CN114740872B - A UUV swarm search and attack decision-making method based on topology and alliance - Google Patents

Abstract

The invention relates to a UUV cluster search attack decision method based on topology and alliance, belonging to the technical field of search attack decision. According to the method, UUV in the cluster are regarded as agents capable of making independent decisions, targets are searched in a formation mode based on a communication topological structure and a alliance income function, and then alliance construction is carried out through information interaction with other UUV, so that tasks for attacking the targets are jointly executed. According to the method, the communication topology of the UUV cluster is considered, the optimality of individual performance and group performance is considered, and intelligent real-time task decision can be realized under a distributed architecture.

Description

A UUV swarm search and attack decision-making method based on topology and alliance

Technical Field

The present invention relates to search and attack decision-making technology, which specifically adopts mechanisms such as cluster intelligence, alliance mechanism, and task decision-making. It acts in the field of distributed decision-making and studies and solves problems such as autonomy and intelligence of UUV cluster task decision-making.

Background technique

Traditional UUV mission decision-making is mainly based on a single individual. However, with the expansion of the scale of UUV clusters and the increasing demand for collaboration in complex tasks, traditional methods have certain limitations for UUV cluster decision-making. First, there is a lack of effective cluster collaborative decision-making model. Second, the solution algorithm is not efficient when the number of UUVs and targets expands.

In order to overcome the non-cooperation and weak scalability of traditional methods, multi-agent technology has begun to be applied to the field of task decision-making. Multi-agent systems are composed of multiple autonomous and interacting agents, which have homogeneous or heterogeneous task objectives and sensor information. Through simple information interaction, agents can perceive the environment and make decisions, select appropriate agents to perform tasks, and thus achieve intelligent decision-making. As a basic collaborative method in multi-agent systems, alliance formation can accurately and completely describe the interactive relationship between agents in the cluster network, combine multiple agents to complete tasks together according to the formation rules, maximize the performance of cluster execution tasks in a shorter time, and thus improve the autonomous decision-making ability of UUV clusters.

Summary of the invention

Technical issues to be solved

In order to avoid the shortcomings of the prior art, the present invention proposes a UUV cluster search attack decision method based on topological structure and alliance mechanism.

Technical solutions

A UUV cluster search attack decision method based on topology and alliance is characterized by the following steps:

Step 1: Initialize the mission scenario and set related parameters, including setting the UUV cluster initial parameters, setting the target initial parameters, setting the sampling time interval and mission decision matrix, and setting the UUV cluster communication topology;

Step 2: The UUV cluster starts to sail along the preset route from the initial moment, and turns on the sonar to search for targets. The relative distance between the i-th UUV and the j-th target at the current moment is recorded as d _ij . When d _ij ＜r _search , it means that the UUV has found the target within the search range, and the target's position L _j , value Value _j , and resource R _j can be obtained. If the UUV does not find the target, it continues to sail in the current direction.

Step 3: The first UUV that discovers the target acts as the coalition initiator and broadcasts the target information and the resources required for the attack mission to its neighboring UUVs in the form of a message, requests the formation of a coalition, and switches from the search state to the waiting state; the coalition requested to be formed when UUVi discovers target j is recorded as The alliance formation request message contains the target's serial number j, location L _j , and the resources R _j required to attack the target;

Step 4: Based on the communication topology of the UUV cluster, the neighbor UUV that can communicate with the alliance initiator acts as the alliance responder. After receiving the alliance request information, it calculates the local benefit value that may be generated by executing the attack task, and sends a response alliance message to the alliance initiator. At the same time, it switches from the search state to the waiting state, waiting for the next request from the alliance initiator; the UUV's response alliance message contains its own resources and the local benefits that may be generated by the attack target;

The local benefit function of UUV includes the task reward function and the task cost function. Taking the i-th UUV performing the attack task on the j-th target as an example, the task reward function includes the reward obtained by performing the task and the cost incurred. The task reward calculation is shown in formula (2):

f _benefit (ij) = Value _j · P _{th_U} (ij) · a _ij (1)

The task cost calculation is shown in formula (3):

f _cost (ij) = Value _i · P _{th_T} (ji) · a _ij (2)

