EEMD-PE-LSTM Based Traffic State Prediction Method for Freeway Section
-
摘要: 高速公路网络的快速发展与交通需求的多样化，使得交通拥堵、道路承载能力瓶颈、道路设计优化等问题愈发凸显，严重制约了出行者的出行体验和交通管理部门的服务效能。因此为了精准量化交通条件与气象条件对交通流参数预测性能的增益，本文构建基于集合经验模态分解（ensemble empirical modal decomposition，EEMD）、排列熵（permutation entropy，PE）和长短期记忆神经网络（long short-term memory，LSTM）的高速公路路段交通流参数组合预测模型。研究运用EEMD算法分解平均行驶速度序列，通过PE算法筛选整合分解后的分量，并对交通和气象数据进行时空匹配和特征分组，识别出最具影响力的因子及其相互作用模式，结合滑动时间窗策略，动态调整输入配置；以LSTM网络为核心，经迭代优化确定最优的历史序列长度和特征组合，进而得到目标路段平均行驶速度最优值；同时，提出区域特性导向的交通状态判定机制，即采用路段平均行驶速度85%分位数作为常态速度基准。以湖北省某高速公路为例，实证结果显示：预测精度方面，相较于单一的LSTM模型，组合预测模型的平均绝对误差显著降低73.4%；运算效率方面，较EEMD-LSTM模型提升67%；特别是在滑动时间窗长度为40 min时，组合模型在各类出行场景及多样化特征的输入下，均保持最低预测误差，展现出良好的稳定性和鲁棒性；此外，纳入交通条件的模型相较于仅依赖历史速度序列的模型，预测误差范围降低了约60%，凸显了交通因素在速度预测中的关键作用。本研究可为交通管理部门在交通高峰期、特殊活动、交通事故突发等期间提供科学的管理决策支持。Abstract: The rapid development of the highway network and the diversification of traffic demand have made traffic congestion, road carrying capacity bottlenecks, road design optimization and other issues more and more prominent. This seriously restricts the travelling experience of travelers and the service effectiveness of traffic management departments. To accurately quantify the gain of traffic conditions and weather conditions on the prediction performance of traffic flow parameters, we a combined prediction model of traffic flow parameters of highway sections construct based on the ensemble empirical modal decomposition (EEMD), permutation entropy (PE) and long short-term memory (LSTM). The study applies the EEMD algorithm to decompose the average travelling speed sequence, screens and integrates the decomposed components through the PE algorithm and proposes to perform spatio-temporal matching and feature grouping of the traffic and weather data to identify the most influential factors and their interaction modes; combined with the sliding time window strategy, the input configurations are dynamically adjusted. With the LSTM network as the core, the optimal history sequence length and feature combination are determined by iterative optimizations, and then the optimal value of the average driving speed of the target road section is obtained. At the same time, the regional characteristic-oriented traffic state determination mechanism is proposed, i.e., the 85% quartile of the average driving speed of the road section is adopted as the normal speed benchmark. Taking a highway in Hubei Province as a case study, the empirical results show that: in terms of prediction accuracy, compared with a single LSTM model, the average absolute error of the combined prediction model is significantly reduced by 73.4%; in terms of computational efficiency, it is improved by 67% compared with the EEMD-LSTM model. Especially when the length of the sliding time window is 40 min, the combined model maintains the lowest prediction error under various types of travelling scenarios and diversified feature inputs, showing good stability and robustness. In addition, the model incorporating traffic conditions reduces the prediction error range by about 60% compared to the model relying only on historical speed sequences, highlighting the key role of traffic factors in speed prediction. This study can provide scientific management decision support for traffic management departments during peak traffic periods, special events, and traffic accident emergencies.
