基于相关性参数选择的飞行数据异常检测

钟杰; 罗冲; 张恒; 苗强

doi:10.13700/j.bh.1001-5965.2022.0574

基于相关性参数选择的飞行数据异常检测

doi: 10.13700/j.bh.1001-5965.2022.0574

四川大学电气工程学院，成都 610065

基金项目: 国家自然科学基金(52075349)；航空科学基金(201905019001)

详细信息

通讯作者:
E-mail：hengzhang27@scu.edu.cn

中图分类号: V241；V267
计量
- 文章访问数: 274
- HTML全文浏览量: 96
- PDF下载量: 7
- 被引次数: 0
出版历程
- 收稿日期: 2022-07-02
- 录用日期: 2022-08-12
- 网络出版日期: 2022-12-14
- 整期出版日期: 2024-05-29

Flight data anomaly detection based on correlation parameter selection

School of Electrical Engineering，Sichuan University，Chengdu 610065，China

Funds: National Natural Science Foundation of China (52075349); Aeronautical Science Foundation of China (201905019001)

More Information

Corresponding author: E-mail：hengzhang27@scu.edu.cn

摘要

摘要:
随着无人机(UAV)技术的成熟，其在军民领域的应用越来越广泛，安全问题也逐渐受到重视。UAV飞行数据能直接反映其飞行健康状态。针对UAV飞行数据开展异常检测研究是提升UAV整体安全性的重要手段之一。基于此，提出了一种基于相关性参数选择与卷积神经网络(CNN)的异常检测方法。利用最大信息系数(MIC)和Pearson相关系数法挖掘参数之间的相关性，并建立相关性飞行参数集合；利用与待检测飞行参数相关的飞行参数数据训练卷积神经网络预测模型，根据模型预测值与真实值之间的残差判定异常。利用真实UAV飞行数据对所提方法进行验证，结果显示：所提方法的假阳率、假阴率、准确率指标均值分别为0%、0.19%、99.6%，证明了方法的有效性。
- 异常检测 /
- 飞行数据 /
- 最大信息系数 /
- Pearson相关系数法 /
- 卷积神经网络
Abstract:
With the maturity of unmanned aerial vehicle (UAV) technology, UAVs are being used more and more widely in the military and civilian fields. Meanwhile, the safety of UAV is gradually being emphasized. The health of the UAV’s flight can be immediately reflected in its flight data. Anomaly detection research for UAV flight data is one of the important ways to improve the overall safety of UAVs. In this paper, we propose a convolutional neural network (CNN) anomaly detection method based on correlation parameter selection for flight data. Firstly, we use the maximal information coefficient (MIC) and Pearson correlation coefficient method to explore the correlation among flight parameters and establish a set of correlations between flight parameters. Then, we use the correlation flight parameters to train the convolutional neural network regression model. Finally, the anomalies were determined based on the residuals between the true and predicted values of the model. The false positive rate, false negative rate, and accuracy indexes of the approach suggested in this work were 0%, 0.19%, and 99.6%, respectively, demonstrating the method’s superiority. The method was confirmed using actual UAV flight data.
- anomaly detection /
- flight data /
- maximal information coefficient /
- Pearson correlation coefficient method /
- convolutional neural network

HTML全文

图 1 一维卷积层运算示例

Figure 1. Example of 1D convolutional layer operations

下载: 全尺寸图片幻灯片

图 2 飞行数据异常检测流程

Figure 2. Flight data anomaly detection flowchart

下载: 全尺寸图片幻灯片

图 3 滑窗处理示意图

Figure 3. Schematic of sliding window treatment

下载: 全尺寸图片幻灯片

图 4 卷积神经网络异常检测模型

Figure 4. Convolutional neural network anomaly detection model

下载: 全尺寸图片幻灯片

图 5 “Clouds”固定翼UAV

Figure 5. Fixed-wing UAV “Clouds”

