Task allocation method for cluster cooperative patrol of unmanned surface vehicles with variable number

Abstract

The invention relates to a task allocation method for collaborative patrol of a variable-number unmanned ship cluster. The method comprises the following steps: modeling unmanned ships; establishing the patrol map as a connected graph of a plurality of nodes; setting evaluation indexes of the nodes, wherein the evaluation indexes comprise a group revenue function and an individual revenue function; the distribution of patrol tasks is modeled as a multi-index optimization problem, an individual revenue function is optimized through distributed calculation, so that a group revenue function is optimized, a task distribution scheme is solved, and meanwhile, distribution nodes are dynamically updated in the task distribution process; the responsible node of each unmanned ship is dynamically updated according to the current map state and position. Compared with the prior art, a distributed optimization scheme is adopted, dynamic distribution of patrol nodes can be achieved, and patrol tasks can be executed more efficiently and evenly in a complex marine environment.

Description

Technical Field

[0001] The invention relates to the field of intelligent control, and in particular to a task allocation method for coordinated patrolling by a variable number of unmanned boat clusters. Background Art

[0002] There is a huge demand for near-coastal and island defense missions. The swarm technology of unmanned ships currently plays the role of preliminary reconnaissance, preliminary contact, and enemy situation investigation in marine defense, which is of great significance to the marine defense system.

[0003] Unmanned surface boats are a kind of surface sports platform with certain autonomous capabilities, which can sail autonomously and complete operational tasks under unmanned conditions. They are low-cost, concealed, and flexible. They can perform environmental surveys, anti-submarine detection, mine detection, search and rescue, material transportation and other tasks, and are widely used in marine surveys and research. The coordinated operation of multiple unmanned boats will have higher operating efficiency. The marine cluster system is a system engineering involving multiple key technologies, among which task allocation technology is equally important as autonomous communication networking, track planning, formation control and other technologies, and is one of the key technologies of marine unmanned system clusters.

[0004] However, the existing unmanned boat swarm task allocation still has problems with low solution efficiency and task allocation effectiveness. Summary of the invention

[0005] The purpose of the present invention is to provide a task allocation method for cooperative patrol of a variable number of unmanned boat clusters, solve the task allocation problem of cooperative patrol of unmanned boat clusters at sea based on discrete particle swarm optimization algorithm, dynamically allocate tasks when the number of unmanned boat clusters changes, realize cooperative patrol tasks, and complete rapid coverage of dangerous sea areas.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] A task allocation method for coordinated patrol of a variable number of unmanned boat swarms comprises the following steps:

[0008] S1, modeling of unmanned boat;

[0009] S2, establishing the patrol map as a connected graph of several nodes;

[0010] S3, setting evaluation indicators of the node, wherein the evaluation indicators include group benefit function and individual benefit function;

[0011] S4, the allocation of patrol tasks is modeled as a multi-index optimization problem. The individual benefit function is optimized through distributed computing, thereby optimizing the group benefit function, solving the task allocation plan, and dynamically updating the allocation nodes during the task allocation process.

[0012] The group benefit function includes coverage rate, load balancing degree of multiple unmanned boat systems and average node access interval. The coverage rate measures the coverage of the entire patrol area by the multiple unmanned boat system, the load balancing degree of the multiple unmanned boat system measures the balance of the number of tasks assigned to each executable unmanned boat, and the average node access interval measures the speed at which the multiple unmanned boat system predicts and eliminates risk nodes, wherein the risk node is defined as a node that has not been visited for a long time.

[0013] The coverage rate is defined as the ratio of the number of patrol nodes visited in the past time window of length T to the total number of patrol nodes:

[0014]

[0015] Where r is the coverage rate, W represents the set of patrol nodes visited in the time window T, V represents the total set of patrol nodes, and |·| represents the number of nodes in the node set.

[0016] The load balancing degree of the multi-unmanned boat system is defined as the standard deviation of the number of tasks assigned to each unmanned boat;

[0017] The value of the average node access interval is a value close to the time window length T.

[0018] The individual benefit function includes a risk elimination reward benefit function, a critical distance penalty function and a neighborhood load balancing degree function. The risk elimination reward benefit function measures the estimated benefit of a single unmanned boat being assigned a node for the boat, the critical distance penalty function measures the overlap of the unmanned boat's responsible area, and the neighborhood load balancing degree function measures the load balancing of each boat in the neighborhood of each unmanned boat.

