V-Type Multi-Agent Aircraft Drag Reduction Control Algorithm

Abstract

The present invention discloses a drag reduction control algorithm for V-shaped multi-agent aircraft, which specifically includes the following steps: Step 1: First, introduce a V-shaped formation configuration according to the biological dynamics model, obtain the extended V-shaped formation control law, and establish a second-order dynamic system to obtain multiple sets of V-shaped formation control laws; Step 2: According to the extended V-shaped formation control law, divide the agent cluster into multiple subgroups and perform multi-agent subgroup dynamic consensus determination; Step 3: Continuously iterate and update the selection of the leader agent of each subgroup in the current state according to the leader agent selection algorithm, so as to perform drag reduction control based on the V-shaped multi-agent aircraft. The present invention promotes the mutual integration between multiple disciplines, can stabilize the robustness and scalability of the aircraft formation flight process, determines the leader agent by judging the subgroup order, and considers the selection strategy of the leader agent under different circumstances.

Description

Technical Field

[0001] The present invention belongs to the technical field of aircraft control, and particularly relates to a drag reduction control algorithm for V-shaped multi-agent aircraft. Background Art

[0002] In recent years, multi-agent technology has developed rapidly and achieved remarkable development in many applications. The cooperation among multiple agents can be applied to tasks in some complex environments. At the same time, integrating a certain formation shape can bring the multi-agent technology into full play. Since the aircraft is affected by the resistance in the atmosphere during flight, resulting in different resistances at different positions, how to effectively ensure the robustness and scalability of formation flight and how to improve the safety during flight are the problems that need to be studied currently. Summary of the Invention

[0003] The purpose of the present invention is to provide a drag reduction control algorithm for V-shaped multi-agent aircraft, which solves part of the problems of how to effectively ensure the robustness and scalability of formation flight of current aircraft and how to improve the safety during flight.

[0004] The technical solution adopted by the present invention is as follows:

[0005] The drag reduction control algorithm for V-shaped multi-agent aircraft specifically includes the following steps:

[0006] Step 1: First, introduce a V-shaped formation configuration according to the biological dynamics model, obtain the control law of the extended V-shaped formation, and establish a second-order dynamic system to obtain multiple sets of control laws for the V-shaped formation;

[0007] Step 2: According to the control law of the extended V-shaped formation, divide the agent cluster into multiple subgroups and perform multi-agent subgroup dynamic consensus determination;

[0008] Step 3: Continuously iterate and update to select the leader agent of each subgroup in the current state according to the leader agent selection algorithm, so as to perform drag reduction control for the V-shaped multi-agent aircraft.

[0009] The characteristics of the present invention also lie in;

[0010] Step 1 is specifically as follows: Assume that the aircraft is a second-order dynamic system, defined by the following formula (1):

[0011]

[0012] Wherein, is the position of the aircraft and is the velocity of the aircraft and t is time, u i(t) is the control input of agent i at time t, and the extended V-shaped formation control law can be derived as shown in the following formula (2):

[0013]

[0014] where, is the difference between the actual position and the desired position of agent j, and v j (t) is the velocity of agent j at time t, is defined as input, and ω is the intelligence of agent j.

[0015] In step 2, according to formula (2), a swarm of agents is divided into multiple agent subgroups as shown in the following formula (3):

[0016]

[0017] In step 2, the dynamic consensus determination of multiple agent subgroups is specifically as follows: by calculating the limit difference between adjacent agent subgroups, when agents i and j in different subgroups both satisfy the following formula (4), it is determined that the agent subgroup reaches dynamic consensus:

[0018]

[0019] where, x i (t) and x j (t) are the positions of agent i and agent j respectively, and v i (t) and v j (t) are the velocities of agent i and agent j respectively.

