People / unmanned aerial vehicle cooperative intelligent attack decision-making method and system based on short-distance air support, and unmanned aerial vehicle

Abstract

The invention relates to a manned / unmanned aerial vehicle cooperative intelligent attack decision-making method and system based on short-distance air support and an unmanned aerial vehicle, and belongs to the technical field of manned / unmanned aerial vehicle cooperative decision-making. The method comprises the following steps: constructing a man-machine cooperation DFCM intelligent decision model, carrying the man-machine cooperation DFCM intelligent decision model on an unmanned aerial vehicle, deploying a manned aerial vehicle and unmanned aerial vehicle formation according to a battlefield situation of a close-range air support task, and enabling the unmanned aerial vehicle to directly make a decision action on a target according to an output result of the man-machine cooperation DFCM intelligent decision model. Or making a decision action on the target under the assistance of the manned aerial vehicle, the decision action including attack, exit and standby, and after the unmanned aerial vehicle attacks the target, evaluating and feeding back the combat effect to determine whether to attack the target again. According to the invention, near-distance air support can be implemented by using cooperative intelligent decision of the manned aerial vehicle and the unmanned aerial vehicle, accurate strike on the target is realized, and the problem that the demand of strong confrontation operation cannot be met by only using the manned aerial vehicle to implement support at present is solved.

Description

Technical Field

[0001] The present invention relates to the field of manned / unmanned aerial vehicle (UAV) collaborative decision-making technology, and in particular to a manned / unmanned aerial vehicle (UAV) collaborative intelligent attack decision-making method and system based on close air support, and a UAV. Background Art

[0002] The close air support combat operation process includes multiple steps and tasks, which requires the coordination and cooperation of various departments and relevant personnel to ensure the achievement of combat objectives. At the same time, the process needs to be flexible to respond to various changes and pay attention to ensuring safety and avoiding losses while ensuring combat effectiveness.

[0003] Traditional close air support, relying solely on manned aircraft, can no longer meet the demands of modern, highly contested battlefields. With the continuous development of intelligent technology, unmanned combat platforms, represented by drones, offer advantages such as low cost and versatility. They can play a significant role in reconnaissance and surveillance missions, becoming a crucial component of information warfare. Collaborative operations with manned aircraft can also expand close air support combat modes. In this collaborative operation, intelligent attack decision-making between manned and drone aircraft plays a crucial role.

[0004] Human-machine collaborative intelligent attack decision-making mainly covers two aspects: UAV intelligent attack decision-making and manned limited intervention decision-making. UAV intelligent attack decision-making specifically refers to the ability of drones to autonomously select appropriate maneuvering and firepower strategies in actual combat, based on different battlefield situations and our own different combat intentions, to preserve their own strength while maximizing damage to enemy targets. Manned limited intervention technology refers to the situation where drones request intervention and make suggestions to manned aircraft when drones have difficulty making intelligent decisions. After receiving the drone's intervention request, the manned aircraft will intervene in the drone to assist the drone in completing the mission. By complementing the advantages of UAV intelligent attack decision-making technology and manned limited intervention technology, the combat effectiveness and battlefield response capabilities of drones can be improved, further promoting the development and application of human-machine collaborative intelligent attack decision-making technology.

[0005] Research on collaborative decision-making between manned and unmanned aircraft requires a theoretical foundation based on a range of technologies, including control science and cognitive science, to develop a set of feasible analytical methods and modeling tools. Fuzzy cognitive maps (FCMs), as a qualitative reasoning technique and software calculation method, can be used to establish conceptual models of systems and simulate their behavior. FCMs can simulate the dynamic evolution of a phenomenon or system, predicting future states by modifying the strength of connections between nodes. FCMs have strong data processing capabilities. Furthermore, FCM data typically does not require rigorous mathematical processing, and the modeling approach is simple and flexible, making them highly adaptable and flexible.

[0006] Therefore, based on the FCM model and combined with the changes in combat scenarios of UAVs and manned aircraft in close air support missions, this paper studies the intelligent decision-making method under the collaboration of UAVs and manned aircraft, so that UAVs can successfully complete intelligent attack decisions. Summary of the Invention

[0007] In response to the above-mentioned defects in the prior art, the present invention aims to provide a manned / unmanned aerial vehicle collaborative intelligent attack decision-making method and system and a drone based on close air support to solve the problems existing in the background technology.

[0008] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0009] In a first aspect, the present invention discloses a method for manned / unmanned aerial vehicle (UAV) collaborative intelligent attack decision-making based on close air support, comprising the following steps:

[0010] S1: Deploy manned and unmanned aerial vehicle (UAV) swarm formations based on the battlefield situation for close air support missions;

[0011] S2: Build a human-machine collaborative DFCM intelligent decision-making model and install it on manned and unmanned aircraft;

[0012] S3: The UAV obtains a decision action for the target based on the output of the human-machine collaborative DFCM intelligent decision model. Based on the decision action, it selectively requests the human-machine to assist the UAV and implement the final decision action on the target. The decision actions include attack, exit, and standby.

