设为首页
加入收藏
Expressway Pavement Temperature Forecast Based on LSTM and Prior Knowledge
-
摘要: 为了精准预报高速公路路面温度, 为车辆安全行驶提供气象保障, 采用2019—2022年南京市绕城高速公路上9个交通气象站及ERA5-land再分析数据, 通过构建时间序列特征工程、引入物理机制相关数据两类方法结合先验知识, 运用长短期记忆神经网络模型建立研究区域内4个交通气象站未来3 h逐10 min路面温度多步预报模型并进行验证; 在此基础上, 将已建立的模型应用于其他交通气象站, 探究模型的适用性。结果表明: 结合先验知识后, 模型预报性能明显提高, 准确率在85%以上, 且随着预报时效的延长, 性能提升更为明显, 准确率最高提升36%; 模型能较为准确地预报路面极端低温发生的时间和极值, 且在预报时效较短时对路面极端高温的预报也具有一定参考价值; 利用已建立的模型对其他交通气象站的路面温度进行预报时, 准确率在62%以上, 在预报时效较短时效果较好, 准确率在80%以上, 且交通气象站所处的下垫面背景类型对模型的选择起关键作用。Abstract: The variation of road surface temperature along highways is a crucial indicator for traffic meteorological conditions and constitutes a significant focus in the research on meteorological disasters related to transportation. Accurate forecast of pavement temperature, timely issuance of pavement condition warnings, and alerting relevant personnel to take defensive measures are of paramount importance for ensuring the safety of people's lives and property. Observations from 4 expressway meteorological stations along Nanjing City Ring Expressway and the corresponding ERA5-land reanalysis data from 2019 to 2022 are analyzed. Utilizing feature engineering techniques that consider the daily and seasonal temperature variations as well as temperature trends, a long-short-term memory (LSTM) neural network model, incorporating prior knowledge, is established for multi-step pavement temperature forecasting at 10 min intervals for the next 3 hours. The models are validated under different scenarios including extreme high and low pavement temperature conditions. They are further transferred and applied to 5 additional meteorological stations to investigate the model universality. This approach addresses the challenge of pavement temperature forecasting for stations with limited historical data due to new construction or equipment maintenance. Results indicate that the incorporation of prior knowledge facilitates a more comprehensive consideration of environmental influences by maximizing the feature extraction capabilities of LSTM. All forecasting performance metrics of the model exhibit significant improvements, with the accuracy exceeding 85%. As the forecast lead time extends, the enhancement in various forecast metrics becomes more pronounced, reaching a maximum accuracy improvement of 36%. The model accurately predicts the occurrence time and extremities of extreme low temperatures, but it exhibits relatively weaker capabilities in forecasting extreme high temperatures, with approximately 1 h advance in occurrence time and an underestimation of about 4 ℃. Despite this generally lower forecasting efficacy, the model still provides valuable information. When applying models to forecast pavement temperatures at other meteorological stations, the accuracy exceeds 62%. The forecast performance is better for short lead times, with the accuracy surpassing 80%. The underlying surface type plays a crucial role in the selection of different models. The suburban station model performs relatively optimally for urban meteorological stations and suburban meteorological stations, while the rural station model performs relatively optimally for rural meteorological stations.
