Mathematical Problems in Engineering

Mathematical Problems in Engineering / 2018 / Article

Research Article | Open Access

Volume 2018 |Article ID 6481635 | 14 pages | https://doi.org/10.1155/2018/6481635

Improved Ant Colony Optimization for Weapon-Target Assignment

Academic Editor: Mengqi Hu
Received24 May 2018
Revised08 Sep 2018
Accepted26 Sep 2018
Published11 Oct 2018

Abstract

Weapon-target assignment (WTA) which is crucial in cooperative air combat explores assigning weapons to targets with the objective of minimizing the threats from those targets. Based on threat functions, there are four WTA models constrained by the payload and other tactical requirements established. The improvements of ant colony optimization are integrated with respect to the rules of path selection, pheromone update, and pheromone concentration interval, and algorithm AScomp is proposed based on the elite strategy of ant colony optimization (ASrank). We add garbage ants to ASrank; when the pheromone is updated, the elite ants are rewarded and the garbage ants are punished. A WTA algorithm is designed based on the improved ant colony optimization (WIACO). For the purpose of demonstration of WIACO in air combat, a real-time WTA simulation algorithm (RWSA) is proposed to provide the results of average damage, damage rate, and kill ratio. The following conclusions are drawn: (1) the third WTA model, considering the threats of both sides and hit probabilities, is the most effective among the four; (2) compared to the traditional ant colony algorithm, the WIACO requires fewer iterations and avoids local optima more effectively; and (3) WTA is better conducted when any fighter is shot down or any fighter’s missiles run out than along with the flight.

1. Introduction

Weapon-target assignment (WTA) is a dynamic multivariable and multiconstraint problem, which is characterized by antagonism, initiative, and uncertainty. So far, there are bulks of studies on the solution to WTA, such as the use of genetic algorithm (GA), simulated annealing (SA), and particle swarm optimization (PSO) algorithm. Additionally, many scholars use ant colony optimization.

In [1, 2], GA was used to solve the problem. In GA, a population of individuals, which encode the problem solutions, are manipulated according to their fitness values through genetic operators, such as reproduction, mutation, and crossover. GA has delightful global searching ability and can find all feasible solutions. However, its local searching ability is poor; to be exact, it is prone to premature convergence. Moreover, it is time consuming as well. Reference [3, 4] used SA in WTA. SA starts from a high initial temperature, and as the temperature falls down, the global optimal solution is found randomly; even when the searching falls into a local optimal solution, SA has a probability to jump out and eventually goes to the global optimum. But this algorithm converges slowly and takes time. PSO was applied to WTA in [5, 6]. PSO builds a swarm intelligence model, initializes a set of random solutions, and searches for the optimal solution by iterations. PSO is simple to implement and converges fast, but it is easy to fall into local optima.

In addition, reference [7, 8] developed a novel multiobjective optimization method based on the evolutionary game theory in real-time WTA. Darryl et al. [9] proved that the dynamic programming method could also solve WTA problem. Liang et al. [10] presented an objective optimization approach based on clonal selection algorithm to solve the problems of WTA in warship formation antiaircraft application. Based on the auction algorithm, Fei et al. [11] brought forward a new distributed multiaircraft cooperative fire assignment method. Considering the complexity and strict time constraints, Sahin et al. [12] proposed a fuzzy decision method to aid commanders in making decisions for WTA.

As is known, ant colony optimization has the characteristics of distributed computing, self-organization, and positive feedback. The complicated WTA process in air combat can be mapped to ant foraging behavior.

The ant system was first proposed by Dorigo [13] in 1991 in his doctoral thesis. In 1994, Lumer and Faita [14] took the idea of ant colony clustering to data analysis and proposed the LF algorithm. Considering the good performance of ant colony optimization in solving discrete combinatorial optimization problems, Lee et al. [15] first applied it to WTA in 2002. The basic ant colony algorithm was applied to the target assignment problem of the air defense C3I systems by Huang et al. [16] in 2005. In recent years, ant colony optimization was widely used. For example, in 2013, Olmo et al. [17] used it in the research on association task of data mining, and the results obtained were very exciting. In 2015, Ariyasingha et al. [18] analyzed the performance of multiobjective ant colony optimization for the traveling salesman problem and concluded that the algorithm performed better in problems with more than two objectives and its performance depended slightly on the number of objectives, iterations, and ants. Lately, in 2017, Li et al. [19] designed a biobjective WTA optimization model which maximized the expected damage of the enemy and minimized the cost of missiles; a modified Pareto ant colony optimization algorithm was used in the solution, which produced better results than two multiobjective optimization algorithms NSGA-II and SPEA-II.

Generally, WTA in air combat aims to minimize the threat from opponent fighters, in other words, to maximize the defused threat, constrained by the payload and other tactical requirements. This paper focuses on the WTA modeling, solution, and simulation in air combat scenario. Based on threat functions, four WTA models are established. Among them, Model 3 is proposed for the first time considering the threats of both sides and hit probabilities. For the solution to the WTA models, a WTA algorithm based on improved ant colony optimization (WIACO) is designed, integrating the improvements of traditional ant colony optimization with respect to the rules of path selection, pheromone update, and pheromone concentration interval, and proposes algorithm . Through a comparative experiment, it is concluded that WIACO requires fewer iterations than traditional ant colony algorithm, and it avoids local optima more effectively. Furthermore, in order to demonstrate and exemplify the effectiveness of WIACO in air combat, a real-time WTA simulation algorithm (RWSA) is presented to simulate real-time WTA in air combat, with the results of average damage, damage rates, and kill ratios.

