基于IGWO-BP算法的轨道交通短时客流预测

张艺铭; 陈明明; 石磊; 康蓉桂

doi:10.3963/j.jssn.1674-4861.2021.03.011

基于IGWO-BP算法的轨道交通短时客流预测

doi: 10.3963/j.jssn.1674-4861.2021.03.011

兰州交通大学交通运输学院兰州 730070

基金项目:

甘肃省自然科学基金项目 21JR1RA244

详细信息

作者简介:
张艺铭（1995—），硕士研究生.研究方向：交通运输系统管理与优化.E-mail：272089435@qq.com

通讯作者:
陈明明（1982—），博士，副教授.研究方向：交通运输系统管理与优化.E-mail：lzcmm2000@mail.lzjtu.cn

中图分类号: U293.1+3
计量
- 文章访问数: 278
- HTML全文浏览量: 153
- PDF下载量: 12
- 被引次数: 0
出版历程
- 收稿日期: 2020-11-18

A Forecast of Short-term Passenger Flow of Rail Transit Based on IGWO-BP Algorithm

School of Traffic and Transportation, Lanzhou Jiaotong University, Lanzhou 730070, China

摘要

摘要: 轨道交通短时客流具有随机性和非线性的特点。为提高轨道交通短时客流预测结果的准确度，研究了基于改进的灰狼优化算法（IGWO）与BP神经网络的短时客流预测算法（IGWO-BP）。计算轨道交通客流不同时间序列的相关系数，确定了BP神经网络的输入和输出方式；用余弦思想和动态权重策略对原始灰狼优化算法改进，提高算法的全局搜索能力和寻优效率；用IGWO算法优化BP神经网络的初始权值和阈值，提高短时客流预测结果的准确性。预测了西安轨道交通2号线龙首原站周三早高峰15 min时间粒度的短时客流量，并将IGWO-BP算法的预测结果与其他5种模型（KF，GM，SVM，BPNN，GWO-BP）比较。结果表明，IGWO-BP算法的均方根误差为89.65，平均绝对百分比误差为1.16%，预测结果的精度和稳定性均为最优。
- 轨道交通 /
- 短时客流 /
- 相关系数 /
- IGWO算法 /
- BP神经网络
Abstract: Short-term passenger flow of rail transit has the characteristics of randomness and nonlinearity. An IGWO-BP algorithm is developed to forecast short-term passenger flow based on improved grey wolf optimization (IGWO) and BP neural network to improve the accuracy of predicting the short-term passenger flow of rail transit. The correlation coefficients of different time series of the rail-transit passenger flow are calculated to determine the input and output modes of the BP neural network. The cosine thought and dynamic weighting strategy are used to improve the orginal grey wolf optimization algorithm, thus enhancing the algorithm's global search and optimization. The IGWO algorithm is used to optimize the initial weights and thresholds of the BP neural network, which can improve the accuracy of predicting the short-term passenger flow. The work predicts the short-term passenger flow at the 15-min time granularity of the LONGSHOUYUAN Station of Xi'an Rail Transit Line 2 on Wednesday morning peak. The predicting results of the IGWO-BP algorithm are compared with those of the other five models (KF, GM, SVM, BPNN, and GWO-BP). For the IGWO-BP algorithm, the RMSE is 89.65, and the MAPE is 1.16%. The results show that the IGWO-BP algorithm has optimal accuracy and stability.
- rail transit /
- short-term passenger flow /
- correlation coefficient /
- IGWO algorithm /
- BP neural network

HTML全文

图 1 轨道交通的短时客流量预测

Figure 1. Forecasting the short-term passenger flow for rail transit

下载: 全尺寸图片幻灯片

图 2 IGWO-BP算法流程图

Figure 2. Flow of the IGWO-BP algorithm

下载: 全尺寸图片幻灯片

图 3 GWO算法寻优过程

Figure 3. Optimization process of the GWO algorithm

下载: 全尺寸图片幻灯片

图 4 收敛因子收敛曲线

Figure 4. Convergence curve of convergence factors

下载: 全尺寸图片幻灯片

图 5 不同日期进站客流量对比

Figure 5. Comparison of the inbound passenger flow on different days

下载: 全尺寸图片幻灯片

图 6 IGWO-BP算法预测结果最优值和平均值

Figure 6. Optimal and average values of predicting the IGWO-BP algorithm

下载: 全尺寸图片幻灯片

图 7 IGWO-BP多次预测结果对比

Figure 7. Predicting results of the IGWO-BP algorithm

下载: 全尺寸图片幻灯片

图 8 不同模型预测结果对比

Figure 8. Comparison of the predicting results of different models

下载: 全尺寸图片幻灯片

表 1 同1 d工作日不同时段客流量相关系数

Table 1. Correlation coefficient of the passenger flow at different time of the same working day

