Unmanned aerial vehicle group multi-target attack task allocation method based on multiple rules

Abstract

The invention discloses an unmanned aerial vehicle group multi-target attack task allocation method based on multiple rules. Comprising the following steps: an unmanned aerial vehicle calculates an income value in real time according to a multi-target attack task based on a voyage cost, an attack threat cost and a target strategic value, adjusts the position of the unmanned aerial vehicle according to the income value obtained through real-time calculation, and determines a cluster task allocation scheme; and in the process that the unmanned aerial vehicle cluster approaches the target, the unmanned aerial vehicle judges whether to update a cluster task allocation scheme or reach a termination condition according to a profit value calculated in real time, and the leading unmanned aerial vehicle guides the pursuit unmanned aerial vehicle and the exploration unmanned aerial vehicle to approach the target and strike the target. On the premise of considering the profit value, the multi-suicide unmanned aerial vehicle attack target model with the maximum profit value as the target is constructed, the multi-rule unmanned aerial vehicle algorithm is provided, global exploration and local utilization are effectively balanced, the unmanned aerial vehicle search efficiency is improved, and the efficiency and accuracy of task allocation are ensured.

Description

Technical Field

[0001] The invention belongs to the technical field of unmanned aerial vehicle control, and in particular relates to a method for allocating multi-target strike tasks of an unmanned aerial vehicle fleet based on multiple rules. Background Art

[0002] In modern warfare, drones have become a key combat tool. With its advantages of flexibility, high efficiency and low cost, it has significantly improved the accuracy and maneuverability of military operations. Therefore, drone swarm collaborative operations have become a new trend. By sharing information and collaborative task execution, it has greatly improved combat effectiveness, can cover a wider battlefield space, and enhance the ability to reconnaissance, interference and strike enemy targets. Collaborative operations not only improve the success rate of missions, but also effectively disperse the enemy's defensive firepower and enhance combat survivability. This kind of drone swarm collaborative mechanism has laid a solid foundation for future intelligent warfare. Therefore, the research on drone swarm target strike missions is of great strategic value. However, in the complex scenarios of existing drone swarm collaborative multi-target task allocation, it still faces problems such as insufficient robustness and easy to fall into local optimality, making it difficult to cope with complex and changing battlefield environments. Summary of the invention

[0003] In view of the above-mentioned deficiencies in the prior art, the present invention invents a method for allocating multi-target strike tasks for a UAV fleet based on multiple rules.

[0004] In order to achieve the above object, the present invention adopts the following technical solution:

[0005] A multi-rule based UAV swarm multi-target strike task allocation method comprises the following steps:

[0006] S1: The UAV calculates the independent benefits of each multi-target strike mission in real time based on the range cost, attack threat cost and target strategic value, and calculates the real-time global benefits of the UAV in the entire mission allocation based on the independent benefits;

[0007] S2: adjusting the positions of the drones according to the independent benefits and global benefits obtained by real-time calculation and determining the swarm task allocation plan, wherein the swarm task allocation plan is that the drone with the highest benefit value is used as the leading drone, and the remaining drones are allocated to the siege drones and the exploration drones according to their benefit values;

[0008] S3: When the drone group approaches the target, the drone determines whether to update the group task allocation plan or reach the termination condition based on the real-time calculated benefit value. If the termination condition is reached, the target strike allocation plan is output. If the termination condition is not reached, the group task allocation plan is updated according to the benefit value.

[0009] S4: The leading UAV guides the siege UAV and exploration UAV to approach the target and strike the target according to the target strike allocation plan.

[0010] Furthermore, the voyage cost is

[0011]

[0012] Where i represents the drone set U = {U1,U2,…,U N}, j represents the set of enemy targets T = {T1, T2, …, T M} any enemy target, d ij represents the distance between UAV i and enemy target j, d min is the minimum flight distance of the drone swarm to perform each target task, d max The maximum range of the drone swarm to perform various target tasks.

[0013] Furthermore, the attack threat cost is

[0014]

[0015] Among them, C j For enemy target T j The attack threat coefficient brought by min is the minimum value of all target threats, C max The maximum value of all target threats.