The paper is organized as follows: Section 2 discusses the air combat threat functions and four WTA models. In Section 3, the WIACO algorithm is introduced, and then in Section 4 a comparative analysis is offered. The RWSA algorithm, three experiments, and result analysis are given in Section 5. Finally, the conclusions of this paper are drawn in Section 6.

2. Weapon-Target Assignment Model

The objective of WTA is to maximize the expected impact on the opponents and to minimize the risk we face [20] in terms of threat. In this case, we need to measure the threat in air combat first.

2.1. Air Combat Threat

It is assumed that the red side has fighters and the blue side has fighters and that the red side’s early warning aircraft can accurately identify the model, speed, spatial position, and other basic information of the blue fighters. This section uses the air combat situation in reference [21] as shown in Figure 1 for threat modeling.

In Figure 1, is the th red fighter, is the th blue fighter, is the direction of , is the speed of , is the off-axis angle of relative to , is the distance between and , and the other parameters are defined similarly.

2.1.1. Angle Threat Function

The angle threat function [22] is given as follows:where is the angle threat of to , with and -1≤ 1. In particular, when the off-axis angle is 180° and the off-axis angle is 0°, that is, when chases from behind, the angle threat of to is 1.

2.1.2. Distance Threat Function

The distance threat function [23] is given as follows:where represents the missile range of the blue fighter and is the maximum detection range of the blue radar. When the red fighter is within the blue attack range, the distance threat of to takes the maximum value of 1; when the red fighter is out of the detection range of blue, the distance threat takes the minimum value of 0.

2.1.3. Speed Threat Function

The speed threat function [22] is given as follows:where and are the speeds of the red and blue fighters, respectively. The greater the than the , the smaller the threat of to .

2.1.4. Ability Threat Function

In this paper, we embrace the air combat capability formula in [24] given as follows:where is the maneuverability parameter of the fighter; is the fire attack capability parameter; is the radar detection capability parameter; is the pilot's control capability coefficient; is the fighter survivability coefficient; is the fighter range coefficient; and is the Electronic Counter Measures (ECM) capability coefficient. The ability threat function is defined as [22]:

For example, supposing the air combat capability of is 17.9, if the capabilities of , , and are 15.8, 18.8, and 17.9, the ability threat degrees of blue to red are 0.25, 0.75, and 0.5, respectively.

As a combination of the above threat functions, the threat degree of to is

where   (i = 1,2,3,4) is the weight, = 1.

The overall threat degree of to all the red fighters is

2.2. Weapon-Target Assignment

Let carry missiles and the number of missiles assigned to :

Four WTA models are presented in this subsection.

Model 1. where , an integer from 0 to 3, represents the missile number of assigned to attack ; is the threat of to ; is the number of red missiles actually used to attack the target; is the number of red fighters; and is the number of blue fighters. The first constraint indicates that the number of missiles launched by each red fighter cannot exceed its carrying capacity; the second constraint implies that the number of missiles assigned to each blue fighter cannot exceed the value obtained in formula (8); the third constraint indicates that the number of fired missiles takes the smaller value between the total number of missiles carried by the red side and the total number of missiles assigned to the blue side.

Model 2. Model 4 was first proposed in reference [25]. The constraints are the same as in Model 1, and is the probability that hits , which is calculated with the two-step adjudication model [26] in this paper. Compared to Model 1, the objective here takes the hit probability of the blue fighters into account; the larger the , the greater the threat.

Model 3. The constraints are the same as in Model 1, and is the threat of to . Model 3 considers the blue side threat while also considering the advantages of the red side for the blue side. The objective signifies that red missiles will be preferentially assigned to the blue fighters which pose high threat to red, get high threat from red, and have high hit probability.

Model 4. Model 4 was first proposed in reference [27], where is the overall hit probability of the red fighter on ; is the threshold of hit probability of ; and is the penalty factor. Model 4 calculates the average hit probability for each missile, maximizing the threat of each missile resolution. At the same time, the minimum hit probability requirement is set for each blue fighter, and if is lower than , it will be punished. The objective here is to maximize the average hit probability, considering the blue fighter’s hit probability threshold.

3. Improvements of the Ant Colony Optimization for Weapon-Target Assignment

To solve the problem of WTA, some researches tried to make improvements on traditional ant colony optimization with respect to the rules of path selection, pheromone update, and pheromone concentration interval. We consider basic ant colony algorithm (ACO), ant system (AS), elitist-rank ACO algorithm (), and max-min ant system (MMAS) and improved the elite strategy in , named . We add garbage ants to the algorithm of ; when the pheromone is updated, the elite ants are rewarded and the garbage ants are punished, as seen in the following sections (Sections 3.13.3). Section 3.4 integrates these improvements in the application of air combat WTA and proposes a WTA algorithm based on the integration.

3.1. Path Selection Rule

In the basic ACO, the path is selected according to the probability , calculated as follows:

To avoid search stagnation, the path selection in AS [2830] uses a combination of deterministic and random selections and dynamically adjusts the state transition probability in searching. The specific path selection rules are as follows:where represents the th red fighter and , , are indices of blue fighters, respectively; is determined by , the same as the ACO,

where

τ(, ) is the pheromone concentration on the path between the current missile and the assigned position .

