一种基于SVM的低空飞行冲突探测算法

韩冬; 张学军; 聂尊礼; 管祥民

doi:10.13700/j.bh.1001-5965.2017.0159

一种基于SVM的低空飞行冲突探测算法

doi: 10.13700/j.bh.1001-5965.2017.0159

1.
北京航空航天大学电子信息工程学院, 北京 100083
2.
中国民航管理干部学院, 北京 100102

基金项目:

国家自然科学基金 U1533119

详细信息

作者简介:
韩冬男, 博士研究生。主要研究方向:航空监视、航空电子

张学军男, 博士, 教授, 博士生导师。主要研究方向:航空数据通信系统、航空电子技术、现代空中交通管理技术

通讯作者:
张学军, E-mail: zhxj@buaa.edu.cn

中图分类号: V355
计量
- 文章访问数: 906
- HTML全文浏览量: 189
- PDF下载量: 476
- 被引次数: 0
出版历程
- 收稿日期: 2017-03-17
- 录用日期: 2017-04-01
- 网络出版日期: 2018-03-20

A conflict detection algorithm for low-altitude flights based on SVM

1.
School of Electronic and Information Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100083, China
2.
Civil Aviation Management Institute of China, Beijing 100102, China

Funds:

National Natural Science Foundation of China U1533119

More Information

Corresponding author: ZHANG Xuejun, E-mail: zhxj@buaa.edu.cn

摘要

摘要:
随着低空飞行密度不断增加，低空航行安全已引起广泛关注，由于低空环境复杂，低空飞行受地面障碍物和天气影响比商用航空显著，传统的空中交通警戒与防撞系统（TCAS）和其他冲突探测方法并不适用于低空密集飞行环境。针对传统探测方法计算量大、适用性差的不足，引入支持向量机（SVM）的二元分类方法，通过对本机和周边飞机航迹归一化处理，采用智能优化算法对关键参数进行优化，利用模拟数据对分类器进行预先训练，实现了适用于低空飞行的高效冲突探测。以大量的仿造数据对算法有效性进行了测试验证，结果表明漏警率和误警率分别控制在约0.1%和6%，克服了传统确定型方法与概率型方法难以兼顾效率与适用性的缺陷。
- 通用航空 /
- 冲突探测 /
- 支持向量机(SVM) /
- GA-PSO /
- 智能优化算法
Abstract:
With the continuous increasing of flight density, the aviation safety in low altitude has caused extensive concern. Low-altitude environment is complex, and ground obstacles and weather have more significant impact on low-altitude flight than commercial aviation. Traditional traffic alert and collision avoidance system (TCAS) and other methods may not be applicable to low-altitude intensive flight environment. In view of the computational complexity and lack of applicability of traditional detection methods, a binary classification method of support vector machine (SVM) was introduced. By normalizing the trajectories of own and surrounding aircraft, optimizing the key parameters by intelligent optimization algorithm, and pre-training the classifier through simulation data, efficient conflict detection for low-altitude flight was carried out. Various sets of artificial data were utilized to verify the effectiveness of the algorithm. The results show that the missed alarm rate and false alarm rate are controlled at about 0.1% and 6% respectively, which proves that the proposed algorithm can overcome the shortcomings of traditional deterministic and probabilistic methods which are difficult to take both the efficiency and applicability into account.
- general aviation /
- conflict detection /
- support vector machine(SVM) /
- GA-PSO /
- intelligent optimization algorithm

HTML全文

图 1 检测系统结构

Figure 1. Detection system structure

下载: 全尺寸图片幻灯片

图 2 保护区模型

Figure 2. Protection zone model

下载: 全尺寸图片幻灯片

图 3 GA-PSO混合算法训练流程

Figure 3. GA-PSO hybrid algorithm training process

下载: 全尺寸图片幻灯片

图 4 w变化曲线

Figure 4. Variation curve of w

下载: 全尺寸图片幻灯片

图 5 粒子变异流程

Figure 5. Particle variation process

下载: 全尺寸图片幻灯片

图 6 交叉验证平均正确率

Figure 6. Average accuracy of cross-validation

下载: 全尺寸图片幻灯片

表 1 无冲突判定准则

Table 1. No conflict judgment criteria

卦限编号	卦限区间	无冲突判定准则
1	x_R＞0, y_R＞0, z_R＞0
2	x_R＜0, y_R＞0, z_R＞0
3	x_R＜0, y_R＜0, z_R＞0
4	x_R＞0, y_R＜0, z_R＞0
5	x_R＞0, y_R＞0, z_R＜0
6	x_R＜0, y_R＞0, z_R＜0
7	x_R＜0, y_R＜0, z_R＜0
8	x_R＞0, y_R＜0, z_R＜0

下载: 导出CSV

表 2 3种算法参数

Table 2. Parameters of three algorithms

参数	GA算法	PSO算法	GA-PSO混合算法
训练代数	100	100	100
种群大小	30	30	30
c₁	N/A	1.5	1.5
c₂	N/A	1.7	1.7
代沟	0.9	N/A	0.5
交叉概率	0.7	N/A	0.7
变异概率	0.02	N/A	0.02
注：N/A表示不适用。

下载: 导出CSV

表 3 归一化飞行状态

Table 3. Normalized flight status

飞机序号	相对经度x_Ri	相对纬度y_Ri	相对高度z_Ri	相对速度‖V_Ri‖	相对航向角θ_Ri
1	0.34	0	0.212	-1	0.2
2	0.78	-0.5	-0.261	0.2	0.5
3	-0.88	1	0.921	0	0.1
					
