冰雪天气下基于LSTM的跑道温度数据-机理联合预测

陈斌; 刘悦; 尹开浪; 方珣

doi:10.13700/j.bh.1001-5965.2022.0579

冰雪天气下基于LSTM的跑道温度数据-机理联合预测

doi: 10.13700/j.bh.1001-5965.2022.0579

陈斌^{1, 2, ,},
刘悦^{1, 2},
尹开浪^{1, 2},
方珣^{1, 2}

1.
中国民航大学电子信息与自动化学院，天津 300300
2.
中国民航大学航空地面特种设备研究基地，天津 300300

详细信息

通讯作者:
E-mail：chenbindavid@163.com

中图分类号: V219
计量
- 文章访问数: 141
- HTML全文浏览量: 107
- PDF下载量: 5
- 被引次数: 0
出版历程
- 收稿日期: 2022-07-05
- 录用日期: 2022-09-25
- 网络出版日期: 2022-10-09
- 整期出版日期: 2024-07-18

Runway temperature data mechanism joint prediction based on LSTM under ice and snow

CHEN Bin^{1, 2
, ,},
LIU Yue^{1, 2},
YIN Kailang^{1, 2},
FANG Xun^{1, 2}

1.
School of Electronic Information and Automation，Civil Aviation University of China，Tianjin 300300，China
2.
Department of Aviation Ground Special Equipment，Civil Aviation University of China，Tianjin 300300，China

More Information

Corresponding author: E-mail：chenbindavid@163.com

摘要

摘要:
温度是跑道结冰的重要因素，针对跑道除冰运行的跑道热特性参数瞬态变化问题和温度周期序列缓慢变化特性，建立冰雪天气下基于长短时记忆(LSTM)的跑道温度数据-机理联合预测模型。通过最大信息系数法选择数据模型的输入特征变量，采用动态时间弯曲法进行跑道温度数据聚类划分，建立基于LSTM的数据预测模型；通过跑道热力学知识获取跑道温度预测机理模型，采用最小误差赋权法建立跑道温度数据-机理联合预测模型。仿真预测显示，预测时长为20 min、残差阈值为±0.5℃时，数据-机理联合预测模型优于单独的数据预测模型和机理模型，预测准确率可达99.34%；横向对比显示，在相同边界条件下，数据-机理联合预测模型优于BP神经网络、多元回归模型和支持向量机模型，平均准确率提高26.11%。研究表明，基于LSTM的跑道温度数据-机理联合预测模型契合冰雪天气下跑道除冰运行实际，可获得较好的跑道温度短时预测结果。
- 跑道温度 /
- 冰雪天气 /
- 预测 /
- 长短时记忆神经网络 /
- 动态时间弯曲
Abstract:
Runway temperature is an important factor in runway icing. Fully considering the transient characteristics of the temperature mechanism model and the time sequence of temperature multivariate time series data, the paper has developed a joint model based on the long short term memory (LSTM) neural network and temperature mechanism model. Firstly, the influencing elements with a greater correlation with runway temperature were selected by the study using the maximum information coefficient approach to serve as the model’s input. Secondly, the paper uses the dynamic time warping method to cluster temperature data under different snowfall conditions, and then develops an LSTM model adapted to different snowfall or icing situations. Finally, to solve the disadvantage of LSTM which can not be characterized by the runway parameters that change irregularly and frequently, the paper developed a joint model based on the LSTM neural network and temperature mechanism model by using the minimum error method. The joint model’s degree of accuracy is 99.34%, which is superior than both the data model and the mechanism model, when the prediction time step is 20 minutes and the residual threshold is ±0.5°C, according to the simulation’s result based on the ice and snowfall weather condition data. With the same condition, the joint model has better accuracy than the BP model, the regression model and the support vector machine model. Average accuracy increased by 26.11%. It proved the joint model based on the LSTM neural network and temperature mechanism model has better accuracy according to the transient characteristics of the mechanism model and the periodic time sequence of the multivariate time series of pavement temperature.
- runway temperature /
- snowy and icy weather /
- predict /
- long short term memory neural network /
- dynamic time warping