下载: 全尺寸图片幻灯片

图 6 Corr-CNN异常检测结果

Figure 6. Corr-CNN anomaly detection results

下载: 全尺寸图片幻灯片

表 1 相关参数对

Table 1. Relevant parameters

相关飞行参数对		相关飞行参数对
侧向过载-真侧滑角		右副翼指令-侧向过载
侧向过载-俯仰角		右副翼指令-真侧滑角
地速-空速		右副翼指令-俯仰角
地速-高度		右方向舵指令-偏航角速率
俯仰角-天向速度		右方向舵指令-横滚角
俯仰角-法向过载		右方向舵指令-轴向过载
俯仰角-攻角		右方向舵指令-俯仰角
攻角-空速		右方向舵指令-攻角
滚转角速率-横滚角		真侧滑角-俯仰角
滚转角速率-侧向过载		轴向过载-俯仰角
滚转角速率-真侧滑角		轴向过载-攻角
滚转角速率-俯仰角		轴向速度-天向速度
横滚角-侧向过载		左副翼指令-滚转角速率
横滚角-真侧滑角		左副翼指令-侧向过载
横滚角-俯仰角		左副翼指令-真侧滑角
空速-高度		左副翼指令-俯仰角
偏航角速率-横滚角		左方向舵指令-右方向舵指令
偏航角速率-轴向过载		左方向舵指令-偏航角速率
偏航角速率-俯仰角		左方向舵指令-横滚角
偏航角速率-攻角		左方向舵指令-轴向过载
右副翼指令-左副翼指令		左方向舵指令-俯仰角
右副翼指令-滚转角速率		左方向舵指令-攻角
右副翼指令-横滚角

下载: 导出CSV

表 2 异常检测结果

Table 2. Abnormal detection results

方法	异常类型	假阴率/%	假阳率/%	准确率/%
KNN	偏差异常	1.2	96.0	48.7
	静态异常	0.8	100	65.5
	点异常	0.5	0	99.5
SVR	偏差异常	18.2	2.5	90.1
	静态异常	19.7	2.0	86.9
	点异常	17.3	0	82.8
Corr-CNN	偏差异常	0.18	0	99.6
	静态异常	0.2	0	99.6
	点异常	0.2	0	99.6

下载: 导出CSV

参考文献(21)