η(, ) is a heuristic function; the greater the threat of the opponent, the greater the probability of launching missiles.

is the information heuristic factor used to measure the influence of pheromones on the path.

is the expected heuristic factor used to measure the influence of the threat degree.

allowedk represents the set of available targets. As the search progresses, allowedk is getting smaller.

(i) is the set of nodes that the th ant needs to access after node has been accessed.

q is a uniform distributed random number in , and is a constant, .

is the probability of selecting .

When is less than or is equal to , the path with the highest pheromone concentration is selected. When is greater than , the selection probability of each node is obtained by formula (14).

3.2. Pheromone Update Rule

In the ASO, the pheromone is updated after all ants have iterated, and the concentration of all passing paths is updated, which is called partial update:where △τ(, ) is the pheromone concentration increment on the optimal path (, ); ρ is pheromone volatilization rate, ρ (0,1); is a constant used to regulate the pheromone concentration; and is the path length of the ant .

In AS, both partial update and global update are performed; global update is the same as formula (16). Only the pheromone on the optimal path is updated in the global update [31, 32] to enhance the effect of positive feedback. The update rules are as follows:where △τ is the pheromone concentration increment on the optimal path (, ) and is the shortest path length of the current cycle.

Traditional algorithms may lead to the elimination of the most adapted ants, and is proposed to preserve them [33]. can find better solutions and find these solutions for a shorter period. Set elite ants; the global update rules for are as follows:

Next, the is improved to make it easier to find the optimal path. Garbage ants are added to the algorithm . When the pheromone is updated, the elite ants are rewarded while punishing the garbage ants. To prevent the penalty from causing the pheromone concentration to be too low, the penalty factor is set. The improved AS formula is as follows (named ):where is the same as formula (21) and is the number of ant garbage ants.

3.3. Pheromone Concentration Interval Rule

(1) In MMAS, to avoid local stagnation in searching, the pheromone concentration interval is set as follows [34]:where , are two constants used to regulate the pheromone concentration.

(2) Another way to adjust pheromone is to smooth the concentration. By increasing the probability of selecting low pheromone paths, the ability to explore new solutions can be improved:where (i, j) is the amount of pheromone after smoothing; is a constant used to regulate the concentration; and is the same as formula (24).

3.4. Algorithm Performance Comparison

Considering the rules of path selection, pheromone update, and pheromone concentration interval, 24 sets of algorithms are obtained, as shown in Table 1. These algorithms are brought into the WTA problem for performance comparison analysis. The average optimal solution and convergence of the 100 trials are counted.


Path Selection RulePheromone Update RulePheromone Interval

Pij of basic ACOPartial update of ACONone

Random selections of ASPartial and global update of ASMMAS’s interval

Elite ant strategy of Smooth the concentration

Punish rule of

Algorithm 1 selects of basic ACO, partial update of ACO, and none. Algorithm 2 selects of basic ACO, partial update of ACO, and MMAS’s interval, and other algorithms are selected according to the same method. The test results are as shown in Table 2.


Algorithm123456

convergence21.3619.6019.0314.434.4316
optimal15.2715.4515.4319.7715.9620.05

Algorithm789101112

convergence234.3018.5312.903.0312.56
optimal19.6814.3519.98320.1013.1720.03

Algorithm131415161718

convergence5.666.106.2317.3616.8615.76
optimal14.2314.2214.2118.1219.5217.98

Algorithm192021222324

convergence23.101522.3316.239.767.13
optimal18.6118.4018.4919.7517.9920.15

Call the threat model;
Initialize parameters:
1. While  NC - 1<NCmax (The largest iteration number)
2. For  i=1:m (The number of ants)
3. Put m ants on red missiles randomly;
4. End
5. For  k=1:count (The total number of ants’ movements)
6.For j=1:m
7. For i=1:Kb (Number of attackable blue fighters)
8. Generate a random number q;
9.Select path with pseudorandom probability using formula ((14), (15));
10.Add the target to selected set;
11.Assignable target number minus 1;
12.The number of remaining missiles in red fighters minus 1;
13.allowedk minus the assigned target;
14.End
15.End
16.For  j=1:m
17.For  i=1:Nrl (The remaining amount of red missiles)
18.Generate a random number q;
19.Select path with pseudorandom probability using formula ((14), (15));
20.Add the target to selected set;
21.allowedk minus the assigned target;
22.End
23.End
24.Determine the path of the maximum threat in this cycle using WTA model;
25.Global update of pheromone using formula ((22), (23));
26.Update the pheromone interval on the paths using formula (26);
27.NC=NC+1;
28.End
29.End
30.Get the optimal WTA solution.
1.For  time=1:T
2. Call the WIACO algorithm;
3.For  i=1:Nr
4. For  j=1:Nb
5. If  Dij<Ti (Dij is the distance between two sides, Ti is the range of the red fighters)
Call the two-step adjudication model to get the probability of damage Pr, Pb;
6. End
7. End
8. End
9.
11. For k_r=1:Nr
12. Get random number Pkr;
13. If Pkr<Pr
14.Red fighter k_r is shot down;
15. End
16. End
17. For  k_b=1:Nb
18. Get random number Pkb;
19. If  Pkb<Pb
20. Blue fighter k_b is shot down;
21. End
22. End
23Update the air combat situation;
24. If one termination condition is met, end the loop;
25. time=time+1;
26End
37Output simulation results.