HTML全文

图 1 跑道温度变化曲线

Figure 1. Changing curves of runway temperature

下载: 全尺寸图片幻灯片

图 2 聚类过程示意图

Figure 2. Diagram of clustering process

下载: 全尺寸图片幻灯片

图 3 最优规划路径示意图

Figure 3. Schematic diagram of optimal planning path

下载: 全尺寸图片幻灯片

图 4 LSTM单元结构

Figure 4. LSTM cell structure

下载: 全尺寸图片幻灯片

图 5 滚动时间窗口结构

Figure 5. Structure of rolling time window

下载: 全尺寸图片幻灯片

图 6 联合预测原理结构框图

Figure 6. Block diagram of joint prediction principle

下载: 全尺寸图片幻灯片

图 7 跑道温度聚类结果

Figure 7. Clustering result of runway temperature

下载: 全尺寸图片幻灯片

图 8 LSTM预测模型预测性能结果

Figure 8. Prediction performance results of LSTM prediction model

下载: 全尺寸图片幻灯片

图 9 LSTM预测模型的预测准确率

Figure 9. Prediction accuracy of LSTM prediction model

下载: 全尺寸图片幻灯片

图 10 联合预测流程

Figure 10. Joint prediction flow

下载: 全尺寸图片幻灯片

图 11 相似跑道温度预测结果

Figure 11. Prediction results of similar runway temperature

下载: 全尺寸图片幻灯片

图 12 联合预测模型预测结果

Figure 12. Prediction results of joint prediction model

下载: 全尺寸图片幻灯片

图 13 多元回归模型预测结果

Figure 13. Prediction results of multiple regression model

下载: 全尺寸图片幻灯片

表 1 输入变量最大信息系数值

Table 1. MIC value for input variables

序号	影响因素	MIC值
1	大气温度	0.6820
2	露点温度	0.6815
3	大气湿度	0.6346
4	风速	0.5362
5	大气压强	0.6156
6	跑道下10 cm温度	0.6870
7	跑道下20 cm温度	0.6486
8	跑道下40 cm温度	0.5230
9	大气温度（天气预报）	0.6566
10	露点温度（天气预报）	0.6283
11	大气湿度（天气预报）	0.6021

下载: 导出CSV

表 2 冰雪天气下跑道温度样本分类

Table 2. Runway temperature samples classification under ice and snow

类别	样本
Ⅰ	1、2
Ⅱ	3、4、5、6
Ⅲ	7
Ⅳ	8

下载: 导出CSV

表 3 LSTM网络参数

Table 3. Parameters of LSTM network

参数	数值
输入层维数	10
隐藏层维数	2
输出层维数	1
每个隐藏层神经元数目	80
初始学习率	0.01
训练批次	64
迭代轮数	10

下载: 导出CSV

表 4 不同预测时长下的联合预测模型权重

Table 4. Weight of joint prediction model in different prediction time step

预测时长/min	机理模型权重	LSTM预测模型权重
5	0.63	0.37
10	0.55	0.45
20	0.34	0.66
30	0.33	0.67

下载: 导出CSV

表 5 不同模型预测值的RMSE

Table 5. RMSE of different model predictions

模型	预测时长/min	RMSE
机理模型	5	0.023
	10	0.32
	20	0.64
	30	1.02
LSTM预测模型	5	0.057
	10	0.064
	20	0.32
	30	0.55
联合预测模型	5	0.002
	10	0.037
	20	0.15
	30	0.21

下载: 导出CSV

表 6 多元模型预测值的RMSE

Table 6. RMSE of multivariate model predictions

模型	预测时长/min	RMSE	准确率/%
多元回归模型	5	0.0447	86.75
	10	0.5806	87.42
	20	0.9807	88.08
	30	1.04	88.08
BP神经网络	5	0.0699	88.08
	10	0.0725	88.74
	20	0.1303	88.08
	30	0.11	87.42
支持向量机模型	5	0.018	71.52
	10	0.2243	60.93
	20	0.2549	60.93
	30	0.3872	60.26
联合预测模型	5	0.002	100
	10	0.037	100
	20	0.15	99.34
	30	0.21	99.34

