A method for evaluating target threat of unmanned vehicle cluster based on CRITIC and AHP group decision

Abstract

The application provides a UAV cluster target threat evaluation method based on CRITIC and AHP group decision, which fully considers the characteristics of the UAV cluster target, selects indexes capable of describing the overall threat characteristics of the cluster, and uses expert experience to distinguish different UAV cluster types to calculate a threat degree membership function, so that the cluster target threat degree membership index based on expert experience is realized. Through the threat degree index objective weight solving method based on the CRITIC method, the objective weight is determined according to the real-time situation, so as to reduce the influence of the correlation between indexes. In order to realize real-time decision, the application constructs a threat degree index subjective weight model based on the group AHP method in advance for the type of UAV cluster, and fuses the judgments of multiple experts on the weight of the threat degree index. Since the expert experience and knowledge needed to be used are partially completed in advance, the method can give the target threat evaluation result of the UAV cluster in real time.

Description

Technical Field

[0001] This invention relates to the field of target threat assessment technology, specifically to a method for assessing the target threat of unmanned aerial vehicle (UAV) swarms based on CRITIC and AHP swarm decision-making. Background Technology

[0002] Unmanned aerial vehicle (UAV) swarm targets, due to their low cost, high saturation, and strong system-of-systems combat capabilities, have become a new and significant threat in future air defense operations. Threat assessment of UAV swarm targets is a crucial aspect of UAV swarm operations and mission execution, and its research is of great importance.

[0003] Existing research on threat assessment of aerial targets mainly focuses on the threat assessment of individual targets. Commonly used threat indicators include target flight shortcuts, flight altitude, flight speed, and arrival time. However, there is relatively little research on swarm targets. In the literature "Huang Darong, Jiang Hui. Threat assessment model of swarm targets based on information entropy [J]. Computer Engineering and Design, 2010, 31(4): 829-831", the threat level of individuals in the swarm was assessed and ranked. However, this study is essentially still a threat assessment of individual targets and does not assess the swarm as a whole, nor does it consider the overall threat characteristics of the swarm. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention proposes a method for assessing the threat of UAV swarm targets based on CRITIC and AHP swarm decision-making, specifically including the following steps:

[0005] Step 1: Based on the threat characteristics of the swarm targets, determine the threat level indicators of the swarm targets, including swarm route shortcuts, swarm density, number of swarm individuals, average swarm flight altitude, average swarm flight speed, and fastest arrival time of the swarm.

[0006] Step 2: Calculate the threat membership index:

[0007] Using a cluster target threat membership function based on expert experience, the threat value membership degree corresponding to each threat index is calculated;

[0008] Step 3: Calculate the weights of the threat level indicators:

[0009] Using the threat value membership degree obtained in step 2, the objective weight of the threat index is calculated based on the CRITIC method; the subjective weight of the threat index is calculated based on the group AHP method; and the objective weight and subjective weight of the threat index are weighted and summed to obtain the final weight of the threat index.

[0010] Step 4: Using the threat value membership degree obtained in Step 2 and the final weight of the threat index obtained in Step 3, calculate the comprehensive threat evaluation value of the drone swarm target.

[0011] Furthermore, for group route shortcuts, the threat value membership function is:

[0012]

[0013] in Here are the shape parameters of the group route shortcut, where p is the group route shortcut and p0 is the group route shortcut corresponding to a threat level of 0.5.

[0014] For group density, the threat value membership function is:

[0015]

[0016] in For the group density shape parameter, For group density, This represents the group density corresponding to a threat level of 0.5.

[0017] For the number of individuals in the group, the threat value membership function is:

[0018]

[0019] in Let n be the shape parameter representing the number of individuals in the population, and n be the number of individuals in the population. This represents the number of individuals in the group when the threat level is 0.5.

[0020] For the average altitude of the group flight, the threat value membership function is:

[0021]

[0022] in Let h be the average altitude shape parameter of the group flight, and h be the average altitude of the group flight. The average altitude of the group flight corresponds to a threat level of 0.5;

[0023] For the average speed of group flight, the threat value membership function is:

[0024]

[0025] in Let v be the shape parameter of the group's average flight velocity, and v be the group's average flight velocity. The average speed of the group flight when the threat level is 0.5;

[0026] For the fastest arrival time of the group, the threat value membership function is:

[0027]

[0028] in, Let be the shape parameter of the group's fastest arrival time, and t be the group's fastest arrival time. This represents the fastest arrival time of the group when the threat level is 0.5.

[0029] Furthermore, for group route shortcuts, when the group route shortcut p=2km, the threat level u=1; when p=20 km, u=0.5; when p is greater than 50km, u=0.

[0030] For group density, when group density ρ = 1 km, threat level u = 1; when ρ = 5 km, u = 0.6; when ρ ≥ 10 km, u = 0.

[0031] For the number of individuals in the group, the threat level u = 0.1 when the number of individuals in the group n = 1; u = 0.6 when the number of individuals in the group n = 5; and u = 1 when the number of individuals in the group n = 10.