[0016] Furthermore, the target strategic value is

[0017]

[0018] Among them, v j For enemy target T j The strategic value of min is the minimum value of all target strategic values, v max It is the maximum value of all target strategic values.

[0019] Furthermore, the benefit value is the sum of the independent benefit and the global benefit. The independent benefit is

[0020] E ij =ω1·NV j -ω2·NC j -ω3·D ij

[0021] Among them, ω1, ω2, and ω3 are the weight factors of value cost, attack cost, and range cost, respectively, and ω1+ω2+ω3=1.

[0022] The global benefit is

[0023]

[0024] All enemy targets are attacked by drones, which is represented by Any UAV i can attack at most one enemy target, which is expressed as x ij is the decision variable, that is, the target T j Assigned to drone i.

[0025] Furthermore, the updating rules of the swarm task allocation scheme are the wandering rule and the siege rule; the exploration drone updates its position according to the wandering rule, and the wandering rule is

[0026]

[0027] Among them, v i,d is the position of the ith exploration drone after a random walk in the dth dimension, x i,d The position of the i-th exploration drone in the d-th dimension, x k,d is the random position of other exploration drones, x best,d is the optimal solution in the local domain in the d dimension, ε represents a small positive number that controls the scope of the search domain, and λ is a probability factor that takes a random value according to the uniform distribution U[0,1];

[0028] The siege rule is that the leading drone summons the siege drones to quickly approach its position. During the approach process, the siege drones are dynamically adjusted according to the siege rule. The specific form is:

[0029]

[0030] Among them, step init is the initial siege step length, α d represents the position of the leading UAV in the dth dimension, δ represents the decay rate of the step length, t is the number of iterations, is the position of the drone in the dth dimension at the kth iteration, is the position of the UAV in the dth dimension at the k+1th iteration.

[0031] Furthermore, the termination condition is whether the difference between the ideal optimal benefit value and the current maximum benefit value of the drone is less than a preset threshold or reaches a maximum limit number of searches.

[0032] Furthermore, in the updating rules of the swarm task allocation scheme, a certain number of drones with lower profit values ​​are eliminated in each round, and the same number of drones are randomly generated for replenishment.

[0033] Furthermore, after the siege ends, the leading drone of the drone swarm autonomously roams around to further explore and search the surrounding area.

[0034] The advantages of the present invention are: the technical scheme of the present invention considers the range cost, attack threat cost and target strategic value, constructs a multi-suicide drone attack target model with the maximum benefit value as the goal, proposes a multi-rule drone algorithm, establishes the leader drone generation rule of the autonomous roaming mechanism, the multi-strategy exploratory drone roaming rule and the adaptive step-size siege drone siege rule, effectively balances the global exploration and local utilization, improves the drone search efficiency, and ensures the efficiency and accuracy of task allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flow chart of a method for allocating multi-target strike tasks of a UAV fleet based on multiple rules of the present invention;

[0036] Figure 2 This is a comparison chart of the errors of four different algorithms;

[0037] Figure 3 This is a comparison chart of the minimum fitness of four different algorithms under different scale clusters;

[0038] Figure 4 This is a comparison chart of the average fitness of four different algorithms under different scale clusters. DETAILED DESCRIPTION

[0039] like Figure 1 As shown, a multi-rule based UAV swarm multi-target strike task allocation method includes the following steps:

[0040] S1: The UAVs calculate their own independent benefits in real time based on the range cost, attack threat cost and target strategic value according to the multi-target strike mission, and calculate the real-time global benefits of the UAV in the entire mission allocation based on the independent benefits; during the UAV swarm search process, each UAV will update and calculate its own independent benefits and global benefits in real time, and compare them with the independent benefits and global benefits of other UAVs in the same communication network.

