Manned aerial vehicle / unmanned aerial vehicle cooperative task allocation method based on insect positive and negative feedback mechanism

Abstract

The invention discloses a manned aerial vehicle / unmanned aerial vehicle cooperative task allocation method based on an insect positive and negative feedback mechanism. The method comprises the steps of 1, establishing a manned aerial vehicle / unmanned aerial vehicle heterogeneous system and a task target model; 2, establishing a collaborative task allocation multi-traveling salesman task constraint model; 3, establishing a manned aerial vehicle / unmanned aerial vehicle task decision network based on an insect social division positive feedback mechanism; 4, establishing a decision network optimization strategy; 5, building an overall collaborative task allocation process; step 6, establishing a collaborative task allocation evaluation index; and step 7, outputting a manned aerial vehicle / unmanned aerial vehicle cooperative task allocation result graph based on the insect positive and negative feedback mechanism. According to the method, under the condition that scenes and tasks change continuously, cooperation and mutual coordination of manned aerial vehicle / unmanned aerial vehicle clusters are enhanced in task allocation, and the minimum time and resource cost is achieved in a large-scale manned aerial vehicle / unmanned aerial vehicle cooperative task allocation system.

Description

Technical Field

[0001] The present invention discloses a man-machine / unmanned aerial vehicle collaborative task allocation method based on insect positive and negative feedback mechanisms, and belongs to the field of autonomous control of man-machine hybrid clusters. Background Art

[0002] In recent years, unmanned aerial vehicles (UAVs) have developed rapidly. As a new type of combat force, they have developed from traditional information support and information assurance fields such as reconnaissance, surveillance, hydrology and meteorology to firepower strikes with attack and killing capabilities. They will undertake more important combat tasks in future wars. Therefore, with the development of UAV technology and the increase in the number of UAV equipment, how to truly integrate this new type of combat force into the combat system and give full play to the advantages of system combat in future wars has become an important topic that countries are competing to study. However, facing the increasingly complex and fierce battlefield environment, as well as the multiplicity and complexity of tactical tasks, it is difficult for UAVs to complete high-intensity confrontation combat tasks alone. Therefore, according to the combat mission requirements of UAVs in future wars, combined with the combat characteristics of manned aerial vehicles (MAVs) and UAVs, in the informationized, networked, intelligent, and system confrontation combat environment, the coordinated combat of manned and UAV formations will realize the sharing of information and resources between manned and UAVs and the entire combat system, and give full play to their respective advantages, which has great military application value for improving the combat effectiveness of the entire system.

[0003] Heterogeneous systems of manned / unmanned aircraft usually complement each other in performance and functionality, and can effectively perform real-time tasks in complex military mission scenarios, such as electronic reconnaissance and jamming in hostile environments. Manned / unmanned aircraft can benefit in many ways, especially when the system consists of multiple entities distributed in different locations or spaces. Through cooperation, manned / unmanned aircraft can work as a team to complete activities that are difficult for them to complete individually. In such missions, with the increase in mission complexity and multiple expansions of manned / unmanned aircraft, flexible and autonomous deployment optimization is required. The mission planning scheme can be further refined into an optimal allocation scheme, which aims to achieve the minimum time and resource cost in a large-scale manned / unmanned aircraft collaborative task allocation system. In general, the optimization problem under multiple constraints including manned / unmanned aircraft and targets is a typical non-deterministic polynomial (NP) hard problem.

[0004] Many traditional solutions require constant communication interactions, which leads to increased communication overhead and directly increases computing costs. The goal of an advantageous mission planning strategy is to complete the given task at the lowest cost by strengthening the coordination and cooperation of manned / unmanned aerial vehicle clusters in task allocation. Therefore, it is very important for manned / unmanned aerial vehicle clusters to provide more decision-making autonomy to achieve dynamic control and management. Manned / unmanned aerial vehicle clusters hope to complete all tasks in the shortest time under the condition of constantly changing scenes and tasks, which involves internal competition and cooperation. Similarly, social insects achieve a balance between competition and cooperation through social division of labor behavior. Social insects divide tasks according to individual strengths and abilities. Different individuals assume different roles and tasks, coordinate with each other, and complete complex tasks together.