[0032] For the average altitude of the swarm flight, the threat level u = 1 when the average altitude h = 0.1 km; u = 0.5 when h = 5 km; and u = 0 when h ≥ 10 km.

[0033] For the average speed of the group flight, when the average speed of the group flight v ≤ 20 km / h, the threat level u = 0; when v = 100 km / h, u = 0.5; when v ≥ 250 km / h, u = 1.

[0034] For the fastest arrival time of the group, the threat level u=1 when the fastest arrival time t=2 min; u=0.8 when t=5 min; and u=0 when t≥20 min.

[0035] Furthermore, for the fastest arrival time of a group, the arrival time is defined as: Where d represents the distance between the target and our side, and r represents the target's fire range. For attack drones, r is the range of the missiles carried by the drone. For reconnaissance and jamming drones, r = 0.

[0036] Furthermore, in step 3, the objective weight vector of the threat index is calculated based on the CRITIC method. Subjective weight vector of threat index calculated based on group AHP method And perform weighted summation.

[0037]

[0038] The final weight vector of the threat level index is obtained. .

[0039] Furthermore, in step 4, according to the formula

[0040]

[0041] The threat membership values ​​obtained in step 2 are weighted and summed to obtain the comprehensive threat evaluation result of the drone swarm target, where W represents the final weight vector of the threat index. U = [ m p , m r , m n , m h , m v , m t ] This represents a vector of membership values ​​for the six threat indicators.

[0042] Furthermore, based on the comprehensive threat assessment results, the threat levels of drone swarm targets are categorized into different types:

[0043] .

[0044] Beneficial effects

[0045] This invention proposes a threat assessment method for UAV swarm targets based on CRITIC and AHP swarm decision-making. This method fully considers the characteristics of UAV swarm targets, selects indicators that can describe the overall threat characteristics of the swarm, and utilizes expert experience to differentiate between different UAV swarm types to calculate threat membership functions, thus realizing a swarm target threat membership index based on expert experience. Through an objective weighting method for threat indicators based on the CRITIC method, objective weights are determined according to the real-time situation to reduce the influence of correlations between indicators. To achieve real-time decision-making, this invention pre-constructs a subjective weighting model for threat indicators based on the swarm AHP method for different UAV swarm types, integrating the judgments of multiple experts on the weighting of threat indicators. Since the required expert experience and knowledge are completed in advance, this method can provide real-time target threat assessment results for UAV swarms.

[0046] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0047] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0048] Figure 1 Threat assessment process for drone swarm targets;

[0049] Figure 2 : Relationship between threat level and group flight shortcuts;

[0050] Figure 3 : Relationship between threat level and group density;

[0051] Figure 4 : A graph showing the relationship between threat level and the number of individuals in a group;

[0052] Figure 5 : Relationship between threat level and flight altitude;

[0053] Figure 6 : Relationship between threat level and flight speed;

[0054] Figure 7 : Relationship between threat level and fastest arrival time of the swarm;

[0055] Figure 8 Threat assessment results for drone swarm targets. Detailed Implementation

[0056] The embodiments of the present invention are described in detail below. These embodiments are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0057] The proposed method for assessing the threat of unmanned aerial vehicle (UAV) swarm targets based on CRITIC and AHP swarm decision-making includes the following steps:

[0058] Step 1: Based on the threat characteristics of the swarm targets, determine the threat level indicators of the swarm targets, including swarm route shortcuts, swarm density, number of swarm individuals, average swarm flight altitude, average swarm flight speed, and fastest arrival time of the swarm.

[0059] Step 2: Calculate threat membership indices: Using the cluster target threat membership function based on expert experience, calculate the threat value membership corresponding to each threat indices.

[0060] The threat membership curves for different types of targets exhibit significant differences. For example, for the same arrival time, the threat membership of a transport aircraft is clearly different from that of a fighter jet. This indicates that different target types correspond to different membership function curves. Although the specific shapes of the membership curves are not entirely consistent, their changing patterns are generally similar. Taking the arrival time index as an example, when the index value is large, the threat level increases relatively slowly; however, as the index value decreases, the threat level increases exponentially. This shows that the basic shape of the threat membership function curve for arrival time is consistent for different types of targets, with only the shape parameters of the curve differing. Based on this, this paper proposes a cluster target threat membership calculation method based on expert experience. Specifically, experts pre-determine the threat membership values ​​corresponding to some typical values ​​based on different target types, and then solve the parameters of the membership curve through reverse calculation, ultimately obtaining the membership function curve. The specific process is as follows:

[0061] (1) Shortcut to group routes

[0062] The horizontal projection distance of the flight path from the protected target to the geometric center of the cluster is defined as the cluster flight path shortcut. Generally, the smaller the flight path shortcut and the shorter the target's arrival time, the more likely the attacking aircraft is to take action against the protected target, and therefore its threat level should increase accordingly. When the flight path shortcut is zero, the probability of the target attack is highest, and the potential threat to the protected target is strongest; at this point, the threat value is set to 1. When the flight path shortcut reaches a certain value Ka, the threat level is considered to be a. This information is the basis for threat level judgment provided by experts based on target type. Generally speaking, the more threat level judgment values ​​given, the higher the fitting accuracy of the membership function curve.