[0013] S4: After the UAV carries out the target attack, the collected combat effect information will be sent to the manned aircraft through the communication link for evaluation. The manned aircraft will then feedback the evaluation results to the ground command platform, and the next decision-making action on the target will be determined based on the evaluation results.

[0014] As a further preferred solution, in step S1, the steps of constructing a human-machine collaborative DFCM intelligent decision-making model include:

[0015] S11: Determine the input node according to the UAV intelligent attack decision-making mode;

[0016] S12: Input the input node into the FCM model, and design the adjacency weight matrix to be dynamically adjusted. After inference, the output node is dynamically calculated. The output node is represented by C i , which is expressed as follows:

[0017]

[0018] Where w ij (t) represents the adjacency weight matrix, i, j represent nodes 1, 2...n respectively, Represents positive impact, Represents negative impact;

[0019] Among them, the dynamic adjustment formula of the adjacency weight matrix is:

[0020]

[0021] In the formula, V represents the external environment information set, V k ∈V represents an element in V (V e ,V1,V f ), the adjustment coefficient is set to a k , a k ∈[-1,1],w ij (t) represents the value of the adjacent weight at time t, and the adjustment of the adjacent weight is determined by a k and V k Decision, a k Determine the adjustment range, V k Determine the extent to which the external environment dynamically influences decision-making;

[0022] S13: Construct an inference algorithm based on the input nodes and the output nodes, and infer an auxiliary decision node. The inference algorithm is expressed as follows:

[0023]

[0024] Where, represents the state value of node i at time t, represents the weighted value of the state value of node i at time t,

[0025] S14: Utilize the output nodes and auxiliary decision nodes to construct an auxiliary decision model, and obtain the human-machine collaborative DFCM intelligent decision model through multiple iterative reasoning of the auxiliary decision model.

[0026] A further preferred solution is: in step S11, the input nodes include terrain complexity, target attack difficulty, enemy air defense strength, enemy radar performance, and target distance, which are represented by C3 to C7 respectively.

[0027] A further preferred solution is: in step S12, the output nodes include our attack advantage and target threat level, which are represented by C1 and C2 respectively. The result of reasoning and calculating C1 and C2 using formula (1) is as follows:

[0028]

[0029] As a further preferred solution, in step S3, the UAV makes corresponding decision actions according to the output results of the human-machine collaborative DFCM intelligent decision model. The decision actions here are calculated by the following inference function:

[0030]

[0031]

[0032] Where w(C1) and w(C2) represent the dynamic adjacency weights of C1 and C2, respectively. δ and ε are the threshold parameters of the decision results. The values of δ and ε are predetermined or dynamically adjusted according to the actual situation during the task. The value range of δ is [-1, 0], and the value range of ε is [0, 1]. The obtained decision function results are TC (i.e., "exit"), DM (i.e., "stand by"), and GJ (i.e., "attack"), respectively.

[0033] Furthermore, the drone's decision-making action is determined by the output results, specifically:

[0034] When the output result is TC or GJ, the UAV makes an autonomous decision and executes the corresponding decision action;

[0035] If the output result is DM, the UAV sends an intervention request to the manned aircraft and provides auxiliary interference information at the same time. The manned aircraft releases interference to suppress C2 based on the received auxiliary interference information. The intensity of the interference to be released is represented by C8, and C8 is adjusted according to the dynamic changes of C2.

[0036] Furthermore, when the manned aircraft releases interference to change the battlefield situation, and the UAV's decision-making action under the new battlefield situation is still DM, the manned aircraft supplements the UAV with target mission priority information or adjusts the threshold parameters to intervene in the decision-making, specifically:

[0037] When the target threat level is high, the UAV is moved towards TC by increasing the parameter δ; conversely, when the target threat level is low, the UAV is moved towards GJ by decreasing the parameter ε.

[0038] When the target is important, that is, when the target task priority is high, the UAV is moved towards GJ by reducing ε; conversely, when the target task priority is low, the UAV is moved towards TC by increasing δ;

[0039] Among them, when the target task priority is high, the adjustment functions of δ and ε are:

[0040]

[0041] In the formula, δ0 represents the initial value of δ, ε0 represents the initial value of ε, T is the fuzzy value of the target threat level, T0 is the minimum value of T, and the adjustment range of δ is [δ0,δ max ], the adjustment range of ε is [ε0,ε max ].