[0005] In summary, the present invention proposes a manned / unmanned aerial vehicle collaborative task allocation method based on the positive and negative feedback mechanism of insects, with the manned / unmanned aerial vehicle collaborative task allocation as the background. First, the problem is described, and the task allocation problem is modeled. Next, a manned / unmanned aerial vehicle collaborative task allocation method that imitates the positive and negative feedback mechanism of insect social division of labor is proposed. Similar to the positive feedback mechanism of insect social division of labor behavior, the decision matrix corresponding to the manned / unmanned aerial vehicle cluster is constructed by imitating the process of mutual communication and interaction between individual insects under external task stimulation. The decision matrix is ​​used to find all strategies that can be used to complete the task without causing conflicts. At the same time, similar to the fact that individual insects will adjust their own behavioral decisions according to the group's benefits and needs, the manned / unmanned aerial vehicle cluster can select the most appropriate task allocation plan from the decision matrix and solve the problem at the lowest cost. Summary of the invention

[0006] 1. Purpose of the invention:

[0007] The present invention provides a manned-machine / unmanned-machine collaborative task allocation method based on the positive and negative feedback mechanism of insects. Its purpose is to achieve the minimum time and resource cost in a large-scale manned-machine / unmanned-machine collaborative task allocation system by strengthening the cooperation of manned-machine / unmanned-machine clusters in task allocation and coordinating with each other under the condition of constantly changing scenes and tasks.

[0008] 2. Technical solution:

[0009] The present invention aims at the problem of manned-machine / unmanned-machine collaborative task allocation and provides a manned-machine / unmanned-machine collaborative task allocation method based on insect positive and negative feedback mechanism. The specific steps of the method are as follows:

[0010] Step 1: Establish a manned / unmanned aircraft heterogeneous system and mission objective model

[0011] Assume existence Manned / unmanned aircraft and m dynamic targets T = {T1, T2, ..., T m}. Manned / unmanned aircraft include The fastest manned aircraft with the largest detection range, A fast attack drone (slower than the aforementioned manned aircraft) with the smallest detection range and The slowest drone (slower than the aforementioned attack drone) and a reconnaissance drone with a larger detection range. It will gradually decrease as the unit time goes by until it reaches the minimum value F min Reconnaissance missions are generally considered to be staying close to a target at a safe distance, tracking it for a period of time using radar or optical sensors. b For reconnaissance targets, they include known targets T k 、Unknown target T u and deceive the target T t .

[0012] Considering the route design that simplifies the top-level decision-making of manned / unmanned aircraft, a timestamp model is used to generate trajectories with constant speed and altitude, such as Figure 1 The timestamp model of the i-th manned / unmanned aircraft is as follows:

[0013]

[0014] In the time stamp model, there are two components, the velocity component and the position component P i (t) = [P i x (t),P i y (t)] T . Indicates the connection of the current point S i (t) and the target point E i The angle between the line (t) and the x-axis of the Cartesian coordinate system. Indicates the location of the target, Indicates the manned / unmanned aircraft mission switching position. Figure 1 In, determine E i After the target is positioned (t), the schematic diagram shows that there is a certain range of threat area. Starting from its current position, after task allocation, manned / unmanned aircraft will converge towards the target and arrive near the threat area at a time interval t b After completing the designated reconnaissance time, the manned / unmanned aircraft will quickly transition to a new reconnaissance phase, starting from its mission switching point Si (t+t b ) and move to the subsequent target position E i (t+t b )move.

[0015] If the Euclidean distance D between the i-th manned / unmanned aircraft and the j-th target ij Equal to the detection range R d , this target will be detected and its authenticity can be confirmed. For the moving ship target at sea, it is assumed to be a two-dimensional plane moving target without height change. For example, the jth target model is designed as:

[0016]

[0017] Where r1, r2∈(-1,1), u and v represent the constant rate of the ship's speed. Assume that the target ship's motion is a completely unpredictable random motion in two dimensions, with a speed range within a certain range. The speed magnitudes on the two coordinate axes are determined by multiplying the constant values ​​u and v by their respective corresponding random numbers, denoted as r1 and r2, respectively.