[0019] The risk elimination reward function is defined as the ratio of the risk value of a node to the estimated time to reach the node:

[0020]

[0021] Among them, i represents the i-th mission node, j represents the j-th unmanned boat, and interval i represents the risk value, which is defined as the square of the difference between the current time of the i-th task node and the last visit time. The estimated time is defined as the distance d between the j-th unmanned boat and the i-th task node. ij and the speed of the unmanned boat v j The ratio of ,P j is the set of mission points that the jth unmanned boat is responsible for.

[0022] The proximity penalty function is defined as:

[0023]

[0024] Among them, max is the preset upper bound of the penalty score, k1 is a fixed constant, and d inear It refers to the distance from the i-th node to the nearest node that is different from the unmanned boat it is responsible for.

[0025] The neighborhood load balancing function is defined as:

[0026]

[0027] Among them, N j represents the set of several unmanned boat neighbors closest to the jth executable unmanned boat, S j represents the average number of responsible nodes of each unmanned boat in the neighborhood of the unmanned boat, S j Represents the number of responsible nodes of the i-th unmanned boat.

[0028] During patrol task allocation, for dynamic task nodes, the multi-unmanned boat system allocates tasks according to the following process at fixed intervals:

[0029] Each unmanned boat abandons some nodes in its own responsible node set according to the value of the individual benefit function;

[0030] Each unmanned boat samples from the risk nodes to be allocated, and uses the discrete particle swarm optimization algorithm to optimize the allocation pair of the node to be allocated and the unmanned boat in the neighborhood composed of multiple unmanned boats closest to itself;

[0031] Each unmanned boat updates its own responsible node set and selects the nearest node from it. According to the adjacency information and the shortest path between nodes in the pre-stored patrol map, the unmanned boat moves and visits nodes with a greedy strategy to perform patrol tasks.

[0032] The discrete particle swarm optimization algorithm optimizes the allocation of patrol nodes to be assigned through crossover and mutation operations. In the mutation phase, a random position is mutated instead of randomly selecting two positions for exchange; the position of the particle is represented by an m-dimensional vector X i =(x1,x2,...,x m ), then the position update formula of discrete particle swarm optimization algorithm is as follows:

[0033]

[0034] in, Represents the probability w of the original position X i(t) Perform mutation operation: randomly generate a random number r in the range [0,1]. If r is less than w, perform mutation operation, randomly generate an integer x in the range [1,m], and randomly change the element at position x to an integer in the range [1,L+1]. L+1 is the number of executable unmanned boats calculated in the current neighborhood; and Indicates that X is changed with probability c1 and c2 i (t) and individual extreme value position P i (t), the global extreme position G(t) performs a crossover operation: randomly generate a random number r in the range [0,1]. If r is less than c, a crossover operation is performed to randomly generate two integers ls and rs in the range [1,m]. i The values ​​in the position range [ls, rs] in (t) are replaced by the individual extreme value position P i (t) or the value of the position corresponding to the global extreme value position G(t).

[0035] In the patrol task allocation process, the allocation vector is used to encode the allocation scheme between the nodes to be allocated and the executable unmanned boats. For the allocation problem of X nodes to be allocated and Y executable unmanned boats at the current moment, an X×1 allocation vector is used to represent the solution of the problem. If the i-th node is allocated to the j-th unmanned boat, the value of the i-th position of the allocation vector is j.

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

[0037] (1) The present invention adopts a distributed allocation architecture, allowing each unmanned boat to sample from the global patrol point set and then perform local allocation within a neighborhood consisting of several adjacent unmanned boat "neighbors". This allows the overall large-scale task solving problem to be divided into task allocations in the local neighborhoods of several intelligent agents, thereby reducing the computational overhead of task allocation and avoiding the problem of insufficient solvability when the task scale is too large in the centralized approach.

[0038] (2) The present invention adopts discrete particle swarm algorithm for allocation, which can meet the discrete form optimization problem of patrol node-executable ship. In the optimization of local neighborhood, the number of executable ships is small, and the discrete particle swarm algorithm has strong search efficiency. In addition, the conventional particle swarm algorithm is added with crossover and mutation operations to optimize the search efficiency, improve the algorithm's solution efficiency, and ensure the optimality of task allocation.