[0042] In a second aspect, the present invention discloses a manned / unmanned aerial vehicle (UAV) collaborative intelligent attack decision-making system based on close air support, comprising:

[0043] The model building module is used to build a human-machine collaborative DFCM intelligent decision-making model and carry the human-machine collaborative DFCM intelligent decision-making model on the UAV;

[0044] The fighter deployment module plans the manned aircraft positions and UAV routes according to the battlefield situation of the close air support mission, and deploys manned and UAV formations;

[0045] Decision-making action execution module: Based on the output of the human-machine collaborative DFCM intelligent decision-making model, the UAV makes decision-making actions independently or with the assistance of human-machine;

[0046] The evaluation module is used to evaluate the damage caused by the UAV after the attack and send the evaluation results to the manned aircraft or ground command platform to determine whether to attack the target again;

[0047] Among them, the decision-making system is used to implement the above-mentioned manned / unmanned aerial vehicle collaborative intelligent attack decision-making method based on close air support.

[0048] In a third aspect, the present invention discloses a drone on which the above-established human-machine collaborative DFCM intelligent decision-making model is deployed.

[0049] Fourthly, the present invention also discloses the application of drones in the process of manned / drone collaborative intelligent attack decision-making under close air support.

[0050] In addition, a non-transitory computer-readable storage medium is also disclosed, on which a computer program is stored. When the computer program is executed by a processor, it implements the above-mentioned manned / unmanned aerial vehicle collaborative intelligent attack decision-making method based on close air support.

[0051] Compared with the prior art, the present invention can produce the following beneficial effects:

[0052] 1. In view of the dynamic changes in the actual combat environment, this invention is based on the FCM model and optimizes and improves three aspects: the determination of intelligent decision nodes, the design of the inference algorithm, and the design of the dynamic adjustment algorithm of the adjacency weight matrix. The constructed human-machine collaborative DFCM intelligent decision-making model can better analyze dynamic behavior. According to the model data, the drone can independently judge the attack strategy of the target, attack or exit, and complete the drone's attack decision on the target.

[0053] 2. The design of the human-machine collaborative DFCM intelligent decision-making model of the present invention incorporates a limited intervention strategy for manned aircraft on drones. Based on the initial state of the target, the drone will autonomously make corresponding attack decisions. When the drone has difficulty making an attack decision, it will request the manned aircraft to intervene. At this time, the manned aircraft will provide the drone with new situation information based on the received auxiliary intervention information, or update the drone's decision threshold. After receiving the feedback information from the manned aircraft, the drone will update and correct it based on this information to generate a more accurate decision result, so that the drone can successfully complete the intelligent attack decision. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention.

[0055] Figure 1 This is a flowchart of the overall process of the manned / unmanned aerial vehicle collaborative intelligent attack decision-making method under close air support of the present invention;

[0056] Figure 2 This is a modeling design idea diagram of the human-machine collaborative DFCM intelligent decision-making model of the present invention;

[0057] Figure 3 It is an intelligent decision node of the human-machine collaborative DFCM intelligent decision model of the present invention;

[0058] Figure 4 is the target distance advantage function curve under different α values of the present invention;

[0059] Figure 5 The auxiliary decision model of the present invention (iterative reasoning of output nodes and auxiliary decision nodes);

[0060] Figure 6 Graph showing simulation results of iterative reasoning of the decision-making support model of the present invention;

[0061] Figure 7 The adjustment and optimization diagram of the decision-making through human-machine intervention of the present invention;

[0062] Figure 8 A schematic diagram of threshold parameters of the decision result of the present invention;

[0063] Figure 9 This is a flow chart of the intelligent attack decision-making of manned-machine collaborative intervention with unmanned aerial vehicles of the present invention;

[0064] Figure 10 This is a simulation result diagram of the adjustment function of the present invention with a given value. DETAILED DESCRIPTION

[0065] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0066] Reference Figures 1-9 The present invention provides a manned / unmanned aerial vehicle (UAV) collaborative intelligent attack decision-making method and system based on close air support, as well as a UAV. By coordinating close air support missions between manned and unmanned aircraft, precise strikes on targets can be achieved, providing support and vitality guarantee for combat operations, greatly reducing casualties of our front-line troops, and striving to ensure the safety of our front-line troops.

[0067] Reference Figure 1 The present invention discloses a method for intelligent attack decision-making based on manned / unmanned aerial vehicle (UAV) collaboration under close air support, comprising the following steps:

[0068] S1: Deploy manned and unmanned aerial vehicle (UAV) swarm formations based on the battlefield situation for close air support missions;

[0069] S2: Build a dynamic fuzzy cognitive map (DFCM) intelligent decision-making model for human-machine collaboration and install it on manned and unmanned aircraft.

