基于改进YOLOX-s的机场跑道冰雪状态感知

doi:10.16183/j.cnki.jsjtu.2022.303

上海交通大学学报 ›› 2023, Vol. 57 ›› Issue (10): 1292-1304.doi: 10.16183/j.cnki.jsjtu.2022.303

所属专题：《上海交通大学学报》2023年“交通运输工程”专题

基于改进YOLOX-s的机场跑道冰雪状态感知

邢志伟¹, 阚犇¹, 刘子硕², 李彪¹(), 罗谦³

1.中国民航大学电子信息与自动化学院, 天津 300300
2.天津航空机电有限公司, 天津 300308
3.中国民航局第二研究所工程技术研究中心, 成都 610041

收稿日期:2022-08-04 修回日期:2022-11-23 接受日期:2022-12-01 出版日期:2023-10-28 发布日期:2023-10-31
通讯作者: 李彪 E-mail:1833022770@163.com.
作者简介:邢志伟(1970-),教授,从事机场运行控制,民航装备与系统的应用研究.
基金资助:
国家重点研发计划(2018YFB1601200)

Airport Pavement Snow and Ice State Perception Based on Improved YOLOX-s

XING Zhiwei¹, KAN Ben¹, LIU Zishuo², LI Biao¹(), LUO Qian³

1. School of Electronic Information and Automation, Civil Aviation University of China, Tianjin 300300, China
2. AVIC Tianjin Aviation Machinery and Electricity Co., Ltd., Tianjin 300308, China
3. Engineering Technology Research Centre, Second Research Institute of Civil Aviation Administration of China, Chengdu 610041, China

Received:2022-08-04 Revised:2022-11-23 Accepted:2022-12-01 Online:2023-10-28 Published:2023-10-31
Contact: LI Biao E-mail:1833022770@163.com.

摘要/Abstract

摘要：

针对机场冰雪跑道安全性和适航性状态感知能力不足及跑道表面状况报告交互的新需求,提出一种面向多尺度特征融合的机场跑道冰雪状态感知模型.以YOLOX-s模型为基础,在主干特征提取网络中引入全局上下文模块,获取更丰富的浅层与深层特征;将颈部结构中路径聚合网络替换为双向特征金字塔,以提升特征融合能力;在加强特征提取网络尾部添加自适应空间特征融合结构,进一步增强特征融合效果;使用α-EIoU优化损失函数,提高模型收敛速度与精度.实验结果表明,改进后的YOLOX-s模型在跑道冰雪实验系统所得的冰雪污染物数据集上平均精度达到了91.53%,比原始的YOLOX-s模型提高了4.68%,能够为机场跑道除冰雪作业提供决策支持.

关键词: 跑道冰雪状态感知, YOLOX-s, 全局上下文模块, 双向特征金字塔网络, 自适应空间特征融合结构, α-EIoU损失函数

Abstract:

Aimed at the lack of awarness of safety and airworthiness state perception ability of airport ice runway and the new demand of interaction of runway surface condition report, a multi-scale feature fusion based ice and snow state perception model of airport runway is proposed. Based on the YOLOX-s model, first, the global context block (GC block) is introduced into the backbone feature extraction network to obtain more abundant shallow and deep features. Then, the PANet networks in neck are replaced with the bi-directional feature pyramid network (BiFPN) to improve the feature fusion ability. Afterwards, an adaptive spatial feature fusion (ASFF) structure is added to the tail of the enhanced feature extraction network to further enhance the feature fusion effect. Finally, α-EIoU is used to optimize the loss function to improve the convergence speed and accuracy of the model. The experimental results show that the improved YOLOX-s model has an average accuracy of 91.53% in the snow and ice pollutant data set obtained from the runway snow and ice experimental system, which is 4.68% higher than the original YOLOX-s model, and can provide decision-making support for airport runway snow removal operations.

Key words: pavement snow and ice state perception, YOLOX-s, global context block (GC block), bi-directional feature pyramid network (BiFPN), adaptive spatial feature fusion (ASFF), α-EIoU loss function

中图分类号:

TP391.9

邢志伟, 阚犇, 刘子硕, 李彪, 罗谦. 基于改进YOLOX-s的机场跑道冰雪状态感知[J]. 上海交通大学学报, 2023, 57(10): 1292-1304.

XING Zhiwei, KAN Ben, LIU Zishuo, LI Biao, LUO Qian. Airport Pavement Snow and Ice State Perception Based on Improved YOLOX-s[J]. Journal of Shanghai Jiao Tong University, 2023, 57(10): 1292-1304.

