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

翁剑成; 陈旭蕊; 潘晓芳; 孙宇星; 柴娇龙

doi:10.3963/j.jssn.1674-4861.2023.05.015

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

doi: 10.3963/j.jssn.1674-4861.2023.05.015

1.
北京工业大学交通工程北京市重点实验室北京 100124
2.
烟台大悦城有限公司山东烟台 264099
3.
北京市运输事业发展中心北京 100161

基金项目:

国家自然科学基金项目 52072011

北京市博士后工作经费资助项目 2022-ZZ-087

详细信息

通讯作者:
翁剑成（1981—），博士，教授，研究方向：交通数据挖掘、交通出行行为建模等.E-mail：youthweng@bjut.edu.cn

中图分类号: U491.14
计量
- 文章访问数: 331
- HTML全文浏览量: 176
- PDF下载量: 14
- 被引次数: 0
出版历程
- 收稿日期: 2023-04-06
- 网络出版日期: 2024-01-18

A Forecasting Method for Arrival Passenger Flow Based on Hyperparametric Optimization WOA-Bi-LSTM Model for Passenger Hubs

1.
Beijing Key Laboratory of Traffic Engineering, Beijing University of Technology, Beijing 100124, China
2.
Yantai Dayue City Co., Yantai 264099, Shandong, China
3.
Beijing Transportation Development Center, Beijing 100161, China

摘要

摘要: 实现城市对外客运枢纽抵站客流的精准预测，是增强枢纽接续运输运力调度科学性的重要前提。为提高枢纽抵站客流的预测精度，研究了基于超参数优化的鲸鱼算法与双向长短期记忆神经网络模型（whale optimization algorithm and bi-directional long short-term memory，WOA-Bi-LSTM）组合的客流预测方法。融合历史抵站客流数据及天气、日期、时段等多源信息，分析抵站客流的时变特性，并开展不同影响因素与枢纽抵站客流量间的相关性分析。改进了传统双向长短期记忆神经网络模型（bi-directional long short-term memory，Bi-LSTM）的参数设置方法，用鲸鱼算法（whale optimization algorithm，WOA）代替手动调参，选取学习效率（η）与隐藏神经元个数（H）2个对模型预测精度具有较大影响的超参数进行最优超参数组合搜寻，通过计算其适应度函数进行循环逻辑判断，实现参数自适应优化。通过不断寻优，获取最优参数组合值，确定设置η为0.060 3、H为120，并输出预测结果和3个模型精度评价指标（R²判定系数，平均绝对误差与均方根误差）；同时构建了3种不同超参数优化算法改进的Bi-LSTM组合模型、2种基于WOA算法改进的其他组合模型，以及2种未改进的神经网络模型与WOA-Bi-LSTM模型使用相同的抵站客流数据集进行多维度对比，验证所建模型的优越性与鲁棒性。结果表明：WOA-Bi-LSTM模型在节假日、工作日与非工作日等不同枢纽抵站客流预测场景下均体现出良好的适用性，与其他模型相比，R²相关系数最大，达到0.951 4，表示所建模型的拟合效果最好；平均绝对误差与均方根误差最小，分别为762.96与556.25，误差相较于其他模型至少减少5.6%和3.2%。
- 综合运输 /
- 对外客运枢纽 /
- 抵站客流预测 /
- 双向长短时记忆神经网络 /
- 超参数优化
Abstract: Accurate prediction of arrival passenger flows at external passenger transportation hubs is an important prerequisite for enhancing the scientific scheduling of the transferring transport capacity of hubs. In order to improve the prediction accuracy of arrival passenger flows, a combination model of the whale optimization algorithm and bi-directional long short-term memory (WOA-Bi-LSTM) is proposed. Integration of historical arrival passenger flow data with multi-source information such as weather, date, and time of day, the time-varying characteristics of arrival passenger flows are analyzed, and correlation analysis is conducted between different influencing factors and arrival passenger flows at the hub. The parameter setting of the traditional bi-directional long short-term memory (Bi-LSTM) model is modified with the whale optimization algorithm (WOA) optimization algorithm. Learning rate (η) and the number of hidden neurons (H) are significant hyperparameters on model prediction accuracy and are determined by searching optimal values. The search procedure is performed to achieve adaptive parameter optimization by calculating their fitness functions through iterative logic. Through continuous optimization, set the η as 0.060 3 and H as 120. The performance of the proposed model is evaluated using three indicators: R² value, mean absolute error (MAE), and root mean square error (RMSE). Simultaneously, the WOA-Bi-LSTM model is compared with several baseline models across multiple dimensions based on the same dataset, including three Bi-LSTM models modified by different hyperparameter optimization algorithms, two other combination models based on the WOA algorithm and two unmodified neural network models. The results show that the WOA-Bi-LSTM model shows better performance of predicting arrival passenger flows in different scenarios involving holiday, workday and non-workday. Compared to other models, the WOA-Bi-LSTM model achieves the highest R² of 0.951 4, indicating that the proposed model has the best fit. The RMSE and MAE are both the lowest, at 762.96 and 556.25, respectively, with errors reduced by at least 5.6% and 3.2% compared to other models.
- integrated transportation /
- external passenger hub /
- arrival passenger flow prediction /
- bi-directional long-short term memory neural network /
- hyperparameter optimization