[0070] S3: The UAV obtains a decision action for the target based on the output of the human-machine collaborative DFCM intelligent decision model. Based on the decision action, it selectively requests the human-machine to assist the UAV and implement the final decision action on the target. The decision actions include attack, exit, and standby.

[0071] S4: After the UAV carries out the target attack, the collected combat effect information will be sent to the manned aircraft through the communication link for evaluation. The manned aircraft will then feedback the evaluation results to the ground command platform, and the next decision-making action on the target will be determined based on the evaluation results.

[0072] Specifically, the human-machine collaborative DFCM intelligent decision-making model in step S1 is an extended model based on the improved traditional FCM model. The UAV intelligent attack decision is a dynamic process, which is manifested in that the UAV needs to receive a comprehensive set of external environment information during actual combat and receive the value of the node vector at the same time, thereby dynamically adjusting the DFCM intelligent decision-making model; in this application, if Figure 2 As shown in Figure 2, the steps for building a human-machine collaborative DFCM intelligent decision-making model include the determination of intelligent decision nodes, the design of reasoning algorithms, and the design of dynamic adjustment algorithms for adjacency weight matrices.

[0073] The intelligent decision node is the hub connecting manned and unmanned aerial vehicles in the fuzzy cognitive map model (i.e., FCM model). It can serve as the output of the unmanned aerial vehicle's intelligent decision-making results, passing the decision results to the unmanned aerial vehicle for execution. At the same time, it can also serve as the input of manned and unmanned collaborative decision-making, receiving feedback information and requests from the manned and unmanned aerial vehicle, and updating and correcting them according to the feedback information and requests to generate more accurate decision results.

[0074] like Figure 3 As shown, first, according to the UAV intelligent attack decision-making mode, the intelligent decision node is determined. The intelligent decision node of this application includes input nodes, output nodes and auxiliary decision nodes. The determined intelligent decision nodes and their definitions are shown in Table 1.

[0075] Table 1 Intelligent decision nodes and definitions

[0076]

[0077]

[0078] Among them, C3~C7 are the input nodes of the model, C8 is the auxiliary decision-making node of the model (in this embodiment, it is the interference intensity to be released), and C8 is responsible for reserving an interface for manned-machine intervention auxiliary decision-making; C1 and C2 are the output nodes of the model, C1 is the attack advantage to our side, and C2 is the threat level to the target.

[0079] For the target distance, assume that the range between us and the target for missile launch is R min to R max First, when the distance between the UAV and the target is greater than R max When R=0.5(R max +R min ), the advantage is considered to be the greatest, and the attack is most appropriate at this time. Finally, as the distance between our aircraft and the target decreases further, the advantage will gradually decrease. Therefore, using the fuzzy function with extreme points, we can construct the target distance advantage function of our attack aircraft, which is expressed as follows:

[0080]

[0081] Where, R0=0.5(R max +R min ),δ=α(R max +R min ), α is the fuzzy factor. max =80km, R min =10km, the attack distance advantage function curves under different α values are as follows Figure 4 shown.

[0082] Then input nodes C3~C7 are input into the FCM model to obtain output nodes. At the same time, the adjacency weight matrix in the FCM model is designed to be dynamically adjusted. After inference, the output node is dynamically calculated. The output node is represented by C i , C i The expression is as follows:

[0083]

[0084] Where w ij (t) represents the adjacency weight matrix, i, j represent nodes 1, 2...n respectively, Represents positive impact, Represents negative impact;

[0085] Among them, the dynamic adjustment formula of the adjacency weight matrix is:

[0086]

[0087] In the formula, V represents the external environment information set, V k ∈V represents an element in V (V e ,V1,V f ), the adjustment coefficient is set to a k , a k ∈[-1,1],w ij It is a value that changes dynamically over time, so w ij (t) represents the value of the adjacent weight at time t, and the adjustment of the adjacent weight is determined by a k and V k Decision, a k Determine the adjustment range, V k Determine the degree to which the external environment dynamically influences decision making.

[0088] In the present invention, there are two output nodes, C1 and C2. Among them, C3, C4, C5, and C6 have a negative impact on C1, C7 has a positive impact on C1, and C4, C5, and C6 have a positive impact on C2. The results of reasoning and calculating C1 and C2 using the above formula (1) are as follows:

[0089]

[0090] The above formula (4) has the following characteristics in the reasoning process:

[0091] ① The inference result always remains in [0,1]. The inference result is positively correlated with the comprehensive impact on the node and can be adjusted dynamically.

