-
摘要: 时间尺度大于15 min的城市交通流预测模型已无法满足交通信号实时控制和交通信息实时发布的需求, 通过对广州市中心区交叉路口交通流长期观察和数据采集, 分析了各种时间尺度的交通流特性, 提出以路口信号周期作为时间尺度, 绿灯流率作为变量的ARIMA (p, d, q) 短时交通预测模型。以1个和3个信号周期的时间尺度为例, 对城市交叉路口不同时间段交通流进行建模和预测。结果表明ARIMA (p, d, q) 预测模型结构稳定, 算法简单, 时间尺度为3个信号周期的预测模型可以很好地保持交通流特征, 均方根误差为0.015 9, 预测精度较高。Abstract: The long-term prediction models (time scale is larger than 15 min) of traffic flow did not satisfy the demands of traffic signal control and traveler information real-time dissemination.After observing and measuring actual intersection traffic flow in Guangzhou city center area for long time, the paper analyzed the characteristics of varied-time-scale traffic flow, and proposed an ARIMA (p, d, q) model, in which signal periods were taken as time intervals, and the flow rates of green phase were taken as model variables for predicting the short-term traffic flow of urban intersection.By the case study of traffic flow modeling and predicting with different time-sections separately in 1 signal period and 3 signal periods, it is validated that the ARIMA (p, d, q) prediction model construction is stable, its arithmetic is simple, the prediction model in 3 signal periods can well keep the characteristics of short-term traffic flow, its accuracy is quite satisfied, its root-mean-squareerror is 0.0159.
-
图 2 差分前相关函数
Figure 2. Autocorrelation and Partial-Autocorrelation Functions of Original Time Series
图 3 一阶差分后相关函数
Figure 3. Autocorrelation and Partial-Autocorrelation Functions of Differenced Time Series
图 4 拟合曲线、预测曲线与实测曲线比较
Figure 4. Curves & apos; Comparison of Model, Prediction Volume and Actual Volume
图 5 早高峰时段数据曲线比较
Figure 5. Curves & apos; Comparison of Traffic Volumes at Morning Peak-Hour
表 1 p与q取不同组合值所得模型的AIC值
Table 1. AIC Values with Different p and q
p q 0 1 2 3 0 392.64 393.23 373.89 375.62 1 394.13 379.81 372.01 377.87 2 387.58 378.69 373.41 376.62 3 389.52 384.44 374.63 389.11 
下载: 导出CSV
表 2 p与q取不同组合值所得模型的AIC值
Table 2. AIC Values with Different p and q
p q 0 1 2 3 0 495.57 494.58 473.19 470.33 1 497.36 477.34 470.35 472.21 2 484.89 471.93 472.44 468.64 3 484.91 473.34 474.10 470.85
下载: 导出CSV
表 3 p与q取不同组合值所得模型的AIC值
Table 3. AIC Values with Different p and q
p q 0 1 2 3 0 483.51 436.36 440.40 436.36 1 484.78 461.91 435.55 448.86 2 467.42 455.69 439.73 437.84 3 469.38 456.49 439.41 440.35
下载: 导出CSV
表 4 预测精度比较
Table 4. Comparison of Predict Accuracies
3个信号周期间隔 1个信号周期间隔 1个信号周期间隔 (00:00~23:59) (08:00~12:00) (16:00~19:00) 实测值 预测值 误差/% 实测值 预测值 误差/% 实测值 预测值 误差/% 0.289 0.295 2.046 0.175 0.197 13.023 0.284 0.294 3.498 0.322 0.315 2.201 0.185 0.213 15.287 0.225 0.269 19.154 0.319 0.308 3.499 0.339 0.285 16.067 0.273 0.290 6.176 0.340 0.311 8.520 0.244 0.199 18.686 0.343 0.333 2.872 0.311 0.312 0.277 0.227 0.227 0.032 0.200 0.242 21.120 0.325 0.323 0.383 0.314 0.264 15.936 0.281 0.318 13.387 0.293 0.303 3.152 0.211 0.208 1.408 0.211 0.243 15.011 0.330 0.335 1.618 0.151 0.143 5.397 0.250 0.295 17.818 0.314 0.310 1.342 0.333 0.315 5.429 0.314 0.292 7.028 0.253 0.284 2.016 0.200 