[0039] (3) The patrol allocation problem of multiple unmanned boat swarms for multiple mission points is an NP-hard problem, and the complexity of directly calculating and solving it is relatively large. In the present invention, when each unmanned boat calculates its own allocation plan, it only selects a part of the unmanned boats from the set of all unmanned boats for allocation, which greatly reduces the complexity of solving the problem.

[0040] (4) The present invention uses allocation vectors to encode the allocation scheme between the nodes to be allocated and the executable unmanned ships; and, at fixed intervals, each unmanned ship chooses to abandon a portion of nodes with low individual benefit functions and samples from the risk nodes to be allocated, and updating the allocation scheme is conducive to the redistribution of task nodes. When the number of unmanned boat clusters changes, the system dynamically updates the responsible nodes of each ship according to the current map status and position, which is conducive to improving the efficiency of task allocation in the multi-unmanned ship system.

[0041] (5) The present invention regards the task as a multi-index optimization problem, establishes a group benefit function as an overall evaluation indicator, and an individual benefit function for distributed computing, and explains their corresponding physical meanings and performance requirements to improve the overall patrol efficiency of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is a flow chart of the method of the present invention;

[0043] Figure 2 A schematic diagram of a map to be patrolled in an embodiment of the present invention;

[0044] Figure 3 A schematic diagram of the initial state of the unmanned boat cluster coordinated patrol involved in an embodiment of the present invention;

[0045] Figure 4 A schematic diagram of a coordinated patrol process of an unmanned boat swarm involved in an embodiment of the present invention;

[0046] Figure 5 It is a schematic diagram of the optimization process of the discrete particle swarm algorithm of the present invention. DETAILED DESCRIPTION

[0047] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0048] This embodiment provides a task allocation method for coordinated patrol of a variable number of unmanned boat swarms, such as Figure 1 As shown, the following steps are included:

[0049] S1, modeling of unmanned boat.

[0050] According to Newtonian kinematics and hydrodynamics related equations, the unmanned boat involved in the present invention is modeled as follows:

[0051]

[0052] where η = [x, y, ψ] T Represents the 2D coordinates and orientation angle of the ship in the world coordinate system, V = [u, v, r] T represents the velocity vector of the hull in its own coordinate system, J(η) represents the rotation matrix from the world coordinate system to the hull, τ is the force vector of the hull, M is the mass matrix, C(V) represents the centripetal force matrix and the Cody-Austrian matrix, and D(V) represents the damping matrix.

[0053] Since the patrol distance at sea is long and the distance between patrol nodes is much larger than the turning radius of the unmanned ship, the unmanned ship under the patrol problem can be regarded as an approximate all-wheel drive system at the decision-making level, and its motion speed u is assumed to be uniform, and there is

[0054] x(t+1)=x(t)+u·cosψ

[0055] y(t+1)=y(t)+u·sinψ

[0056] Therefore, we only need to consider optimizing the movement direction ψ of the unmanned ship at the next moment to update the position of the unmanned ship.

[0057] S2, establish the patrol map as a connected graph of several nodes.

[0058] The patrol map is obtained by image processing technology as a binary grid map with both ocean and land areas. The value of the obstacle (island, land) area is 1, and the value of the feasible area (ocean) is 0. Then the Sobel edge detection method is used to extract the boundaries of each island. According to the distribution of each boundary point, the boundary point set is sampled according to the principle of uniform distribution of two-dimensional area, so that patrol nodes are marked along the boundary as map node information for simulation verification, such as Figure 2 The goal of the patrol mission is to have the unmanned boat cluster patrol regularly. Figure 2 Task nodes in the task can be maximized to maximize coverage and eliminate dangerous nodes in time.

[0059] S3, setting evaluation indicators of the nodes, wherein the evaluation indicators include a group benefit function and an individual benefit function.

[0060] The group benefit function includes coverage rate, load balancing degree of multi-unmanned boat system and average node access interval. The coverage rate measures the coverage of the entire patrol area by the multi-unmanned boat system, the load balancing degree of the multi-unmanned boat system measures the balance of the number of tasks assigned to each executable unmanned boat, and the average node access interval measures the speed at which the multi-unmanned boat system predicts and eliminates risk nodes. Risk nodes are defined as nodes that have not been visited for a long time.