下载: 导出CSV

参考文献(24)

[1]	孙立军. 沥青路面结构行为理论[M]. 北京: 人民交通出版社, 2005: 51-95. SUN L J. Theory of structural behavior of asphalt pavement[M]. Beijing: People’s Traffic Publishing House, 2005: 51-95(in Chinese).
[2]	NUIJTEN A D. Runway temperature prediction, a case study for Oslo Airport, Norway[J]. Cold Regions Science and Technology, 2016, 125: 72-84. doi: 10.1016/j.coldregions.2016.02.004
[3]	NUIJTEN A D, HØYLAND K V. Modelling the thermal conductivity of a melting snow layer on a heated pavement[J]. Cold Regions Science and Technology, 2017, 140: 20-29. doi: 10.1016/j.coldregions.2017.04.008
[4]	HERMANSSON Å. Mathematical model for calculation of pavement temperatures: Comparison of calculated and measured temperatures[J]. Transportation Research Record: Journal of the Transportation Research Board, 2001, 1764(1): 180-188. doi: 10.3141/1764-19
[5]	HERMANSSON Å. Simulation model for calculating pavement temperatures including maximum temperature[J]. Transportation Research Record: Journal of the Transportation Research Board, 2000, 1699(1): 134-141. doi: 10.3141/1699-19
[6]	HO C H, ROMERO P. Low design temperatures of asphalt pavements in dry-freeze regions: Predicting by means of solar radiation, transient heat transfer, and finite element method[J]. Transportation Research Record: Journal of the Transportation Research Board, 2009, 2127(1): 60-71. doi: 10.3141/2127-08
[7]	张丽娟, 黄建武, 许薛军. 基于拉普拉斯变换的路面一维时变温度场预测[J]. 华南理工大学学报(自然科学版), 2017, 45(11): 10-16. ZHANG L J, HUANG J W, XU X J. Prediction of time-dependent one-dimension temperature field of pavement on the basis of Laplace transform[J]. Journal of South China University of Technology (Natural Science Edition), 2017, 45(11): 10-16(in Chinese).
[8]	陈斌, 刘悦, 李庆真, 等. 冰雪天气下基于MFOA-KELM残差修正的跑道温度混合预测[J]. 北京航空航天大学学报, 2022, 48(11): 2153-2164. CHEN B, LIU Y, LI Q Z, et al. Runway temperature hybrid prediction based on MFOA-KELM residual correction under ice and snow[J]. Journal of Beijing University of Aeronautics and Astronautics, 2022, 48(11): 2153-2164(in Chinese).
[9]	孙立军, 秦健. 沥青路面温度场的预估模型[J]. 同济大学学报(自然科学版), 2006, 34(4): 480-483. doi: 10.3321/j.issn:0253-374X.2006.04.011 SUN L J, QIN J. Prediction model on temperature field in asphalt pavement[J]. Journal of Tongji University (Natural Science), 2006, 34(4): 480-483(in Chinese). doi: 10.3321/j.issn:0253-374X.2006.04.011
[10]	NANTASAI B, NASSIRI S. Winter temperature prediction for near-surface depth of pervious concrete pavement[J]. International Journal of Pavement Engineering, 2019, 20(7): 820-829. doi: 10.1080/10298436.2017.1353389
[11]	LIU B, YAN S, YOU H L, et al. Road surface temperature prediction based on gradient extreme learning machine boosting[J]. Computers in Industry, 2018, 99: 294-302. doi: 10.1016/j.compind.2018.03.026
[12]	邱欣, 洪皓珏, 杨青, 等. 基于APRIORI-GBDT算法的沥青路面路表温度预测[J]. 公路交通科技, 2019, 36(5): 1-10. QIU X, HONG H J, YANG Q, et al. Prediction of temperature of asphalt pavement surface based on APRIORI-GBDT algorithm[J]. Journal of Highway and Transportation Research and Development, 2019, 36(5): 1-10(in Chinese).