The task reward function is a weighted operation of the task reward and the task cost, and the calculation is shown in formula (4):

Among them, w ₁ and w ₂ represent weighting coefficients, and their values are related to the decision maker’s preferences;

The mission cost function is the time consumed by the UUV when performing the attack mission, which is related to the Dubins distance between the UUV and the target at the current moment and the movement speed of the UUV. The calculation is shown in formula (5):

The local benefit value generated by the UUV when performing an attack mission is calculated as shown in formula (6):

The sum of the local benefits of all UUVs in the same coalition C is the global benefit when performing the attack mission on the jth target at the current moment, expressed as: v(j) = ∑ _i∈C v(ij);

Step 5: The alliance initiator arranges the local benefit values of all alliance responders in order according to the alliance response information, selects the UUVs with the highest local benefit and sufficient task resources to form an alliance, and sends the information of successful formation to these UUVs; at this time, the number of UUV individuals included in the alliance should be the least, and the resources should not be less than the attack task requirements of the corresponding target, that is, Moreover, the global benefit of executing the attack task on target T _j should satisfy/>

Step 6: After receiving the information that the alliance is successfully formed, the alliance responder abandons the current task, adjusts the heading angle, and sails towards the target together, and performs the attack task when the distance from the target is r _attack ; after the task is completed, the target disappears, and the status of the corresponding task changes from the assigned state to the completed state; the UUV disbands the alliance, changes from the attack state to the search state, continues to sail according to the current heading angle, and searches for the target again; the UUV that has not received the formation information changes from the waiting state to the search state, turns to step 2, and continues to search for the target along the current heading;

Step 7: If all targets have been searched and attacked within the sampling time, the mission is completed and the intelligent allocation decision ends; otherwise, go to step 2, the UUV continues to search for unattacked targets in the scene, re-forms the alliance and executes the attack mission.

A further technical solution of the present invention: the initial parameters of the UUV cluster are set as described in step 1, specifically as follows:

There are M UUVs in the mission scenario, represented by the set U = {U ₁ ,U ₂ ,L, _UM }; the position of the i-th UUV is recorded as _Li , the heading angle is recorded as _Hi , the speed is recorded as _Vi , i∈U, and the resources carried are recorded as _Ri . The total resources of the cluster are The effective target search range of UUV is denoted as r _search , the attack range is denoted as r _attack , and the minimum turning radius is denoted as r _turn ; UUV has different attack capabilities on targets, and its value is expressed as Value _i , i∈U , and the threat probability matrix is expressed as P _{th_U} ＝[p _ij ] ^M×N , 0≤p _ij ≤1, where p _ij represents the threat probability of target j when UUVi performs the attack mission.

A further technical solution of the present invention: the setting of the target initial parameters described in step 1 is as follows:

There are N targets in the mission scenario, represented by the set T = {T ₁ ,T ₂ ,L, _TN }; the position of the jth target is recorded as _Lj , and the resources required to execute the attack mission on it are recorded as _Rj . The total resources required for all targets are The value of the target is expressed as Value _j , j∈T, and the threat probability matrix is expressed as _{Pth_T} = [ _pji ] ^N×M , _0≤pji≤1 , where _pji represents the threat probability when UUVi attacks target j.

A further technical solution of the present invention: the setting of the sampling time interval and the task decision matrix described in step 1 is as follows:

UUV only performs one attack task on each target at the sampling time, so the cluster's task allocation decision matrix is expressed as A = [a _ij ] ^{M × N} . When UUVi performs an attack task on target j, a _ij = 1, otherwise a _ij = 0. The allocation decision variable should satisfy the constraint condition of collaborative execution of attack tasks, that is, at each attack decision moment, a UUV is assigned to at most one target, and a target can be assigned to multiple UUVs at the same time, which is specifically described as formula (1):

A further technical solution of the present invention: the UUV cluster communication topology structure set in step 1 is as follows:

The adjacency matrix of the topological graph is represented by G(τ) = [ _gik (τ)] ^M×M. When the i-th UUV and the k-th UUV can communicate at time τ, _gik (τ) = 1, otherwise _gik (τ) = 0, and the two connected UUVs are called neighbors. Since each UUV can connect to itself, _gii (τ) = 1.