As can be seen from Table 2, the average optimal solution of algorithm 24 is the largest, which is 20.15. So, it can resolve the largest threat, and the average convergence is 7.13. Algorithm 11 has the worst effect, and the optimal solution is only 13.17, which is the easiest to fall the local optimal solution.

Algorithm 24 selects rules of random selections of AS, punishes the rule of , smooths the concentration, and achieves the best results. Algorithm 24 is used as the research algorithm of this paper.

3.5. WIACO Algorithm

When the ant colony optimization is used to solve the WTA problem, the assignment process needs to be modeled with an ant colony network. For example, in Figure 2, each red missile is represented by a small node and each red and blue fighter is represented by a big node; the two sides both dispatch two fighters; red fighter 1 carries four missiles and red fighter 2 carries three; blue fighter 1 is assigned two missiles and blue fighter 2 is assigned one missile.

The ants follow the path from the red nodes to the blue nodes, and then, according to the same strategy, take a virtual path back to the red nodes until the assignment is completed.

The number of ants in the population is set to [35]:where is the total number of missiles carried by red fighters and is the number of missiles assigned to . In the beginning of the iteration, ants are placed randomly on the red missiles, and the initial pheromone concentration on each path is set to 1.

The ants move according to the following rules:

Rule 1. An ant can only move to a blue fighter whose missile assignment is insufficient; red fighters launch the remaining number of missiles at most.

Rule 2. Each ant can only reach one node at a time; that is, each missile can only attack a single target.

Rule 3. The ants do not interfere with each other. Ants return to red fighter nodes with the same pseudorandom probability, and the targets are the red fighters who still have missiles.

Rule 4. All ants need to update the pheromone by the end of a cycle and generate new pheromones only on the optimal path; pheromone is partially volatile.

Pseudocode of the WIACO algorithm is shown as in algorithm 1.

4. Examples of Comparative Analysis

4.1. Model Parameter Analysis

The parameters of the WTA model called by WIACO algorithm are discussed with a control variable experiment in this subsection, including the , , , , and . Set to 0.001 and set to 0.1. The number of iterations is 200, and the means of the results are taken.

Taking Model 3 in Section 2.2 for example, the parameter variation curves of Model 3 are shown in Figure 3. The horizontal axis represents the parameter; the vertical axis represents the maximum defused threat. As we can see in Figure 3, when = 1.5, = 3, = 0.7, Q = 0.4, and = 0.7, the curves reach their respective maximum defused threat. Therefore, Model 3 takes the final parameters as = 1.5, = 3, = 0.7, Q = 0.4, and = 0.7. Similarly, the final parameters of the other three models are set as in Table 3.


ParameterModel 1Model 2Model 3Model 4

1.51.51.51.5
3232
0.40.70.70.6
0.70.50.40.6
0.60.70.70.5

The obtained parameters are substituted into the models for the experimental analysis, and the results of one of these experiments are as in Figure 4. The horizontal axis represents the iterations; the vertical axis represents the maximum defused threat.

Repeat the experiment 100 times and take the average for analysis. The results are presented in Table 4. As seen in Table 4, the variances of the four models are not large and the iterations converge at the 20th iteration, earlier or later, and that is acceptable.


MaxMinMeanVar

Model 117.289617.283117.28590.0021
Model 212.218611.836812.12670.0109
Model 322.120718.685820.45730.5852
Model 46.087645.88626.05330.0024

4.2. Comparison with Traditional Algorithm

It is assumed that the red side adopts WIACO algorithm with WTA Model 3.

According to Table 5 and the air combat capability in formula (4), the air combat capability of each red fighter is = 20.79 and that of each blue fighter is = 21.97.


ParameterRedBlue

Number of fighters88
Number of missiles44
Flight speed (m/s)340340
Missile range (km)8090
Radar detection distance (km)120120
Maneuvering parameter25.525.65
Fire parameter27612832
Detection capability parameter15141553
Manipulation efficiency factor0.90.95
Viability coefficient0.9150.995
Voyage coefficient1.141.14
Electronic countermeasure coefficient1.051.1

Based on reference [25], assuming the fighter performance has the greatest impact on air combat, we let = 0.2, = 0.2, = 0.2, and = 0.4 and take the value of =3, =40.

The traditional ant colony optimization and the WIACO proposed in this paper are both simulated 100 times, and the best convergence results of the two algorithms are obtained.

The convergence of the traditional algorithm is shown in Figure 5(a); thereinto, the results fluctuate. After the algorithm goes through the maximum defused threat degree of 26.06, it falls into the local optimum and gets a stable defused threat degree of 25.9. In light of this, the traditional algorithm cannot jump out of local optimum, and it does not produce the global optimal WTA solution.

The WIACO convergence is shown in Figure 5(b). Figure 5(b) shows that at the 3rd iteration the improved algorithm finds a locally optimal assignment with the defused threat degree of 26.19; then it jumps out of the local optimum quickly and converges to a stable optimal assignment with a maximum defused threat degree of 26.29 at the 12th iteration. Table 6 shows the WTA solution in detail. For example, as we can see on the first row of Table 6, the red fighter 1 launches two missiles to attack the blue fighter 1, one missile to attack blue fighter 2, and one missile to attack the blue fighter 6, and the missiles all run out; on the seventh row, none of the red fighter 7’s missiles are assigned.