-
图 1 基于EEMD-PE-LSTM的交通流参数预测模型框架示意图
Figure 1. Schematic diagram of the model framework for predicting traffic flow parameters based on EEMD-PE-LSTM
图 2 交通指标提取流程
注：T为通过当前门架的时间，G为通过当前门架的编号。
Figure 2. Traffic indicator extraction process
图 3 路段平均行驶速度序列EEMD分解结果示意图
Figure 3. Schematic diagram of the EEMD decomposition results of the average travel speed sequence of the road segment
图 5 速度基准组：误差指标在滑动时间窗内的变化趋势
Figure 5. Velocity benchmark group: trends in error metrics over a sliding time window
图 6 交通条件辅助组：误差指标在滑动时间窗内的变化趋势
Figure 6. Transportation conditions auxiliary group: trends in error indicators over a sliding time window
图 7 气象条件辅助组：误差指标在滑动时间窗内的变化趋势
Figure 7. Auxiliary group of Meteorological conditions: trends in error metrics over a sliding time window
图 8 综合交通环境组：误差指标在滑动时间窗内的变化趋势
Figure 8. Integrated transportation environment group: trends in error indicators over a sliding time window
图 9 速度基准组的损失函数变化图
Figure 9. Variation of the loss function for the velocity benchmark group
图 10 工作日场景：行驶速度预测值与真实值对比图
Figure 10. Workday scenario: comparison of predicted and true values of travel speeds
图 11 双休日场景：行驶速度预测值与真实值对比图
Figure 11. Weekend scenario: comparison of predicted and true values of travel speeds
图 12 行驶速度误差对比图
Figure 12. Weekend scenario: comparison of predicted and true values of travel speeds
表 1 交通环境特征类别及说明
Table 1. Transportation environmental characterization categories and descriptions
特征类型 序号 变量名称 变量符号 交通条件 1 目标路段平均行驶速度（/km/h） V 2 目标路段平均车流量（/veh/h） Q 3 目标路段与上游路段平均流量差（/veh/h） Q_diff 4 目标路段与上游路段平均行驶速度差（/km/h） V_diff 5 相邻门架断面流量差（/veh/h） Q_gan_diff 6 目标路段客货比 PTR 气象条件 1 风速（/m/s） WS 2 湿度/% Hum 3 降雨量等级 RC 4 能见度等级 VC 
下载: 导出CSV
表 2 模型训练组合
Table 2. Model training combinations
训练组合 特征组成 速度基准组 V 交通条件辅助组 V、Q、Q_diff、V_diff、Q_gan_diff、PTR 气象条件辅助组 V、WS、Hum、RC、VC 综合交通环境组 V、Q、Q_diff、V_diff、Q_gan_diff、PTR、WS、Hum、RC、VC
下载: 导出CSV
表 3 高速公路路段交通状态等级标准
Table 3. Traffic condition rating criteria for freeway sections
拥挤度 设计速度（/km/h） 120 100 80 畅通（绿色） ≥90 ≥80 ≥60 基本畅通（蓝色） [70，90) [60，80) [50，60) 轻度拥堵（黄色） [50，70) [40，60) [35，50) 中度拥堵（橙色） [30，50) [20，40) [20，35) 严重拥堵（红色） [0，30) [0，20) [0，20)
下载: 导出CSV
表 4 湖北省高速公路路段交通状态等级标准
Table 4. Traffic condition level standards for Freeway sections in Hubei Province
交通状态指数 拥挤度 路段常态速度/（km/h） 100 90 80 70 60 1 畅通 ≥95 ≥85 ≥75 ≥65 ≥55 2 基本畅通 [75，95) [70，85) [65，75) [55，65) [50，55) 3 轻度拥堵 [55，75) [50，70) [45，65) [40，55) [35，50) 4 中度拥堵 [35，55) [30，50) [25，45) [20，40) [15，35) 5 严重拥堵 [0，35) [0，30) [0，25) [0，20) [0，15)
下载: 导出CSV
表 5 不同参数搜索范围及最优结果
Table 5. Search range and optimal results for different parameters
参数说明 搜索范围 调整步长 最优参数 速度基准组 交通条件辅助组 气象条件辅助组 综合交通环境组 LSTM1 LSTM2 LSTM1 LSTM2 LSTM1 LSTM2 LSTM1 LSTM2 隐藏层数 [1，2，3] 1 1 1 1 1 1 1 1 1 神经元数 [50，500] 50 450 300 450 400 250 400 350 400 优化器 [Adam，SGD，RMSProp] - RMSProp Adam Adam Adam RMSProp Adam Adam Adam
下载: 导出CSV
表 6 工作日场景：交通状态时段数预测值与真实值对比
Table 6. Workday scenario: comparison of predicted and real values of the number of traffic condition periods
交通状态指数 拥挤度 真实值 预测值 1 畅通 57 55 2 基本畅通 11 9 3 轻度拥堵 4 3 注：数据样本中未出现中度拥堵和重度拥堵情形。
下载: 导出CSV
表 7 双休日场景：交通状态时段数预测值与真实值对比
Table 7. Weekend scenario: comparison of predicted and real values of the number of traffic state periods
交通状态指数 拥挤度 真实值 预测值 1 畅通 48 46 2 基本畅通 14 12 3 轻度拥堵 10 9 注：数据样本中未出现中度拥堵和重度拥堵情形。
下载: 导出CSV
表 8 模型预测误差对比
Table 8. Comparison of model prediction errors
模型 EMAE EMSE EMAPE/% ERMSE 时间/s LSTM 2.203 7.708 2.678 2.776 5 693 EEMD-LSTM 0.924 1.017 1.204 1.008 20 164 EEMD-PE-LSTM 0.586 0.685 0.726 0.828 6 721
下载: 导出CSV
-