时段	07：00—07：15	> 07：15—07：30	> 07：30—07：45	> 07：45—08：00	> 08：00—08：15	> 08：15—08：30	> 08：30—08：45	> 08：45—09：00
07：00—07：15	1.00
> 07：15—07：30	0.55	1.00
> 07：30—07：45	0.24	0.74	1.00
> 07：45—08：00	0.42	0.71	0.75	1.00
> 08：00—08：15	-0.52	-0.02	0.23	-0.02	1.00
> 08：15—08：30	-0.67	-0.32	0.06	-0.12	0.63	1.00
> 08：30—08：45	-0.42	-0.61	-0.14	-0.36	0.41	0.29	1.00
> 08：45—09：00	-0.69	-0.61	-0.67	-0.41	0.08	0.47	-0.02	1.00

下载: 导出CSV

表 2 同1 d工作日同一时段客流量相关系数

Table 2. Correlation coefficient of the passenger flow at the same time of the same working day

工作日	11月1日	11月8日	11月15日	11月22日	11月29日	12月6日	12月13日	12月20日	12月27日
11月1日	1.00
11月8日	1.00	1.00
11月15日	0.99	0.98	1.00
11月22日	1.00	0.99	0.99	1.00
11月29日	0.98	0.98	0.99	0.99	1.00
12月6日	1.00	0.99	0.99	1.00	0.99	1.00
12月13日	0.99	0.98	0.99	0.99	0.99	1.00	1.00
12月20日	0.97	0.96	0.98	0.97	0.98	0.99	0.99	1.00
12月27日	0.97	0.96	0.99	0.97	0.99	0.98	0.99	0.99	1.00

下载: 导出CSV

表 3 IGWO-BP算法权值和阈值计算值

Table 3. Calculated values of the weight and threshold of the IGWO-BP algorithm

输入层到隐含层权值			隐含层到输出层权值	隐含层阈值	输出层阈值
-0.02	0.04	0.26	0.10	-0.78	0.02
-0.02	0.24	0.03	0.10	-0.02
0.25	0.88	-0.39	0.19	1.00
0.01	0.02	-0.04	0.09	0.34
-0.03	0.12	-0.03	-0.05	0.08
-0.31	-0.03	0.11	-0.06	0.19
-0.42	0.16	-0.03	0.01	-0.03
-0.02	-0.06	1.00	-0.05	-0.07
-0.09	-1.00	-0.03	0.29	-0.03

下载: 导出CSV

表 4 IGWO-BP算法预测结果评价指标

Table 4. Evaluation of the forecast results of the IGWO-BP algorithm

评价指标	最大值	最小值	平均值	标准差
MAE	82.75	64.84	75.02	4.93
RMSE	98.53	77.92	89.65	5.62
MAPE	1.97%	0.46%	1.16%	0.01

下载: 导出CSV

表 5 预测结果评价指标对比

Table 5. Comparison of evaluation indices for predicting results

评价指标	KF	GM	SVM	BPNN	GWO-BP	IGWO-BP
MAE	131.97	244.25	119.83	113.14	80.42	75.02
RMSE	149.36	302.55	128.37	136.92	97.76	89.65
MAPE/%	3.68	4.97	3.18	2.75	1.24	1.16

下载: 导出CSV

参考文献(19)