[0018] To ensure safety during joint reconnaissance operations, manned and unmanned aircraft can only fly within a certain range. L The inner surrounding and following ships must have an angle distribution θ with respect to each other. ik , the formula is:

[0019]

[0020] Among them, X (k,j) = 1 means that the kth manned / unmanned aircraft executes the jth target, and X (k,j) =0, the opposite is true.

[0021] Step 2: Establish a collaborative task allocation multi-traveling salesman task constraint model

[0022] Generally speaking, radar sensors have greater detection capabilities than optical sensors, and the number of required payloads N and the duration of reconnaissance t b It needs to change proportionally. Assume that each manned / unmanned aircraft can only be equipped with one payload. For example, when performing a reconnaissance mission, a manned / unmanned aircraft equipped with a radar payload needs Radar payload completion time t R reconnaissance; or equipped with optical payloads, requiring The optical payload completion time t J of reconnaissance.

[0023] For the constraints of the multi-traveling salesman problem, it should be ensured that each goal must be executed once by a manned / unmanned aircraft. The formula is expressed as:

[0024]

[0025] in, and They represent the number of sensors required for manned / unmanned aerial vehicles to detect a target. Represents the number of manned / unmanned aircraft when executing target j.

[0026] Step 3: Establish a manned / unmanned aerial vehicle task decision network based on the positive feedback mechanism of insect social division of labor

[0027] When a task appears, the execution of the task is broken down into a single manned / unmanned aircraft carrying a payload, following the group discussion process in insect society, based on the set of manned / unmanned aircraft, the requirements of the task, and the set of goals. The payload type affects the number of payloads carried. Based on the response relationship between the task volume and the threshold of the manned / unmanned aircraft, the response function is designed as:

[0028]

[0029] Among them, s w represents the number of tasks, n represents a fixed constant, θ i Is a manned / unmanned aircraftU i The threshold of one's own ability, Indicates manned aircraft / unmanned aircraftU i The probability of responding to external stimuli. Assuming that the number of individuals responding after external environmental stimulation is q, the vertex set of its selection scheme is Construct the adjacency matrix G = {V, E, W}, and the edge set is E = {e ij |i,j∈V}, W is the vertex weight.

[0030] According to the competitive interaction mode of communication and coordination between insects, the following three rules are also designed to solve the conflict problem in the process of executing tasks. The following three rules are used to restrict as follows:

[0031] Rule 1: Ensure that each manned / unmanned aircraft can only complete a single mission of the target. For scheme j and scheme k, if the sequence of manned / unmanned aircraft selected is the same, then G (j×k)&(k×j) ←1.

[0032] Rule 2: Ensure that different task allocation schemes for each goal do not conflict. If the goals of scheme j and scheme k are the same, then G (j×k)&(k×j) ←1.

[0033] Rule 3: Ensure that each mission of the target can only be completed by one manned / unmanned aircraft. For scheme j and scheme k, if the target, sequence of schemes, and choice of payload are the same, then G (j×k)&(k×j) ←1.

[0034] According to the implementation of the above three rules, there will be related allocation schemes and unrelated allocation schemes in the adjacency matrix G. The related allocation schemes j and k can choose each other. And the unrelated division of labor schemes can all be selected.

[0035] Taking the capabilities of the manned / unmanned aircraft and the cost-benefit of the target as the vertex weight W, the cost C(i, j) of the i-th manned / unmanned aircraft from the current position to the pre-assigned position of the j-th target is considered, which can be expressed as:

[0036]

[0037] Among them, v i represents the speed of the i-th manned / unmanned aircraft, D ij represents the Euclidean distance between the i-th manned / unmanned aircraft and the j-th target. In addition, the normalized cost function is set to the vertex weight value W, expressed as:

[0038]

[0039] Among them, in the jth scheme, the serial numbers of the manned / unmanned aircraft and the target are recorded by α(j) and β(j), respectively. In other words, after establishing the game matrix, vertex j represents a strategy involving the allocation of manned / unmanned aircraft and target tasks, and the identities of the deployed manned / unmanned aircraft and the target are represented by the symbol α(j) and the symbol β(j), respectively.