[0063] Considering that the smaller the shortcut, the greater the threat level, and that the threat level approaches zero when the shortcut is large, a sigmoid curve can be used to fit it.

[0064] The Sigmoid function is a common S-shaped curve function widely used in machine learning, neural networks, and other fields. It has the following characteristics: when the input value is small, the output value is close to 0; when the input value is large, the output value is close to 1; and the middle part transitions smoothly. To meet different needs, the general form of the Sigmoid function can be expressed as:

[0065] (1)

[0066] Where: k is a parameter that controls the steepness of the curve, a larger k value will make the curve steeper; x0 is the midpoint of the curve, that is, when x=x0, the output value is 0.5.

[0067] If the change in threat level follows a reverse S-curve, then the Sigmoid function can be adjusted as follows:

[0068] (2)

[0069] The function approaches 1 when x is small and gradually decreases to 0 as x increases. Clearly, for the group route shortcut indicator, formula (2) is more suitable for representing its threat level. Therefore, the membership function of the group route shortcut is defined as:

[0070]

[0071] in, For the shape parameters of the group route shortcut, These are the group route shortcuts corresponding to a threat level of 0.5. These two parameters can be obtained using nonlinear least squares fitting based on the threat level values ​​provided by experts. p represents the route shortcut, in kilometers (km).

[0072] In this embodiment, based on expert experience, the threat level u=1 when the group route shortcut p=2km; u=0.5 when p=20km; and u=0 when p is greater than 50km. Using the lsqcurvefit function in Matlab for fitting, k=0.4368 and p0=20 were obtained. The threat level membership function curve for the group route shortcut is plotted as follows: Figure 2 As shown.

[0073] (2) Group density

[0074] This invention selects the proximity divergence index to describe the density of individuals in a cluster. Proximity divergence characterizes the density of a cluster by calculating the average minimum distance between individuals at each moment during the cluster's movement. The calculation method is as follows:

[0075] (3)

[0076] in, This represents the minimum distance from individual i to any other individual in the cluster, where n is the total number of individuals in the cluster.

[0077] The higher the swarm density, the higher the threat level. Its influence on threat level is similar to that of shortcut routes and can also be represented by an inverted S-curve. Therefore, the membership function for swarm density is defined as:

[0078] (4)

[0079] in, For the group density shape parameter, ρ represents the swarm density when the threat level is 0.5. These two parameters can be obtained by fitting nonlinear least squares values ​​based on expert-provided threat levels. ρ is the swarm density in kilometers (km).

[0080] In this embodiment, based on expert experience, for drone swarms, the threat level u = 1 when the swarm density ρ = 1 km; u = 0.6 when ρ = 5 km; and u = 0 when ρ ≥ 10 km. Through nonlinear least squares fitting, k = 1.1977 and ρ0 = 5.34 are obtained. The swarm density threat level membership function curve plotted accordingly is shown below. Figure 3 As shown.

[0081] (3) Number of individuals in the group

[0082] The number of individuals in the swarm represents the number of individuals detected in the swarm. The relationship between the number of individuals (n) in the swarm and the overall threat level exhibits an S-shaped growth trend. Therefore, the membership function for the number of individuals in the swarm is defined as:

[0083] (5)

[0084] in, For the shape parameters of the number of individuals in the group, The number of individuals in the population corresponds to a threat level of 0.5. These two parameters can be obtained by fitting the threat level values ​​given by experts using a nonlinear least squares method. n is the number of individuals in the population.

[0085] For example, in the case of an attack drone swarm, the threat level is low when the number of swarm members is small; the threat level increases exponentially when the number of swarm members exceeds 5; and the threat level tends to 1 when the number of swarm members reaches around 10. In this embodiment, based on expert experience, it is believed that when the number of swarm members n=1, the threat level u=0.1; when n=5, u=0.6; and when n=10, u=1. Through nonlinear least squares fitting, k=0.6753 and n0=4.37 are obtained. The swarm member number threat level membership function curve is plotted as follows: Figure 4 As shown.

[0086] (4) Average altitude of group flight

[0087] Due to limitations in communication power and formation requirements, UAV swarms typically fly within a certain altitude layer, and their flight altitude can be represented by the average altitude of the swarm. Generally, the higher the target's flight altitude, the lower the threat level; conversely, the lower the flight altitude, the greater the likelihood of performing low-altitude stealth penetration missions, and therefore the higher the threat level. Thus, the membership function for flight altitude threat is defined as follows:

[0088] (6)

[0089] in, The average altitude shape parameter of the group flight The two parameters are the average altitude of the swarm flight when the threat level is 0.5. These two parameters can be obtained by fitting the threat level values ​​given by experts using the nonlinear least squares method. h is the average altitude of the swarm flight, in kilometers (km).