Beneficial Effects

The present invention is based on a distributed decision-making architecture for autonomous task allocation of UUV clusters, and provides an effective and reliable decision-making method for multiple UUVs to collaboratively perform complex tasks. The present invention can be used to solve group decision-making problems when UUV clusters perform various complex tasks such as reconnaissance and surveillance, environmental surveys, target tracking, and coordinated attacks, and provides an efficient and intelligent decision-making technology solution for the development of UUV clusters. Compared with the prior art, the present invention has the following advantages:

1. The communication network topology of the UUV cluster is considered, which is suitable for the situation where the communication of some UUV nodes is limited.

2. The two factors of task reward and task cost are considered as the benefit function, while taking into account both local and global benefits, and an effective cluster collaborative decision-making model is established.

3. The multi-agent alliance theory is introduced for task decision-making, which can reduce the number of UUV platforms used while ensuring cluster benefits, with high real-time performance and good distribution.

BRIEF DESCRIPTION OF THE DRAWINGS

The drawings are only for the purpose of illustrating specific embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

FIG1 is a schematic diagram of a UUV cluster search attack decision process of the present invention;

FIG2 is a schematic diagram of the conversion relationship of UUV interaction information types according to the present invention;

FIG3 is a schematic diagram of the state transition relationship between the UUV and the task of the present invention;

FIG4 is a diagram of the communication topology structure of full connection and non-full connection during the simulation process of the present invention: (a) full connection topology; (b) non-full connection topology (UUV3 and 4 are not connected);

FIG5 is a diagram of the UUV cluster decision process under the fully connected topology structure during the simulation process of the present invention: (a) t=64; (b) t=107; (c) t=126; (d) t=167;

FIG6 is a cluster decision process in the simulation process of the present invention when UUV3 and 4 are not connected: (a) t=64; (b) t=107; (c) t=157; (d) t=167;

FIG7 is a list of UUV mission times under a fully connected topology during the simulation process of the present invention;

FIG8 is a list of the mission times of each UUV when UUV3 and 4 are not connected during the simulation process of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in each embodiment of the present invention described below can be combined with each other as long as they do not conflict with each other.

The present invention designs a UUV cluster search and attack decision method based on topological structure and alliance mechanism. The UUV cluster is composed of a group of UUVs carrying different resources and having different task execution capabilities. For the stationary targets distributed in the scene, the UUVs first search in formation. After searching for the target, according to their own capabilities and alliance formation requirements, they establish an alliance benefit function including task rewards and task costs, and exchange information with other UUVs based on the communication topological structure, and finally generate the optimal alliance to attack the target in a coordinated manner.

In order to realize the coordinated execution of search and attack tasks by UUV clusters, the topology and alliance-based UUV cluster search and attack decision method of the present invention, as shown in FIG1 , mainly includes the following steps:

Step 1: Initialize the task scenario and set related parameters.

Set initial parameters for the UUV cluster. There are M UUVs in the mission scenario, represented by the set U = {U ₁ ,U ₂ ,L,U _M }. The position of the i-th UUV is recorded as _Li , the heading angle is recorded as _Hi , the speed is recorded as _Vi (i∈U), and the resources carried are recorded as _Ri . The total resources of the cluster are The effective target search range of UUV is recorded as r _search , the attack range is recorded as r _attack , and the minimum turning radius is recorded as r _turn . UUV has different attack capabilities on targets, and its value is expressed as Value _i (i∈U), and the threat probability matrix is expressed as P _{th_U} ＝[p _ij ] ^M×N (0≤p _ij ≤1), where p _ij represents the threat probability of target j when UUVi performs the attack mission.

Set initial parameters for the target. There are N targets in the mission scenario, represented by the set T = {T ₁ ,T ₂ ,L,T _N }. The position of the jth target is recorded as L _j , and the resources required to perform the attack mission on it are recorded as R _j . The total resources required for all targets are The value of the target is expressed as Value _j (j∈T), and the threat probability matrix is expressed as P _{th_T} = [p _ji ] ^N×M (0≤p _ji ≤1), where p _ji represents the threat probability when UUVi attacks target j.