[0092] ②For example: when the input sequence is C3~C7=[1 1 1 1 0], that is, when all nodes that are not conducive to attack are at their maximum value, the target threat level C2 is the largest, and our attack advantage C1 is the smallest, which is not conducive to attack;

[0093] ③For example: when the input sequence is C3~C7=[0 0 0 0 1], that is, when the nodes that are conducive to attack are at the maximum value, the target threat level C2 is the minimum, and our attack advantage C1 is the maximum, which is conducive to attack.

[0094] Next, based on the interaction between the input nodes and the output nodes, an inference algorithm is constructed to derive the auxiliary decision node (represented by C8). The inference algorithm is expressed as follows:

[0095]

[0096] Where, represents the state value of node number i at time t, Represents the weighted value of the state value of node number i at time t,

[0097] Finally, the output nodes and auxiliary decision nodes are used to construct the auxiliary decision model, such as Figure 5 As shown in the figure, in the auxiliary decision-making model, the increase of C1 will reduce C8, the increase of C2 will increase C8, and C2 will reduce C1. The increase of C8 will increase C1 and reduce C2 at the same time. Through multiple iterative reasoning of the auxiliary decision-making model, the human-machine collaborative DFCM intelligent decision-making model is obtained.

[0098] Figure 6 The evolution of the key nodes C1, C2, and C8 is shown, among which, Figure 6 (a) represents the node state evolution. Figure 6 (b) represents the state difference between node C1 and node C2. Figure 6(b) represents our relative advantage. Set the minimum interference intensity that the drone needs to release to 0.4 units. Figure 6 As can be seen from the figure, the state of the interference intensity C8 that the node needs to release will change dynamically according to the output node C1 (our attack advantage) and the output node C2 (target threat level). When C1 becomes smaller, the release of C8 will increase accordingly, thereby reducing C2; when C1 is relatively large, the release of C8 will weaken, causing C2 to increase. Figure 6 As shown in the figure, the auxiliary decision-making model can simulate the process of suppressing target threats by releasing a certain degree of interference to the UAV. At the same time, the interference intensity C8 to be released is adjusted according to the dynamic changes of C2.

[0099] Before the mission begins, drone data is planned, including weapon selection, route planning, and target information. Once the mission begins, drones are deployed over the designated airspace via manned aircraft. All drones follow the predetermined route to the mission area. Within the target area, intelligent attack decisions are made. During this intelligent attack decision-making process, the drones rely on their sensors and other equipment to identify the target, perceive and assess the combat situation, and decide whether to attack or withdraw. The battlefield situation is then translated into the drone's attack advantage and the target's threat level by the human-machine collaborative DFCM intelligent decision-making model. Simultaneously, the manned aircraft dynamically adjusts the drone's decision based on the status of C1 and C2.

[0100] In step S3, the UAV makes corresponding decision actions according to the output results of the human-machine collaborative DFCM intelligent decision model. When our attack advantage is very large, our UAV combat situation is good, and the output result is biased towards attack. When the target threat level is extremely high, our UAV combat situation is extremely poor, and the output result is biased towards exit. When the output result is uncertain, it will trigger human-machine intervention to provide auxiliary decision support, such as Figure 7 As shown in Figure 1, this ensures that uncertainties are taken into account during the decision-making process, and that the decision is adjusted and optimized through the intervention of human-controlled aircraft. According to the intensity of the interference released by the drone, w(C1) and w(C2) are designed as follows:

[0101]

[0102] The decision action here is calculated by the following inference function:

[0103]

[0104] Where w(C1) and w(C2) represent the dynamic adjacency weights of C1 and C2 respectively, δ and ε are the threshold parameters of the decision result, which determine the output of the decision. The values of δ and ε can be predetermined or dynamically adjusted according to the actual situation during the task. δ∈[δ min ,δ max ],ε∈[ε min ,ε max ], δ max and δ min , ε max and ε min They are used to represent the upper and lower limits of δ and ε respectively. DecisionResult is the decision result of the UAV intelligent attack. TC stands for "exit", DM stands for "uncertain, standby", and GJ stands for "attack". The threshold parameters of the decision result are as follows: Figure 8 shown.

[0105] In this invention, the drone's decision-making action is determined by the output result. When the output result is TC or GJ, the drone makes an autonomous decision and executes the corresponding decision-making action. If the output result is DM, the drone sends an intervention request to the manned aircraft and simultaneously provides auxiliary interference information. The manned aircraft releases interference based on the received auxiliary interference information to suppress C2. C8 is adjusted according to the dynamic changes of C2. The confidence level of the decision result is calculated using the following formula:

[0106]

[0107] In the formula, confidence G and T belong to GJ (attack) and TC (exit) respectively. When the decision result is TC, the closer the value obtained by inference calculation is to δ min , the corresponding confidence is higher; similarly, when the decision result is GJ, the value obtained by inference calculation is closer to ε max , the higher the confidence level.