[0090] In this embodiment, based on expert experience, the threat level u = 1 when the average altitude of the swarm flight is h = 0.1 km; u = 0.5 when h = 5 km; and u = 0 when h ≥ 10 km. Through nonlinear least squares fitting, k = 1.0787 and h0 = 5 are obtained. The resulting swarm flight average altitude threat level membership function curve is shown below. Figure 5 As shown.

[0091] (5) Average speed of group flight

[0092] To maintain formation, the speeds of individual drones in a swarm are usually relatively close. If an individual's speed is significantly higher than that of the others in the swarm, that individual may immediately break away from the swarm and should be analyzed as a separate entity. This is an important manifestation of the variability and dynamism of swarms.

[0093] Generally, the threat level of aerial targets increases exponentially with increasing speed. Therefore, the threat membership function based on flight speed is defined as follows:

[0094] (7)

[0095] in, The shape parameter of the average velocity of the group flight. The two parameters are the average speed of the swarm flight when the threat level is 0.5. These two parameters can be obtained by fitting the threat level value given by experts using the nonlinear least squares method. v is the average speed of the swarm flight, in kilometers per hour (km / h).

[0096] Higher flight speeds generally indicate that the aircraft may be high-value targets such as missile swarms or attack swarms, or advanced supersonic aircraft or hypersonic reconnaissance aircraft. In either case, it signifies a higher threat level. In this embodiment, based on expert experience, the threat level u = 0 when the average speed of the swarm flight v ≤ 20 km / h; u = 0.5 when v = 100 km / h; and u = 1 when v ≥ 250 km / h. Through nonlinear least squares fitting, k = 0.0574 and v0 = 100 are obtained. The membership function curve of the threat level based on the average speed of the swarm flight is shown below. Figure 6 As shown.

[0097] (6) Fastest arrival time of the group

[0098] The fastest arrival time of an individual in a cluster reflects the reaction time available to our forces and is an important indicator of threat level. Generally, the shorter the arrival time of an aerial target, the higher the threat level. Therefore, the threat membership function based on arrival time is defined as follows:

[0099] (8)

[0100] in, The shape parameter is the fastest arrival time of the group. The fastest arrival time of the swarm is denoted by t, which represents the swarm's fastest arrival time when the threat level is 0.5. These two parameters can be obtained by fitting the threat level values ​​provided by experts using a nonlinear least squares method. t represents the swarm's fastest arrival time in minutes (min).

[0101] In this embodiment, based on expert experience, the threat level u=1 when the fastest arrival time of the swarm is t=2 min; u=0.8 when t=5 min; and u=0 when t≥20 min. Through nonlinear least squares fitting, k=1.256 and t0=6.1 are obtained. The swarm's fastest arrival time threat membership function curve is plotted as follows: Figure 7 As shown.

[0102] It is important to note that the arrival time should be the result of dividing the target's distance from the threat area by its speed *v*, and cannot be set to a negative value. This is because if set to a negative value, the threat level will increase abruptly when the target suddenly changes direction, rather than changing continuously. As the target approaches, the threat level increases exponentially with shorter arrival times; conversely, the threat level increases slowly with greater distance. Therefore, the arrival time is defined as: Where d represents the distance between the target and our side, and r represents the target's firing range. For attack drones, r is represented by the range of the main missiles they are equipped with; for reconnaissance / jamming drones, r = 0.

[0103] Step 3: Calculate the weights of the threat index: Using the threat value membership degree obtained in Step 2, calculate the objective weights of the threat index based on the CRITIC method; calculate the subjective weights of the threat index based on the group AHP method; and sum the objective and subjective weights of the threat index to obtain the final weights of the threat index.

[0104] Among the six indicators in threat assessment, the fastest arrival time of a swarm is correlated with the average flight speed and the shortest flight path (generally, the faster the flight speed and the shorter the shortest flight path, the shorter the arrival time). Furthermore, there may also be a correlation between the number of swarm members and the swarm density (generally, the more members, the higher the density). However, these are all critical indicators, and none can be arbitrarily removed. For this type of highly correlated weight calculation problem, the CRITIC method (CRiteria Importance Through IntercriteriaCorrelation) is more suitable. Therefore, this invention selects this method to obtain objective weights.

[0105] The CRITIC method is an objective weighting method proposed by Diakoulaki

[10] in 1995. This method calculates the objective weight of indicators by measuring the contrast strength of threat indicators and the conflict between indicators. The contrast strength reflects the difference between different targets of the same indicator, and is expressed in the form of standard deviation. The larger the standard deviation, the greater the threat difference between the targets, and the higher the corresponding weight. The conflict between indicators is expressed by the correlation coefficient. If there is a strong positive correlation between two indicators, the conflict is smaller, and the corresponding weight is lower. The CRITIC method is suitable for situations where there is a correlation between indicators, indicators are both correlated and need to be analyzed independently, and the data fluctuates significantly.

[0106] (1) Dimensionless data

[0107] Standardize the raw data to eliminate differences in dimensions and orders of magnitude between different indicators, ensuring comparability among them. It is important to note that standardization should avoid using the method of subtracting the mean from the raw value and then dividing by the standard deviation (z-score), as this eliminates differences in standard deviation and is detrimental to subsequent calculations of comparative strength.