Set the sampling time interval and task decision matrix. UUV only performs one attack task on each target at the sampling time, then the cluster's task allocation decision matrix is expressed as A = [a _ij ] ^{M × N} , when UUVi performs an attack task on target j, a _ij = 1, otherwise a _ij = 0. The allocation decision variable should satisfy the constraints of collaborative execution of attack tasks, that is, at each attack decision moment, a UUV is assigned at most one target, and a target can be assigned multiple UUVs at the same time, specifically described as formula (1):

Set the UUV cluster communication topology. Use G(τ) = [g _ik (τ)] ^{M × M} to represent the adjacency matrix of the topology graph. When the ith UUV and the kth UUV can communicate at time τ, g _ik (τ) = 1, otherwise g _ik (τ) = 0, and the two connected UUVs are called neighbors. Since each UUV can connect to itself, g _ii (τ) = 1.

Step 2: The UUV cluster starts to sail along the preset route from the initial moment, and turns on the sonar to search for targets. The relative distance between the i-th UUV and the j-th target at the current moment is recorded as d _ij . When d _ij ＜r _search , it means that the UUV has found the target within the search range, and the target's location L _j , value Value _j , resource R _j and other information can be obtained; if the UUV does not find the target, it continues to sail in the current direction.

Step 3: The first UUV that discovers the target acts as the coalition initiator and broadcasts the target information and the resources required for the attack mission to its neighboring UUVs in the form of a message, requests the formation of a coalition, and switches from the search state to the waiting state. The coalition requested to be formed when UUVi discovers target j is denoted as The alliance formation request message includes the target's serial number j, location L _j , and the resources R _j required to attack the target.

Step 4: Based on the communication topology of the UUV cluster, the neighboring UUV that can communicate with the alliance initiator acts as the alliance responder. After receiving the alliance request information, it calculates the local benefit value that may be generated by executing the attack task, and sends a response alliance message to the alliance initiator. At the same time, it switches from the search state to the waiting state, waiting for the next request from the alliance initiator. The UUV's response alliance message contains its own resources and the local benefits that may be generated by the attack target.

The local benefit function of UUV includes the task reward function and the task cost function. Taking the i-th UUV performing the attack task on the j-th target as an example, the task reward function includes the reward obtained by performing the task and the cost incurred. The task reward calculation is shown in formula (2):

f _benefit (ij) = Value _j · P _{th_U} (ij) · a _ij (8)

The task cost calculation is shown in formula (3):

f _cost (ij) = Value _i · P _{th_T} (ji) · a _ij (9)

The task reward function is a weighted operation of the task reward and the task cost, and the calculation is shown in formula (4):

Among them, _w1 and _w2 represent weighting coefficients, and their values are related to the decision maker's preferences.

The mission cost function is the time consumed by the UUV when performing the attack mission, which is related to the Dubins distance between the UUV and the target at the current moment and the movement speed of the UUV. The calculation is shown in formula (5):

The local benefit value generated by the UUV when performing an attack mission is calculated as shown in formula (6):

The sum of the local benefits of all UUVs in the same coalition C is the global benefit when performing the attack mission on the jth target at the current moment, expressed as: v(j) = ∑ _i∈C v(ij).

Step 5: The alliance initiator arranges the local benefit values of all alliance responders in order according to the alliance response information, selects the UUVs with the highest local benefit and sufficient mission resources to form an alliance, and sends a successful formation message to these UUVs. At this time, the number of UUV individuals included in the alliance should be the least, and the resources should not be less than the attack mission requirements of the corresponding target, that is, Moreover, the global benefit of executing the attack task on target T _j should satisfy/>

Step 6: After receiving the information that the alliance is successfully formed, the alliance responder abandons the current task, adjusts the heading angle, and sails towards the target together, and performs the attack task when the distance from the target is r _attack . After the task is completed, the target disappears, and the status of the corresponding task changes from the assigned state to the completed state; the UUV disbands the alliance, changes from the attack state to the search state, continues to sail according to the current heading angle, and searches for the target again. The UUV that has not received the formation information changes from the waiting state to the search state, turns to step 2, and continues to search for the target along the current heading.