-
图 1 南京市绕城高速公路交通气象站地理位置分布
Fig. 1 Geographical location distribution of traffic meteorological stations along Nanjing Ring Expressway
图 2 4个交通气象站不同方案预报能力评估
Fig. 2 Evaluation of forecasting performance using different schemes at 4 traffic meteorological stations
图 3 2022年交通气象站M9522 1 h预报时效5种方案模型预报效果及对比
(a)方案1，(b)方案2，(c)方案3，(d)方案4，(e)方案5，(f)各方案预报能力评价指标对比(各指标均进行归一化处理)
Fig. 3 Forecasting performance and comparison of 1 h lead time for M9522 traffic meteorological station in 2022
(a)Scheme 1, (b)Scheme 2, (c)Scheme 3, (d)Scheme 4, (e)Scheme 5, (f)normalized comparison of forecasting performance for each scheme
图 5 2022年2月21日路面极端低温个例路面温度观测值与方案5预报值对比
Fig. 5 Comparisons between observation and forecast of Scheme 5 for extreme low pavement temperature cases on 21 Feb 2022
图 6 2022年8月8日路面极端高温个例路面温度观测值与方案5预报值对比
Fig. 6 Comparisons between observation and forecast of Scheme 5 for extreme high pavement temperature cases on 8 Aug 2022
表 1 模型方案设定
Table 1 Model scheme setting
模型方案 交通气象站观测数据 物理机制相关变量 特征变量 方案1 √ 方案2 √ 方案3 √ √ 方案4 √ √ 方案5 √ √ √ 
下载: 导出CSV
表 2 低温和高温路面状况下方案5准确率和TS评分
Table 2 Accuracy rate and threat score of Scheme 5 under low and high pavement temperature conditions
路面状况 交通气象站 1 h时效 2 h时效 3 h时效 准确率/% TS评分 准确率/% TS评分 准确率/% TS评分 低温 M9518 100 0.64 99.77 0.63 99.77 0.60 M9520 100 0.56 100 0.52 100 0.53 M9522 100 0.53 98.41 0.49 94.42 0.42 M9526 99.40 0.71 100 0.74 99.28 0.69 高温 M9518 27.82 0.41 11.55 0.17 6.45 0.07 M9520 58.10 0.56 25.05 0.28 13.75 0.16 M9522 67.67 0.52 45.65 0.39 22.02 0.21 M9526 51.27 0.62 43.79 0.51 28.74 0.41
下载: 导出CSV
表 3 路面极端低温和极端高温个例中方案5模型评估
Table 3 Evaluation of Scheme 5 in extreme high and extreme low pavement temperature cases
个例 交通气象站 1 h时效 3 h时效 准确率/% 平均绝对偏差/℃ 均方根误差/℃ 准确率/% 平均绝对偏差/℃ 均方根误差/℃ 极端低温 M9518 100 0.80 0.88 97.91 0.96 1.24 M9520 100 0.53 0.71 100 0.94 1.18 M9522 100 0.69 0.84 93.75 1.42 1.75 M9526 97.92 0.77 1.08 92.36 1.11 1.50 极端高温 M9518 69.44 2.27 2.95 53.47 4.19 5.64 M9520 88.89 0.92 1.40 68.06 2.68 3.27 M9522 90.27 1.01 1.50 67.36 2.97 3.72 M9526 79.86 1.63 2.84 61.80 3.88 5.94
下载: 导出CSV
表 4 模型迁移应用预报效果评估
Table 4 Evaluation of model transfer application forecasting performance
模型站 模型站类型 验证站 验证站类型 1 h时效 3 h时效 准确率/% 平均绝对偏差/℃ 均方根误差/℃ 准确率/% 平均绝对偏差/℃ 均方根误差/℃ M9521 城市 86.09 1.47 2.30 73.10 2.40 3.53 M9126 城市 87.17 1.47 2.31 72.12 2.54 3.67 M9518 城郊 M9293 城郊 80.18 1.86 2.92 67.95 2.88 4.23 M9513 乡村 78.87 1.99 2.97 59.44 3.86 5.65 M9516 乡村 80.39 1.93 2.93 59.95 3.80 5.62 M9521 城市 88.20 1.43 2.05 70.22 2.60 3.75 M9126 城市 87.35 1.49 2.18 68.28 2.83 4.06 M9520 城郊 M9293 城郊 84.11 1.65 2.41 66.88 2.92 4.24 M9513 乡村 83.78 1.65 2.71 62.80 3.37 5.01 M9516 乡村 85.32 1.65 2.87 62.82 3.39 5.12 M9521 城市 86.10 1.58 2.22 70.23 2.77 4.08 M9126 城市 85.29 1.72 2.35 66.55 3.06 4.34 M9522 城市 M9293 城郊 80.97 1.90 2.73 65.96 3.24 4.86 M9513 乡村 81.92 1.81 2.82 57.21 4.01 6.03 M9516 乡村 81.35 1.85 2.80 55.69 4.06 6.02 M9521 城市 86.39 1.52 2.17 58.39 3.36 4.61 M9126 城市 85.09 1.63 2.31 57.56 3.62 5.15 M9526 乡村 M9293 城郊 81.69 1.79 2.68 64.73 3.12 4.51 M9513 乡村 84.17 1.70 2.44 63.65 3.20 4.55 M9516 乡村 84.18 1.75 2.48 62.16 3.12 4.38