Red fighterAssignment (missiles)Remaining missiles

Red1Blue1 ; Blue2 ; Blue6 0
Red2Blue3 ; Blue6 0
Red3Blue5 ; Blue4 1
Red4Blue5 ; Blue4 1
Red5Blue4 ; Blue71
Red6Blue4 3
Red74
Red8Blue6 ; Blue8 0

The above analysis indicates the advantages of the WIACO algorithm, which can provide better solution than traditional algorithm for the WTA. Comparatively speaking, it can be considered that the traditional algorithm takes longer time to convergence, and it is harder to jump out of local optimum.

5. Simulation Analysis of WTA in Air Combat

During air combat, the combat situation is constantly changing, and the WTA needs to be continuously updated. This section simulates the complete air combat process, identifies the final effectiveness, and makes relevant experimental analysis.

5.1. Simulation Strategy

Assuming the red side performs WTA using the WIACO algorithm, the air combat simulation in this section is based on the following rules:

Rule 1. Due to the early warning aircraft, air combat situation is accessed in real time. During the simulation, both sides constantly approach each other. To reflect the randomness and flexibility of the formation and the air combat situation, the fighters move with a random and constant speed under their respective limits. Assuming that both sides are constantly adjusting their direction of flight, the off-axis angles of two side fighters are decreasing [36].

Rule 2. Once the blue fighters move into range, the red fighters launch their missiles.

Rule 3. The data are put into the two-step adjudication model [26] to calculate the effectiveness.

Rule 4. The termination conditions of the simulation are as follows: one side runs out of missiles; all fighters on either side are shot down; and the air combat move beyond the horizon (within 20 km).

The simulation steps are as follows:

Step 1. Obtain the initial data of the red and blue sides and initialize the parameters.

Step 2. Call the WIACO algorithm.

Step 3. Determine whether or not the blue fighters are within range of red side. Once they are within range, launch the red missiles and call the two-step adjudication model to assess the effectiveness.

Step 4. Update the air combat situation.

Step 5. Repeat Steps 24 and output the final effectiveness when one of the termination conditions is reached.

In the simulations, the number of red fighters is and that of blue fighters is , and the number of simulations is . The pseudocode of the RWSA algorithm is given as shown in Algorithm 2.

5.2. Experiment 1

In experiment 1, both sides dispatch 8 fighters. The parameters used in the two-step adjudication model are shown in Table 7.


ParameterRedBlue

Command factor0.60.7
Fighter vulnerability0.450.4
Multitarget attack capability44
Sorting ratios11
Allow war to match0.80.8
Fighter radar reflection cross section (m2)3.22
Probability of target discovery0.70.7
Missile score0.80.8
Airborne radar maximum discovery distance (km)120120
Airborne missile maximum effective range (km)8090

Because of the randomness of fighter kill and air combat situation, each simulation result is different, which reflects the uncertainty of air combat. Table 8 lists the results of one simulation as an example. As seen in Table 8, by the end of the simulation, 4 red fighters (3,6,7, and 8) are shot down, 4 red fighters (1,2,4, and 5) run out of missiles, 3 blue fighters (1,3, and 7) are shot down, and 2 blue fighters (2 and 4) run out of missiles.


RedBlue

clock 6Fighters 6 and 8 shot down; fighters 4 and 5 out of missilesFighters 1 and 7 shot down; Fighter 4 out of missiles

clock 8Fighter 3 shot down; fighter 2 out of missilesFighter 3 shot down

clock 15Fighter 7 shot down; 1 out of missilesFighter 2 out of missiles

totalShot down: 4; Missiles exhausted: 4Shot down: 3, Missiles exhausted: 2

In this paper, we run a large number of simulations and statistical results with formula ((28)-(30)). Denote the amount of damage of red side in the th simulation with Lir and that of blue side with Lib, and the damage rates of red and blue sides, Pr and Pb, are calculated with formula ((28), (29)):

The kill ratio is calculated with formula (23):

The smaller the kill ratio is, the greater the advantage of the red side is in air combat.

First, we run 1000 times simulations using Model 1, and the relationship between the kill ratio and the number of iterations is shown in Figure 6. Figure 6 shows that, as the number of iterations increases, the kill ratio tends to stabilize. The results of the air combat are shown in Figure 7 and Table 9. From Figure 7 we can see that 4 to 6 red fighters are killed mostly and 2 to 4 blue fighters are killed in most cases. The kill ratio in Table 9 is 1.6921; that is, the blue side is ascendant.


RedBlue

Average damage7.39454.1864
Damage rate0.92430.5233
Kill ratio1.7663

And then additional experiments are conducted using Models 2, 3, and 4 (M = 100, = 0.2), respectively. Among the results of the four simulations seen in Tables 912, the kill ratio of Model 3 is the smallest, so Model 3 gets the best effectiveness for WTA in air combat.


RedBlue

Average damage 7.28214.3632
Damage rate0.91030.5454
Kill ratio1.6690


RedBlue

Average damage 7.24924. 5326
Damage rate0.90620.5665
Kill ratio1.5993


RedBlue

Average damage7.32154. 0854
Damage rate0.91510.5106
Kill ratio1.7921

5.3. Experiment 2

In the air combat simulations, there are two possible options of WTA timing. The first is to reassign at each time step of the simulation so that the WTA can be adjusted according to the real-time air combat situation. However, this increases the requirements of the pilots’ target locating capability. The second option is to reassign when any fighter is shot down or any fighter’s missiles are used up. This option allows the pilot to focus on the attacks on located targets but reduces the ability to adapt to the battlefield.