In the alliance formation process of steps 3 to 6, the interactive information conversion relationship between UUVs is shown in Figure 2. The interaction types include requesting alliance, responding to alliance, and forming alliance. Information interaction is mainly carried out in the form of messages. The state conversion relationship between UUV and task is shown in Figure 3, where the waiting state of UUV includes waiting when requesting alliance and waiting when responding to alliance.

Step 7: If all targets have been searched and attacked within the sampling time, the mission is completed and the intelligent allocation decision ends; otherwise, go to step 2, the UUV continues to search for unattacked targets in the scene, re-forms the alliance and executes the attack mission.

In order to further illustrate the technical means adopted by the present invention to achieve the predetermined purpose, the UUV cluster search attack decision under two conditions of fully connected topology and non-fully connected topology is simulated, and the detailed description is as follows with reference to the accompanying drawings.

First, a 100m×100m scenario is set up, where the specific task is for four UUVs to search and attack two stationary targets in turn. Parameters such as position, speed, heading angle, resources, value, and threat probability matrix are preset for each UUV, so that it can sail in a straight formation from the initial moment. Two stationary targets are set up in the scenario, and parameters such as resources, value, and threat probability matrix are preset for them.

Assume that the initial speed of the UUV is 1m/s, the heading angle is π/3rad, the search range is 5m, the attack range is 2m, and the minimum turning radius is 4m. The position, value, and resource settings of the UUV are shown in Table 1.

Table 1

The threat probability matrix setting of UUV is shown in Table 2.

Table 2

The location, value, and resource settings of the target are shown in Table 3.

table 3

The threat probability matrix setting of the target is shown in Table 4.

Table 4

Secondly, the sampling time interval is set to 0.5s, and the sampling number is represented by an integer sequence {0, 1, 2, L}. Then, two communication topologies, fully connected and non-fully connected, are set for the UUV cluster, as shown in Figure 4. The non-fully connected topology is analyzed by taking UUV3 and 4 as an example. Its adjacency matrix is expressed as

The method proposed in the present invention is used to search and attack the UUV cluster with a fully connected topology, and the decision-making process is shown in Figure 5. When t=0, the four UUVs are distributed in a straight line in the scene, sailing along the preset route according to the heading angle, and turning on the sonar to search for the target. As shown in Figure 5(a), when t=64, that is, the actual sampling time is 32s, UUV3 finds target 1 within its search range, and begins to broadcast information containing target resources 80 and target location (50,50), and sends a message requesting the establishment of an alliance. Due to the fully connected topology structure, all UUVs can receive the alliance request and send back their own resources and local benefits. The resources and local benefits corresponding to the four UUVs are (75,0.065), (25,14.986), (60,0.005), and (80,21.048) respectively. According to the needs of both the alliance initiator and the alliance responder, UUV3 selects UUV4 with the highest benefit and satisfactory task resources to independently perform the attack task. The alliance it forms is The global benefit corresponding to the alliance is 21.048, and the UUVs that are not involved in the alliance continue to sail in the current direction.

As shown in Figure 5(b), at t=107, UUV3 searches for target 2 again and initiates a request to form an alliance. Since UUV4 is still performing the attack mission, UUV1, 2, and 3 respond to the request and send messages of (75, 4.040), (25, 5.700), and (60, 8.694) respectively. After comparing all local benefits, UUV3 obtains the highest benefit from performing the attack mission and meets the task resource requirements, so the alliance is formed. The corresponding global benefit is 8.694. As shown in Figure 5(c), when t=126, UUV4 completes the task of target 1 and searches for the target again along the current direction. As shown in Figure 5(d), when t=167, UUV3 completes the task of target 2. At this time, all targets in the mission scenario have been attacked, and the cluster search attack decision process ends.