[0108] For positive indicators (the larger the indicator value, the higher the threat level):

[0109] (9)

[0110] in, Let max(j) represent the j-th index value of the i-th target, where i = 1, 2, ..., m, j = 1, 2, ..., n; ) and min( ) are the maximum and minimum values ​​of the j-th indicator, respectively; .

[0111] For negative indicators (the smaller the indicator value, the higher the threat level):

[0112] (10)

[0113] In this paper, the values ​​of each indicator have been converted to the [0,1] interval through threat membership calculation, thus completing the standardization. Since the positive and negative aspects of the indicators have been considered when calculating the threat membership, it is only necessary to calculate according to formula (9).

[0114] (2) Calculate the contrast intensity

[0115] Contrast intensity is expressed in terms of standard deviation, S j The standard deviation of the j-th indicator is:

[0116] (11)

[0117] in, Let j be the mean of the index. .

[0118] (3) Calculate conflict

[0119] Conflictability is a measure of the correlation between indicators, expressed by a correlation coefficient. The closer the correlation coefficient is to 1, the stronger the correlation between the two indicators, meaning they provide more similar information and therefore have a smaller role in threat assessment. In the CRITIC method, conflictability is represented by subtracting the sum of the correlation coefficients from 1:

[0120] (12)

[0121] (13)

[0122] Where n is the number of indicators, This represents the correlation coefficient between index i and index j.

[0123] (4) Calculate the amount of information

[0124] Information content is the product of contrast strength and conflict level, representing the importance of each indicator in threat assessment. The greater the information content, the greater the role of the indicator in threat assessment.

[0125] (14)

[0126] (5) Calculate the objective weights

[0127] The information content is normalized to obtain the weights of each indicator. The normalization formula is:

[0128] (15)

[0129] Commonly used methods for determining subjective weights include the Analytic Hierarchy Process (AHP), Group AHP, pecking order graph method, and Best-Worst Weighting (BWM). Subjective weights reflect experts' subjective judgments on the contribution of the six indicators to the threat level. To improve the scientific rigor and reliability of decision-making, this paper considers using the Group AHP method to determine subjective weights. This process can be completed before real-time threat assessment, so that the subjective weight information formed by experts can be directly utilized during the real-time assessment.

[0130] When implementing the group AHP method, 5 to 10 experts (including pilots, commanders, technical experts, and other personnel from different fields) can be selected to score the indicators based on their respective professional knowledge and experience. For AHP group decision problems, Saaty (1980) and Aczel and Saaty (1983) pointed out that the weighted geometric mean (WGMM) method is the only group aggregation method that satisfies unimodality, homogeneity, and reciprocity.

[0131] set up A [ k ] = ( a ij [ k ] ) This is the judgment matrix provided by the k-th decision-maker when comparing n elements, where i, j=1,2,…,n. oh [ k ] = ( oh 1 [ k ] , oh 2 [ k ] , ⋯ , oh n [ k ] ) Its weight vector ( oh i [ k ] ≥ 0 , ∑ i = 1 n oh i [ k ] = 1 ),and The weight of the kth decision-maker in the group ( , Using WGMM as the aggregation process, the group decision matrix and group weight vector are as follows:

[0132] (16)

[0133] in, a ij G = ∏ k = 1 m ( a ij [ k ] ) β k ( i , j = 1 , … , n ) .

[0134] (17)

[0135] in, oh i G = ∏ k = 1 m ( oh i [ k ] ) β k ( i , j = 1 , … , n ) .

[0136] When using the Row Geometric Mean Method (RGMM) to determine weights, the final weights of the alternative methods can be obtained through two aggregation methods (AIJ and AIP).

[0137] (1) AIJ method: from the individual judgment matrix A [ k ] Starting from this point, the group judgment matrix is ​​obtained using the WGMM method (i.e., formula (16)). Then, the group weights are calculated using the RGMM method. .

[0138] (2) AIP method: judging the matrix of individuals A [ k ] Starting with RGMM, we obtain individual weights. oh [ k ] Then, the group weights are obtained by using WGMM aggregation (i.e., formula (17)). .

[0139] Barzilai and Golan (1994) proved that when using RGMM to determine weights, the weight results obtained by the two aggregation methods are the same. Therefore, Escobar et al. (2004) suggested that the simpler method of the two methods can be chosen to determine the weights, such as the AIP method.

[0140] In AHP (Analytical Power of Choice), it is often necessary to test the consistency of expert judgments. The eigenvalue method (EM) and the geometric mean row method (RGMM) are two of the most widely used methods for determining weights, employing the consistency index (CI) proposed by Saaty for EM and the geometric consistency index (GCI) proposed by Aguarón and Moreno-Jiménez for RGMM, respectively. The expressions for these two indices are:

[0141] (18)

[0142] (19)

[0143] in, , This is the weight vector.

