L i Extract a task sequence L0 from the task list and delete L i The same task points in L0; then add the task points in L0 to L i The positions added are those that minimize ΔF. The formula for ΔF is defined as follows:

[0086]

[0087] in, Indicates adding the jth task in L0 to the task sequence L i The kth position in each task sequence L i All the tasks corresponding to L0 are added to L with the minimum increment of ΔF. i Then, we get the task sequence L′ after task rearrangement i , re-encode the m groups of rearranged task sequences to obtain the updated population X′.

[0088] This step can transform the disordered task population into an ordered, accurate task sequence through clustering, which can greatly improve the convergence speed of the algorithm. This step can efficiently transform the originally disordered task population into a structured, ordered, and reasonable task execution sequence, effectively improving the feasibility and execution efficiency of the task allocation scheme. By clustering and sorting individual tasks, it not only speeds up the algorithm's identification and convergence of high-quality solutions, but also reduces redundant searches and unnecessary waste of computing resources. At the same time, this process improves the coherence and stability of the task sequence, helps reduce execution costs, improves resource utilization, and enhances the algorithm's global optimization capabilities and robustness in complex task environments.

[0089] The technical principles of the present invention have been described above with reference to specific embodiments. These descriptions are intended solely to illustrate the principles of the present invention and are not to be construed in any way as limiting the scope of protection of the present invention. Based on the explanations herein, those skilled in the art will readily conceive of other specific embodiments of the present invention without inventive effort, and such embodiments will fall within the scope of protection of the present invention.

Claims

1. A method for allocating unmanned boat swarm tasks based on an improved discrete grey wolf algorithm, characterized in that: The steps include: Step 1: Input map information, UAV cluster information, mission point information, maximum number of iterations T, and population size N in advance, model various information, and design a cost function to measure the quality of the mission; Step 2: Call the improved discrete grey wolf algorithm to perform population initialization encoding on all task points; Step 3: Update and iterate the initialized population. First, use the individual and group interaction strategy to cluster the scattered task points so that the task points with similarity are aggregated into one class, which is conducive to the subsequent allocation of task points to the unmanned boat. Step 4: Update the leader guidance strategy for the clustered task points; Step 5: Finally, update the individual wolf wandering strategy of the population; Step 6: If the current number of iterations t is greater than the maximum number of iterations T, output the alpha wolf population sequence, otherwise return to step 3; Step 7: Decode the alpha wolf population sequence and assign the decoded task sequence to each unmanned boat to complete the unmanned boat task allocation.

2. The unmanned boat swarm task allocation method based on the improved discrete grey wolf algorithm according to claim 1 is characterized in that: In step 2: Applying the encoding to the algorithm transforms the GWO algorithm into a discrete GWO algorithm to be applicable to the task allocation problem. The operation is as follows: During the encoding initialization process, the number of encodings of the improved discrete GWO algorithm is N, where N represents the population size, also called chromosomes. Each chromosome consists of n+m-1 genes, n is the total number of mission points, and m is the number of unmanned boats performing the mission. The genes of the mission points are numbered {1, 2, 3, …, n}, and the genes of the split points are numbered {n+1, n+2, …, n+m-1}. The encoding process is the process of randomly combining these n+m-1 genes into a chromosome. In order to make the initial encoding more diverse, this chapter uses the SPM chaotic mapping method for encoding. The specific process is as follows: the n+m-1 genes on each chromosome of N chromosomes are first generated by SPM for random numbers, and then the n+m-1 genes are assigned corresponding values ​​according to the size of the generated random numbers; the formula of SPM chaotic mapping is as follows: Among them, η is a random number between [0, 0.5]. Decoding is the process of untangling each chromosome to generate m sub-chromosomes. The chromosome can be divided into m groups of sub-chromosomes through m-1 split points, and these m groups of sub-chromosomes are the execution task sets to which each unmanned boat will be assigned.

