Prediction model of pilot maneuver stability based on LSTM

doi:10.16265/j.cnki.issn1003-3033.2024.12.0145

Abstract

Abstract:

To predict unsafe events for pilots in real time, a LSTM neural network was used to assess pilot maneuver stability and pilot maneuvering quality was improved by optimizing indicators. Firstly, a set of human-machine maneuvering factors presenting the pilot's maneuvering behavior characteristics was proposed by analyzing the pilot's stability maneuvering QAR data in flight. Secondly, the factors affecting the stability maneuvering of the aircraft were analyzed, and a gray correlation analysis method was used to determine the 15 characteristic parameters of associated risks from the 37 monitoring parameters closely related to the stability of the aircraft. Then, the LSTM model was used to train and test the data to predict the pilot's maneuvering stability, and indicators were proposed to evaluate safety stability quality. Finally, ML was used to rank the importance of relevant influencing factors to improve model validity. The results indicated that the time series model effectively eliminated the interference of parameters with little or no correlation with the prediction results in the original parameters. The stability model can predict risks with high accuracy and provide pilots with a 3-4 s time margin to take preventive measures and reduce unsafe incident occurrence during flight.

Key words: long short-term memory (LSTM), pilot, maneuvering stability, prediction model, quick access recorder (QAR), machine learning (ML)

CLC Number:

X949

WANG Wenchao, HE Jian, SONG Baisheng, WANG Lei. Prediction model of pilot maneuver stability based on LSTM[J]. China Safety Science Journal, 2024, 34(12): 48-55.

Figures/Tables 10

Table 1

Fig.1

Table 2

Table 3

Fig.2

Fig.3

Table 4

Table 5

Fig.4

Fig.5

References 20

[1]	中国民航总局航空安全办公室. 中国民航调查报告,2023[R], 2023.
[2]	中国民用航空局. 中国民用航空"十四五"发展规划,2021[R], 2021.
[3]	祁明亮, 邵雪焱, 池宏. QAR超限事件飞行操作风险诊断方法[J]. 北京航空航天大学学报, 2011, 37(10):1207-1210.
	QI Mingliang, SHAO Xueyan, CHI Hong. Flight operations risk diagnosis method on quick-access-record exceedance[J]. Journal of Beijing University of Aeronautics and Astronautics, 2011, 37(10):1207-1210.
[4]	邵雪焱, 池宏, 高敏刚. 基于copula函数的飞行操作风险分析[J]. 数理统计与管理, 2012, 7(5):35-37.
	SHAO Xueyan, CHI Hong, GAO Mingang. Risk analysis of operations in flight based on copula[J]. Application of Statistics and Management, 2012, 7(5):35-37.
[5]	冯兴杰, 吴稀钰, 赵杰, 等. QAR数据仓库在Hive中的构建[J]. 计算机工程与应用, 2017, 53(11):90-94. doi: 10.3778/j.issn.1002-8331.1603-0290
	FENG Xingjie, WU Xiyu, ZHAO Jie, et al. Data warehouse of QAR based on Hive[J]. Computer Engineering and Applications, 2017, 53(11):90-94. doi: 10.3778/j.issn.1002-8331.1603-0290
[6]	HUNTER P A. Flight measurements of the flying qualities of five light airplanes[J]. Atmospheric Environment, 1948, 34(20):3271-3280.
[7]	COOPER G E, HARPER R P. The use of pilot rating in the evaluation of aircraft handing qualities[J]. Moffett Field,CA, 1969, 36(9):1557-1562.
[8]	杜红兵, 王雪莉. 基于贝叶斯网络的可控飞行撞地事故原因分析方法[J]. 安全与环境学报, 2009, 9(5):136-139.
	DU Hongbing, WANG Xueli. Cause analysis of the controlled flight bumping into terrain based on the Bayesian network[J]. Journal of Safety and Environment, 2009, 9(5):136-139.
[9]	何鹏, 孙瑞山. 中国民航典型征候的趋势研究[J]. 中国安全科学学报, 2023, 33(7):133-139. doi: 10.16265/j.cnki.issn1003-3033.2023.07.1459
	HE Peng, SUN Ruishan. Study on trend of typical incidents of civil aviation in China[J]. China Safety Science Journal, 2023, 33(7):133-139. doi: 10.16265/j.cnki.issn1003-3033.2023.07.1459
[10]	张光远, 邓龙, 王亚伟, 等. 基于脑电信号特征的高铁调度员疲劳状态识别[J]. 中国安全科学学报, 2024, 34(6):235-246. doi: 10.16265/j.cnki.issn1003-3033.2024.06.1674
	ZHANG Guangyuan, DENG Long, WANG Yawei, et al. Recognition of fatigue state of high-speed rail dispatchers based on EEG signal characteristics[J]. China Safety Science Journal, 2024, 34(6):235-246. doi: 10.16265/j.cnki.issn1003-3033.2024.06.1674
[11]	潘卫军, 张衡衡, 吴天祎, 等. 基于神经网络的飞机着陆速度预测模型[J]. 舰船电子工程, 2022, 42(1):73-78.
	PAN Weijun, ZHANG Hengheng, WU Tianyi, et al. Prediction model of aircraft landing speed based on neural network[J]. Ship Electronic Engineering, 2022, 42(1):73-78.
[12]	汪磊, 孙瑞山, 吴昌旭, 等. 基于飞行QAR数据的重着陆风险定量评价模型[J]. 中国安全科学学报, 2014, 24(2):88-92.
	WANG Lei, SUN Ruishan, WU Changxu, et al. A flight QAR based model for hard landing risk quantitative evaluation[J]. China Safety Science Journal, 2014, 24(2):88-92.
[13]	NAGI A, MOHAMED D, ISMAIL E, et al. Risk assessment of LNG and FLNG vessels during manoeuvring in open sea[J]. Journal of Ocean Engineering and Science, 2018, 26(8):67-69.
[14]	WANG Lei, SUN Ruishan, WU Changxu. An analysis of flight quick access recorder (QAR) data and its applications in preventing landing incidents[J]. Reliability Engineering and System Safety, 2014, 23(11):89-91.
[15]	ZAZZI A, DAS D, HUSSEN L, et al. Scalable orthogonal delay-division multiplexed OEO artificial neural network trained for TI-ADC equalization[J]. Photonics Research, 2024, 12(1):85-105.
[16]	刘莉雯, 张天伟, 茹斌. 多参数融合的飞行品质评估模型的建立[J]. 计算机工程与科学, 2016, 38(6):1262-1268.
	LIU Liwen, ZHANG Tianwei, RU Bin. A flying qualities assessment model based on multi-parameter integration[J]. Computer Engineering & Science, 2016, 38(6):1262-1268.
[17]	LIU Jiaqi, FENG Yunwen, TENG Da, et al. Operational reliability evaluation and analysis framework of civil aircraft complex system based on intelligent extremum machine learning model[J]. Reliability Engineering & System Safety, 2023, 23(35):684-686.
[18]	LI Chongfeng, SUN Ruishan, PAN Xing. Takeoff runway overrun risk assessment in aviation safety based on human pilot behavioral characteristics from real flight data[J]. Safety Science, 2023, 158(69):533-536.
[19]	TANG Ziyang, LI Zuotian, HOU Tieying. SiGra:single-cell spatial elucidation through an image-augmented graph transformer[J]. Nature Communication, 2023, 14(1):23-37.
[20]	戢晓峰, 乔新. 建成环境对行人交通事故严重程度的非线性影响[J]. 交通运输系统工程与信息, 2023, 23(1):314-323.
	JI Xiaofeng, QIAO Xin. Nonlinear influence of built environment on pedestrian traffic accident severity[J]. Journal of Transportation Systems Engineering and Information Technology, 2023, 23(1):314-323.