[0061] The coverage rate is defined as the ratio of the number of patrol nodes visited in the past time window of length T to the total number of patrol nodes:

[0062]

[0063] Among them, r is the coverage rate, W represents the set of patrol nodes visited within the time window T, V represents the total patrol node set, and |·| represents the number of nodes in the node set. Ideally, when the multi-unmanned boat patrol system achieves efficient coverage of the target area, most of the risk nodes should be discovered and cleared quickly, so that the overall coverage rate fluctuates less, rapidly decreases from the initial state, and finally stabilizes at a lower level. Therefore, the first-order derivative of the coverage rate should be larger at the beginning and smaller later.

[0064] The load balancing degree of the multi-UAV system is defined as the standard deviation of the number of tasks assigned to each UAV;

[0065] Ideally, the average node access interval takes a value close to the time window length T.

[0066] The individual benefit function includes the risk elimination reward benefit function, the critical distance penalty function and the neighborhood load balancing degree function. The risk elimination reward benefit function measures the estimated benefit of a single unmanned boat being assigned a node for the boat, the critical distance penalty function measures the overlap of the unmanned boat's responsible area, and the neighborhood load balancing degree function measures the load balancing of each ship in the neighborhood of each unmanned boat.

[0067] The risk value is defined as the square of the time interval from the last visit to the present. For a single node, the longer it has not been visited, the higher its risk value. The risk elimination reward function is defined as the ratio of the risk value of a node to the estimated time to reach the node:

[0068]

[0069] Among them, i represents the i-th mission node, j represents the j-th unmanned boat, and interval i represents the risk value, which is defined as the square of the difference between the current time of the i-th task node and the last visit time. The estimated time is defined as the distance d between the j-th unmanned boat and the i-th task node. ijand the speed of the unmanned boat v j The ratio of ,P j is the set of mission points that the jth unmanned boat is responsible for.

[0070] In order for the multi-unmanned boat patrol system to make full use of the resources of each unmanned boat and achieve efficient coverage of the target area, each unmanned boat needs to have an appropriate division of labor. Ideally, each single ship should maintain a certain distance between them, and the intersection size of each responsible area needs to be at a low level, while their union can be close to full coverage of the target area risk point set. When a node contains nodes assigned to different executors, a corresponding penalty score will be generated. In this embodiment, the proximity distance penalty function is defined as:

[0071]

[0072] Among them, max is the preset upper bound of the penalty score, k1 is a fixed constant, and d inear It refers to the distance from the i-th node to the nearest node that is different from the unmanned boat it is responsible for.

[0073] Corresponding to the global load balancing function, the load balancing degree within the neighborhood of a single unmanned boat is an important factor in the optimization process. When each boat in the neighborhood of each unmanned boat reaches a load balancing state, the multi-unmanned boat cluster patrol system as a whole can reach a load balancing state. In this embodiment, the neighborhood load balancing degree function is defined as:

[0074]

[0075] Among them, N j represents the set of several unmanned boat neighbors closest to the jth executable unmanned boat, S j represents the average number of responsible nodes of each unmanned boat in the neighborhood of the unmanned boat, S j Indicates the number of responsible nodes for the i-th unmanned boat. Adding a negative sign in front of the task load distribution indicates that the load variance is controlled to prevent the variance from being too large during the optimization process.

[0076] S4, the allocation of patrol tasks is modeled as a multi-index optimization problem. The individual benefit function is optimized through distributed computing, thereby optimizing the group benefit function, solving the task allocation plan, and dynamically updating the allocation nodes during the task allocation process.

[0077] Regarding the task allocation problem in the coordinated patrol of multiple unmanned boat systems involved in the present invention, Figure 3As shown in the figure, by analyzing the cooperation and competition relationship of each unmanned ship during the patrol process, a group-individual relationship description of the group benefit function-individual benefit function is established. At the same time, the problem is regarded as a multi-index optimization problem. The individual benefit function is optimized through distributed computing, thereby optimizing the group benefit function and solving and optimizing the task allocation plan.

[0078] In the patrol task allocation process, the allocation vector is used to encode the allocation scheme between the nodes to be allocated and the executable unmanned boats. For the allocation problem of X nodes to be allocated and Y executable unmanned boats at the current moment, an X×1 allocation vector is used to represent the solution of the problem. If the i-th node is allocated to the j-th unmanned boat, the value of the i-th position of the allocation vector is j.

