基于QAR数据的飞行员飞行操作风格研究

doi:10.16265/j.cnki.issn1003-3033.2022.12.1169

摘要/Abstract

摘要：

为减少或避免由飞行员操作导致的不安全事件发生,保障民航航班运行安全,深入探究飞行员飞行操作过程中的整体倾向性和飞行员操作风格。基于某B737-800型机队快速存储记录器(QAR)数据,首先,提取多个飞行阶段飞行员操作风格的特征参数,构建数据集;然后,采用高斯混合模型(GMM)聚类的方法对数据集分类;最后,运用方差分析方法,分析不同操作风格影响下的飞机状态。结果表明:该机队飞行员驾驶风格能够划分为有明显差异的3类,分别为激进型、普通型和保守型;相同操作风格飞行员的飞行操作表现出整体相似的特征;不同操作风格飞行员操控下飞机飞行状态具有显著性差异。利用QAR数据能够较为准确地判断机队飞行员的操作风格,识别飞行员操作整体特征,并评估不同操作风格飞行员在飞行实践中的操作风险,进一步辅助规范飞行员操作行为,提升飞行员操作能力。

关键词: 快速存储记录器(QAR)数据, 飞行员, 飞行操作风格, 高斯混合模型(GMM), 聚类算法

Abstract:

In order to reduce or avoid the occurrence of unsafe events caused by pilot operation and ensure the safety of civil aviation flight operation, the overall orientation and pilot operating style during flight operation were studied in depth. Based on QAR data of a B737-800 fleet, firstly, the characteristic parameters of the pilot's operation style in multiple flight phases were extracted to build a data set. Then, GMM clustering method was used to classify the data sets. Finally, the analysis of variance was used to analyze the aircraft status under the influence of different operating styles. The results show that the pilots' operation styles can be divided into three distinct categories: radical, ordinary and conservative. The flight operations of pilots with the same operating style show similar characteristics as a whole. There are significant differences in aircraft flight status under the control of pilots with different operation styles. QAR data can be used to accurately judge the operation style of the pilots in the fleet, identify the overall characteristics of the pilots' operation, and evaluate the operational risks of pilots with different operation styles in flight practice, so as to further help standardize the pilot's operation behavior and improve the pilot's operation ability.

Key words: quick access recorder(QAR) data, pilot, flight operation style, Gaussian mixture model (GMM), clusteri algorithm

孙瑞山, 刘胤甫. 基于QAR数据的飞行员飞行操作风格研究[J]. 中国安全科学学报, 2022, 32(12): 63-69.

SUN Ruishan, LIU Yinfu. Research on pilots' flight operation style based on QAR data[J]. China Safety Science Journal, 2022, 32(12): 63-69.