N	-0.44	0.34	0	-0.3	0.8
注：x_Ri、y_Ri、z_Ri、‖V_Ri‖的区间为[-1, 1];θ_Ri区间为[0, 1]。

下载: 导出CSV

表 4 SVM模型的平均正确率

Table 4. Average accuracy of SVM models

%
训练集(冲突-非冲突)	GA算法	PSO算法	GA-PSO混合算法
100-100	82.016 9	81.903	82.789
150-150	87.649 8	87.603 4	89.029
200-200	82.713 1	83.641 4	84.244
300-300	85.779 7	87.139 8	87.411

下载: 导出CSV

表 5 CESSNA 172的尺寸和性能

Table 5. Dimensions and performance of CESSNA 172

尺寸	数值	主要性能	数值
长/ft	27.2	最大巡航速度/(km·h^-1)	230
高/ft	8.9	最大航程/nmi	640(1 185 km)
翼展/ft	36.1	最大爬升率/(ft·mim^-1)	730(223 m/min)
		极限速度/(km·h^-1)	302
		失速速度/(km·h^-1)	89
注：最大承载人数为4。

下载: 导出CSV

表 6 用于检测的统计数据

Table 6. Statistical data for detection

检测总次数	冲突航迹检测次数	不冲突航迹检测次数
104 069	51 680	52 439

下载: 导出CSV

表 7 检测系统性能的统计数据

Table 7. Statistical data for performance of detection system

性能	初始检测状态	10 s内检测
性能	初始检测状态	不经过移动加权平均	经过移动加权平均
漏警数	10	57	52
误警数	363	3 280	3 277
漏警率/%	N/A	0.110 3	0.100 6
误警率/%	N/A	6.254 9	6.249 2

下载: 导出CSV

参考文献(15)

[1]	GARIEL M, HANSMAN R, FRAZZOLI E. Impact of GPS and ADS-B reported accuracy on conflict detection performance in dense traffic: AIAA-2011-6893[R]. Reston: AIAA, 2011.
[2]	FULTON N L.Airspace design:Towards a rigorous specification of complexity based on computational geometry[J].Aeronautical Journal, 1999, 103(1020):75-84. doi: 10.1017/S0001924000027779
[3]	CHIANG, YI J, KLOSOWSKI J, et al. Geometric algorithms for conflict detection and resolution in air traffic management[C]//36th IEEE Conference on Decision and Control. Piscataway, NJ: IEEE Press, 1997, 2(2): 1835-1840.
[4]	MCDONALD J, VIVONA R. Strategic airborne conflict detection of air traffic and area hazards: NAS2-98005[R]. Washington, D. C. : NASA, 2000.
[5]	PRANDINI M, HU J, SASTRY S.A probabilistic approach to aircraft conflict detection[J].IEEE Transactions on Intelligent Transportation Systems, 2000, 1(4):199-220. doi: 10.1109/6979.898224
[6]	JARDIN M R. Grid-based strategic air traffic conflict detection: AIAA-2005-5826[R]. Reston: AIAA, 2005.
[7]	HU J. Aircraft conflict detection in presence of spatially correlated wind perturbations: AIAA-2003-5339[R]. Reston: AIAA, 2003.
[8]	李彬, 吴珍珍.基于航迹预测的飞行冲突预测[J].微处理机, 2011(2):73-80. http://kns.cnki.net/KCMS/detail/detail.aspx?filename=wclj201102022&dbname=CJFD&dbcode=CJFQ LI B, WU Z Z.Flight conflict detection based on flight path prediction[J].Microprocessors, 2011(2):73-80(in Chinese). http://kns.cnki.net/KCMS/detail/detail.aspx?filename=wclj201102022&dbname=CJFD&dbcode=CJFQ
[9]	韩艳茹, 敬忠良, 龚嘉琦.空中交通预警与防撞系统(TCAS)风险及对策研究[J].计算机测量与控制, 2012, 20(3):737-740. http://kns.cnki.net/KCMS/detail/detail.aspx?filename=jzck201203052&dbname=CJFD&dbcode=CJFQ HAN Y R, JING Z L, GONG J Q.Research of traffic alert and collision avoidance system(TCAS) risk and countermeasure[J].Computer Measurement & Control, 2012, 20(3):737-740(in Chinese). http://kns.cnki.net/KCMS/detail/detail.aspx?filename=jzck201203052&dbname=CJFD&dbcode=CJFQ
[10]	WILLIAMSON T, SPENCER N A.Development and operation of the traffic alert and collision avoidance system(TCAS)[J].Proceedings of the IEEE, 1989, 77(11):1735-1744. doi: 10.1109/5.47735
[11]	林熙. 密集飞行条件下的间隔自主保持方法研究[D]. 北京: 北京航空航天大学, 2011: 11-13. LIN X. Research on self-separation assurance methods in condition of intensive flight[D]. Beijing: Beihang University, 2011: 11-13(in Chinese).
[12]	LIN C E.Collision avoidance solution for low-altitude flights[J].Proceedings of the Institution of Mechanical Engineers, Part G:Journal of Aerospace Engineering, 2011, 225(7):779-790. doi: 10.1177/0954410011399211
[13]	JIAO Y L, ZHANG X J, GUAN X M.An algorithm for airborne conflict detection based on support vector machine[J].Applied Mechanics and Materials, 2012, 229-231:1140-1145. doi: 10.4028/www.scientific.net/AMM.229-231
[14]	CORTES C, VAPNIK V.Support-vector networks[J].Machine Learning, 1995, 20(3):273-276. http://ci.nii.ac.jp/naid/10021013062
[15]	KOHAVI R. A study of cross-validation and bootstrap for accuracy estimation and model selection[C]//Proceedings of the 14th International Joint Conference on Artificial Intelligence. New York: ACM, 1995: 1137-1143.