[0079] During patrol task allocation, for dynamic task nodes, the multi-unmanned boat system allocates tasks according to the following process at fixed intervals:

[0080] Each unmanned boat abandons some nodes in its own responsible node set according to the value of the individual benefit function;

[0081] Each unmanned boat samples from the risk nodes to be allocated, and uses the discrete particle swarm optimization algorithm to optimize the allocation pair of the node to be allocated and the unmanned boat in the neighborhood composed of multiple unmanned boats closest to itself;

[0082] Each unmanned boat updates its own responsible node set and selects the nearest node from it. According to the adjacency information in the pre-stored patrol map and the shortest path between nodes (pre-calculated by the Floyd algorithm), the unmanned boat moves and visits nodes with a greedy strategy to perform patrol tasks, such as Figure 4 shown.

[0083] The search space of the classic particle swarm algorithm is continuous and cannot be directly used for discrete optimization problems. The present invention redefines the operation operator in the particle swarm algorithm, draws on the ideas of mutation and crossover in biological genetics, and improves the classic particle swarm optimization algorithm into a discrete particle swarm algorithm in which the particle positions have a certain probability of mutation and crossover, so that the algorithm can be applied to discrete optimization problems. The algorithm flow is as follows: Figure 5 shown.

[0084] Specifically, the present invention randomly mutates a position in the mutation stage instead of randomly selecting two positions to exchange, which makes the search efficiency of the task allocation algorithm higher.

[0085] The position of the particle is represented by an m-dimensional vector X i =(x1,x2,...,x m ), then the position update formula of discrete particle swarm optimization algorithm is as follows:

[0086]

[0087] in, Represents the probability w of the original position X i (t) Perform mutation operation: randomly generate a random number r in the range [0,1]. If r is less than w, perform mutation operation, randomly generate an integer x in the range [1,m], and randomly change the element at position x to an integer in the range [1,L+1]. L+1 is the number of executable unmanned boats calculated in the current neighborhood; and Indicates that X is changed with probability c1 and c2 i (t) and individual extreme value position P i (t), the global extreme position G(t) performs a crossover operation: randomly generate a random number r in the range [0,1]. If r is less than c, a crossover operation is performed to randomly generate two integers ls and rs in the range [1,m]. i The values ​​in the position range [ls, rs] in (t) are replaced by the individual extreme value position P i (t) or the value of the position corresponding to the global extreme value position G(t).

[0088] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning, or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A task allocation method for coordinated patrol of a variable number of unmanned boat swarms, characterized in that: The following steps are involved: S1, modeling of unmanned boat; S2, establishing the patrol map as a connected graph of several nodes; S3, setting evaluation indicators of the node, wherein the evaluation indicators include a group benefit function and an individual benefit function; S4, the allocation of patrol tasks is modeled as a multi-index optimization problem. The individual benefit function is optimized through distributed computing, thereby optimizing the group benefit function, solving the task allocation plan, and dynamically updating the allocation nodes during the task allocation process.

2. The task allocation method for coordinated patrol of a variable number of unmanned boat swarms according to claim 1 is characterized in that: The group benefit function includes coverage rate, load balancing degree of multiple unmanned boat systems and average node access interval. The coverage rate measures the coverage of the entire patrol area by the multiple unmanned boat system, the load balancing degree of the multiple unmanned boat system measures the balance of the number of tasks assigned to each executable unmanned boat, and the average node access interval measures the speed at which the multiple unmanned boat system predicts and eliminates risk nodes, wherein the risk node is defined as a node that has not been visited for a long time.

3. The task allocation method for coordinated patrol of a variable number of unmanned boat swarms according to claim 2 is characterized in that: The coverage rate is defined as the ratio of the number of patrol nodes visited in the past time window of length T to the total number of patrol nodes: W={v|interval(v) <T,v∈V} Where r is the coverage rate, W represents the set of patrol nodes visited in the time window T, V represents the total set of patrol nodes, and |·| represents the number of nodes in the node set. The load balancing degree of the multi-unmanned boat system is defined as the standard deviation of the number of tasks assigned to each unmanned boat; The value of the average node access interval is a value close to the time window length T.