[0040] Step 4: Establish a manned / unmanned aerial vehicle decision network optimization strategy based on the negative feedback mechanism of insect social division of labor

[0041] According to the division of labor in insect society, individuals will have two reactions after making a specific decision: "disappointed" and "positive". In the manned / unmanned aerial vehicle collaborative task allocation, assuming that there are N individuals in the insect society P, there are N decision-making schemes. Consider the network G = (V, {X i},{u i}), the i-th individual X i There are m nodes {x1,x2,...,x m} definition, where x i =1 means that the i-th sub-strategy is selected, and x i =0 On the contrary, Indicates that it is different from x i The decision behavior of the program, X-i represents the decision-making behavior of other individuals. i represents the local utility function. In the randomly initialized insect society P t＝0 Before the iteration begins, in the social division of labor of insects, individuals will adjust their behavioral decisions according to the group's benefits and needs. In manned / unmanned aircraft, through the individual's loss or positive degree of the decision made, the individual with the largest loss value can also learn from the individual with the largest positive value. Therefore, it is required that individual x j Each node is updated iteratively to meet the following constraints:

[0042]

[0043] Among them, b i represents the maximum positive value obtained when node i makes a specific decision, η i Indicates the loss value when the decision is not the best. Indicates that the choice is different from the current decision x i The positive value of the behavior, Indicates that the choice is different from the current decision x i The decision of the behavior makes the choice of the loss value that the current decision regrets. Then the best node k is selected and its nodes are updated. When the loss value of all nodes in the entire system is 0, the learning process terminates.

[0044] In the division of labor in insect society, some individuals with stronger abilities will replace other individuals and take on more workloads. In addition, individuals have differentiated abilities and behavioral preferences, and individuals will learn from the best individuals. Therefore, it is necessary to find the global optimal solution. t * , and the cost function is designed by the weight value W in formula (6) and the strategy that individuals with stronger capabilities can replace other individuals. This process is shown in Algorithm 1 shown in the following table.

[0045]

[0046] As in the social division of labor in insects, when individuals observe the successful behaviors of other individuals in a social environment, they will imitate and learn these behaviors. Based on the behaviors learned and observed by individuals, individuals will move towards the optimal result. However, the random learning and observation process is diverse for individuals, and this diversity becomes smaller and smaller over time. Therefore, Algorithm 2 shown in the following table is designed to imitate the random learning and observation behavior in the division of labor in insect society. Where τ is a 1×m random vector, each element of which is 0 or 1 with equal probability, and represents the Hadamard product operator, which is a mathematical operation performed by multiplying corresponding elements of two matrices or vectors of the same size. and diversity parameter ρ, from each individual Generate new individuals

[0047]

[0048] In addition, in order to obtain a decision solution with a maximum loss value of 0 for each node, when the individual When the decision is made, it is necessary to re-update the satisfaction value according to its own decision, as shown in formula (7), so as to obtain the individual Then, we imitated the selective memory of successful experience in the division of labor in insect society and put it in the insect social P i t The δ individuals with higher cost function values ​​in the insect society are memorized until the next insect society P i t+1 The node is being updated.

[0049] Step 5: Build the overall process of manned / unmanned aerial vehicle collaborative task allocation

[0050] When manned / unmanned aerial vehicle (UAV) swarms perform reconnaissance mission assignment tasks, they must deal with a series of real-time problems. These problems include the constant changes in the position and state information of known targets, the sudden appearance of unknown targets, and the continuous changes in the position and state information of manned / unmanned aerial vehicle heterogeneous systems. Therefore, a task assignment cooperative decision framework is designed, such as Figure 3 , which aims to solve the unpredictable combinatorial optimization problem, determine whether the target has been assigned and update the status of manned / unmanned aircraft. Since the difference between the tasks of reconnaissance mission assignment is only reflected in the type, quantity and start time of the payload, the stage of status update and mission planning is applicable to the type of manned / unmanned aircraft reconnaissance mission, and the purpose of the whole process is to complete these tasks as quickly as possible.

[0051] After the state update phase is completed, it is necessary to enter the mission planning phase. First, it is necessary to determine whether the task allocation has been resolved. If it has been completed, the manned / unmanned aircraft group that has set out will return to the base, the manned / unmanned aircraft group that has arrived at the base will be shut down, and the manned / unmanned aircraft group with the minimum fuel quantity will start to return. Finally, maximize the objective function in the real-time mission planning algorithm 1.