[0041] S2: According to the independent benefits and global benefits obtained by real-time calculation, the position of the drone is adjusted and the swarm task allocation plan is determined. The swarm task allocation plan is that the drone with the highest benefit value (independent benefit + global benefit) is the leading drone, and the remaining drones are divided into siege drones and exploration drones according to the benefit value; the swarm task allocation plan mainly determines the status of the drones in the swarm. The leading drone is used to guide the siege drones and exploration drones to approach the target. During the approach process, the status of the drones in the swarm is adjusted at any time according to their real-time independent benefits and global benefits; the siege drone introduces an adaptive step size mechanism to dynamically adjust the step size according to the current search status, so that the flexibility of fine search can be maintained during the siege process, and the precision strike capability against the target can be improved; the exploration drone is used to enhance the search capability of the entire group, expand the search space, improve the global optimization performance, and ensure that the potential global optimal solution is found in a larger range. The status of the siege drone and the exploration drone is determined by comparing the calculated benefit value with the benefit value calculated by the leading drone. For example, the calculated benefit value of the leading drone is 5.0, the benefits of 4.9-4.0 are siege drones, the benefits of 3.9-3.0 are exploration drones, and the remaining drones are considered to be non-attack drones.

[0042] S3: In the process of the drone swarm approaching the target, the drone determines whether to update the swarm task allocation plan or reach the termination condition based on the real-time calculated benefit value. If the termination condition is reached, the target attack allocation plan is output. If the termination condition is not reached, the swarm task allocation plan is updated according to the benefit value; the updating of the swarm task allocation plan is to change the position of the drone in the swarm at any time according to the benefit value calculated by the drone, until the optimal solution for the drone to approach the target is obtained or the number of iterations of the drone reaches the upper limit. The termination condition is whether the difference between the ideal optimal benefit value and the current maximum benefit value of the drone is less than the preset threshold or the maximum number of searches is reached. After the termination condition is reached, the swarm task allocation plan will no longer be updated until the leading drone is attacked and destroyed. In the update rule of the swarm task allocation plan, a certain number of drones with lower benefit values ​​are eliminated in each round, and the same number of drones are randomly generated to supplement, to ensure that the number of drones attacking the target will not decrease.

[0043] S4: The leading drone guides the siege drone and exploration drone to approach the target and attack the target according to the target attack allocation plan. After the target attack is completed, the leading drone will move around on its own to find a new target, that is, after the siege is over, the leading drone will move around autonomously to further explore and search the surrounding area.

[0044] The target attack allocation plan is that the leading drone guides the siege drones. If the leading drone is destroyed, the fleet will re-determine a new leading drone based on the benefit value and continue to attack the target until the target is destroyed.

[0045] The range cost is that each drone carries a certain power source and has a limited range. If it flies for a long time, the longer it is exposed to the enemy's field of vision, the higher the probability of being discovered. Therefore, the shorter the range, the higher the benefit to the drone. Therefore, the range cost is constructed as

[0046]

[0047] Where i represents the drone set U = {U1,U2,…,U N}, j represents the set of enemy targets T = {T1, T2, …, T M} any enemy target, d ij represents the distance between UAV i and enemy target j, d min is the minimum flight distance of the drone swarm to perform each target task, d max The maximum range of the drone swarm to perform various target tasks.

[0048] The attack threat cost is that when the drone attacks the enemy target, it may be attacked by different equipment carried by the enemy target, and this attack will have different degrees of impact on the benefits. So the attack threat cost is expressed as

[0049]

[0050] Among them, C j For enemy target T j The attack threat coefficient brought by min is the minimum value of all target threats, C max The maximum value of all target threats.

[0051] The target strategic value is the target's own strategic value, so the target strategic value is expressed as

[0052]

[0053] Among them, v j For enemy target T j The strategic value of j Normalized, v min is the minimum value of all target strategic values, v max It is the maximum value of all target strategic values.

[0054] According to the range cost, attack threat cost and target strategic value, the benefit E obtained by any UAV i after executing target mission j can be obtained. ij , the independent income is

[0055] E ij =ω1·NVj -ω2·NC j -ω3·D ij

[0056] Among them, ω1, ω2, and ω3 are the weight factors of value cost, attack cost, and range cost, respectively, and ω1+ω2+ω3=1.

[0057] Assuming that M = N, and each suicide drone can only attack one target, the drone swarm collaborative task allocation model can be described as maximizing the total benefit. The global benefit is

[0058]

[0059] All enemy targets are attacked by drones, which is represented by Any UAV i can attack at most one enemy target, which is expressed as x ij is the decision variable, that is, the target T j Assigned to drone i.

[0060] The benefit value is the sum of the independent benefit and the global benefit, and the position of the drone in the fleet is determined by the benefit value.