图/表 19

图1

图2

图3

表1

图4

图5

图6

图7

图8

图9

图10

图11

图12

图13

表2

图14

表3

图15

图16

参考文献 5

[6]	HOSHINO S, HASHIMOTO K, TATEYAMA K, et al. Snow and ice monitoring technique for the contaminated runway[C]// AIAA SciTech 2020 Forum. New York, USA: AIAA, 2020: 767-777.
[7]	任宏宇, 苑丹丹, 桂康, 等. 复阻抗式结冰探测技术的温度补偿方法研究[J]. 仪器仪表学报, 2021, 42(6): 88-94.
	REN Hongyu, YUAN Dandan, GUI Kang, et al. A temperature compensation method for complex impedance ice detection[J]. Chinese Journal of Scientific Instrument, 2021, 42(6): 88-94.
[8]	HONG S B, LEE B W, KIM C H, et al. System dynamics modeling for estimating the locations of road icing using GIS[J]. Applied Science-Basel, 2021, 11(18): 8537-8547.
[9]	勾一, 李清英, 刘森云. 基于闪光红外热波探测的积冰界线识别算法研究[J/OL]. 实验流体力学. http://kns.cnki.net/kcms/detail/11.5266.V.20220621.1044.002.html.
	GOU Yi, LI Qingying, LIU Senyun. Flash infrared thermal wave detection of ice surface edge[J/OL]. Journal of Experiment in Fluid Mechanics. http://kns.cnki.net/kcms/detail/11.5266.V.20220621.1044.002.html.
[10]	QIN F G F, CHEN X D, FARID M M. Growth kinetics of ice films spreading on a subcooled solid surface[J]. Separation & Purification Technology, 2004, 39(1/2): 109-121.
[11]	QIN F G F, ZHAO J C, RUSSELL A B, et al. Simulation and experiment of the unsteady heat transport in the onset time of nucleation and crystallization of ice from the subcooled solution[J]. International Journal of Heat and Mass Transfer, 2003, 46(17): 3221-3231. doi: 10.1016/S0017-9310(03)00097-8 URL
[12]	CHEN B, ZHOU C, LIU Y, et al. Correlation analysis of runway icing parameters and improved PSO-LSSVM icing prediction[J]. Cold Regions Science and Technology, 2022, 54(15): 193-198. doi: 10.1016/j.coldregions.2008.06.005 URL
[13]	COSTA M, MONIACI W, PASERO F. INFO: An artificial neural system to forecast ice formation on the road[C]// IEEE International Symposium on Computational Intelligence for Measurement Systems & Applications. Lugano, Switzerland: IEEE, 2003: 127-132.
[14]	ENVELOPE Z D. Uncertainty-aware accurate insulator fault detection based on an improved YOLOX model[J]. Energy Reports, 2022, 8: 12809-12821. doi: 10.1016/j.egyr.2022.09.195 URL
[15]	ISHIKAWA N, NARUSE R, MAENO K. Heat balance characteristics of road snow and ice[J]. Low Temperature Science (Physics), 1987, 46(2): 151-162.
[16]	MAENO K, NARITA E, NISHIMURA K, et al. Road snow and ice structure and new classification[J]. Low Temperature Science (Physics), 1987, 46(3): 119-133.
[17]	王雪莹. 道路冰雪与路面粘附特性及除雪机理研究[D]. 吉林: 吉林大学, 2019.
	WANG Xueying. Study on adhesion characteristics and snow removal mechanism of road ice, snow and pavement[D]. Jilin: Jilin University, 2019.
[18]	CAO Y, XU J, LIN S, et al. GCNet: Non-Local networks meet Squeeze-Excitation networks and beyond[C]//2019 IEEE/CVF International Conference on Computer Vision Workshop. Seoul, Koera: IEEE, 2020: 390-401.
[19]	WANG X, GIRSHICK R, GUPTA A, et al. Non-local neural networks[C]//2018 IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City, USA: IEEE, 2018: 225-235.
[20]	HU J, SHEN L, ALBANIE S, et al. Squeeze-and-Excitation networks[J]. IEEE Transactions on Pattern Recognition and Machine Intelligence, 2017, 42(8): 2011-2023. doi: 10.1109/TPAMI.34 URL
[21]	TAN M, PANG R, LE Q V. EfficientDet: Scalable and efficient object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle, USA: IEEE, 2020: 10778-10787.
[22]	李登攀, 任晓明, 颜楠楠. 基于无人机航拍的绝缘子掉串实时检测研究[J]. 上海交通大学学报, 2022, 56(8): 994-1003.
	LI Dengpan, REN Xiaoming, YAN Nannan. Real-time detection of insulator drop string based on UAV aerial photography[J]. Journal of Shanghai Jiao Tong University, 2022, 56(8): 994-1003.
[23]	LI X LIW, REN D, et al. Enhanced blind face restoration with multi-exemplar images and adaptive spatial feature fusion[C]// 2020 IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City, USA: IEEE, 2020: 235-245.
[1]	International Civil Aviation Organization. Global Reporting Format[EB/OL]. (2020-05-26) [2022-07-15]. https://www.icao.int/safety/Pages/GRF.aspx.
[2]	KIM H G, JANG M S, LEE Y S, et al. A black ice detection method using infrared camera and YOLO[J]. Journal of the Korea Institute of Information and Communication Engineering, 2020, 25(12): 1874-1881.
[3]	MA X, CHI R. Method for black ice detection on roads using tri-wavelength backscattering measurements[J]. Applied Optics, 2020, 59(24): 7242-7246. doi: 10.1364/AO.398772 URL
[4]	BABY K C, GEORGE B. A capacitive ice layer detection system suitable for autonomous inspection of runways using an ROV[C]// Robotic and Sensors Environment. Magdeburg, Germany: IEEE, 2012: 127-132.
[24]	程换新, 蒋泽芹, 程力, 等. 基于改进YOLOX-S的安全帽反光衣检测算法[J]. 电子测量技术, 2022, 45(6): 130-135.
	CHENG Huanxin, JIANG Zeqin, CHENG Li, et al. Helmet and reflective clothing detection algorithm based on improved YOLOX-S[J]. Electronic Measurement Technology, 2022, 45(6): 130-135.
[25]	ZHANG Y F, REN W, ZHANG Z, et al. Focal and efficient IOU loss for accurate bounding box regression[J]. Neurocomputing, 2022, 506: 146-157. doi: 10.1016/j.neucom.2022.07.042 URL
[26]	HE J, ERFANI S, MA X, et al. Alpha-IoU: A family of power intersection over union losses for bounding box regression[C]// 2021 IEEE Conference on Computer Vision and Pattern Recognition. Shenzhen, China: IEEE, 2021: 280-291.
[5]	TROIANO A, PASERO E, MESIN L. New system for detecting road ice formation[J]. IEEE Transactions on Instrumentation & Measurement, 2011, 60(3): 1091-1101.