The method proposed in the present invention is used to search and attack the UUV cluster with non-fully connected topology, and the decision-making process is shown in Figure 6. In the time period [0,64), the navigation process of the UUV cluster is consistent with the situation of the fully connected topology. As shown in Figure 6(a), when t=64, UUV3 searches for target 1 and sends a request to other UUVs to form an alliance. Since UUV3 and 4 are not connected, UUV1, 2 and 3 respond to the request, and the message information sent is (75, 0.065), (25, 14.986), and (60, 0.005) respectively. According to the principle of maximizing local and global benefits, and based on the task resource requirements, an alliance is formed. The global benefit of executing the attack task is the sum of the local benefits of UUV1 and 2, which is 15.051. As shown in Figure 6(b), when t=107, UUV3 forms an alliance against target 2/> The obtained task benefit is 8.694. As shown in Figures 6(c) and 6(d), at t=157 and t=167, target 1 and target 2 are attacked successively, and all tasks are completed.

Figures 7 and 8 are the execution schedules of each UUV in the case of fully connected topology and non-fully connected topology, respectively. Corresponding to the decision-making process in Figure 5, in Figure 7, UUV1 and 2 experienced two states: searching for targets and responding to alliances, and did not participate in the alliance to perform the attack task. UUV3 searched for the target and requested an alliance at t=64 and t=107, respectively, and attacked target 2 at t=109 until the task was completed at t=167. UUV4 responded to UUV3's request to form a team at t=65, and attacked target 1 at t=66 until the task was completed at t=126, and then entered the search state until all tasks were completed. The schedule in Figure 8 corresponds to the decision-making process in Figure 6. UUV4 could not communicate with UUV3, so it did not participate in the alliance formation and searched for targets along the preset route within the sampling time. UUV1 and 2 formed an alliance at t=65 and jointly performed the task of attacking target 1. UUV3 experienced four states: searching for targets, requesting alliances, responding to alliances, and attacking targets, and completed the task of attacking target 2 at t=167.

The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should be included in the protection scope of the present invention.

Claims (5)

1. A UUV cluster search attack decision method based on topology and alliance is characterized by comprising the following steps:

step 1: initializing a task scene and setting related parameters, wherein the method comprises the steps of setting UUV cluster initial parameters, setting target initial parameters, setting sampling time intervals and task decision matrixes, and setting UUV cluster communication topological structures;

Step 2: starting the UUV cluster to navigate along a preset route from the initial moment, and starting a sonar to search a target; the relative distance between the ith UUV and the jth target at the current moment is recorded as d _ij, when d _ij＜r_search, the UUV finds the target in the searching range, and then the position L _j, the Value _j and the resource R _j of the target can be obtained; if the UUV does not find the target, continuing to navigate according to the current direction;

Step 3: the first UUV for finding the target is used as a alliance initiator, target information and resources required by the attack task are broadcast to neighbor UVs in the form of messages, alliance construction is requested, and the state is converted into a waiting state from a searching state; the federation that is requested to be built when UUVi finds target j is noted as The request building message of the alliance comprises a sequence number j and a position L _j of a target, and a resource R _j required for attacking the target;

Step 4: based on the communication topology of the UUV cluster, a neighbor UUV which can communicate with the alliance initiator is used as an alliance responder, after the alliance request information is received, a local profit value possibly generated by executing an attack task by the neighbor UUV is calculated, a response alliance message is sent to the alliance initiator, and meanwhile, the search state is converted into a waiting state, and the next request of the alliance initiator is waited; the response alliance message of the UUV contains the self resources and local benefits possibly generated by an attack target;

The UUV local benefit function comprises a task rewards function and a task cost function; taking an example of an attack task executed by an ith UUV on a jth target, a task reward function comprises rewards obtained by executing the task and generated cost; the task rewards calculation is shown in the formula (2):

f_benefit(ij)＝Value_j•P_{th_U}(ij)•a_ij (1)

The task cost calculation is shown in the formula (3):

f_cost(ij)＝Value_i·P_{th_T}(ji)·a_ij (2)

The task reward function is a weighted operation of task rewards and task cost, and the calculation is shown in a formula (4):

Wherein w ₁ and w ₂ represent weighting coefficients, the values of which are related to the decision maker's preference;