[0108] like Figure 9 As shown in the figure, the intervention of manned aircraft on UAVs is divided into direct control, intervention environment and intervention decision. The implementation of the intervention method depends on the actual battlefield environment. The purpose is to make full use of the autonomous intelligence of UAVs and reduce the task load of manned aircraft.

[0109] ① Direct control: Directly commanding the drone is the best option when the pilot's mission load is light, and the drone's autonomy level is low. When the drone is in command control mode, the drone provides the pilot with battlefield situation information to assist in analyzing the situation and making attack decisions;

[0110] ② Intervention environment: When the UAV has difficulty making a decision, it will send an intervention request to the manned aircraft and provide auxiliary interference information at the same time. After analyzing the battlefield environment, the UAV will autonomously determine the intensity of the interference that needs to be released. When the interference is released, the combat situation will change. At this time, the human-machine collaborative DFCM intelligent decision-making model will be used to regulate the state of the node and update the UAV's target data and information, thereby affecting the output of the UAV's intelligent decision-making results, achieving limited intervention in the UAV's intelligent attack decision-making;

[0111] ③ Intervention decision: When manned aircraft release interference to change the battlefield situation and UAVs are still unable to make decisions under the new battlefield situation, manned aircraft supplement the UAVs with target mission priority information or adjust threshold parameters to intervene in the decision-making. Specifically:

[0112] When the target threat level is high, the UAV is moved towards TC by increasing the parameter δ; conversely, when the target threat level is low, the UAV is moved towards GJ by decreasing the parameter ε.

[0113] When the target is important, that is, when the target task priority is high, the UAV is moved towards GJ by reducing ε; conversely, when the target task priority is low, the UAV is moved towards TC by increasing δ;

[0114] The target mission priority is a prerequisite. If the target priority is low, the drone chooses TC. The target threat level is an important influencing factor. If the mission priority is high and the threat is large, then DM is selected. If the mission priority is high and the threat is small, then GJ is selected.

[0115] According to the above decision-making principle, the decision result should be TC when the target task priority is low. When the target task priority is high, the adjustment functions of δ and ε are:

[0116]

[0117] In the formula, δ0 represents the initial value of δ, ε0 represents the initial value of ε, T is the fuzzy value of the target threat level, T0 is the minimum value of T, and the adjustment range of δ is [δ0,δ max ], the adjustment range of ε is [ε0,ε max ]. When the target threat level changes, the regulation function is simulated by giving the values of δ and ε. The simulation results are as follows Figure 10 shown.

[0118] Depend on Figure 10As can be seen, as the battlefield situation continues to change, the curves of the drone's attack decision results (δ and ε) divide the entire plane into three regions: GJ, DM, and TC. The human-machine collaborative DFCM intelligent decision-making model of this invention combines our attack advantage with the target's threat level to analyze the drone's current attack status. The decision results will also vary depending on the different regions the drone is located in.

[0119] When the target mission priority is high and the target threat is low, the drone’s intelligent decision-making will fall into the GJ area, thus executing the attack command;

[0120] The drone's attack risk increases as the threat level of the target increases, and its intelligent decision-making approaches the DM zone. At this point, the drone will appropriately reduce its autonomy level, request intervention from the manned aircraft, and await further intervention instructions from the manned aircraft. This maximizes the drone's intelligent decision-making capabilities, ensuring operational safety while improving combat efficiency and accuracy. When the threat level reaches a certain limit and the target mission has a lower priority, the drone's intelligent decision-making will fall into the TC zone, executing a self-protection return. However, standby is only an intermediate state when the drone faces decision-making difficulties; attack or retreat is the final decision. Therefore, if the drone fails to receive timely and correct intervention feedback from the manned aircraft during the waiting period, it may choose to enter the TC zone on its own and execute the retreat decision.

[0121] After the UAV completes the target attack mission, it will report the battle situation to the manned aircraft. The manned aircraft or the ground command platform will evaluate the attack effect to confirm whether another strike is needed. If the attack effect is not good, it will return to step S3 to carry out another strike. If the combat effect is achieved, the UAV will exit.

[0122] Example:

[0123] After the mission begins, our manned aircraft will carry the drones to the designated airspace for deployment. All drones will follow pre-planned routes to the battlefield where they will provide close air support. This ensures that during the mission, the manned aircraft remain in a safe position and close enough to the battlefield to maintain a connection with the drones. Any special circumstances will be promptly reported to the ground command center. In this example, four batches of drones were deployed. After arriving at their respective battlefields, the drones identified their four corresponding targets. The recognition results are shown in Table 2.