[0061] The update rules of the swarm task allocation scheme are the wandering rule and the siege rule. According to the rules, the target is optimized to achieve the optimal task allocation; the exploration drone updates its position according to the wandering rule. The wandering rule is

[0062]

[0063] Among them, v i,d is the position of the ith exploration drone after a random walk in the dth dimension, x i,d The position of the i-th exploration drone in the d-th dimension, x k,d is the random position of other exploration drones, x best,d is the optimal solution in the local domain in the d dimension, ε represents a very small positive number that controls the scope of the search domain, and λ is a probability factor that randomly takes values ​​according to the uniform distribution U[0,1].

[0064] when The rule is based on the neighbor position of the exploration drone's wandering rule. This rule retains the basic search capability of the original swarm algorithm, and guides the wandering of the exploration drone by using the position of other drones, which improves the algorithm's exploration capability and effectively enhances the algorithm's initial exploration capability, which helps to find better local solutions in the solution space.

[0065] when When the exploration drone is guided by the local optimal information, it allows the exploration drone to more fully explore the current area, which helps to conduct a more detailed search near the known optimal solution.

[0066] when When , the exploration drone’s wandering rule is based on neighborhood search, which allows the exploration drone to conduct a fine search near the current solution, which helps to further explore the potential good solutions that have been found, thereby improving the accuracy of the solution and the local optimization capability.

[0067] In the walking rule, the current profit value of the exploration drone is calculated. If it is greater than the leading drone’s profit value, it becomes the new leading drone; otherwise, it walks according to the multi-strategy walking rule, that is, the element position is exchanged until the profit value is greater than the leading drone’s profit value or the number of walks reaches the upper limit. Therefore, the multi-strategy walking rule of the exploration drone can enhance the algorithm’s search ability and global optimization performance.

[0068] The siege rule is that the leading UAV summons the siege UAV to quickly approach its position. During the approach process, the siege UAV is dynamically adjusted according to the siege rule. That is, by introducing an adaptive step length mechanism, when the siege UAV runs towards the leading UAV, the step length will gradually decrease as the iteration proceeds, or dynamically adjust according to the distance between the siege UAV and the leading UAV. The specific form is

[0069]

[0070] Among them, step init is the initial siege step length, α d represents the position of the leading UAV in the dth dimension, δ represents the decay rate of the step length, t is the number of iterations, is the position of the drone in the dth dimension at the kth iteration, is the position of the UAV in the dth dimension at the k+1th iteration.

[0071] In the siege rule, the leading drone summons the siege drones to quickly approach its position. The siege drones make adaptive adjustments according to the siege rule. If the benefit value is greater than the current leading drone benefit value, the siege drone becomes the new leading drone; otherwise, the siege continues until the prey is captured. The adaptive step size (siege rule) can not only improve the convergence speed of the algorithm, but also effectively avoid the situation where the step size is too large and the convergence is difficult.

[0072] In order to verify the feasibility of the allocation method of the present invention (referred to as MR-WPA), a comparative analysis is performed with the genetic algorithm (GA), particle swarm optimization (PSO) and wolf pack algorithm (WPA). The difference between the benefit value obtained by each algorithm and the optimal benefit value obtained based on linear programming can be obtained to obtain the error change diagram of different algorithms under different iteration times, as shown in Figure 2. Figure 2 As shown in the figure, it can be seen intuitively that the MR-WPA algorithm has great advantages in both running speed and accuracy. The improved MR-WPA algorithm can make the error with the optimal return value less than 0.1 at a faster speed, and under the same iteration rounds, the final error is smaller than that of the other three algorithms.

[0073] For small-scale cluster problems, due to the small solution space, the four algorithms have similar performance and similar benefits. However, as the cluster size increases, the performance differences of different algorithms also gradually increase, especially the gap between MR-WPA algorithm and WPA algorithm and genetic algorithm and particle swarm algorithm. Figure 3 and Figure 4 shown.

[0074] Although the specific embodiments of the present invention have been described and illustrated in detail above, it should be noted that various changes and modifications may be made to the above embodiments without departing from the spirit of the present invention and the scope of the appended claims.