[0052] Step 6: Establish evaluation indicators for manned / unmanned aerial vehicle collaborative task allocation

[0053] In the process of manned / unmanned aerial reconnaissance task allocation, due to the existence of deceptive targets that may interfere with normal behavior, these targets can only be distinguished when they are within the detection range in formula (3). Therefore, while considering the efficiency of the optimization algorithm in solving the task allocation problem, it is also necessary to evaluate the ability to obtain the authenticity of all target information.

[0054] In addition, information entropy can be used to analyze information transmission and reception, and has gradually developed into a tool for quantifying the information content of the system. Therefore, in addition to evaluating the effectiveness of the optimization algorithm in solving the reconnaissance problem, the ability to obtain the authenticity of all target information should also be evaluated. This paper adopts the absolute information entropy method. Given a set of group events, the prior probability and entropy function H(t) of these events are expressed as:

[0055]

[0056] Among them, H(t) represents the information content. The more chaotic the system is, the higher the H(t) value is. The traditional method based on binary hypothesis assumes that each target has two states (real or false) with equal probability. Therefore, H(t) will change with the actions of the manned / unmanned aerial vehicle group. In addition, by observing the changes in information entropy H(t) over a certain period of time, the impact of different algorithms on the orderliness of the system can be demonstrated, thereby evaluating the effectiveness of the selected algorithm.

[0057] Step 7: Output the result graph of manned / unmanned aerial vehicle collaborative task allocation based on insect positive and negative feedback mechanism

[0058] Determine whether the iteration time reaches the maximum iteration time T max ,If it is reached, the loop is jumped out and the result diagram of manned / unmanned aerial vehicle collaborative task allocation based on the insect positive and negative feedback mechanism is output. Otherwise, the task allocation process is continued and dynamic task allocation is performed on the target area.

[0059] The present invention is a manned-machine / unmanned-machine collaborative task allocation method based on the positive and negative feedback mechanism of insects, and its advantages and functions are: First, it provides a manned-machine / unmanned-machine collaborative task allocation method that simulates the positive and negative feedback mechanism of insect social division of labor. Through the positive feedback mechanism, external task stimulation encourages manned-machine / unmanned-machine with different capabilities to participate in decision-making, and generates a variety of non-conflicting strategies. At the same time, according to the negative feedback mechanism, individuals adjust their behavioral decisions according to group benefits and their own needs to generate the optimal task allocation plan; second, a manned-machine / unmanned-machine collaborative task allocation decision-making framework is designed. The framework introduces information entropy evaluation, task completion rate indicators and various timeliness inspection indicators to evaluate the timeliness and orderliness of task allocation; third, it has multiple expansions with the increase of task complexity and the number of manned-machine / unmanned machines, and due to the internal competition and cooperation mechanism, the optimization ability becomes more prominent. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 Schematic diagram of the manned / unmanned aircraft timestamp model.

[0061] Figure 2 Framework diagram of the decision process for manned / unmanned aerial vehicle collaborative task allocation.

[0062] Figure 3 Simulation diagram for collaborative task allocation of manned / unmanned aircraft carrying radar payloads.

[0063] Figure 4 Simulation diagram for collaborative task allocation of manned / unmanned aircraft carrying optoelectronic payloads.

[0064] Figure 5 Simulation diagram of task completion rate for manned / unmanned aerial vehicle collaborative task allocation.

[0065] Figure 6 Simulation diagram of information entropy value allocation for manned / unmanned aerial vehicle collaborative tasks. DETAILED DESCRIPTION

[0066] The effectiveness of the method proposed in the present invention is verified by a specific example of manned-machine / unmanned-machine collaborative task allocation. The experimental computer is configured with an Intel Core i7-10700 processor, 2.90Ghz main frequency, 16G memory, and the software is MATLAB R2022b version. The specific steps of a manned-machine / unmanned-machine collaborative task allocation method based on insect positive and negative feedback mechanism are as follows:

[0067] Step 1: Establish a manned / unmanned aircraft heterogeneous system and mission objective model

[0068] The entire scenario is set as a square area with a side length of 1650m. The initial positions of manned / unmanned aircraft are randomly distributed in a square base area with a side length of 200m, and the initial positions of dynamic ship targets are also randomly distributed in a square target area with a side length of 1000m. The scenarios and constraints of the manned / unmanned aircraft heterogeneous system and targets are shown in Tables 1 and 2.