[0144] Corresponding to Saaty's threshold for the EM method (CR < 0.1), Aguarón and Moreno-Jiménez gave the following thresholds for GCI: for n=3, GCI < 0.31; for n=4, GCI < 0.35; for n>4, GCI < 0.37. Indicators representing the coherence of a group G C I [ k ] This represents the consistency index of the judgment of the k-th decision-maker.

[0145] The specific steps of the group AHP method are as follows:

[0146] Step a: Invite experts: Invite m experts to perform pairwise comparisons of the six threat indicators of the drone swarm target to obtain their respective judgment matrices. A [ k ] , k=1,2,…,m.

[0147] Step b: Determine individual weights and perform consistency checks: Use RGMM to obtain individual weights. oh [ k ] = ( oh 1 [ k ] , oh 2 [ k ] , … , oh 6 [ k ] ) and consistency indicators G C I [ k ] .like G C I [ k ] ≥ 0 . 3 7 If the consistency of the judgment matrix given by the k-th decision-maker is unacceptable, then consistency adjustment is required. The adjustment method can be found in references [26, 27].

[0148] Step c: Aggregate to obtain group weights: Using the AIP method, the group weights are obtained through WGMM aggregation (i.e., formula (17)). and group consistency index Among them, the weights of m experts The appropriate level of expertise can be determined using methods such as Analytic Hierarchy Process (AHP) or grey relational analysis, based on the experts' experience and cognitive abilities. When the experts' experience and cognitive abilities are similar, one approach can be taken. =1 / m, k=1,2,…,m.

[0149] Finally, the objective weights obtained above will be used... With subjective weight By performing a weighted summation, the final weight of the threat level index can be obtained:

[0150] (20)

[0151] Where α is the influence factor of subjective weight, and its value ranges from (0, 1).

[0152] The objective weights of the threat level indicators are closely related to the real-time status of the cluster targets and are dynamically adjusted according to changes in the target status. The subjective weights, on the other hand, are predetermined by experts and are usually kept constant. By combining these objective and subjective weights, the influence of changes in the target cluster's situation is reflected, while also incorporating the experience and judgment of experts.

[0153] Step 4: Using the threat value membership degree obtained in Step 2 and the final weight of the threat index obtained in Step 3, calculate the comprehensive threat evaluation value of the drone swarm target.

[0154] By weighted summing of the threat membership values ​​obtained in step 2, the comprehensive threat assessment result of the drone swarm target can be obtained:

[0155] (twenty one)

[0156] Where W represents the final weight of the threat level indicator. U = [ m p , m r , m n , m h , m v , m t ] This represents the membership value of the six threat indicators.

[0157] Based on the comprehensive threat assessment results above, the threat levels of drone swarm targets are categorized into different types, as shown in Table 1.

[0158] Table 1 Threat Level Classification of Unmanned Aerial Vehicle (UAV) Swarms

[0159]

[0160] Specific implementation examples are given below:

[0161] Suppose that at a certain moment, there are 5 drone swarm targets. The known conditions are shown in Table 2. Based on the method described above, the threat level of each swarm target is assessed.

[0162] Table 2. Known conditions for drone swarms

[0163]

[0164] 1. Determine objective weights

[0165] (1) Calculate the threat membership index:

[0166] Using the expert-based cluster target threat membership function (formulas (1)~(8)), the membership matrix corresponding to each threat feature is calculated:

[0167] U = [ m p , m r , m n , m h , m v , m t ] = [ 0.98750.83270.97820.74620.50000.0074 0.95510.90060.60480.89640.99900.7992 0.97050.31210.92070.96220.24090.0000 0.98080.60040.95800.83451.00000.9332 0.99940.99700.99920.99230.94630.7992 ]

[0168] The data is dimensionless and standardized using formula (9) to obtain the standardized membership matrix:

[0169] U ' = [ 0.7308 0.7602 0.9466 0 0.3414 0.0079 0 0.8593 0 0.6102 0.9987 0.8564 0.3481 0 0.8008 0.8777 0 0 0.5790 0.4210 0.8954 0.3588 1 1 1 1 1 1 0.9293 0.8564 ]

[0170] (2) Calculate the contrast intensity

[0171] Based on the standardized membership matrix, the contrast strength of each indicator is calculated using formula (11):

[0172] S = [ 0.38, 0.401, 0.414, 0.403, 0.458, 0.497 ]

[0173] (3) Calculate conflict

[0174] Using formulas (12) and (13), the conflict between indicators is calculated:

[0175] R = [ 2.786, 2.326, 2.289, 3.555, 2.001, 2.068 ]

[0176] (4) Using formula (14), calculate the information content of each indicator:

[0177] C = [ 1.059, 0.937, 0.947, 1.439, 0.917, 1.027 ]

[0178] (5) Using formula (15), the objective weights of each indicator are calculated:

[0179] oh o = [ 0.1676 ， 0.1478 ， 0.15 ， 0.227 ， 0.1451 ， 0.1626 ]

[0180] The calculation results show that the objective weight of the average altitude of the group flight is relatively high, while the objective weight of the average speed of the group flight is relatively low.

