Fig. 1. Relationship between data volume and error
Fig. 2. Battlefield scenario of UAV threat assessment
Fig. 3. UAV threat assessment framework diagram
Fig. 4. BN structure matrix expression
Fig. 5. Construction flow chart of the constraint matrix
Fig. 6. Flow chart of BN structure learning algorithm based on matrix constraints
Fig. 7. Flow chart of parameter learning algorithm based on prior constraints
Fig. 8. Total BIC score of the algorithm proposed in this paper and the K2 algorithm
Fig. 9. Average BIC score of the algorithm proposed in this paper and K2 algorithm
Fig. 10. Total Hamming distance between the algorithm in this paper and the K2 algorithm
Fig. 11. Average Hamming distance between the algorithm in this paper and the K2 algorithm
Fig. 12. Threat assessment network based on 20 sets of data by our algorithm
Fig. 13. Threat assessment network based on 30 sets of data by our algorithm
Fig. 14. Threat assessment network based on 40 sets of data by our algorithm
Fig. 15. Threat assessment network based on 50 sets of data by our algorithm
Fig. 16. UAV threat assessment network model structure
Fig. 17. Threat assessment network derived from 40 sets of data by K2 algorithm
Fig. 18. Threat assessment network derived from 50 sets of data by K2 algorithm
Fig. 19. Threat assessment network derived from 60 sets of data by K2 algorithm
Fig. 20. Threat assessment network derived from 70 sets of data by K2 algorithm
Fig. 21. Relationship between data volume and threat probability
Target node name | Meaning | Value range and meaning |
---|
Type information I | Which of the two types does the target belong to | {1, 2}={I1: Warning radar, I2: Antiaircraft Artillery Position(AAP)} | Confrontation information K | Combat capability of target weapon | {1, 2}={K1: yes, K2: no} | Location information L | Whether UAV is in the detection range of target attack and search | {1, 2}={L1: no, L2: yes} | Threat level information T | Threat level of the UAV relative to the target at the current moment | {1, 2}={T1: low, T2: high} |
|
Table 1. Discretization of target node information
Target threat(T1: low,T2: high) | Target type(I1: Warning radar,I2: AAP) | Target confrontation (K1: yes,K2: no) | Target location (L1: far,L2: near) |
---|
T1(low) | [0.3,0.7] | ([0.3,0.7], [0.3,0.7]) | ([0.7,0.9], [0.1,0.3]) | ([0.3,0.7], [0.3,0.7]) | T2(high) | [0.3,0.7] | ([0.8,1], [0,0.2]) | ([0.2,0.4], [0.6,0.8]) | ([0.5,0.7], [0.3,0.5]) |
|
Table 2. Interval constraints corresponding to parameters
Target threat(T1: low,T2: high) | Target type(I1: Warning radar,I2: AAP) | Target confrontation (K1: yes,K2: no) | Target location (L1: far,L2: near) |
---|
T1(low) | 0.5 | (0.5,0.5) | (0.8,0.2) | (0.5,0.5) | T2(high) | 0.5 | (0.9,0.1) | (0.3,0.7) | (0.6,0.4) |
|
Table 3. Initial parameters of threat assessment model
Target threat(T1: low,T2: high) | Target type(I1: Warning radar,I2: AAP) | Target confrontation(K1: yes,K2: no) | Target location (L1: far,L2: near) |
---|
T1(low) | 0.5173 | (0.4841,0.5159) | (0.7866,0.2134) | (0.5086,0.4914) | T2(high) | 0.4827 | (0.9008,0.0992) | (0.2240,0.7760) | (0.64050.3595) |
|
Table 4. 40 sets of data model parameters obtained by our algorithm
Target threat(T1: low,T2: high) | Target type(I1: Warning radar,I2: AAP) | Target confrontation (K1: yes,K2: no) | Target location (L1: far,L2: near) |
---|
T1(low) | 0.4867 | (0.4391,0.5609) | (0.8094,0.1906) | (0.4454,0.5546) | T2(high) | 0.5133 | (0.8851,0.1149) | (0.2189,0.7811) | (0.5343,0.4657) |
|
Table 5. 50 sets of data model parameters obtained from our algorithm
Target threat(T1: low,T2: high) | Target type(I1: Warning radar,I2: AAP) | Target confrontation (K1: yes,K2: no) | Target location (L1: far,L2: near) |
---|
T1(low) | 0.4657 | (0.4583,0.5417) | (0.8146,0.1854) | (0.4484,0.5516) | T2(high) | 0.5343 | (0.8866,0.1134) | (0.2380,0.7620) | (0.5069,0.4931) |
|
Table 6. 60 sets of data model parameters obtained by our algorithm
Target threat(T1: low,T2: high) | Target type(I1: Warning radar,I2: AAP) | Target confrontation (K1: yes,K2: no) | Target location (L1: far,L2: near) |
---|
T1(low) | 0.4750 | (0.4737,0.5263) | (0.6184,0.3452) | (0.4474,0.5526) | T2(high) | 0.5250 | (0.6548,0.3452) | (0.3929,0.6071) | (0.4643,0.5357) |
|
Table 7. Model parameters obtained from 60 sets of data by MLE algorithm
Target threat(T1: low,T2: high) | Target type(I1: Warning radar,I2: AAP) | Target confrontation(K1:yes,K2:no) | Target location(L1: far,L2: near) |
---|
T1(low) | 0.5286 | (0.3784,0.6216) | (0.8649,0.1351) | (0.4595,0.5405) | T2(high) | 0.4714 | (0.7879,0.2121) | (0.1212,0.8788) | (0.6061,0.3939) |
|
Table 8. Model parameters obtained from 70 sets of data by MLE algorithm
Target threat(T1: low,T2: high) | Target type(I1: Warning radar,I2: AAP) | Target confrontation(K1: yes,K2: no) | Target location(L1: far,L2: near) |
---|
T1(low) | 0.5875 | (0.4255,0.5745) | (0.8298,0.1702) | (0.3830,0.6170) | T2(high) | 0.4125 | (0.9394,0.0606) | (0.1515,0.8485) | (0.6061,0.3939) |
|
Table 9. Model parameters obtained from 80 sets of data by MLE algorithm
40 groups | 50 groups | 60 groups |
---|
0.0163 | 0.0161 | 0.0156 |
|
Table 10. KL-divergence results of this algorithm
60 groups | 70 groups | 80 groups |
---|
0.0188 | 0.0160 | 0.0132 |
|
Table 11. KL-divergence results of MLE algorithm