[0069] Table 1 Scenarios and restriction parameters of manned / unmanned aerial vehicle collaborative task allocation

[0070]

[0071] Table 2 Reconnaissance target restriction parameters

[0072]

[0073]

[0074] Step 2: Establish a collaborative task allocation multi-traveling salesman task constraint model

[0075] Radar sensors have greater detection capabilities than optical sensors, the number of required payloads N and the duration of reconnaissance t b It needs to change proportionally. Assume that each manned / unmanned aircraft can only be equipped with one payload. For example, when performing a reconnaissance mission, a manned / unmanned aircraft equipped with a radar payload requires two radar payloads to complete a 10-second reconnaissance mission, or equipped with an optical payload requires four optical payloads to complete a 30-second reconnaissance mission.

[0076] Steps 3 and 4: Establishing a manned / unmanned aerial vehicle task decision network and optimization strategy based on the positive feedback mechanism of insect social division of labor

[0077] According to the test 1 and 2 of the same target state. Then it is compared with various existing algorithms, such as Population-based Game-theoretic Optimizer (PGTO) and Population-based Iterated Greedy (PBIG), which have significant effects in obtaining optimization results for decision network problems. Table 3 shows the parameters of each optimization method.

[0078] Table 3 Parameter settings of this algorithm and the comparison algorithm

[0079]

[0080]

[0081] Step 5: Build the overall process of manned / unmanned aerial vehicle collaborative task allocation

[0082] according to Figure 3 First, it is necessary to determine whether the task allocation has been resolved. If it has been completed, the manned / unmanned aircraft group that has set off will return to the base, the manned / unmanned aircraft group that has arrived at the base will be shut down, and the manned / unmanned aircraft group that has reached the minimum fuel quantity will start to return. Finally, the algorithm proposed in this paper is used to maximize the objective function in step three and step four of real-time task planning algorithm 1.

[0083] Step 6: Establish evaluation indicators for manned / unmanned aerial vehicle collaborative task allocation

[0084] The task completion rate and the absolute information entropy value are calculated according to the parameters set in steps one, three and four to judge the completion effect of the task and evaluate the ability to obtain the authenticity of the target information in an orderly manner.

[0085] Step 7: Output the result graph of manned / unmanned aerial vehicle collaborative task allocation based on insect positive and negative feedback mechanism

[0086] First, the total iteration time is set to 50 and the sampling time step is 0.1. When the iteration step reaches 250, the simulation result diagram of manned / unmanned aerial vehicle collaborative task allocation based on the positive and negative feedback mechanism of insects is output. When the iteration step reaches 500, the simulation test diagrams under different algorithms are output.

[0087] The present invention adopts two test simulations, firstly analyzing the effect of manned / unmanned aerial vehicle collaborative task allocation with photoelectric sensors, and then analyzing the effect of manned / unmanned aerial vehicle collaborative task allocation with radar sensors, and finally completing the collaborative task allocation task under different scenario settings.