Experiment 2 is carried out using the above two options, where parameters are the same as in experiment 1. The results are shown in Tables 13 and 14. As seen in Tables 13 and 14, there is not much difference in the kill ratio between option 1 and option 2, but the red damage rate in option 2 is lower. Therefore, option 2 is more suitable for WTA regarding survivability and economics.


RedBlue

Average damage 4.52122.8727
Damage rate0.56520.3591
Kill ratio1.5738


RedBlue

Average damage 3.91222.4820
Damage rate0.48900.3103
Kill ratio1.5762

5.4. Experiment 3

Given the number of blue fighters, experiment 3 studies how many red fighters should be dispatched. The number of blue fighters is fixed at 8, while the number of red fighters increases from 3 to 18. The parameters are the same as in experiment 1. A total of 1000 simulations are performed to calculate the kill ratio, and the results are shown in Table 15 and Figure 8.


Number of red fighters3456
Kill ratio3.07262.68452.45871.9752

Number of red fighters78910
Kill ratio1.68721.61541.58861.5754

Number of red fighters11121314
Kill ratio1.57431.51591.52451.5088

Number of red fighters15161718
Kill ratio1.50511.49121.49211.5011

In Table 15 and Figure 8, as the number of red fighters’ increases, the kill ratio declines; however, the downward trend weakens after 12. This implies that the red side achieves a satisfying combat effectiveness if it dispatches 12 fighters to cope with 8 blue fighters at a kill ratio of about 1.5159.

6. Conclusions and Future Directions

In the air combat, the battlefield situation is complex and changeable, and WTA plays a decisive role. In Section 2 of this paper, four threat functions are used to evaluate the threat. Meanwhile, four WTA models are set up with different objectives. Among them, Model 3 is proposed for the first time considering the threats of both sides and hit probabilities.

In order to solve the WTA problem, WIACO algorithm is presented in Section 3 with improvements of the traditional ant colony optimization in the aspects of path selection rule, pheromone updating rule, and pheromone concentration interval rule. The comparative experiment in Section 4 shows that WIACO algorithm which provides the optimal solution for WTA has the advantages of faster convergence and better avoidance from local optima.

For the sake of the demonstration and exemplification of WIACO in air combat, four combat simulation rules and RWSA algorithm are designed in Section 5. Based on this, this section carries out three simulation experiments. Through experiment 1, we can find that Model 3 gets the minimal kill ratio indicating the largest advantage of red side. Hence, we can conclude that Model 3 gets the best effectiveness for WTA in air combat. In experiment 2, we analyze different WTA timings with results showing that WTA is better conducted when air combat situation changes (i.e., any fighter is shot down or any fighter’s missiles are used up) than along with the flight. Finally, experiment 3 shows that, when the blue dispatches 8 fighters, 12 red fighters shall be dispatched accordingly. When the number of red fighters exceeds 12, the decrease in kill ratio is not obvious, if it is not increasing.

In general, from the advantages exemplified by the simulation experiments, it can be concluded that the improved ant colony optimization proposed in this paper can be applied to WTA in air combat.

As future work, we intend to apply other intelligent algorithms to the WTA problem and compare it with the improved ant colony algorithm to further explore the best solution for WTA. At the same time, the air combat simulation process needs to be refined. The current simulation hypothesis is relatively simple. The next step is to make the simulation process closer to actual combat and to make the simulation results more practical.

Notations

:Number of red fighters
:Number of blue fighters
:The th red fighter
:The th blue fighter
:The direction of
:The speed of
:The off-axis angle of relative to
:The distance between and
:The threat of to
:The missile ranges of the blue fighter
:The maximum detection ranges of the blue radar
:The speeds of the red fighter
:The speeds of the blue fighter
:The fire attack capability parameter
:The radar detection capability parameter
:The maneuverability parameter of the fighter
:The air combat capability
:The pilot’s control capability coefficient
:The fighter survivability coefficient
:The fighter range coefficient
:The Electronic Counter Measures capability coefficient
:The number of missiles of
:The number of missiles assigned to
:The missile number of assigned to attack
:The number of red missiles actually used to attack
:The probability that hits
:The overall hit probability of the red fighter on
:The threshold of hit probability of
:The path selection probability
:The pheromone concentration
:The heuristic function
:The information heuristic factor
:The expected heuristic factor
: The set of available targets
:The set of nodes that the th ant needs to access
:A random number in
:A constant,
:The pheromone volatilization rate
:A constant used to regulate the pheromone concentration
:The path length of the ant
:The shortest path length of the current cycle
:The number of ant garbage ants
:A constant used to regulate the pheromone concentration
:A constant used to regulate the concentration
:The penalty factor
:The total number of missiles carried by red fighters
:The amount of damage of red side in the th simulation
:The amount of damage of blue side in the th simulation
:The damage rate of red side
:The damage rate of blue side
:The kill ratio.

Data Availability

The data used to support the findings of this study are included within the supplementary information file(s). Link: https://figshare.com/s/433d02b1d101aa94301d

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.