下载: 导出CSV
-
[1] 冯蕾, 王晓峰, 何晓凤, 等.基于INCA和METRo的江苏省路面高温精细化预报.应用气象学报, 2017, 28(1): 109-118. doi: 10.11898/1001-7313.20170110Feng L, Wang X F, He X F, et al. Fine forecast of high road temperature along jiangsu highways based on INCA system and METRo model. J Appl Meteor Sci, 2017, 28(1): 109-118. doi: 10.11898/1001-7313.20170110 [2] Chen J Q, Wang H, Xie P Y. Pavement temperature prediction: Theoretical models and critical affecting factors. Appl Therm Eng, 2019, 158: 113755. doi: 10.1016/j.applthermaleng.2019.113755 [3] Chen S K, Saeed T U, Labi S. Impact of road-surface condition on rural highway safety: A multivariate random parameters negative binomial approach. Anal Meth Accid Res, 2017, 16: 75-89. [4] Adwan I, Milad A, Memon Z A, et al. Asphalt pavement temperature prediction models: A review. Appl Sci, 2021, 11(9): 3794. doi: 10.3390/app11093794 [5] Barber E S. Calculation of maximum pavement temperatures from weather reports. Highw Res Board Bull, 1957, 168: 1-8. [6] 刘熙明, 喻迎春, 雷桂莲, 等. 应用辐射平衡原理计算夏季水泥路面温度. 应用气象学报, 2004, 15(5): 623-628. doi: 10.3969/j.issn.1001-7313.2004.05.012Liu X M, Yu Y C, Lei G L, et al. Using radiant balance theory to calculate concrete road-surface temperature in summer. J Appl Meteor Sci, 2004, 15(5): 623-628. doi: 10.3969/j.issn.1001-7313.2004.05.012 [7] 朱承瑛, 谢志清, 严明良, 等. 高速公路路面温度极值预报模型研究. 气象科学, 2009, 29(5): 645-650.Zhu C Y, Xie Z Q, Yan M L, et al. Study on the numerical prediction model of extreme temperature on speedway-surface. Sci Meteor Sinica, 2009, 29(5): 645-650. [8] Dempsey B J, Thompson M R. A heat transfer model for evaluating frost action and temperature-related effects in multilayered pavement systems. Highw Res Rec, 1970(342): 39-56. [9] 孟春雷, 张朝林. 路面气象数值预报模型及性能检验. 应用气象学报, 2012, 23(4): 451-458. doi: 10.3969/j.issn.1001-7313.2012.04.008Meng C L, Zhang C L. Development and verification of a numerical forecast model for road meteorological services. J Appl Meteor Sci, 2012, 23(4): 451-458. doi: 10.3969/j.issn.1001-7313.2012.04.008 [10] Asefzadeh A, Hashemian L, Bayat A. Development of statistical temperature prediction models for a test road in Edmonton, Alberta, Canada. Int J Pavement Res Technol, 2017, 10(5): 369-382. doi: 10.1016/j.ijprt.2017.05.004 [11] 董天翔, 包云轩, 袁成松, 等. 三种统计预报模型在江苏省道路低温预警中的应用. 气象科技, 2018, 46(4): 773-784.Dong T X, Bao Y X, Yuan C S, et al. Application of three statistical forecast models in early warning of low-temperature on road surface in Jiangsu and their comparison. Meteor Sci Technol, 2018, 46(4): 773-784. [12] Chandrappa A K, Biligiri K P. Development of pavement-surface temperature predictive models: Parametric approach. J Mater Civ Eng, 2016, 28(3): 04015143.1-04015143.12. [13] 王琨, 郝培文. 不同层位沥青路面温度预估模型. 长安大学学报(自然科学版), 2017, 37(6): 24-30. doi: 10.3969/j.issn.1671-8879.2017.06.004Wang K, Hao P W. Prediction model of temperature in different layers of asphalt pavement. J Chang'an Univ(Nat Sci Ed), 2017, 37(6): 24-30. doi: 10.3969/j.issn.1671-8879.2017.06.004 [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. 