[1]	王明生. 城市轨道交通概论[M]. 北京: 人民交通出版社, 2012. WANG Mingsheng. Introduction to urban rail transit[M]. Beijing: China Communications Press, 2012. (in Chinese)
[2]	张亚运. 基于客流短时预测的城市轨道交通运营组织[D]. 西安: 长安大学, 2016. ZHANG Yayun. Operation organization of urban rail transit based on the passenger flow short-term prediction[D]. Xi'an: Chang'an University, 2016. (in Chinese)
[3]	ROOS J, GAVIN G, BONNEVAY S. A dynamic Bayesian network approach to forecast short-term urban rail passenger flows with incomplete data[J]. Transportation Research Procedia, 2017(26): 53-61. http://www.sciencedirect.com/science/article/pii/S2352146517308670
[4]	白伟华, 张传斌, 张塽旖, 等. 基于异常值识别卡尔曼滤波器的短期交通流预测[J]. 计算机应用研究, 2021, 38(3): 817-821. https://www.cnki.com.cn/Article/CJFDTOTAL-JSYJ202103034.htm BAI Weihua, ZHANG Chuanbin, ZHANG Shuangyi, et al. Outlier-identified kalman filter for short-term traffic flow forecasting[J]. Application Research of Computers: 2021, 38(3): 817-821. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JSYJ202103034.htm
[5]	帅斌, 张玥, 张永超. 我国市郊铁路客流特征分析及需求预测[J]. 铁道工程学报, 2014(1): 20-23+106. doi: 10.3969/j.issn.1006-2106.2014.01.004 SHUAI Bin, ZHANG Yue, ZHANG Yongchao. Analysis of suburban railway passenger flow characteristics and demand forecast in china[J]. Journal of Railway Engineering Society, 2014(1): 20-23+106. (in Chinese) doi: 10.3969/j.issn.1006-2106.2014.01.004
[6]	王兴川, 姚恩建, 刘莎莎. 基于AFC数据的大型活动期间城市轨道交通客流预测[J]. 北京交通大学学报, 2018, 42(1): 87-93. https://www.cnki.com.cn/Article/CJFDTOTAL-BFJT201801013.htm WANG Xingchuan, YAO Enjian, Liu Shasha. Urban rail transit passenger flow forecasting for large special event based on AFC data[J]. Journal of Beijing Jiaotong University, 2018, 42(1): 87-93. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-BFJT201801013.htm
[7]	郭旷, 王雪梅, 张宁. 城市轨道交通短时客流不确定性预测模型[J]. 城市轨道交通研究, 2020, 23(1): 22-26. https://www.cnki.com.cn/Article/CJFDTOTAL-GDJT202001008.htm GUO Kuang, WANG Xuemei, ZHANG Ning. Uncertainty prediction model of short-term passenger flow in urban rail transit[J]. Urban Mass Tranit, 2020, 23(1): 22-26. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-GDJT202001008.htm
[8]	SUN Yuxing, LENG Biao, GUAN Wei. A novel wavelet-SVM short-time passenger flow prediction in Beijing subway sys-tem[J]. Neurocomputing, 2015(166): 109-121. http://dl.acm.org/citation.cfm?id=2794209
[9]	赵阳阳, 夏亮, 江欣国. 基于经验模态分解与长短时记忆神经网络的短时地铁客流预测模型[J]. 交通运输工程学报, 2020, 20(4): 194-204. https://www.cnki.com.cn/Article/CJFDTOTAL-JYGC202004020.htm ZHAO Yangyang, XIA Liang, JIANG Xinguo. Short-term metro passenger flow prediction based on EMD-LSTM[J]. Journal of Traffic and Transportation Engineering, 2020, 20(4): 194-204. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JYGC202004020.htm
[10]	HAO S, LEE D, ZHAO D. Sequence to sequence learning with attention mechanism for short-term passenger flow pediction in large-scale metro system[J]. Transportation Research Part C: Emerging Technologies, 2019(107): 287-300. http://www.sciencedirect.com/science/article/pii/S0968090X19300245
[11]	赵建立, 石敬诗, 孙秋霞, 等. 基于混合深度学习的地铁站进出客流量短时预测[J]. 交通运输系统工程与信息, 2020, 20(5): 128-134. https://www.cnki.com.cn/Article/CJFDTOTAL-YSXT202005019.htm ZHAO Jianli, SHI Jingshi, SUN Qiuxia, et al. Short-time inflow and outflow prediction of metro stations based on hybrid deep learning[J]. Journal of Transportation Systems Engineering and Information Technology, 2020, 20(5): 128-134. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-YSXT202005019.htm
[12]	PENG Yanni, XIANG Wanli. Short-term traffic volume prediction using GA-BP based on wavelet denoising and phase space reconstruction[J]. Physica A: Statistical Mechanics and its Applications, 2020(549): 123913. http://www.sciencedirect.com/science/article/pii/S0378437119321715
[13]	ZHOU Xingyu, LI Hongmei, ZHENG Weihao, et al. Short-term traffic flow prediction based on the IMM-BP-UKF model[J]. Journal of Highway and Transportation Research and Development(English Edition), 2019, 13(2): 56-64. doi: 10.1061/JHTRCQ.0000679
[14]	李建森, 沈齐, 范馨月. 城市道路短时交通流量预测[J]. 数学的实践与认识, 2019, 49(5): 192-197. https://www.cnki.com.cn/Article/CJFDTOTAL-SSJS201905021.htm LI Jiansen, SHEN Qi, FAN Xinyue. Forecast of short-term traffic flow on urban roads[J]. Mathematics in Practice and Theory, 2019, 49(5): 192-197. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-SSJS201905021.htm
[15]	MIRJALILI S, MIRJALILI S M, LEWIS A. Grey Wolf Optimizer[J]. Advances in Engineering Software, 2014, 69(3): 46-61. http://www.sciencedirect.com/science/article/pii/s0965997813001853
[16]	傅蔚阳, 刘以安, 薛松. 基于灰狼算法与小波神经网络的目标威胁评估[J]. 浙江大学学报(工学版), 2018, 52(4): 680-686. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDZC201804010.htm FU Weiyang, LIU Yian, XUE Song. Target threat assessment using grey wolf optimization and wavelet neural network[J]. Journal of Zhejiang University(Engineering Science), 2018, 52(4): 680-686. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDZC201804010.htm
[17]	梁强升, 许心越, 刘利强. 面向数据驱动的城市轨道交通短时客流预测模型[J]. 中国铁道科学, 2020, 41(4): 153-162. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGTK202004018.htm LIANG Qiangsheng, XU Xinyue, LIU Liqiang. Data-driven short-term passenger flow prediction model for urban rail transit[J]. China Rail Way Science, 2020, 41(4): 153-162. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZGTK202004018.htm
[18]	何正风. MATLAB R2015b神经网络技术[M]. 北京: 清华大学出版社, 2016. HE Zhengfeng. MATLAB R2015b neural network technology[M]. Beijing: Tsinghua University Press, 2016. (in Chinese)
[19]	NIU Peifeng, NIU Songpeng, LIU Nan, et al. The defect of the grey wolf optimization algorithm and its verification method[J]. Knowledge-Based Systems, 2019(171): 37-43. http://www.sciencedirect.com/science/article/pii/S0950705119300188