References

  1. Z.-J. Lee, S.-F. Su, and C.-Y. Lee, “Efficiently solving general weapon-target assignment problem by genetic algorithms with greedy eugenics,” IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics, vol. 33, no. 1, pp. 113–121, 2003. View at: Publisher Site | Google Scholar
  2. M. A. Sahin and K. Leblebicioglu, “A genetic algorithm for weapon target assignment problem,” in Proceedings of the Summer Computer Simulation Conference Society for Modeling Simulation International, pp. 61–65, 2009. View at: Google Scholar
  3. Y. Li and Y. Dong, “Weapon-target assignment based on simulated annealing and discrete particle swarm optimization in cooperative air combat,” Acta Aeronautica et Astronautica Sinica, vol. 31, no. 3, pp. 626–631, 2010. View at: Google Scholar
  4. Z. Ding, D.-W. Ma, M.-D. Tang, and X.-F. Zhang, “TSAPSO: a hybrid search algorithm of tabu search and annealing particle swarm optimization for weapon-target assignment,” Xitong Fangzhen Xuebao / Journal of System Simulation, vol. 18, no. 9, pp. 2480–2483, 2006. View at: Google Scholar
  5. M. C. Su, “A PSO-based Decision Aid for Multi-Aircraft Combat Situations,” in Proceedings of the International Journal of Fuzzy Systems, vol. 10, pp. 161–167, 2010. View at: Google Scholar
  6. X. Liu, Z. Liu, W. S. Hou, and J. H. Xu, “Improved MOPSO algorithm for multi-objective programming model of weapon-target assignment,” Systems Engineering & Electronics, vol. 35, no. 2, pp. 326–330, 2013. View at: Google Scholar
  7. C. Leboucher, H.-S. Shin, S. Le Ménec et al., “Novel evolutionary game based multi-objective optimisation for dynamic weapon target assignment,” in Proceedings of the 19th IFAC World Congress on International Federation of Automatic Control, IFAC 2014, pp. 3936–3941, South Africa, August 2014. View at: Google Scholar
  8. Z. Yao, M. Li, Z. Chen, and R. Zhou, “Mission decision-making method of multi-aircraft cooperatively attacking multi-target based on game theoretic framework,” Chinese Journal of Aeronautics, vol. 29, no. 6, pp. 1685–1694, 2016. View at: Publisher Site | Google Scholar
  9. D. K. Ahner and C. R. Parson, “Optimal multi-stage allocation of weapons to targets using adaptive dynamic programming,” Optimization Letters, vol. 9, no. 8, pp. 1689–1701, 2015. View at: Publisher Site | Google Scholar | MathSciNet
  10. H. Liang and F. Kang, “Adaptive chaos parallel clonal selection algorithm for objective optimization in WTA application,” Optik - International Journal for Light and Electron Optics, vol. 127, no. 6, pp. 3459–3465, 2016. View at: Publisher Site | Google Scholar
  11. A. G. Fei, L. Y. Zhang, and Q. J. Ding, “Multi-aircraft cooperative fire assignment based on auction algorithm,” Systems Engineering & Electronics, vol. 34, no. 9, pp. 1829–1833, 2012. View at: Google Scholar
  12. M. A. Şahin and K. Leblebicioğlu, “Approximating the optimal mapping for weapon target assignment by fuzzy reasoning,” Information Sciences, vol. 255, pp. 30–44, 2014. View at: Publisher Site | Google Scholar
  13. A. Colorni, M. Dorigo, and V. Maniezzo, “Distributed Optimization by Ant Colonies,” in Proceedings of the Ecal91 - European Conference on Artificial Life, 1992. View at: Google Scholar
  14. E. D. Lumer and B. Faiet, “Diversity and adaptation in populations of clustering ants,” in Proceedings of the third international conference on Simulation of adaptive behavior : from animals to animats 3: from animals to animats 3, pp. 501–508, 1994. View at: Google Scholar
  15. Z.-J. Lee, C.-Y. Lee, and S.-F. Su, “An immunity-based ant colony optimization algorithm for solving weapon-target assignment problem,” Applied Soft Computing, vol. 2, no. 1, pp. 39–47, 2002. View at: Publisher Site | Google Scholar
  16. S. C. Huang and W. M. Li, “Research of Ant Colony Algorithm for Solving Target Assignment Problem,” Journal of Systems Engineering and Electronics, vol. 27, no. 1, pp. 79-80, 2005. View at: Google Scholar
  17. J. L. Olmo, J. M. Luna, J. R. Romero, and S. Ventura, “Mining association rules with single and multi-objective grammar guided ant programming,” Integrated Computer-Aided Engineering, vol. 20, no. 3, pp. 217–234, 2013. View at: Publisher Site | Google Scholar
  18. I. D. I. D. Ariyasingha and T. G. I. Fernando, “Performance analysis of the multi-objective ant colony optimization algorithms for the traveling salesman problem,” Swarm and Evolutionary Computation, vol. 23, pp. 11–26, 2015. View at: Publisher Site | Google Scholar
  19. Y. Li, Y. Kou, Z. Li, A. Xu, and Y. Chang, “A Modified Pareto Ant Colony Optimization Approach to Solve Biobjective Weapon-Target Assignment Problem,” International Journal of Aerospace Engineering, vol. 2017, 2017. View at: Google Scholar