The task cost function is the time consumed when the UUV executes the attack task, and is related to dubins distance between the UUV and the target at the current moment and the movement speed of the UUV, and the calculation is shown in a formula (5):

the calculation of the local profit value generated by the UUV when the UUV executes the attack task is shown as a formula (6):

The sum of the local benefits of all UUV in the same alliance C is the global benefit when the attack task is executed on the jth target at the current moment, and is expressed as follows: v (j) = Σ _i∈C v (ij);

Step 5: the alliance initiator arranges local income values of all alliance responders in sequence according to the alliance response information, selects UUV with highest local income and meeting task resources to construct an alliance, and sends information of successful construction to the UUV; at this time, the number of UUV individuals included in the alliance should be minimum, and the resources should not be smaller than the resources required by the attack task of the corresponding target, namely And the global benefit generated by executing the attack task on the target T _j should meet/>

Step 6: after receiving information of successful alliance organization, the alliance respondents give up the current task, adjust the course angle, navigate towards the target direction together, and execute the attack task when the distance from the target r _attack; after the task execution is finished, the target disappears, and the state of the corresponding task is converted from the allocation state to the completion state; the UUV breaks up the alliance, is converted into a search state from the attack state, continues to navigate according to the heading angle at the current moment, and searches the target again; the UUV which does not receive the construction information is converted into a searching state from a waiting state, and the method goes to the step 2 to search the target along the current course continuously;

Step 7: in the sampling time, if all targets are searched and attacked, completing task execution, and ending intelligent allocation decision; otherwise turning to step 2, the UUV continues to search for the target which is not attacked in the scene, and the alliance building is performed again and the attack task is executed.

2. The topology and federation-based UUV cluster search attack decision method of claim 1, wherein: setting UUV cluster initial parameters in the step 1, specifically as follows:

M UUV are in the task scene and are represented by a set U= { U ₁,U₂,L,U_M }; the position of the ith UUV is marked as L _i, the course angle is marked as H _i, the speed is marked as V _i, the i epsilon U, the carried resource is marked as R _i, and the total resource of the cluster is The effective target search range of UUV is marked as r _search, the attack range is marked as r _attack, and the minimum turning radius is marked as r _turn; UUV has different attack capacities on targets, the Value of UUV is expressed as Value _i, i epsilon U, the threat probability matrix is expressed as P _{th_U}＝[p_ij]^M×N,0≤p_ij is less than or equal to 1, and P _ij represents the threat probability suffered by the target j when UUVi executes an attack task.

3. The topology and federation-based UUV cluster search attack decision method of claim 1, wherein: the setting target initial parameters in the step 1 are specifically as follows:

N targets exist in the task scene, and are represented by a set T= { T ₁,T₂,L,T_N }; the j-th target is marked as L _j, the resource needed by executing the attack task is marked as R _j, and the total resource needed by all targets is The Value of the target is expressed as Value _j, j epsilon T, and the threat probability matrix is expressed as P _{th_T}＝[p_ji]^N×M,0≤p_ji < 1, wherein P _ji represents the threat probability suffered by UUVi when the target j is attacked.

4. The topology and federation-based UUV cluster search attack decision method of claim 1, wherein: the setting of the sampling time interval and the task decision matrix in the step 1 is specifically as follows:

The UUV only executes an attack task once for each target at the sampling moment, the task allocation decision matrix of the cluster is expressed as A= [ a _ij]^M×N ], when UUVi executes the attack task for the target j, a _ij =1, otherwise a _ij =0; the allocation decision variable should satisfy the constraint condition of cooperatively executing the attack task, that is, at most one target is allocated to one UUV at each attack decision moment, and one target may simultaneously allocate a plurality of UUVs, specifically described as formula (1):

5. The topology and federation-based UUV cluster search attack decision method of claim 1, wherein: the UUV cluster communication topology structure is set in the step 1, which is specifically as follows:

Representing an adjacency matrix of the topological graph by G (τ) = [ G _ik(τ)]^M×M ], when the ith UUV and the kth UUV can communicate at the τ moment, G _ik (τ) =1, otherwise G _ik (τ) =0, and the two connected UUVs are mutually called neighbors; since each UUV can be self-connecting, g _ii (τ) =1.