Claims

1. A multi-rule based UAV swarm multi-target strike task allocation method, characterized in that: The steps include: S1: The UAV calculates the independent benefits of each multi-target strike mission in real time based on the range cost, attack threat cost and target strategic value, and calculates the real-time global benefits of the UAV in the entire mission allocation based on the independent benefits; S2: adjusting the position of the drones according to the independent benefits and global benefits obtained by real-time calculation and determining the swarm task allocation plan, wherein the swarm task allocation plan is that the drone with the highest benefit value is used as the leading drone, and the remaining drones are divided into siege drones and exploration drones according to their benefit values; S3: When the drone group approaches the target, the drone determines whether to update the group task allocation plan or reach the termination condition based on the real-time calculated benefit value. If the termination condition is reached, the target strike allocation plan is output. If the termination condition is not reached, the group task allocation plan is updated according to the benefit value. S4: The leading UAV guides the siege UAV and exploration UAV to approach the target and strike the target according to the target strike allocation plan.

2. The multi-rule-based UAV swarm multi-target strike task allocation method according to claim 1 is characterized in that: The voyage cost is Where i represents the drone set U = {U1,U2,…,U N }, j represents the set of enemy targets T = {T1, T2, …, T M } any enemy target, d ij represents the distance between UAV i and enemy target j, d min is the minimum flight distance of the drone swarm to perform each target task, d max The maximum range of the drone swarm to perform various target tasks.

3. The multi-rule-based UAV swarm multi-target strike task allocation method according to claim 2 is characterized in that: The attack threat cost is Among them, C j For enemy target T j The attack threat coefficient brought by min is the minimum value of all target threats, C max The maximum value of all target threats.

4. The multi-rule-based UAV swarm multi-target strike task allocation method according to claim 3 is characterized in that: The target strategic value is Among them, v j For enemy target T j The strategic value of min is the minimum value of all target strategic values, v max It is the maximum value of all target strategic values.

5. The multi-rule-based UAV swarm multi-target strike task allocation method according to claim 4 is characterized in that: The benefit value is the sum of the independent benefit and the global benefit. The independent benefit is E ij =ω1·NV j -ω2·NC j -ω3·D ij Among them, ω1, ω2, and ω3 are the weight factors of value cost, attack cost, and range cost, respectively, and ω1+ω2+ω3=1. The global benefit is All enemy targets are attacked by drones, which is represented by Any UAV i can attack at most one enemy target, which is expressed as x ij is the decision variable, that is, the target T j Assigned to drone i.

6. The multi-rule-based UAV swarm multi-target strike task allocation method according to claim 5 is characterized in that: The updating rules of the cluster task allocation scheme are wandering rules and siege rules; The exploration drone updates its position according to the wandering rule. The wandering rule is: Among them, v i,d is the position of the ith exploration drone after a random walk in the dth dimension, x i,d The position of the i-th exploration drone in the d-th dimension, x k,d is the random position of other exploration drones, x best,d is the optimal solution in the local domain in the d dimension, ε represents a small positive number that controls the scope of the search domain, and λ is a probability factor that takes a random value according to the uniform distribution U[0,1]; The siege rule is that the leading drone summons the siege drones to quickly approach its position. During the approach process, the siege drones are dynamically adjusted according to the siege rule. The specific form is: Among them, step init is the initial siege step length, α d represents the position of the leading UAV in the dth dimension, δ represents the decay rate of the step length, t is the number of iterations, is the position of the drone in the dth dimension at the kth iteration, is the position of the UAV in the dth dimension at the k+1th iteration.

7. The multi-rule-based UAV swarm multi-target strike task allocation method according to claim 6 is characterized in that: The termination condition is whether the difference between the ideal optimal benefit value and the current maximum benefit value of the drone is less than a preset threshold or reaches the maximum number of searches.

8. The multi-rule-based UAV swarm multi-target strike task allocation method according to claim 7 is characterized in that: In the updating rule of the swarm task allocation scheme, a certain number of drones with lower profit values ​​are eliminated in each round, and the same number of drones are randomly generated for replenishment.

9. The multi-rule-based UAV swarm multi-target strike task allocation method according to claim 8 is characterized in that: After the siege ends, the leading drone of the drone swarm autonomously roams around to further explore and search the surrounding area.

Images