[1]	PARK K H, PARK E, KIM H K. Unsupervised fault detection on unmanned aerial vehicles: Encoding and thresholding approach[J]. Sensors, 2021, 21(6): 2208.
[2]	FOURLAS G K, KARRAS G C. A survey on fault diagnosis and fault-tolerant control methods for unmanned aerial vehicles[J]. Machines, 2021, 9(9): 197. doi: 10.3390/machines9090197
[3]	ALTINORS A, YOL F, YAMAN O. A sound based method for fault detection with statistical feature extraction in UAV motors[J]. Applied Acoustics, 2021, 183: 108325. doi: 10.1016/j.apacoust.2021.108325
[4]	ACKERSON J, DAVE R, SELIYA N. Applications of recurrent neural network for biometric authentication & anomaly detection[J]. Information, 2021, 12(7): 272. doi: 10.3390/info12070272
[5]	THEISSLER A. Detecting known and unknown faults in automotive systems using ensemble-based anomaly detection[J]. Knowledge-Based Systems, 2017, 123: 163-173.
[6]	FREEMAN P, PANDITA R, SRIVASTAVA N, et al. Model-based and data-driven fault detection performance for a small UAV[J]. IEEE/ASME Transactions on Mechatronics, 2013, 18(4): 1300-1309. doi: 10.1109/TMECH.2013.2258678
[7]	WANG B K, WANG Z Y, LIU L S, et al. Data-driven anomaly detection for UAV sensor data based on deep learning prediction model[C]//Proceedings of the Prognostics and System Health Management Conference. Piscataway: IEEE Press, 2019: 286-290.
[8]	CORTES C, VAPNIK V. Support-vector networks[J]. Machine Learning, 1995, 20(3): 273-297.
[9]	WANG J Y, MIAO J G, WANG J L, et al. Fault diagnosis of electrohydraulic actuator based on multiple source signals: An experimental investigation[J]. Neurocomputing, 2020, 417: 224-238. doi: 10.1016/j.neucom.2020.05.102
[10]	LING J, LIU G J, LI J L, et al. Fault prediction method for nuclear power machinery based on Bayesian PPCA recurrent neural network model[J]. Nuclear Science and Techniques, 2020, 31(8): 75.
[11]	刘意, 毛莺池, 程杨堃, 等. 基于邻域一致性的异常检测序列集成方法[J]. 计算机科学, 2022, 49(1): 146-152. doi: 10.11896/jsjkx.201000156 LIU Y, MAO Y C, CHENG Y K, et al. Locality and consistency based sequential ensemble method for outlier detection[J]. Computer Science, 2022, 49(1): 146-152(in Chinese). doi: 10.11896/jsjkx.201000156
[12]	郁滨, 熊俊. 基于平衡迭代规约层次聚类的无线传感器网络流量异常检测方案[J]. 电子与信息学报, 2022, 44(1): 305-313. doi: 10.11999/JEIT201004 YU B, XIONG J. A novel WSN traffic anomaly detection scheme based on BIRCH[J]. Journal of Electronics & Information Technology, 2022, 44(1): 305-313(in Chinese). doi: 10.11999/JEIT201004
[13]	吴蕊, 张安勤, 田秀霞, 等. 基于改进K-means的电力数据异常检测算法[J]. 华东师范大学学报(自然科学版). 2020(4): 79-87. WU R, ZHANG A Q, TIAN X X, et al. Anomaly detection algorithm based on improved K-means for electric power data[J]. Journal of East China Normal University (Natural Science), 2020(4): 79-87(in Chinese).
[14]	MISHRA S, SAGBAN R, YAKOOB A, et al. Swarm intelligence in anomaly detection systems: An overview[J]. International Journal of Computers and Applications, 2021, 43(2): 109-118. doi: 10.1080/1206212X.2018.1521895
[15]	PANG G S, SHEN C H, CAO L B, et al. Deep learning for anomaly detection: A review[J]. ACM Computing Surveys, 2022, 54(2): 1-38.
[16]	高妮, 贺毅岳, 马新成. 基于低频分量EEMD-SVR预测建模的指数择时策略[J]. 统计与决策, 2022, 38(2): 140-145. GAO N, HE Y Y, MA X C. Exponential timing strategy based on EEMD-SVR predictive modeling of low frequency component[J]. Statistics & Decision, 2022, 38(2): 140-145(in Chinese).
[17]	陶君雯, 张韬, 庄雪菲, 等. 动态贝叶斯网络模型和SARIMA模型对手足口病预测效果的比较[J]. 现代预防医学, 2020, 47(21): 3851-3854. TAO J W, ZHANG T, ZHUANG X F, et al. Comparison on prediction accuracy of dynamic Bayesian networks and SARIMA model for hand foot and mouth disease[J]. Modern Preventive Medicine, 2020, 47(21): 3851-3854(in Chinese).
[18]	纪少波, 李洋, 李萌, 等. 纯电动共享汽车驾驶行为对能耗的影响[J]. 吉林大学学报(工学版), 2022, 52(4): 754-763. JI S B, LI Y, LI M, et al. Influence of driving behavior on energy consumption of pure electric shared vehicles[J]. Journal of Jilin University (Engineering and Technology Edition), 2022, 52(4): 754-763(in Chinese).
[19]	EDELMANN D, MORI T F, SZEKELY G J. On relationships between the Pearson and the distance correlation coefficients[J]. Statistics & Probability Letters, 2021, 169: 108960.
[20]	WANG B K, LIU D T, PENG Y, et al. Multivariate regression based fault detection and recovery of UAV flight data[J]. IEEE Transactions on Instrumentation and Measurement, 2020, 69(6): 3527-3537. doi: 10.1109/TIM.2019.2935576
[21]	KEIPOUR A, MOUSAEI M, SCHERER S. Automatic real-time anomaly detection for autonomous aerial vehicles[C]//Proceedings of the International Conference on Robotics and Automation. Piscataway: IEEE Press, 2019: 5679-5685.