同济大学学报(自然科学版), 2017, 45(12): 1824-1829.Tang J J, Guo Z Y. Pavement temperature short-impending prediction based on ARIMA in winter. J Tongji Univ(Nat Sci Ed), 2017, 45(12): 1824-1829. [15] Chapman L, Thornes J E. A geomatics-based road surface temperature prediction model. Sci Total Environ, 2006, 360(1/2/3): 68-80. [16] Thornes J E, Cavan G, Chapman L. XRWIS: The use of geomatics to predict winter road surface temperatures in Poland. Meteor Appl, 2005, 12(1): 83-90. doi: 10.1017/S135048270500157X [17] Shao J. Improving nowcasts of road surface temperature by a backpropagation neural network. Wea Forecasting, 1998, 13(1): 164-171. doi: 10.1175/1520-0434(1998)013<0164:INORST>2.0.CO;2 [18] Abo-Hashema M A. Modeling Pavement Temperature Prediction Using Artificial Neural Networks Airfield and Highway Pavement 2013. Los Angeles, California, USA. Reston, VA: American Society of Civil Engineers, 2013: 490-505. [19] Xu B, Dan H C, Li L. Temperature prediction model of asphalt pavement in cold regions based on an improved BP neural network. Appl Therm Eng, 2017, 120: 568-580. doi: 10.1016/j.applthermaleng.2017.04.024 [20] 雷建军, 韦惠红, 李剑. 一种遗传微粒群算法-支持向量机集成道路结冰预测系统. 华中师范大学学报(自然科学版), 2010, 44(3): 392-396.Lei J J, Wei H H, Li J. A genetic PSO-SVM hybrid algorithm for road icing forecast. J Huazhong Norm Univ(Nat Sci Ed), 2010, 44(3): 392-396. [21] 王可心, 包云轩, 朱承瑛, 等. 随机森林回归法在冬季路面温度预报中的应用. 气象, 2021, 47(1): 82-93.Wang K X, Bao Y X, Zhu C Y, et al. Forecasts of road surface temperature in winter based on random forests regression. Meteor Mon, 2021, 47(1): 82-93. [22] Liu B, Shen L B, You H L, et al. Comparison of algorithms for road surface temperature prediction. Int J Crowd Sci, 2018, 2(3): 212-224. doi: 10.1108/IJCS-09-2018-0021 [23] Karpatne A, Atluri G, Faghmous J H, et al. Theory-guided data science: A new paradigm for scientific discovery from data. IEEE Trans Knowl Data Eng, 2017, 29(10): 2318-2331. doi: 10.1109/TKDE.2017.2720168 [24] 田华, 吴昊, 赵琳娜, 等. 沪宁高速公路路面温度变化特征及统计模型. 应用气象学报, 2009, 20(6): 737-744. doi: 10.3969/j.issn.1001-7313.2009.06.012Tian H, Wu H, Zhao L N, et al. Characteristics and statistical model of road surface temperature on huning expressway. J Appl Meteor Sci, 2009, 20(6): 737-744. doi: 10.3969/j.issn.1001-7313.2009.06.012 [25] 吴晟, 吴兑, 邓雪娇, 等. 南岭山地高速公路路面温度变化特征分析. 气象科技, 2006, 34(6): 783-787. doi: 10.3969/j.issn.1671-6345.2006.06.028Wu S, Wu D, Deng X J, et al. Characteristics of road surface temperature on freeway over Nanling hilly region. Meteor Sci Technol, 2006, 34(6): 783-787. doi: 10.3969/j.issn.1671-6345.2006.06.028 [26] Jun C, Ban Y F, Li S N. Open access to earth land-cover map. Nature, 2014, 514(7523): 434. [27] Zhang K E, Cao C, Chu H R, et al. Increased heat risk in wet climate induced by urban humid heat. Nature, 2023, 617(7962): 738-742. doi: 10.1038/s41586-023-05911-1 [28] Hersbach H, Bell B, Berrisford P, et al. The ERA5 global reanalysis. Q J R Meteor Soc, 2020, 146(730): 1999-2049. doi: 10.1002/qj.3803 [29] Muoz-Sabater J, Dutra E, Agustí-Panareda A, et al. ERA5-Land: A state-of-the-art global reanalysis dataset for land applications. Earth Syst Sci Data, 2021, 13(9): 4349-4383. [30] 张曦, 浩宇, 梁佳, 等. 