0.171 14.541 0.338 0.283 16.091 0.319 0.308 3.522 0.165 0.195 18.162 0.380 0.324 14.800 0.275 0.279 1.266 0.377 0.339 10.347 0.159 0.166 4.135 0.299 0.299 0.113 0.224 0.203 9.372 0.268 0.291 8.777 0.365 0.329 9.842 0.226 0.208 7.980 0.218 0.257 17.815 0.286 0.304 6.042 0.184 0.214 16.600 0.300 0.302 0.780 0.324 0.339 4.573 0.291 0.285 2.096 0.292 0.275 5.833 0.268 0.289 8.125 0.279 0.249 10.644 0.296 0.294 0.706 0.315 0.338 7.503 0.267 0.220 17.665 0.304 0.280 8.025 0.300 0.300 0.170 0.255 0.247 2.826 0.282 0.285 1.017 0.324 0.320 1.250 0.350 0.285 18.460 0.265 0.269 1.445 Emape=3.87% Drmse=0.0159 Emape=10.97% Drmse=0.0328 Emape=9.27% Drmse=0.0301
下载: 导出CSV
-
[1] 贺国光, 李宇, 马寿峰. 基于数学模型的短时交通流预测方法探讨[J]. 系统工程理论与实践, 2000, 20 (12): 51-56. https://www.cnki.com.cn/Article/CJFDTOTAL-XTLL200012007.htmHE Guo-guang, LI Yu, MA Shou-feng. Discussion on Short-Term Traffic Flow Forecasting Methods Based on Mathematical Models[J]. Systems Engineering-Theory and Practice, 2000, 20 (12): 51-56. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-XTLL200012007.htm [2] 刘静, 关伟. 交通流预测方法综述[J]. 公路交通科技, 2004, 21 (3): 82-85. doi: 10.3969/j.issn.1002-0268.2004.03.022LIU Jing, GUAN Wei. A Summary of Traffic Flow Forecasting Methods[J]. Journal of Highway and Transportation Research and Development, 2004, 21 (3): 82-85. (in Chinese) doi: 10.3969/j.issn.1002-0268.2004.03.022 [3] MOHAMMAD M H, HASHEMR A, ZAHI MB. Short-Term Prediction of Traffic Volume in Urban Arterials[J]. Journal of Transportation Engineering, 1995, 121 (3): 249-254. doi: 10.1061/(ASCE)0733-947X(1995)121:3(249) [4] SHERIF I M, HAITHAM A M. Performance Evaluation of Short-Term Time-Series Traffic Prediction Model[J]. Journal of Transportation Engineering, 2002, 128 (6): 490-498. [5] 韩超, 宋苏, 王成红. 基于ARI MA模型的短时交通流实时自适应预测[J]. 系统仿真学报, 2004, 16 (7): 1 530-1 535. https://www.cnki.com.cn/Article/CJFDTOTAL-XTFZ200407043.htmHAN Chao, SONG Su, WANG Cheng-hong. A Real-Time Short-Term Traffic Flow Adaptive Forecasting Method Basedon ARIMA Model[J]. Journal of System Simulation, 2004, 16 (7): 1 530-1 535. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-XTFZ200407043.htm [6] 徐金强, 刘智勇. 交通流的时间序列建模及预测[J]. 五邑大学学报: 自然科学版, 2004, 18 (3): 16-20. https://www.cnki.com.cn/Article/CJFDTOTAL-WYDW200403004.htmXU Jin-qiang, LIU Zhi-yong. Time Series Modeling and Predicting for Urban Traffic Flow[J]. Journal of Wuyi University: Natural Science Edition, 2004, 18 (3): 16-20. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-WYDW200403004.htm [7] 欧晓凌, 裘刚, 张毅, 等. 城市交通流信息相似性分析与研究[J]. 中南公路工程, 2003, 28 (2): 4-7. https://www.cnki.com.cn/Article/CJFDTOTAL-ZNGL200302001.htmOU Xiao-ling, QIU Gang, ZHANG Yi, et al. Analysis of Similarity for Urban Traffic Volume[J]. Central South Highway Engineering, 2003, 28 (2): 4-7. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZNGL200302001.htm [8] 范金城, 梅长林. 数据分析[M]. 北京: 科学出版社, 2002. [9] BOX G E P[美], 顾岚. 时间序列分析: 预测与控制[M]. 北京: 中国统计出版社, 1997. -
点击查看大图
计量
- 文章访问数: 330
- HTML全文浏览量: 119
- PDF下载量: 482
- 被引次数: 0