指标类型	监控参数	指标类型	监控参数
结构类	襟翼位置	轨迹类	航向道偏差
	左发EGT		下滑道偏差
	右发EGT		垂直偏差
	左油门杆		左发N1转速
	右油门杆		右发N1转速
	方向舵		左发N2转速
	左升降舵位置		右发N2转速
	右升降舵位置		风速
	左驾驶盘位置		风向
	右驾驶盘位置		磁航向
	左驾驶杆位置	姿态类	飞行指引坡度偏差
	右驾驶杆位置		飞行指引仰角偏差
限制类	下降率		左攻角
	升降率		右攻角
	垂直载荷		坡度
	气压高度		仰角
	无线电高度		空速
	经度		地速
	纬度		—

参数序号	参数名称	参数序号	参数名称
1	下降率	9	风速
2	地速	10	风向
3	空速	11	飞行指引仰角偏差
4	垂直载荷	12	飞行指引坡度偏差
5	仰角	13	襟翼位置
6	坡度	14	下滑道偏差
7	磁航向	15	航向道偏差
8	垂直偏差

平稳性等级分类	指标标准
优秀	下滑道偏差在[-0.01,0.01],或垂直偏差为0,或飞行指引仰角偏差为0,或航向道偏差在[-0.005,0]∪[0,0.005],或飞行指引坡度偏差在[-0.2,-<0.1]∪[0.1,0.2]
良好	下滑道偏差在[-0.02,0.02],或垂直偏差为0,或飞行指引仰角偏差为0,或航向道偏差在[-0.007,0]∪[0,0.007],或飞行指引坡度偏差在[-0.3,-<0.2]∪[0.2,0.3]
一般	下滑道偏差在[-0.03,0.03],或垂直偏差为0,或飞行指引仰角偏差为0,或航向道偏差在[-0.009,0]∪[0,0.009],或飞行指引坡度偏差在[-0.4,-<0.3]∪[0.3,0.4]
需改进	下滑道偏差在[-0.04,0.04],或垂直偏差为0,或飞行指引仰角偏差为0,或航向道偏差在[-0.01,0]∪[0,0.01],或飞行指引坡度偏差在[-0.5,-<0.4]∪[0.4,0.5]
不合格	下滑道偏差在[-0.05,0.05],或垂直偏差为0,或飞行指引仰角偏差为0,或航向道偏差在[-∞,-0.01]∪[0.01,+∞],或飞行指引坡度偏差在[-∞,-<0.5]∪[0.5,+∞]

输入步长	输出步长	训练集损失值	测试集损失值	测试集平均绝对误差
25	25	0.002 4	0.002 1	0.042 2
25	50	0.002 3	0.002 6	0.050 2
25	75	0.005 7	0.003 9	0.060 1
50	25	0.001 9	0.002 1	0.030 1
50	50	0.004 1	0.003 1	0.039 6
50	75	0.003 6	0.004 5	0.051 1
75	25	0.001 8	0.002 7	0.010 2
75	50	0.003 8	0.001 9	0.015 3
75	75	0.002 2	0.004 2	0.023 3

输入参数数量	测试集损失值	测试集平均绝对误差
37	0.010 2	0.021 5
15	0.000 9	0.011 2