图/表 7

图1

图2

表1

图3

表2

表3

表4

参考文献 20

[1]	中国民用航空局航空安全办公室. 中国民航安全报告2020[R], 2021.
[2]	International Air Transport Association. Safety report 2021[R], 2022.
[3]	班永宽. 航空事故的人为因素[M]. 北京: 中国民航出版社, 2002:30-38.
[4]	PAYNE K H, HARRIS D. A psychometric approach to the development of multidimensional scale to assess aircraft handling qualities[J]. International Journal of Aviation Psychology, 2000, 10(4):343-362. doi: 10.1207/S15327108IJAP1004_3
[5]	汪磊, 任勇. 民机着陆目视判断及操作影响因素研究[J]. 人类工效学, 2017, 23(5):1-8.
	WANG Lei, REN Yong. Study on visual perception and operation influencing factors in landing of civil aircraft[J]. Chinese Journal of Ergonomics, 2017, 23(5):1-8.
[6]	孙瑞山, 李重锋. 基于k-SC聚类的飞行操作模式及危险性分析[J]. 中国安全生产科学技术, 2021, 17(9):150-155.
	SUN Ruishan, LI Chongfeng. Analysis of flight operation patterns and risk based on k-SC clustering[J]. Journal of Safety Science and Technology, 2021, 17(9):150-155.
[7]	段照斌, 杜海龙, 张鹏. 基于QAR2Vec模型的QAR数据特征提取[J]. 中国安全科学学报, 2021, 31(1):145-152. doi: 10.16265/j.cnki.issn 1003-3033.2021.01.021
	DUAN Zhaobin, DU Hailong, ZHANG Peng. Feature extraction of QAR data based on QAR2Vec model[J]. China Safety Science Journal, 2021, 31(1):145-152. doi: 10.16265/j.cnki.issn 1003-3033.2021.01.021
[8]	张兆宁, 陈子辰, 卢飞. 配对进近侧向位置误差QAR参数关联性分析[J]. 中国安全科学学报, 2021, 31(8):47-52. doi: 10.16265/j.cnki.issn1003-3033.2021.08.007
	ZHANG Zhaoning, CHEN Zichen, LU Fei. Correlation analysis of QAR parameters for paired approach lateral position error[J]. China Safety Science Journal, 2021, 31(8):47-52. doi: 10.16265/j.cnki.issn1003-3033.2021.08.007
[9]	孙瑞山, 肖亚兵. 基于QAR记录数据的民航飞行员操作特征指标结构研究[J]. 中国安全生产科学技术, 2012, 8(11):49-54.
	SUN Ruishan, XIAO Yabing. Research on indicating structure for operation characteristic of civil aviation pilots based on QAR data[J]. Journal of Safety Science and Technology, 2012, 8(11):49-54.
[10]	汪磊, 董传亭, 崔美叶. 基于功率谱的飞行操作技能评价方法[J]. 飞行力学, 2018, 36(4):92-96.
	WANG Lei, DONG Chuanting, CUI Meiye. Evalution method of flight operation skills based on power spectrum[J]. Flight Dynamics, 2018, 36(4):92-96.
[11]	SHAPPEL L S, DETWILER C, HOLCOMB K, et al. Human error and commercial aviation accidents: an analysis using the human factors analysis and classification system[J]. Human Factors, 2007, 49(2):227-242. pmid: 17447665
[12]	BOEING. Statistical summary of commercial jet airplane accidents worldwide operations 1959-2020[R], 2021.
[13]	WANG Lei, WU Changxu, SUN Ruishan. An analysis of flight quick access recorder (QAR) data and its applications in preventing landing incidents[J]. Reliability Engineering & System Safety, 2014, 127:86-96. doi: 10.1016/j.ress.2014.03.013
[14]	WANG Lei, REN Yong, WU Changxu. Effects of flare operation on landing safety: a study based on ANOVA of real flight data[J]. Safety Science, 2018, 102: 14-25. doi: 10.1016/j.ssci.2017.09.027
[15]	李经纬, 赵治国, 沈沛鸿, 等. 驾驶风格K-means聚类与识别方法研究[J]. 汽车技术, 2018(12):8-12. doi: 10.19620/j.cnki.1000-3703.20180272
	LI Jingwei, ZHAO Zhiguo, SHEN Peihoong, et al. Research on method of K-means clustering and recognition for driving style[J]. Automotive Technology, 2018(12):8-12.
[16]	刘通, 付锐, 张名芳, 等. 融合K-means与高斯混合模型的驾驶风格聚类研究[J]. 中国安全科学学报, 2019, 29(12):40-45. doi: 10.16265/j.cnki.issn1003-3033.2019.12.007
	LIU Tong, FU Rui, ZHANG Mingfang, et al. Study on driving style clustering based on K-means and Gaussian mixture model[J]. China Safety Science Journal, 2019, 29(12):40-45. doi: 10.16265/j.cnki.issn1003-3033.2019.12.007
[17]	刘迪, 郑建明, 覃斌, 等. 基于自然驾驶场景大数据的驾驶风格研究[J]. 汽车技术, 2022(9):36-42.
	LIU Di, ZHENG Jianming, QIN Bin, et al. Research on driving style based on big data of natural driving scene[J]. Automobile Technology, 2022(9):36-42.
[18]	WANG Zhiqiang, CUNHA C, RITOU M, et al. Comparison of K-means and GMM methods for contextual clustering in HSM[J]. Procedia Manufacturing, 2019, 28:154-159. doi: 10.1016/j.promfg.2018.12.025
[19]	GYS A M M, HERMANUS C M. A review of intelligent driving style analysis systems and related artificial intelligence algorithms[J]. Sensors (Basel, Switzerland), 2015, 15(12):30653-30 682.
[20]	AC-121/135-FS-2012,飞行品质监控(FOQA)实施与管理[S].

飞行阶段	平均杆位/(°)	杆位标准差/(°)	平均移动速度/((°)·s^-1)	移动速度标准差/((°)·s^-1)
起飞阶段	4.396 758 0	2.445 802 5	1.627 770 4	2.409 022 9
	4.762 846 6	2.359 584 8	1.408 652 7	1.780 590 1
	4.096 366 2	2.538 252 1	1.208 687 0	1.703 888 0
着陆阶段	2.520 783 2	1.825 138 8	0.856 684 8	1.109 305 9
	5.777 307 9	1.378 634 7	1.227 036 9	1.651 408 6
	4.723 870 8	1.742 885 2	1.284 697 7	1.961 861 1