[13]	魏鑫. 基于时间序列机场跑道温度预测模型[D]. 大连: 大连海事大学, 2012. WEI X. Temperature prediction model based on time series for airport runway[D]. Dalian: Dalian Maritime University, 2012(in Chinese).
[14]	汤筠筠, 郭忠印. 基于自回归求和移动平均的冬季路温短临预测[J]. 同济大学学报(自然科学版), 2017, 45(12): 1824-1829. doi: 10.11908/j.issn.0253-374x.2017.12.012 TANG J J, GUO Z Y. Pavement temperature short-impending prediction based on ARIMA in winter[J]. Journal of Tongji University (Natural Science), 2017, 45(12): 1824-1829(in Chinese). doi: 10.11908/j.issn.0253-374x.2017.12.012
[15]	YU M, XU F, HU W, et al. Using long short-term memory (LSTM) and Internet of things (IoT) for localized surface temperature forecasting in an urban environment[J]. IEEE Access, 2021, 9: 137406-137418. doi: 10.1109/ACCESS.2021.3116809
[16]	FANG Z, CRIMIER N, SCANU L, et al. Multi-zone indoor temperature prediction with LSTM-based sequence to sequence model[J]. Energy and Buildings, 2021, 245: 111053. doi: 10.1016/j.enbuild.2021.111053
[17]	LIU J, ZHANG T, HAN G J, et al. TD-LSTM: Temporal dependence-based LSTM networks for marine temperature prediction[J]. Sensors, 2018, 18(11): 3797. doi: 10.3390/s18113797
[18]	颜弋凡, 安路达, 吕志民. 基于最大互信息系数属性选择的冷轧产品机械性能预测[J]. 中南大学学报(自然科学版), 2020, 51(1): 68-75. YAN Y F, AN L D, LÜ Z M. Prediction of mechanical properties of cold rolled products based on maximal information coefficient attribute selection[J]. Journal of Central South University (Science and Technology), 2020, 51(1): 68-75(in Chinese).
[19]	李正欣, 刘畅, 吴诗辉, 等. 特征点分段提取的时间序列模式匹配方法[J]. 北京航空航天大学学报2023, 49(7): 1593-1599. LI Z X, LIU C, W S H, et al. A time series matching method for feature point segmentation extraction[J]. Journal of Beijing University of Aeronautics and Astronautic, 2023, 49(7): 1593-1599(in Chinese).
[20]	ALAEE S, MERCER R, KAMGAR K, et al. Time series motifs discovery under DTW allows more robust discovery of conserved structure[J]. Data Mining and Knowledge Discovery, 2021, 35(3): 863-910. doi: 10.1007/s10618-021-00740-0
[21]	张慧. 深度学习中优化算法的研究与改进[D]. 北京: 北京邮电大学, 2018. ZHANG H. Research and improvement of optimization algorithms in deep learning [D]. Beijing: Beijing University of Posts and Telecommunications, 2018(in Chinese).
[22]	顾尔祚. 流体力学中的有限差分法基础[M]. 上海: 上海交通大学出版社, 1988. GU E Z. Fundamentals of finite difference method in fluid mechanics[M]. Shanghai: Shanghai Jiao Tong University Press, 1988(in Chinese).
[23]	中国民用航空局. 运输机场跑道表面状况评估程序[S]. 北京: 中国民用航空局, 2022. Civil Aviation Administration of China. Runway surface condition of transport airport assessment procedure[S]. Beijing: Civil Aviation Administration of China, 2022(in Chinese).
[24]	王志刚, 王业光, 杨宁, 等. 基于LSTM的飞行数据挖掘模型构建方法[J]. 航空学报, 2021, 42(8): 525800. WANG Z G, WANG Y G, YANG N, et al. Construction method of flight data mining model based on LSTM[J]. Acta Aeronautica et Astronautica Sinica, 2021, 42(8): 525800(in Chinese).