[0020] In step 3, the distance between two agents can be obtained from the following formula (5):

[0021]

[0022] where, p x (i) and p y (i) are the positions of agent i on the x-axis and y-axis respectively, p x (j) and p y (j) are the positions of agent j on the x-axis and y-axis respectively, and ξ is a positive weighting factor to prevent the agent from oscillating between the leading agents;

[0023] Through the communication range of each agent and case-by-case analysis, the leading agent in the subgroup is effectively selected, thereby realizing the drag reduction control of the V-shaped multi-agent aircraft.

[0024] The beneficial effects of the present invention are as follows. Based on the V-shaped multi-agent aircraft drag reduction control algorithm, the present invention combines the biological dynamics model with aircraft drag reduction, promoting the mutual integration between multiple disciplines. The present invention divides a V-shaped agent cluster into multiple V-shaped agent subgroups, which can stabilize the robustness and scalability of the aircraft formation flight process. The present invention determines the leader agent by judging the subgroup order and considers the selection strategy of the leader agent under different circumstances. Description of the Drawings

[0025] Figure 1 is a schematic diagram of the V-shaped subgroup sorting in the V-shaped multi-agent aircraft drag reduction control algorithm of the present invention;

[0026] Figure 2 is a flowchart of the leader agent selection algorithm in the V-shaped multi-agent aircraft drag reduction control algorithm of the present invention. Detailed Embodiment

[0027] The V-shaped multi-agent aircraft drag reduction control algorithm of the present invention will be further described in detail below in conjunction with the drawings and specific embodiments.

[0028] The present invention proposes a V-shaped multi-agent aircraft drag reduction control algorithm. By simulating the movement trajectory of biological dynamics, a V-shaped formation configuration is proposed, which can effectively reduce the drag during flight. The extended formation control law is obtained through system assumptions, and then the leader agent in each subgroup is determined according to the algorithm to control the overall formation flight. The detailed steps and processes are as follows:

[0029] As Figure 1 and Figure 2 shown, Step 1: Obtain the extended V-shaped formation control law;

[0030] Assume that the aircraft is a second-order dynamic system, defined as:

[0031]

[0032] where is the position of the aircraft and is the velocity of the aircraft, and t is the time, u i (t) is the control input of agent i at time t, and the extended V-shaped formation control law can be derived as shown in the following formula (2):

[0033]

[0034] In formula (2), is the difference between the actual position and the desired position of agent j, v j (t) is the velocity of agent j at time t, is defined as For the input, ω is the intelligence of agent j. According to formula (2), an agent cluster is divided into multiple agent subgroups as shown in formula (3).

[0035]

[0036] Step 2: Dynamic consensus determination for multiple agent subgroups;

[0037] The purpose of dynamic consensus is to make each extended subgroup finally move in the same direction, which is mainly obtained by calculating the limit difference between adjacent agent subgroups. When agents i and j in different subgroups both satisfy the following formula (4), it is determined that the agent subgroup reaches dynamic consensus:

[0038]

[0039] In formula (4), x i (t) and x j (t) are the positions of agent i and agent j respectively, and v i (t) and v j (t) are the velocities of agent i and agent j respectively.

[0040] Step 3: Leader agent selection algorithm;

[0041] As Figure 1 shown in the schematic diagram of V - type subgroup sorting proposed by the present invention, 1 - 5 are five agents respectively. The distance from the black circle to the center of the circle represents the communication range of each agent, which is an index to measure the network connection strength established by all group agents.

[0042] Assume that the communication range between each agent subgroup is R, and agent i with intelligence ω has subgroup order O i,w , which represents the position of the agent in the formation of its designated subgroup. In the leader agent selection algorithm, assume that all agents in an agent subgroup have the same target path point (i.e., the same end point), and their starting positions are aligned with the direction of the path point. The algorithm process is as follows: First, let each agent subgroup O i,w = 0, that is, initialize the subgroup order of all agents to 0. In this algorithm, the agent with the default order of 0 serves as the leader in the subgroup. So after initialization, each agent is a potential leader. Assume that the target path point is in the positive x - axis direction of the Cartesian coordinate. When the agent subgroup order takes a non - zero value, the algorithm automatically assigns a positive order to the upper part (left wing) of the leader agent and a negative order to the lower part (right wing). Each agent subgroup continuously updates the leader agent in front during the entire flight process. Since the target path point is in the positive x - axis direction, agent i is in the candidate set C i = {j|e ij ∈E,px (i)<p x (j),w i = w j} to determine its leading agent. The distance between two agents can be obtained from formula (5):