项目类型	起飞阶段				着陆阶段
项目类型	平均杆位/(°)	杆位标准差/(°)	平均移动速度/ ((°)·s^-1)	移动速度标准差/ ((°)·s^-1)	平均杆位/(°)	杆位标准差 /(°)	平均移动速度/ ((°)·s^-1)	移动速度标准差/ ((°)·s^-1)
第1类	4.378 7	2.083 4	0.898 1	1.094 4	4.247 3	1.308 6	1.098 3	1.395 9
第2类	4.070 1	2.237 5	1.036 8	1.346 6	4.225 1	1.273 4	1.375 7	1.676 8
第3类	4.053 8	2.626 5	1.288 4	1.639 6	4.350 6	1.863 6	1.796 6	2.331 3

飞行状态关键指标	抬前轮速度/ kn	主轮离地速度/ kn	主轮离地俯仰角/(°)	抬前轮速率/ ((°)·s^-1)	平均初始爬升速度/kn	最大起飞滚转角/ (°)	11~46 m 平均滚转角/ (°)	11~46 m 最大滚转角/ (°)	46~122 m 平均滚转角/ (°)	46~122 m 最大滚转角/ (°)
FOQA监控项目	抬前轮速度大/小	离地速度大/小	离地俯仰角大	抬前轮速率大/小	初始爬升速度大/小	起飞滚转角大	爬升滚转角大
自由度	2.000 0	2.000 0	2.000 0	2.000 0	2.000 0	2.000 0	2.000 0	2.000 0	2.000 0	2.000 0
均方	17.425 2	14.811 9	1.965 2	0.000 4	84.322 0	1.100 5	0.681 5	1.333 2	2.213 2	7.493 1
F	0.648 9	0.813 0	1.910 1	6.877 9	6.095 1	2.506 5	2.304 7	2.110 1	4.411 1	6.241 5
显著性	0.522 9	0.443 9	0.148 7	0.001 1	0.002 4	0.082 2	0.100 5	0.121 9	0.012 4	0.002 0
保守型均值	148.823 4	155.739 0	4.170 6	0.027 2	168.430 0	1.123 7	0.868 0	1.417 3	1.109 3	1.917 4
普通型均值	148.342 6	155.465 0	4.176 8	0.028 3	168.590 0	1.251 4	0.944 4	1.536 5	1.290 3	2.234 5
激进型均值	148.462 7	155.279 0	4.325 1	0.029 6	167.520 0	1.178 0	0.960 1	1.536 7	1.197 1	2.149 8

飞行状态关键指标	152~ 15 m 最大下降率/ (ft·min^-1)	152~ 61 m 平均滚转角/ (°)	152~ 61 m 最大滚转角/ (°)	15 m~ 着陆平均滚转角/(°)	15 m~ 着陆最大滚转角/(°)	305~ 152 m 平均进近速度/kn	152~ 15 m 平均进近速度/kn	15 m~ 接地平均速度/kn		305~ 15 m 平均下滑道偏离/ DOTS	305~ 15 m 最大下滑道偏离/ DOTS	接地点俯仰角/ (°)
FOQA 监控项目	下降率大	进近滚转角大		着陆滚转角大		进近速度小	进近速度大		着陆速度大	ILS下滑道偏离		接地俯仰角大/ 小	15m至接地距离远	着陆垂直过载大
自由度	2.000 0	2.000 0	2.000 0	2.000 0	2.000 0	2.000 0	2.000 0	2.000 0		2.000 0	2.000 0	2.000 0	2.000 0	2.000 0
均方	52 704.380 0	0.389 0	3.746 1	0.092 3	1.295 1	16.737 9	18.021 4	28.383 8		0.022 2	0.090 9	1.283 8	149 443.900 0	0.009 3
F	2.363 5	4.989 2	5.949 3	2.129 1	3.867 7	0.965 1	0.839 4	1.231 5		0.326 7	0.520 5	2.798 5	8.757 4	1.306 2
显著性	0.094 8	0.007 0	0.002 7	0.119 6	0.021 3	0.381 4	0.432 4	0.292 4		0.721 4	0.594 4	0.061 5	0.000 2	0.271 4
保守型均值	-869.914 0	0.737 9	1.961 4	0.556 3	1.374 9	147.756 2	147.967 6	140.177 7		0.067 4	0.392 5	0.938 1	805.400 3	1.086 0
普通型均值	-897.619 0	0.782 0	2.094 2	0.588 2	1.508 2	147.435 1	147.792 4	139.826 7		0.064 9	0.419 9	0.800 8	811.659 8	1.083 3
激进型均值	-878.903 0	0.812 5	2.193 7	0.587 2	1.467 9	147.274 5	147.457 1	139.536 2		0.082 3	0.383 0	0.864 7	766.693 7	1.095 0