[0124] Table 2. UAV state recognition results of attack targets

[0125] Target <![CDATA[C3]]> <![CDATA[C4]]> <![CDATA[C5]]> <![CDATA[C6]]> <![CDATA[C7]]> A 0.9 0.9 0.9 0.9 80 B 0.6 0.4 0.4 0.4 60 C 0.4 0.4 0.6 0.6 55 D 0.09 0.09 0.09 0.09 44

[0126] After the UAV completes the identification of the attack target, it will use the human-machine collaborative DFCM intelligent decision-making model to fuzzify the target's state value according to the current battlefield situation and obtain the inference result, as shown in Table 3. The UAV will perform the corresponding decision action based on the received inference result.

[0127] Table 3 Fuzzification and reasoning results of target state values

[0128] Target <![CDATA[C1]]> <![CDATA[C2]]> ε δ Inference results A 0.20 0.66 0.35 -0.15 -0.98 B 0.42 0.51 0.35 -0.15 -0.10 C 0.47 0.63 0.35 -0.15 0.11 D 0.89 0.09 0.35 -0.15 0.98

[0129] Combining Tables 2 and 3, analyzing the status values of the four targets above reveals that Target A has complex terrain and good radar, resulting in a small range advantage for the drone. Therefore, according to the decision instructions, the drone will choose to withdraw and return home for self-protection. Target D has simple terrain and poor radar, resulting in a large range advantage for the drone, so the drone will choose to attack. For Targets B and C, the drone is unable to make the corresponding intelligent attack decision and will execute the standby decision. At this point, the drone will issue an intervention request to the manned aircraft. The manned aircraft, in response to the drone's request, will release interference to change the battlefield situation. The drone will then make another decision after updating its situation.

[0130] Table 4. UAV intelligent attack decision-making results under human-machine intervention

[0131] Target <![CDATA[C1′]]> <![CDATA[C2′]]> ε′ σ′ Results after intervention B 0.91 0.49 0.25 -0.30 0.82 C 0.64 0.49 0.25 -0.10 0.39

[0132] Tables 3 and 4 show that target B's threat level is 0.51, our attack advantage is 0.42, and the inference result is -0.10. The drone's decision is to standby, requiring human intervention. After the human aircraft releases interference on the target according to the drone's needs, the drone's attack advantage and target threat become 0.91 and 0.49, respectively. The drone's decision output is now to attack.

[0133] After the intervention of the manned aircraft, the UAV still cannot make a decision to attack or withdraw from target C and remains on standby. At this time, the manned aircraft will make the following decisions:

[0134] ① Based on the mission requirements for the target, the manned aircraft directly inputs the withdrawal or strike command to the drone. When the mission requirements are high, the drone is ordered to directly strike the target; when the mission requirements are low, the drone is instructed to withdraw.

[0135] ② Or the manned aircraft can completely take over the UAV and directly control the UAV to attack the target or withdraw.

[0136] After the drone completes its strike on a target, it will provide feedback on the target's status. The manned aircraft will then evaluate and analyze the operational effectiveness based on this feedback. If the close air support mission objective is achieved, it will report mission completion to the ground command center. If the operational effectiveness falls short of the objective, the manned aircraft will report mission failure to the ground command center and inquire whether another strike or support is required.

[0137] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A manned / unmanned aerial vehicle (UAV) collaborative intelligent attack decision-making method based on close air support, characterized by: The following steps are involved: S1: Deploy manned and unmanned aerial vehicle (UAV) swarm formations based on the battlefield situation for close air support missions; S2: Build a human-machine collaborative DFCM intelligent decision-making model and install it on manned and unmanned aircraft; S3: The UAV obtains a decision action for the target based on the output of the human-machine collaborative DFCM intelligent decision model. Based on the decision action, it selectively requests the human-machine to assist the UAV and implement the final decision action on the target. The decision actions include attack, exit, and standby. S4: After the UAV carries out the target attack, the collected combat effect information will be sent to the manned aircraft through the communication link for evaluation. The manned aircraft will then feedback the evaluation results to the ground command platform, and the next decision-making action on the target will be determined based on the evaluation results.