Claims

1. A manned / unmanned aerial vehicle collaborative task allocation method based on insect positive and negative feedback mechanism, characterized by: The steps of this method are as follows: Step 1: Establish a manned / unmanned aircraft heterogeneous system and mission objective model: Assume there are n manned / unmanned aircraft and m dynamic targets T = {T1, T2, ..., T m }; Manned / unmanned aircraft include The fastest and most extensive manned aircraft, r A fast attack drone with the smallest detection range and The slowest reconnaissance drone with a larger detection range; the fuel volume of manned / unmanned aircraft It will gradually decrease as the unit time goes by until it reaches the minimum value F min ; The reconnaissance mission is to keep close to the target within a safe distance and track the target for a period of time using radar or optical sensors. b ; For reconnaissance targets, including known targets T k 、Unknown target T u and deceive the target T t ; The timestamp model of the i-th manned / unmanned aircraft is as follows: In the time stamp model, there are two components, the velocity component and position component Indicates the connection of the current point S i (t) and the target point E i The angle between the line (t) and the x-axis of the Cartesian coordinate system; Indicates the location of the target, Indicates the task switching position of manned aircraft / unmanned aircraft; starting from its current position, after task allocation, manned aircraft / unmanned aircraft will converge towards the target and arrive near the threat area at a time interval t b After completing the designated reconnaissance time, the manned / unmanned aircraft will quickly transition to a new reconnaissance phase, starting from its mission switching point S i (t+t b ) and move to the subsequent target position E i (t+t b )move; If the Euclidean distance D between the i-th manned / unmanned aircraft and the j-th target ij Equal to the detection range R d , this target will be detected and its authenticity will be confirmed; for the mobile ship target at sea, it is set as a two-dimensional plane moving target without height change, and the j-th target model is designed as: Where r1, r2∈(-1, 1), u and v represent the constant speed of the ship; the motion of the target ship is assumed to be a completely unpredictable random motion occurring in two dimensions, and the speed on the two coordinate axes is determined by multiplying the constant values ​​u and v with their corresponding random numbers, expressed as r1 and r2 respectively; To ensure safety during joint reconnaissance operations, manned and unmanned aircraft can only operate within a certain range. L The inner surrounding and following ships must have an angle distribution θ with respect to each other. ik , the formula is: Among them, X (k,j) = 1 means that the kth manned / unmanned aircraft executes the jth target, and X (k,j) =0, the opposite is true; Step 2: Establish a collaborative task allocation multi-traveling salesman task constraint model: The number of required payloads N and the duration of reconnaissance t b needs to change proportionally; assume that each manned aircraft / unmanned aircraft can only be equipped with one payload; that is, when performing a reconnaissance mission, a manned aircraft / unmanned aircraft equipped with a radar payload needs Radar payload completion time t R reconnaissance or equipped with optical payloads, it is necessary The optical payload completion time t J reconnaissance; For the constraints of the multi-traveling salesman problem, it should be ensured that each goal must be executed once by a manned / unmanned aircraft. The formula is expressed as: in, and They represent the number of sensors required for manned / unmanned aerial vehicles to detect a target. represents the number of manned / unmanned aircraft when executing target j; Step 3: Establish a manned / unmanned aerial vehicle task decision network based on the positive feedback mechanism of insect social division of labor: According to the response relationship between the task volume and the threshold of the individual manned / unmanned aircraft, the response function is designed as follows: Among them, s w represents the number of tasks, n represents a fixed constant, θ i Is a manned / unmanned aircraftU i The threshold of one's own ability, Indicates manned aircraft / unmanned aircraftU i The probability of responding to external stimuli; let the number of responding individuals after external environmental stimulation be q, and the vertex set of its selection scheme is Construct the adjacency matrix G = {V, E, W}, and the edge set is E = {e ij ∣i,j∈V}; Taking the capabilities of the manned / unmanned aircraft and the cost-benefit of the target as the vertex weight W, the cost C(i, j) of the i-th manned / unmanned aircraft from the current position to the pre-assigned position of the j-th target is considered, which can be expressed as: Among them, v i represents the speed of the i-th manned / unmanned aircraft, D ij represents the Euclidean distance between the i-th manned / unmanned aircraft and the j-th target. In addition, the normalized cost function is set to the vertex weight value W, expressed as: Among them, in the jth scheme, the serial numbers of the manned / unmanned aircraft and the target are recorded by α(j) and β(j), respectively. In other words, after the game matrix is ​​established, vertex j represents a strategy involving the assignment of manned / unmanned aircraft and target tasks, and the identities of the deployed manned / unmanned aircraft and the target are represented by the symbol α(j) and the symbol β(j), respectively; Step 4: Establish a manned / unmanned aerial vehicle decision network optimization strategy based on the negative feedback mechanism of insect social division of labor: In the manned / unmanned aerial vehicle collaborative task allocation, suppose there are N individuals in the insect society P, that is, there are N decision-making schemes. Consider the network G = (V, {X i },{u i }), the i-th individual X i There are m nodes {x1,x2,...,x m } definition, where x i =1 means that the i-th sub-strategy is selected, and x i =0 On the contrary, Indicates that it is different from x i The decision behavior of the program, X -i represents the decision-making behavior of other individuals; u i represents the local utility function; in the randomly initialized insect society P t＝0 Before starting the iteration, individual x is required j Each node is updated iteratively to meet the following constraints: Among them, b i represents the maximum positive value obtained when node i makes a specific decision, η i Indicates the loss value when it is not the best decision; Indicates that the choice is different from the current decision x i The positive value of the behavior, Indicates that the choice is different from the current decision x i The decision of behavior makes the choice of the loss value that the current decision feels regretful; then the node k with the best performance is selected and its node is updated. When the loss value of all nodes in the entire system reaches 0, the learning process terminates. Step 5: Build the overall process of manned / unmanned aerial vehicle collaborative task allocation: First, it is necessary to determine whether the task allocation has been resolved. If it has been completed, the manned / unmanned aircraft group that has set out will return to the base, the manned / unmanned aircraft group that has arrived at the base will be shut down, and the manned / unmanned aircraft group whose fuel quantity has reached the minimum value will begin to return.