状态	特点	晶体状态	灰度区间
雪	颗粒状,遇热变形	粒径0.05 mm以上互不相连的圆粒	48~80
雪浆	雪半融解产物	粒径0.5~1 mm的圆粒相互成网状连接	81~111
融雪	雪融解后产物	粒径1 mm以上互不相连的颗粒	235~255
冻冰	雨雪冻结而成	粒径0.05~2 mm的多结晶冰,含粒径0.01~0.1 mm气泡	112~144
湿冰	冻冰半融解产物	粒径0.1~0.4 mm的多结晶冰,含粒径0.1~0.5 mm气泡	145~207
水	冰雪完全融解产物含杂质	透明且流动不止,含有大量气泡	—

算法	φ_AP						ψ_mAP	?_mR	FPS/ (帧·s^-1)
算法	雪	雪浆	融雪	冻冰	湿冰	水	ψ_mAP	?_mR	FPS/ (帧·s^-1)
Faster R-CNN	0.789 8	0.842 1	0.814 0	0.833 4	0.794 2	0.922 5	0.832 7	0.702 5	18
SSD	0.764 3	0.820 1	0.801 0	0.805 6	0.779 5	0.911 5	0.813 7	0.683 4	16
YOLOv5-s	0.795 6	0.854 2	0.825 8	0.853 2	0.802 5	0.931 3	0.843 8	0.726 2	25
YOLOX-s	0.810 7	0.882 5	0.876 3	0.868 8	0.826 3	0.946 5	0.868 5	0.790 2	24
IYOLOX-s	0.871 1	0.915 4	0.923 9	0.915 8	0.883 8	0.981 9	0.915 3	0.834 5	23

组别	GC	BiFPN	ASFF	α-EIoU	ψ_mAP	?_mR
G1	×	×	×	×	0.868 5	0.790 2
G2	√	×	×	×	0.881 4	0.801 2
G3	√	√	×	×	0.892 1	0.812 3
G4	√	√	√	×	0.902 9	0.823 9
G5	√	√	√	√	0.915 3	0.834 5

基于改进YOLOX-s的机场跑道冰雪状态感知

Airport Pavement Snow and Ice State Perception Based on Improved YOLOX-s

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 19

参考文献 5

相关文章 1

编辑推荐

Metrics

本文评价