[0181] 2. Determine subjective weights

[0182] Five experts from different fields (such as pilots, commanders, and technical experts) were invited to conduct pairwise comparisons of the importance of cluster threat indicators, resulting in the following judgment matrix:

[0183]

[0184] A 4 = [ 1 1 / 2 1 1 / 3 1 / 4 1 / 5 2 1 1 / 2 1 / 2 1 / 2 1 / 3 1 2 1 1 1 / 3 1 / 2 3 2 1 1 3 1 / 3 4 2 3 1 / 3 1 1 / 2 5 3 2 3 2 1 ] , A 5 = [ 1 1 / 3 1 / 2 1 / 5 1 / 5 1 / 7 3 1 1 1 / 3 1 / 2 1 / 3 2 1 1 2 2 1 5 3 1 / 2 1 1 / 3 1 / 3 5 2 1 / 2 3 1 1 7 3 1 3 1 1 ]

[0185] The weight and consistency index of each expert were calculated using the RGMM method. The calculation results are shown in Table 3.

[0186] Table 3. Calculation results of individual expert weights and consensus indices.

[0187]

[0188] As can be seen from Table 3, the judgment matrix given by the five experts meets the requirements in terms of consistency ratio (CR < 0.1) and geometric consistency index (GCI < 0.37), and therefore can be used to calculate the group weight index.

[0189] (1) When adopting the AIJ method

[0190] The individual judgment matrices are aggregated using the WGMM method (Formula (16)) to obtain the group judgment matrix A. G Subsequently, the group weights ω were calculated using the RGMM method. G The calculation result is:

[0191] A G = [ 1 0 . 699 0 . 407 0 . 257 0 . 334 0 . 217 1 . 431 1 0 . 608 0 . 488 0 . 461 0 . 407 2 . 460 1 . 644 1 1 0 . 699 0 . 660 3 . 898 2 . 048 1 1 1 . 320 0 . 450 2 . 993 2 . 169 1 . 431 0 . 758 1 0 . 660 4 . 618 2 . 460 1 . 516 2 . 221 1 . 516 1 ]

[0192] oh G = [ 0.0619 , 0.0974 , 0.1647 , 0.1924 , 0.1917 , 0.292 ]

[0193] (2) When adopting the AIP method

[0194] Using the individual weight data in Table 3, substitute them into formula (17) and aggregate them using the WGMM method to obtain the group weight ω. G The calculation result is:

[0195] oh G = [ 0.0619 , 0.0974 , 0.1647 , 0.1924 , 0.1917 , 0.292 ]

[0196] Both methods (AIJ and AIP) yielded the same subjective weighting results. The results show that the fastest arrival time of the swarm, the average altitude of the swarm, and the average speed of the swarm have the greatest impact on threat level, while the swarm route shortcut has the least impact. This is consistent with expert experience, indicating that the swarm AHP method used is scientifically sound.

[0197] 3. Final weight calculation

[0198] According to formula (20), assuming the influence factor α of subjective weight is 0.6, the final weights of each threat level indicator are calculated as follows:

[0199] W = 0 . 6 × oh G + ( 1 − 0 . 6 ) × oh o = [ 0.1042, 0.1175, 0.1588, 0.2062, 0.1731, 0.2402 ]

[0200] 4. Calculate the comprehensive threat assessment value of the drone swarm target.

[0201] Using formula (21) and the final weights, the comprehensive threat level evaluation value of each UAV cluster target is calculated. Based on the comprehensive evaluation value, the threat levels of each cluster target are ranked, and the results are as follows:

[0202] P = W ⋅ U T =[0.598, 0.851, 0.524, 0.894, 0.940]

[0203] Threat ranking: Cluster 5 > Cluster 4 > Cluster 2 > Cluster 1 > Cluster 3.

[0204] As can be seen from the standardized membership matrix U′, cluster 5 has high membership values ​​for all threat levels, thus its overall threat level is the highest. Clusters 5, 2, and 4 have overall threat levels between 0.75 and 1.0, and are classified as high threat; clusters 1 and 3 have overall threat levels between 0.5 and 0.75, and are classified as medium threat. Among these, cluster 3 has the lowest overall threat level due to its low membership values ​​for swarm density, average swarm flight speed, and fastest arrival time. The threat assessment results are as follows: Figure 8 As shown.

[0205] This invention conducts in-depth research on the target threat assessment problem of UAV swarms, proposes a threat index system suitable for UAV swarm targets, and establishes a method for calculating the threat membership degree of swarm targets based on expert experience. To achieve real-time threat assessment of UAV swarm targets, this invention proposes a method for determining the objective weights of threat indicators based on real-time situational awareness—the CRITIC method. This method effectively reduces the influence of correlations between threat indicators and can dynamically adjust the objective weights of threat indicators according to real-time situational awareness. Furthermore, to make the threat index weights of UAV swarms more closely reflect actual needs, this invention introduces a subjective weight determination method for threat indicators based on the swarm AHP method. Finally, by combining subjective and objective weights, the final weights of the threat indicators are obtained, and a comprehensive score is given for the threat level of UAV swarm targets, achieving real-time threat classification.