2. The method for intelligent attack decision-making based on manned / unmanned aerial vehicle (UAV) collaboration under close air support according to claim 1 is characterized in that: In step S1, the steps of constructing the human-machine collaborative DFCM intelligent decision model include: S11: Determine the input node according to the UAV intelligent attack decision-making mode; S12: Input the input node into the FCM model, and design the adjacency weight matrix to be dynamically adjusted. After inference, the output node is dynamically calculated. The output node is represented by C i , which is expressed as follows: Where w ij (t) represents the adjacency weight matrix, i, j represent nodes 1, 2...n respectively, Represents positive impact, Represents negative impact; Among them, the dynamic adjustment formula of the adjacency weight matrix is: In the formula, V represents the external environment information set, V k ∈V represents an element in V (V e ,V1,V f ), the adjustment coefficient is set to a k , a k ∈[-1,1],w ij (t) represents the value of the adjacent weight at time t, and the adjustment of the adjacent weight is determined by a k and V k Decision, a k Determine the adjustment range, V k Determine the extent to which the external environment dynamically influences decision-making; S13: Construct an inference algorithm based on the input nodes and the output nodes, and infer an auxiliary decision node. The inference algorithm is expressed as follows: Where, represents the state value of node i at time t, represents the weighted value of the state value of node i at time t, S14: Utilize the output nodes and auxiliary decision nodes to construct an auxiliary decision model, and obtain the human-machine collaborative DFCM intelligent decision model through multiple iterative reasoning of the auxiliary decision model.

3. The method for intelligent attack decision-making based on close air support of manned / unmanned aerial vehicles according to claim 2 is characterized in that: In step S11, the input nodes include terrain complexity, target attack difficulty, enemy air defense strength, enemy radar performance, and target distance, which are represented by C3 to C7 respectively.

4. The method for intelligent attack decision-making based on close air support of manned / unmanned aerial vehicles according to claim 3 is characterized in that: In step S12, the output nodes include our attack advantage and target threat level, which are represented by C1 and C2 respectively. The result of reasoning and calculating C1 and C2 using formula (1) is as follows:

5. The method for intelligent attack decision-making based on close air support of manned / unmanned aerial vehicles according to claim 4 is characterized in that: In step S3, the UAV makes corresponding decision actions according to the output results of the human-machine collaborative DFCM intelligent decision model. The decision actions here are calculated by the following inference function: Where w(C1) and w(C2) represent the dynamic adjacency weights of C1 and C2, respectively. δ and ε are the threshold parameters of the decision results. The values of δ and ε are predetermined or dynamically adjusted according to the actual situation during the task. The value range of δ is [-1, 0], and the value range of ε is [0, 1]. The obtained decision function results are TC, DM, and GJ, respectively.

6. The method for intelligent attack decision-making based on close air support of manned / unmanned aerial vehicles according to claim 5 is characterized in that: The drone's decision-making action is determined by the output results, specifically: When the output result is TC or GJ, the UAV makes an autonomous decision and executes the corresponding decision action; If the output result is DM, the UAV sends an intervention request to the manned aircraft and provides auxiliary interference information at the same time. The manned aircraft releases interference to suppress C2 based on the received auxiliary interference information. The intensity of the interference to be released is represented by C8, and C8 is adjusted according to the dynamic changes of C2.

7. The method for intelligent attack decision-making based on close air support of manned / unmanned aerial vehicles according to claim 6 is characterized in that: When the manned aircraft releases interference to change the battlefield situation and the UAV's decision-making action under the new battlefield situation is still DM, the manned aircraft supplements the UAV with target mission priority information or adjusts the threshold parameters to intervene in the decision-making. Specifically: When C2 is large, the UAV is made to move towards TC by increasing the parameter δ; conversely, when C2 is small, the UAV is made to move towards GJ by decreasing the parameter ε; When the target is important, that is, when the target task priority is high, the UAV is moved towards GJ by reducing ε; conversely, when the target task priority is low, the UAV is moved towards TC by increasing δ; Among them, when the target task priority is high, the adjustment functions of δ and ε are: In the formula, δ0 represents the initial value of δ, ε0 represents the initial value of ε, T is the fuzzy value of the target threat level, T0 is the minimum value of T, and the adjustment range of δ is [δ0,δ max ], the adjustment range of ε is [ε0,ε max ].

8. Based on the manned / unmanned aerial vehicle collaborative intelligent attack decision-making system under close air support, it is characterized by: include: The model building module is used to build a human-machine collaborative DFCM intelligent decision-making model and carry the human-machine collaborative DFCM intelligent decision-making model on the UAV; The fighter deployment module plans the manned aircraft positions and UAV routes according to the battlefield situation of the close air support mission, and deploys manned and UAV formations; Decision-making action execution module: Based on the output of the human-machine collaborative DFCM intelligent decision-making model, the UAV makes decision-making actions independently or with the assistance of human-machine; The evaluation module is used to evaluate the damage caused by the UAV after the attack and send the evaluation results to the manned aircraft or ground command platform to determine whether to attack the target again; The decision-making system is used to implement the manned / unmanned aerial vehicle collaborative intelligent attack decision-making method based on close air support as described in any one of claims 1 to 7.

9. A drone, characterized in that: The UAV is deployed with a human-machine collaborative DFCM intelligent decision-making model established according to claim 4.

10. Application of the UAV according to claim 9 in a process of manned / UAV collaborative intelligent attack decision-making under close air support.

Images