4. The task allocation method for coordinated patrol of a variable number of unmanned boat swarms according to claim 1 is characterized in that: The individual benefit function includes a risk elimination reward benefit function, a critical distance penalty function and a neighborhood load balancing degree function. The risk elimination reward benefit function measures the estimated benefit of a single unmanned boat being assigned a node for the boat, the critical distance penalty function measures the overlap of the unmanned boat's responsible area, and the neighborhood load balancing degree function measures the load balancing of each boat in the neighborhood of each unmanned boat.

5. The task allocation method for coordinated patrol of a variable number of unmanned boat swarms according to claim 4 is characterized in that: The risk elimination reward function is defined as the ratio of the risk value of a node to the estimated time to reach the node: Among them, i represents the i-th mission node, j represents the j-th unmanned boat, and interval i represents the risk value, which is defined as the square of the difference between the current time of the i-th task node and the last visit time. The estimated time is defined as the distance d between the j-th unmanned boat and the i-th task node. ij and the speed of the unmanned boat v j The ratio of ,P j is the set of mission points that the jth unmanned boat is responsible for.

6. The task allocation method for coordinated patrol of a variable number of unmanned boat swarms according to claim 4 is characterized in that: The proximity penalty function is defined as: Among them, max is the preset upper bound of the penalty score, k1 is a fixed constant, and d inear It refers to the distance from the i-th node to the nearest node that is different from the unmanned boat it is responsible for.

7. The task allocation method for coordinated patrol of a variable number of unmanned boat swarms according to claim 4 is characterized in that: The neighborhood load balancing function is defined as: Among them, N j represents the set of several unmanned boat neighbors closest to the jth executable unmanned boat, S j represents the average number of responsible nodes of each unmanned boat in the neighborhood of the unmanned boat, S j Represents the number of responsible nodes of the i-th unmanned boat.

8. The task allocation method for coordinated patrol of a variable number of unmanned boat swarms according to claim 1 is characterized in that: During patrol task allocation, for dynamic task nodes, the multi-unmanned boat system allocates tasks according to the following process at fixed intervals: Each unmanned boat abandons some nodes in its own responsible node set according to the value of the individual benefit function; Each unmanned boat samples from the risk nodes to be allocated, and uses the discrete particle swarm optimization algorithm to optimize the allocation pair of the node to be allocated and the unmanned boat in the neighborhood composed of multiple unmanned boats closest to itself; Each unmanned boat updates its own responsible node set and selects the nearest node from it. According to the adjacency information and the shortest path between nodes in the pre-stored patrol map, the unmanned boat moves and visits nodes with a greedy strategy to perform patrol tasks.

9. The task allocation method for coordinated patrol of a variable number of unmanned boat swarms according to claim 8 is characterized in that: The discrete particle swarm optimization algorithm optimizes the allocation of patrol nodes to be assigned through crossover and mutation operations. In the mutation phase, a random position is mutated instead of randomly selecting two positions for exchange. The position of the particle is represented by an m-dimensional vector X i =(x1,x2,...,x m ), then the position update formula of discrete particle swarm optimization algorithm is as follows: in, Represents the probability w of the original position X i (t) Perform mutation operation: randomly generate a random number r in the range [0,1]. If r is less than w, perform mutation operation, randomly generate an integer x in the range [1,m], and randomly change the element at position x to an integer in the range [1,L+1]. L+1 is the number of executable unmanned boats calculated in the current neighborhood; and Indicates that X is changed with probability c1 and c2 i (t) and individual extreme value position P i (t), the global extreme position G(t) performs a crossover operation: randomly generate a random number r in the range [0,1]. If r is less than c, a crossover operation is performed to randomly generate two integers ls and rs in the range [1,m]. i The value in the position range [ls, rs] in (t) is replaced by the individual extreme value position P i (t) or the value of the position corresponding to the global extreme value position G(t).

10. The task allocation method for coordinated patrol of a variable number of unmanned boat swarms according to claim 8 is characterized in that: In the patrol task allocation process, the allocation vector is used to encode the allocation scheme between the nodes to be allocated and the executable unmanned boats. For the allocation problem of X nodes to be allocated and Y executable unmanned boats at the current moment, an X×1 allocation vector is used to represent the solution of the problem. If the i-th node is allocated to the j-th unmanned boat, the value of the i-th position of the allocation vector is j.

Images