[0206] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention without departing from the principles and spirit of the present invention.

Claims

1. A method for assessing the threat of unmanned aerial vehicle (UAV) swarm targets based on CRITIC and AHP swarm decision-making, characterized in that: Includes the following steps: Step 1: Based on the threat characteristics of the swarm targets, determine the threat level indicators of the swarm targets, including swarm route shortcuts, swarm density, number of swarm individuals, average swarm flight altitude, average swarm flight speed, and fastest arrival time of the swarm. Step 2: Calculate the threat membership index: Using a cluster target threat membership function based on expert experience, the threat value membership degree corresponding to each threat index is calculated; Step 3: Calculate the weights of the threat level indicators: Using the threat value membership degree obtained in step 2, the objective weight of the threat index is calculated based on the CRITIC method; the subjective weight of the threat index is calculated based on the group AHP method. The objective and subjective weights of the threat level index are weighted and summed to obtain the final weight of the threat level index. Step 4: Using the threat value membership degree obtained in Step 2 and the final weight of the threat index obtained in Step 3, calculate the comprehensive threat evaluation value of the drone swarm target.

2. The method for assessing the threat of unmanned aerial vehicle (UAV) swarm targets based on CRITIC and AHP swarm decision-making as described in claim 1, characterized in that: For group route shortcuts, the threat value membership function is: in Here are the shape parameters of the group route shortcut, where p is the group route shortcut and p0 is the group route shortcut corresponding to a threat level of 0.

5. For group density, the threat value membership function is: in For the group density shape parameter, For group density, This represents the group density corresponding to a threat level of 0.

5. For the number of individuals in the group, the threat value membership function is: in Let n be the shape parameter representing the number of individuals in the population, and n be the number of individuals in the population. This represents the number of individuals in the group when the threat level is 0.

5. For the average altitude of group flights, the threat value membership function is: in Let h be the average altitude shape parameter of the group flight, and h be the average altitude of the group flight. The average altitude of the group flight corresponds to a threat level of 0.5; For the average speed of group flight, the threat value membership function is: in Let v be the shape parameter of the group's average flight velocity, and v be the group's average flight velocity. The average speed of the group flight corresponding to a threat level of 0.5; For the fastest arrival time of the group, the threat value membership function is: in, Let be the shape parameter of the group's fastest arrival time, and t be the group's fastest arrival time. This represents the fastest arrival time of the group when the threat level is 0.

5.

3. The method for assessing the threat of unmanned aerial vehicle (UAV) swarm targets based on CRITIC and AHP swarm decision-making as described in claim 2, characterized in that: For group route shortcuts, the threat level u=1 when the group route shortcut p=2km; u=0.5 when p=20km; and u=0 when p is greater than 50km. For group density, when group density ρ = 1 km, threat level u = 1; when ρ = 5 km, u = 0.6; when ρ ≥ 10 km, u = 0. For the number of individuals in the group, the threat level u = 0.1 when the number of individuals in the group n = 1; u = 0.6 when n = 5; and u = 1 when n = 10. For the average altitude of the swarm flight, the threat level u = 1 when the average altitude h = 0.1 km; u = 0.5 when h = 5 km; and u = 0 when h ≥ 10 km. For the average speed of the group flight, when the average speed of the group flight v ≤ 20 km / h, the threat level u = 0; when v = 100 km / h, u = 0.5; when v ≥ 250 km / h, u = 1. For the fastest arrival time of the group, the threat level u=1 when the fastest arrival time t=2 min; u=0.8 when t=5 min; and u=0 when t≥20 min.

4. The method for assessing the threat of unmanned aerial vehicle (UAV) swarm targets based on CRITIC and AHP swarm decision-making as described in claim 2, characterized in that: For the fastest arrival time of a group, the arrival time is defined as: Where d represents the distance between the target and our side, and r represents the target's fire range. For attack drones, r is the range of the missiles carried by the drone. For reconnaissance and jamming drones, r = 0.

5. The method for assessing the threat of unmanned aerial vehicle (UAV) swarm targets based on CRITIC and AHP swarm decision-making as described in claim 1, characterized in that: In step 3, the objective weight vector of the threat index is calculated based on the CRITIC method. Subjective weight vector of threat index calculated based on group AHP method And perform weighted summation. The final weight vector of the threat level index is obtained. .

6. A method for assessing the threat of unmanned aerial vehicle (UAV) swarm targets based on CRITIC and AHP swarm decision-making, as described in claim 5, is characterized in that: In step 4, according to the formula The threat membership values ​​obtained in step 2 are weighted and summed to obtain the comprehensive threat evaluation result of the drone swarm target, where W represents the final weight vector of the threat index. This represents a vector of membership values ​​for the six threat indicators.

7. A method for assessing the threat of unmanned aerial vehicle (UAV) swarm targets based on CRITIC and AHP swarm decision-making, as described in claim 6, is characterized in that: Based on the comprehensive threat assessment results, the threat levels of drone swarm targets are categorized into different types: 。

Images