3. The unmanned boat swarm task allocation method based on the improved discrete grey wolf algorithm according to claim 1 is characterized in that: In step 3, the traditional gray wolf algorithm only communicates with the three alpha wolves, α, β, and γ, and lacks communication with other wolf packs, which leads to insufficient population diversity. Therefore, this study introduces the idea of ​​K-means clustering algorithm on the basis of discrete GWO algorithm, so that individual wolves can communicate with the group of wolves, so that wolves with similar individuals can be clustered together. The K-means algorithm is an unsupervised learning clustering algorithm whose goal is to minimize the sum of the squares of the distances between each task point and the center point of its corresponding cluster. The mathematical expression is as follows: Among them, n is the number of mission points, m is the number of unmanned boats, and x i represents the position of the i-th task, μ k Indicates the center point position of the kth cluster, 1 c(i)＝k is the indicator function, which means when x i When it belongs to cluster k, its value is 1, otherwise it is 0; When the improved discrete grey wolf algorithm uses K-means to perform individual and group communication behavior, it is necessary to first i , i=1,2,…,N is decoded to generate the task sequence L k ,k=1,2,…,m; then for each task sequence L k Perform cluster midpoint calculation to obtain the cluster center point μ corresponding to the task sequence k ,k=1,2,…,m; Then from each task sequence L k A gene sequence is randomly selected from the original sequence to form a new task set S, and then the task set S is deleted from the original sequence, and the cluster k that makes F the smallest is calculated according to formula (2). At the same time, the task set S is redistributed to the task sequence. After the distribution, the new task sequence and the split point gene are updated. Finally, the new task sequence is re-encoded to form a new population X. i ′.

4. The unmanned boat swarm task allocation method based on the improved discrete grey wolf algorithm according to claim 1 is characterized in that: In step 4, the population update formula of the traditional gray wolf algorithm is: d α =|C(P α -P)|,d β =|C(P β -P)|,d γ =|C(P γ -P)| Formula (3) P1=P α -Ad α ,P1=P β -A2d β ,P3=P γ -Ad γ Formula (4) In order to make the population update formula of communication between individuals and the alpha wolf applicable to the discrete gray wolf algorithm, and to avoid the traditional gray wolf algorithm from falling into the local optimal situation, d α ,d β ,d γ Modify to: d α =0.3+0.2 cos(2πt / T)·(2rand-1) Formula (6) d γ =1-d α -d β Formula (8) Where t is the current iteration number of the population, T is the total iteration number, and the update of P1, P2, and P3 is modified as follows: Among them, cross(X i ,X′ i ) represents the population X i With population X′ i (Chromosome X i Chromosome X′ i ) perform crossover operations between them, and the detailed operation process is as follows: Assume that the population X i The numbering sequence is X i (1,2,3,4,5,6,7,8,9,10), population X′ i The numbering sequence is X′ i (5,7,1,3,4,8,10,9,6,2), now randomly select population X′ i Two numbered positions r1 and r2 on the y-axis, assuming r1 = 3 and r2 = 6, then X′ i {r1,r2} is (1,3,4,8), then the population X i Up and X′ i The same genes {r1, r2} are deleted and then from the remaining population X i Randomly select a position r3 and place X′ i {r1, r2} is inserted into this position to generate a new population X i (t+1).

5. The unmanned boat swarm task allocation method based on the improved discrete grey wolf algorithm according to claim 1 is characterized in that: In step 5, in addition to communicating with other individual wolves and the alpha wolf, individual wolves also need to roam and hunt independently, so that the algorithm can better conduct local exploration and accelerate algorithm convergence. This method proposes a task reordering method based on the original algorithm to reorder the decoded unmanned boat task sequence; The detailed operation of task rescheduling is as follows: Randomly select task sequence L i Extract a task sequence L0 from the task list and delete L i The same task points in L0; then add the task points in L0 to L i The positions added are those that minimize ΔF. The formula for ΔF is defined as follows: ΔF=F(L i ⊕ k L0{j})-F(L i ) Formula (10) Among them, L i ⊕ k L0{j} means adding the jth task in L0 to the task sequence L i The kth position in each task sequence L i All the tasks corresponding to L0 are added to L with the minimum increment of ΔF. i Then, we get the task sequence L′ after task rearrangement i , re-encode the m groups of rearranged task sequences to obtain the updated population X′.

Images