[0043]

[0044] In formula (5), p x (i) and p y (i) represent the positions of agent i on the x-axis and y-axis respectively, p x (j) and p y (j) represent the positions of agent j on the x-axis and y-axis respectively, and ξ is a positive weighting factor that can prevent the agent from oscillating between leading agents.

[0045] The leading agent selection algorithm is divided into the following situations:

[0046] 1) O i,w = 0 and C i is an empty set: If C i is an empty set then agent i is at the front of all agents connected to it. Since the agent has been set as a leader (i.e., O i,w = 0), there is no need to change O i,w = 0.

[0047] 2) O i,w ≠ 0 and C i is an empty set: When a non-leading agent serves as a temporary leader, the agent retains its own order O i,w without modification. If an agent directly serves as a leader, it will cause frequent order changes, resulting in unstable oscillations. And an agent that has lost the order 0 will no longer have the opportunity to become a leader.

[0048] 3) O i,w = 0 and C i is not an empty set: That is, when an agent with O i,w = 0 is in the middle of the subgroup. Similar to case 1), the agent is regarded as a lead, so there is no need to change O i,w , unless it encounters a new order 0. If its main agent j ∈ C i has O i,w = 0, O i,w should be changed. Therefore, when O i,w = 0 is modified, O i,w is modified to the following formula (6):

[0049]

[0050] 4) O i,w ≠ 0 and C i is not an empty set: In this case, the agent needs to change its order according to the order of the leading agents. For the leading agent j ∈ Ci, d ij has the minimum value, and the order is changed to the following formula (7):

[0051]

[0052] By analyzing the above four cases, the leading agents in the subgroup can be effectively selected, so as to achieve the purpose of drag reduction control for the extended V - formation of the aircraft.

[0053] The present invention is based on the drag reduction control algorithm for V - type multi - agent aircraft. First, it is combined with the biological dynamics model, which can effectively reduce the drag effect during the flight of the aircraft cluster. Second, the algorithm of the present invention uses multi - agent technology to perform dynamic consensus processing on multiple agent subgroups, making it possible to operate the group separately by aggregating the aircraft cluster into multiple groups for multiple targets. Third, the algorithm of the present invention determines the leading agents by judging the subgroup order and considers the selection strategies of the leading agents in various situations, with high flexibility and a wider application range. Fourth, the algorithm of the present invention can effectively ensure the robustness and scalability of formation flight. Fifth, the present invention continuously updates the leading agents to avoid the global collapse caused by the failure of a single aircraft, and can effectively improve the safety performance of the aircraft.