  20. L.-M. Zhang, Y. Zhang, and W.-B. Liu, “The design of target assignment model based on the reverse mutation ant colony algorithm,” in Proceedings of the 2012 International Workshop on Information and Electronics Engineering, IWIEE 2012, pp. 1554–1558, China, March 2012. View at: Google Scholar
  21. M.-C. Su, S.-C. Lai, S.-C. Lin, and L.-F. You, “A new approach to multi-aircraft air combat assignments,” Swarm and Evolutionary Computation, vol. 6, pp. 39–46, 2012. View at: Publisher Site | Google Scholar
  22. S. C. Huang and L. I. Wei-Min, “Ant colony algorithm for solving scheduling problem in multi-targets attacking,” Omr Ngnrng, 2008. View at: Google Scholar
  23. T. Zhang, L. Yu, X. Wei, and Z. L. Zhou, “Decision-making for cooperative multiple target attack based on adaptive pseudo-parallel genetic algorithm,” Fire Control & Command Control, vol. 38, no. 5, pp. 137–140, 2013. View at: Google Scholar
  24. B. L. Zhu, R. C. Zhu, and X. F. Xiong, Effectiveness Evaluation for Combat Aircrafts, Aviation Industrial Publishing House, Beijing, China, 2006.
  25. H. Cai, J. Liu, Y. Chen, and H. Wang, “Survey of the research on dynamic weapon-target assignment problem,” Journal of Systems Engineering and Electronics, vol. 17, no. 3, pp. 559–565, 2006. View at: Publisher Site | Google Scholar
  26. D. A. Shlapak, D. T. Orletsky, and B. A. Wilson, “Dire strait? Military aspects of the China-Taiwan confrontation and options for US policy,” EMBO Journal, vol. 17, no. 5, pp. 1228–1235, 2000. View at: Google Scholar
  27. J. Yan, X. Li, L. Liu, and F. Zhang, “Weapon-target assignment based on Memetic optimization algorithm in beyond-visual-rang cooperative air combat,” Beijing Hangkong Hangtian Daxue Xuebao/Journal of Beijing University of Aeronautics and Astronautics, vol. 40, no. 10, pp. 1424–1429, 2014. View at: Google Scholar
  28. R. J. Kuo, B. S. Wibowo, and F. E. Zulvia, “Application of a fuzzy ant colony system to solve the dynamic vehicle routing problem with uncertain service time,” Applied Mathematical Modelling: Simulation and Computation for Engineering and Environmental Systems, vol. 40, no. 23-24, pp. 9990–10001, 2016. View at: Publisher Site | Google Scholar | MathSciNet
  29. C.-J. Ting and C.-H. Chen, “A multiple ant colony optimization algorithm for the capacitated location routing problem,” International Journal of Production Economics, vol. 141, no. 1, pp. 34–44, 2013. View at: Publisher Site | Google Scholar
  30. J. Zhang, “ACGA Algorithm of Solving Weapon - Target Assignment Problem,” Open Journal of Applied Sciences, vol. 02, no. 04, pp. 74–77, 2012. View at: Publisher Site | Google Scholar
  31. P. Guo and L. Zhu, “Ant colony optimization for continuous domains,” in Proceedings of the 8th International Conference on Natural Computation (ICNC '12), pp. 758–762, Chongqing, China, May 2012. View at: Publisher Site | Google Scholar
  32. M. Mahi, Ö. K. Baykan, and H. Kodaz, “A new hybrid method based on particle swarm optimization, ant colony optimization and 3-Opt algorithms for traveling salesman problem,” Applied Soft Computing, vol. 30, pp. 484–490, 2015. View at: Publisher Site | Google Scholar
  33. A. C. Zecchin, A. R. Simpson, H. R. Maier, A. Marchi, and J. B. Nixon, “Improved understanding of the searching behavior of ant colony optimization algorithms applied to the water distribution design problem,” Water Resources Research, vol. 48, no. 9, 2012. View at: Google Scholar
  34. M. M. S. Abdulkader, Y. Gajpal, and T. Y. Elmekkawy, “Hybridized ant colony algorithm for the Multi Compartment Vehicle Routing Problem,” Applied Soft Computing, vol. 37, pp. 196–203, 2015. View at: Publisher Site | Google Scholar
  35. J. Guan and G. Lin, “Hybridizing variable neighborhood search with ant colony optimization for solving the single row facility layout problem,” European Journal of Operational Research, vol. 248, no. 3, pp. 899–909, 2016. View at: Publisher Site | Google Scholar | MathSciNet
  36. J. Li, J. Chen, B. Xin, and L. Dou, “Solving multi-objective multi-stage weapon target assignment problem via adaptive NSGAII and adaptive MOEA/D: A Comparison Study,” in Proceedings of the IEEE Congress on Evolutionary Computation (CEC '15), pp. 3132–3139, May 2015. View at: Publisher Site | Google Scholar

Copyright © 2018 Xinwu Hu et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


More related articles

917 Views | 472 Downloads | 5 Citations
 PDF  Download Citation  Citation
 Download other formatsMore
 Order printed copiesOrder

Related articles

We are committed to sharing findings related to COVID-19 as quickly and safely as possible. Any author submitting a COVID-19 paper should notify us at help@hindawi.com to ensure their research is fast-tracked and made available on a preprint server as soon as possible. We will be providing unlimited waivers of publication charges for accepted articles related to COVID-19. Sign up here as a reviewer to help fast-track new submissions.