[1] MAGI A, SIL G, TYAGI A. 85th and 98th percentile speed prediction models of car, light, and heavy commercial vehicles for four-lane divided rural highways[J]. Journal of Transportation Engineering, Part A: Systems, 2018, 144(5): 04018009. doi: 10.1061/JTEPBS.0000136 [2] SIL G, NAMA S, MAGI A, et al. Effect of horizontal curve geometry on vehicle speed distribution: a four-lane divided highway study[J]. Transportation Letters, 2020, 12(10): 713-722. doi: 10.1080/19427867.2019.1695562 [3] SATHIARAJ D, PUNKASEM T, WANG F, et al. Data-driven analysis on the effects of extreme weather elements on traffic volume in Atlanta, GA, USA[J]. Computers, Environment and Urban Systems, 2018, 72: 212-220. doi: 10.1016/j.compenvurbsys.2018.06.012 [4] 袁振洲, 胡嫣然, 杨洋. 考虑多维动态特征交互的高速公路实时事故风险建模[J]. 交通运输系统工程与信息, 2022, 22 (3): 215-223.YUAN Z Z, HU Y R, YANG Y. Real-time highway accident risk modeling considering multi-dimensional dynamic feature interaction[J]. Transportation Systems Engineering and Information, 2022, 22(3): 215-223. (in Chinese) [5] 李硕, 马玉坤, 韩晖, 等. 山区高速公路货车事故严重度致因及随机参数分析[J]. 公路交通科技, 2023, 40(4): 228-236. doi: 10.3969/j.issn.1002-0268.2023.04.028LI S, MA Y K, HAN H, et al. Analysis of causation and random parameters of truck accident severity on mountain expressway[J]. Highway Transportation Science and Technology, 2023, 40(4): 228-236. (in Chinese) doi: 10.3969/j.issn.1002-0268.2023.04.028 [6] 孙静怡, 牟若瑾, 刘拥华. 考虑大型车因素的支持向量机短时交通状态预测模型研究[J]. 公路交通科技, 2018, 35(10): 126-132.SUN J Y, MOU R J, LIU Y H. Research on short-time traffic state prediction model by support vector Machine considering large vehicle factors[J]. Highway Transportation Science and Technology, 2018, 35(10): 126-132. (in Chinese) [7] 李维佳, 王建军, 白骅, 等. 路网环境下考虑大型车混入率的事故疏散诱导模型[J]. 中国公路学报, 2020, 33(11): 275-284.LI W J, WANG JJ, BAI H, et al. Accident evacuation guidance model considering the mixing rate of large vehicles in road network environment[J]. China Journal of Highway and Transportation, 2020, 33(11): 275-284. (in Chinese) [8] SONG Z, GUO Y, WU Y, et al. Short-term traffic speed prediction under different data collection time intervals using a SARIMA-SDGM hybrid prediction model[J]. PloS one, 2019, 14(6): e0218626. doi: 10.1371/journal.pone.0218626 [9] XU D, WANG Y, JIA L, et al. Real-time road traffic state prediction based on ARIMA and Kalman filter[J]. Frontiers of Information Technology & Electronic Engineering, 2017, 18: 287-302. [10] EMAMI A, SARVI M, BAGLOEE S A. Short-term traffic flow prediction based on faded memory Kalman filter fusing data from connected vehicles and bluetooth sensors[J]. Simulation Modelling Practice and Theory, 2020, 102: 102025. doi: 10.1016/j.simpat.2019.102025 [11] 廖荣华, 兰时勇, 刘正熙. 基于混沌时间序列局域法的短时交通流预测[J]. 计算机技术与发展, 2015(1): 1-5.LIAO R H, LAN S Y, LIU Z X. Short-term traffic flow prediction based on chaotic time series local method[J]. Computer Technology and Development, 2015(1): 1-5. (in Chinese) [12] 吴晋武, 张海峰, 冉旭东. 基于数据约减和支持向量机的非参数回归短时交通流预测算法[J]. 公路交通科技, 2020, 37 (7): 129-134.WU J W, ZHANG H F, RAN X D. Non-parametric regression short-term traffic flow prediction algorithm based on data reduction and support vector machine[J]. Journal of Highway and Transportation Science and Technology, 2020, 37 (7): 129-134. (in Chinese) [13] 翁剑成, 陈旭蕊, 潘晓芳, 等. 基于超参数优化WOA-Bi-LSTM模型的客运枢纽抵站客流预测方法[J]. 交通信息与安全, 2023, 41(5): 148-157. doi: 10.3963/j.jssn.1674-4861.2023.05.015WENG J C, CHEN X R, PAN X F, et al. Prediction method of passenger arrivals at passenger hubs based on hyperparameter optimization WOA-Bi-LSTM model[J]. Traffic Information and Safety, 2023, 41(5): 148-157. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2023.05.015 [14] FUKUDA S, UCHIDA H, FUJⅡ H, et al. Short-term prediction of traffic flow under incident conditions using graph convolutional recurrent neural network and traffic simulation[J]. IET Intelligent Transport Systems, 2020, 14(8): 936-946. [15] 张文松, 姚荣涵. 基于时空特性和组合深度学习的交通流参数估计[J]. 交通运输系统工程与信息, 2021, 21(1): 82-89.ZHANG W S, YAO R H. Traffic flow parameter estimation based on spatiotemporal characteristics and combined deep learning[J]. Transportation Systems Engineering and Information, 2021, 21(1): 82-89. (in Chinese) [16] 刘清梅, 万明, 严利鑫, 等. 基于集合经验模态分解降噪和优化LSTM的道路交通事故预测[J]. 交通信息与安全, 2023, 41(5): 12-23. doi: 10.3963/j.jssn.1674-4861.2023.05.002LIU Q M, WAN M, YAN L X, et al. Road traffic accident prediction based on ensemble empirical mode decomposition for noise reduction and LSTM optimization[J]. Traffic Information and Safety, 2023, 41(5): 12-23. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2023.05.002 [17] 田佳, 王德勇, 师文喜. 基于集合经验模态分解和随机森林的短时交通流预测[J]. 科学技术与工程, 2023, 23(29): 12612-12619.TIAN J, WANG D Y, SHI W X. Short-term traffic flow prediction based on ensemble empirical mode decomposition and random forest[J]. Science Technology and Engineering, 2023, 23(29): 12612-12619. (in Chinese) [18] TIAN Z, LI S, WANG Y. A prediction approach using ensemble empirical mode decomposition-permutation entropy and regularized extreme learning machine for short-term wind speed[J]. Wind Energy, 2020, 23(2): 177-206. [19] 过加锦, 李磊, 任俞霏, 等. 融合信号分解与排列熵的高铁线路风速区间预测方法[J]. 交通科技与经济, 2023, 25(4): 74-80.GUO JJ, LI L, REN Y F, et al. Wind speed interval prediction method of high-speed railway line with signal decomposition and arrangement entropy[J]. Transportation Science and Economics, 2023, 25(4): 74-80. (in Chinese) [20] 周翔宇, 吉哲, 王凤武. 基于EEMD-PE-LSTM的短时船舶交通流量预测与航道交通状态可视化[J]. 大连海事大学学报, 2023, 49(2): 58-68.ZHOU X Y, JI Z, WANG F W. Short-time ship traffic flow prediction and waterway traffic state visualization based on EEMD-PE-LSTM[J]. Journal of Dalian Maritime University, 2023, 49(2): 58-68. (in Chinese) [21] 中华人民共和国国家质量监督检验检疫总局、中国国家标准化管理委员会. GB/T 28592-2012降水量等级[S]. 北京: 中国标准出版社, 2012.The General Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of China, the Standardization Administration of China. GB/T 28592-2012 precipitation grade[S]. Beijing: Standards Press of China, 2012. (in Chinese) [22] 中国气象局. QX/T114-2010能见度等级和预报[S]. 北京: 中国标准出版社, 2012.China Meteorological Administration. QX/T114-2010 Visibility rating and forecast[S]. Beijing: Standards Press of China, 2012. (in Chinese) [23] 李秉晨, 于惠钧, 刘靖宇. 基于K-means和CEEMD-PE-LSTM的短期光伏发电功率预测[J]. 水电能源科学, 2021, 39(4): 204-208.LI B C, YU H J, LIU J Y. Short-term PV power prediction based on K-means and CEEMD-PE-LSTM[J]. Hydropower Energy Science, 2021, 39(4): 204-208. (in Chinese) [24] 交通运输部. 公路网运行监测与服务暂行技术要求(2012年第3号公告)[S]. 北京: 人民交通出版社, 12: 25-26.Ministry of Transport. Provisional technical requirements for highway network operation monitoring and service (Announcement No. 3 of 2012)[S]. Beijing: People's Communications Press, 12: 25-26. (in Chinese) [25] 重庆市交通运行监测与应急调度中心. 重庆高速公路交通量综合统计数据标准[EB/OL]. (2021-03-16)[2024-04-30]. https://www.doczj.com/doc/b515781756.html .Chongqing Traffic Operation Monitoring and Emergency Dispatching Center. Chongqing highway traffic comprehensive statistics standard[EB/OL]. (2021-03-16)[2024-04-30].https://www.doczj.com/doc/b515781756.html . (in Chinese)[26] ABDULJABBAR R L, DIA H, TSAI P W. Unidirectional and bidirectional LSTM models for short-term traffic prediction[J]. Journal of Advanced Transportation, 2021, 2021(1): 5589075. -
点击查看大图
计量
- 文章访问数: 16
- HTML全文浏览量: 7
- PDF下载量: 2
- 被引次数: 0