Claims

1. The drag reduction control algorithm for a V-type multi-agent aircraft is characterized in that Specifically, it includes the following steps: Step 1: First, introduce the V - formation configuration according to the biological dynamics model, obtain the extended V - formation control law, and establish a second - order dynamic system to obtain multiple groups of V - formation control laws; Step 2: According to the extended V - formation control law, divide the intelligent agent cluster into multiple subgroups and conduct multi - agent subgroup dynamic consensus determination; Step 3: Continuously iterate and update the selection of the leader agent of each subgroup in the current state according to the leader agent selection algorithm, so as to conduct drag reduction control based on the V - type multi - agent aircraft; Assume that the communication range between each subgroup of agents is \(R\), and agent \(i\) with intelligence \(\omega\) has a subgroup order \(O\). i,w , which represents the position of the agent in the formation of its designated subgroup; in the leader agent selection algorithm, assume that all agents in a subgroup of agents have the same target waypoint, and their starting positions are aligned with the direction of the waypoint; the algorithm process is as follows: First, set each subgroup order \(O\) of the agents i,w = 0, that is, initialize the subgroup order of all agents to 0. In this algorithm, the agent with the default order of 0 serves as the leader in the subgroup. Therefore, each agent is a potential leader after initialization; assume that the target waypoint is in the positive \(x\)-axis direction of the Cartesian coordinate. When the subgroup order of the agents takes a non-zero value, the algorithm automatically assigns a positive order to the upper part of the leader agent and a negative order to the lower part; each subgroup of agents continuously updates the leader agent in front during the entire flight; since the target waypoint is in the positive \(x\)-axis direction, agent \(i\) is in the candidate set \(C\) i = {j|e ij ∈ E, p x (i) < p x (j), w i = w j}, and determines its leader agent in it; the distance between two agents can be obtained from formula (5): In formula (5), p x (i) and p y (i) represent the positions of agent i on the x-axis and y-axis respectively, and p x (j) and p y (j) represent the positions of agent j on the x-axis and y-axis respectively. ξ is a positive weighting factor that can prevent the agent from oscillating between the leading agents; The leader agent selection algorithm is divided into the following situations: 1) O i,w = 0 and C i is an empty set: if C i is an empty set, then agent i is at the front of all agents connected to it; since the agent has been set as a leader, i.e., O i,w = 0, so there is no need to change O i,w = 0; 2) O i,w ≠ 0 and C i is an empty set: when a non - leading agent assumes a temporary leadership role, the agent retains its own order O i,w No modification is required; if an agent directly assumes the leadership, it will cause frequent order changes, resulting in unstable oscillations; and an agent that has lost order 0 will no longer have the opportunity to become a leader; 3) O i,w = 0 and C i is not an empty set: i.e., when the agent with O i,w = 0 is located in the middle of the subgroup; similar to case 1), the agent is regarded as a lead, so there is no need to change O i,w , unless it encounters a new order 0; If its main agent j ∈ C i has O i,w = 0O i,w = 0, O i,w should be changed; thus, when O i,w = 0 is modified, O i,w is modified to the following formula (6): 4) O i,w ≠ 0 and C i is not an empty set: In this case, the agent needs to change its order according to the order of the dominant agent; For the leading agent j ∈ C i , d ij has the minimum value, and the order is changed to the following formula (7): By analyzing the above four situations, the leader agent in the subgroup can be effectively selected, so as to achieve the purpose of drag reduction control for the extended V - formation of the aircraft.

2. The drag reduction control algorithm for V-type multi-agent aircraft according to claim 1, characterized in that The specific content of Step 1 is: Assume that the aircraft is a second - order dynamic system, defined as the following formula (1): Among them, is the position of the aircraft and is the indicated aircraft speed and t is time, u i (t) is the control input of agent i at time t, and the extended V-shaped formation control law can be derived as shown in the following formula (2): where, is the difference between the actual position and the desired position of agent j, and v j (t) is the velocity of agent j at time t, is defined as the input of, and ω is the intelligence of agent j.

3. The drag reduction control algorithm for V-shaped multi-agent aircraft according to claim 2, characterized in that, In Step 2, according to formula (2), an intelligent agent cluster is divided into multiple intelligent agent subgroups as shown in formula (3):

4. The drag reduction control algorithm for V-type multi-agent aircraft according to claim 3, characterized in that In Step 2, the multi - agent subgroup dynamic consensus determination is specifically: By calculating the limit difference between adjacent intelligent agent subgroups, when the intelligent agents i and j in different subgroups both satisfy the following formula (4), it is determined that the intelligent agent subgroup reaches dynamic consensus: where x i (t) and x j (t) are the positions of agent i and agent j respectively, and v i (t) and v j (t) are the velocities of agent i and agent j respectively.

Images