2. The method according to claim 1, characterized in that: In step 3, according to the competitive interaction mode of communication and coordination between insects, the following rules are designed to restrict and solve the conflict problem in the task execution process. The following three rules are used: Rule 1: Ensure that each manned / unmanned aircraft can only complete a single mission of the target. For scheme j and scheme k, if the sequence of manned / unmanned aircraft selected is the same, then G (j×k)&(k×j) ←1; Rule 2: Ensure that different task allocation schemes for each goal do not conflict. If the goals of scheme j and scheme k are the same, then G (j×k)&(k×j) ←1; Rule 3: Ensure that each mission of the target can only be completed by one manned / unmanned aircraft. For scheme j and scheme k, if the target, sequence of schemes, and choice of payload are the same, then G (j×k)&(k×j) ←1.

3. The method according to claim 1, characterized in that: The step 4 also includes the following contents: solving the global optimal solution And the cost function is designed by the weight value W in formula (6) and the strategy that individuals with stronger abilities can replace other individuals.

4. The method according to claim 1, characterized in that: The fourth step also includes the following: imitate the social division of labor in insects, according to the behavior of individual learning and observation, individuals will move towards the optimal result. Learning is carried out, but in the random learning and observation process, there is diversity for individuals, but this diversity becomes smaller and smaller over time. Therefore, a task allocation scheme that simulates the random learning and observation behavior of insect social division of labor is designed to update nodes.

5. The method according to claim 4, characterized in that: The step 4 also includes the following: In order to obtain a decision solution with a maximum loss value of 0 for each node, when the individual When the decision is made, it is necessary to re-update the satisfaction value according to the decision itself, so as to obtain the individual Then, we will imitate the selective memory of successful experience in the division of labor in insect society. The δ individuals with higher cost function values ​​in the insect society will be memorized until the next insect society The node is being updated.

6. The method according to claim 1, characterized in that: This method also The following steps are included: Step 6: Establish evaluation indicators for manned / unmanned aerial vehicle collaborative task allocation: While considering the efficiency of the optimization algorithm in solving the task allocation problem, it is also necessary to evaluate the ability to obtain the authenticity of all target information; The absolute information entropy method is used; given a set of group events, the prior probability and entropy function H(t) of these events are expressed as: Among them, H(t) represents the information content. The more chaotic the system is, the higher the H(t) value is. Using the traditional method based on binary assumptions, it is assumed that each target has two states - real or false, with equal probability. Therefore, H(t) will change with the actions of the manned / unmanned aerial vehicle group.

7. The method according to claim 6, characterized in that: The method further includes the following steps: Step 7: Outputting a manned / unmanned aerial vehicle collaborative task allocation result diagram based on the insect positive and negative feedback mechanism: Determine whether the iteration time reaches the maximum iteration time T max ,If it is reached, the loop is jumped out and the result diagram of manned / unmanned aerial vehicle collaborative task allocation based on the insect positive and negative feedback mechanism is output. Otherwise, the task allocation process is continued and dynamic task allocation is performed on the target area.

Images