陕西省高速公路路面温度特征及预报模型. 干旱气象, 2019, 37(6): 1028-1034.Zhang X, Hao Y, Liang J, et al. Characteristics of road surface temperature of Shaanxi expressway and its prediction model. J Arid Meteor, 2019, 37(6): 1028-1034. [31] 李昊. 基于机器学习的兰州市空气质量预报方法研究. 兰州: 兰州大学, 2017.Li H. Study of Statistical Methods Based on Machine Learning Algorithms for Air Quality Forecasting in Lanzhou. Lanzhou: Lanzhou University, 2017. [32] Hochreiter S, Schmidhuber J. Long short-term memory. Neural Comput, 1997, 9(8): 1735-1780. [33] 米前川, 高西宁, 李玥, 等. 深度学习方法在干旱预测中的应用. 应用气象学报, 2022, 33(1): 104-114. doi: 10.11898/1001-7313.20220109Mi Q C, Gao X N, Li Y, et al. Application of deep learning method to drought prediction. J Appl Meteor Sci, 2022, 33(1): 104-114. doi: 10.11898/1001-7313.20220109 [34] 谢舜, 孙效功, 张苏平, 等. 基于SVD与机器学习的华南降水预报订正方法. 应用气象学报, 2022, 33(3): 293-304. doi: 10.11898/1001-7313.20220304Xie S, Sun X G, Zhang S P, et al. Precipitation forecast correction in South China based on SVD and machine learning. J Appl Meteor Sci, 2022, 33(3): 293-304. doi: 10.11898/1001-7313.20220304 [35] 孔雯, 车权, 赵慧荣, 等. 基于奇异谱分析与长短时记忆神经网络的电厂存煤量短期预测. 信息与控制, 2020, 49(6): 742-751.Kong W, Che Q, Zhao H R, et al. Short-term prediction of coal stock in power plant based on singular spectrum analysis and long short-term memory neural network. Inf Control, 2020, 49(6): 742-751. [36] 胡莹莹, 庞林, 王启光. 基于深度学习的7~15 d温度格点预报偏差订正. 应用气象学报, 2023, 34(4): 426-437. doi: 10.11898/1001-7313.20230404Hu Y Y, Pang L, Wang Q G. Application of deep learning bias correction method to temperature grid forecast of 7-15 days. J Appl Meteor Sci, 2023, 34(4): 426-437. doi: 10.11898/1001-7313.20230404 [37] Liu B, Yan S, You H, et al. An Ensembled RBF Extreme Learning Machine to Forecast Road Surface Temperature//2017 16th IEEE International Conference on Machine Learning and Applications(ICMLA), 2017: 977-980. [38] 王玉虹, Bica Benedikt. 不同天气背景下京津冀降水临近外推预报. 应用气象学报, 2022, 33(3): 270-281. doi: 10.11898/1001-7313.20220302Wang Y H, Bica B. Precipitation extrapolation nowcasting in Being-Tianjin-Hebei under different weather backgrounds. J Appl Meteor Sci, 2022, 33(3): 270-281. doi: 10.11898/1001-7313.20220302 [39] 赵琳娜, 卢姝, 齐丹, 等. 基于全连接神经网络方法的日最高气温预报. 应用气象学报, 2022, 33(3): 257-269. doi: 10.11898/1001-7313.20220301Zhao L N, Lu S, Qi D, et al. Daily maximum air temperature forecast based on fully connected neural network. J Appl Meteor Sci, 2022, 33(3): 257-269. doi: 10.11898/1001-7313.20220301 [40] 中国气象局. 高速公路交通气象条件等级: QX/T 111—2010. 北京: 气象出版社, 2010.China Meteorological Administration. Grade of Weather Conditions for Freeway Transportation: QX/T 111-2010. Beijing: China Meteorological Press, 2010. [41] 中国气象局. 公路交通高影响天气预警等级: QX/T 414—2018. 北京: 气象出版社, 2018.China Meteorological Administration. The Warning Levels of High-impact Weather on Highway Traffic: QX/T 414-2018. Beijing: China Meteorological Press, 2018. -
图(6) / 表(4)
计量
- 摘要浏览量: 751
- HTML全文浏览量: 151
- PDF下载量: 111
- 被引次数: 0