HTML全文

图 1 抵站客流周分布特征

Figure 1. The distribution characteristics of arriving passengers in different weeks

下载: 全尺寸图片幻灯片

图 2 不同日期类型抵站客流分布特征

Figure 2. The distribution characteristics of arriving passengers on different date types

下载: 全尺寸图片幻灯片

图 3 不同时段（以小时维度）抵站客流分布特征

Figure 3. The distribution characteristics of arriving passengers at different times（in hourly dimension）

下载: 全尺寸图片幻灯片

图 4 LSTM神经网络结构

Figure 4. LSTM neural network structure

下载: 全尺寸图片幻灯片

图 5 Bi-LSTM神经网络原理

Figure 5. Bi-LSTM neural network principle

下载: 全尺寸图片幻灯片

图 6 基于超参数优化方法的WOA-Bi-LSTM算法流程图

Figure 6. Flowchart of WOA-Bi-LSTM algorithm for hyperparameter optimization methods

下载: 全尺寸图片幻灯片

图 7 最优个体适应度值变化

Figure 7. Optimal individual fitness value change

下载: 全尺寸图片幻灯片

图 8 WOA-Bi-LSTM模型的抵站客流预测结果

Figure 8. Arrival passenger flow forecast results of WOA-Bi-LSTM model

下载: 全尺寸图片幻灯片

图 9 节假日的预测模型结果

Figure 9. Prediction model results for holidays

下载: 全尺寸图片幻灯片

图 10 工作日的预测模型结果

Figure 10. Prediction model results for Working Days

下载: 全尺寸图片幻灯片

图 11 非工作日的模型预测结果

Figure 11. Prediction results of the optimization model for Non-Working Days

下载: 全尺寸图片幻灯片

图 12 不同模型评价指标对比

Figure 12. Comparison of evaluation indexes of different models

下载: 全尺寸图片幻灯片

表 1 对外综合客运枢纽抵站客流量数据集示例

Table 1. A sample data set of arrival passenger flow at comprehensive hubs

时间（2021年4月1日）	时段	工作日类型	节假日类型	风速/(m/s)	天气类型	客流/人
06：00—06：59	6	0	0	2	3	745
07：00—07：59	7	0	0	2	3	4 217
08：00—08：59	8	0	0	1	3	5 441
⋮	⋮	⋮	⋮	⋮	⋮	⋮

下载: 导出CSV

表 2 枢纽抵站客流的影响因素分类变量定义

Table 2. Definition of classification variables of influencing factors of Arrival Passenger Flow in hubs

名称	分类变量定义
时段	00：00—00：59为0；01：00—01：59为1；…；23：00—23：59为23
工作日类型	工作日为0；非工作日且非节假日为1；节假日为2
节假日类型	非节假日为0；节假日前1 d为1；节假日第1 d为2；节假日最后1 d为3；节假日后1 d为4
天气类型	雨天为1；多云为2；霾为3；晴朗为4

下载: 导出CSV

表 3 各因素与客流量的皮尔森相关系数

Table 3. Pearson's correlation coefficient between each factor and passenger flow

影响因素	相关系数
时段	0.666^**
工作日类型	-0.135^**
节假日类型	0.058^*
风速	0.468^**
天气	0.087^*
注：^*-在0.01级别（单尾）；相关性显著。^-在0.05级别（单尾），相关性显著。

下载: 导出CSV

表 4 最优参数组合

Table 4. Optimal combination of parameters

参数	WOA
学习效率η	0.060 3
隐藏神经元个数H	120

下载: 导出CSV

表 5 不同模型评价指标数值

Table 5. Value of evaluation indexes of different models

优化模型	R²	W_MAE	W_RMSE
BP	0.766 1	1 724.90	1 078.28
LSTM	0.859 6	1 321.82	907.06
WOA-BP	0.941 2	822.36	589.38
WOA-LSTM	0.945 8	808.32	574.62
GA-Bi-LSTM	0.925 5	984.18	653.98
PSO-Bi-LSTM	0.945 2	810.17	593.85
SSA-Bi-LSTM	0.943 9	817.91	591.65
WOA-Bi-LSTM	0.951 4	762.96	556.25

下载: 导出CSV

参考文献(20)

[1]	汪鸣, 向爱兵, 杨宜佳. "十四五"我国交通运输发展思路[J]. 北京交通大学学报(社会科学版), 2022, 21(2): 68-75. WANG M, XIANG A B, YANG Y J. Analysis of China's transportation development during the 14th Five-Year plan period[J]. Journal of Beijing Jiaotong University(Social Sciences Edition), 2022, 21(2): 68-75. (in Chinese)
[2]	北京交通发展研究院. 2021北京市交通发展年度报告[R]. 北京: 北京交通发展研究院, 2021. Beijing Transport Institute. 2021 Beijing transportation development annual report[R]. Beijing: Beijing Transport Institute, 2021. (in Chinese)
[3]	祈伟, 刘清祥, 赵宁宇. 深圳福田综合交通枢纽客流预测研究[C]. 2014第九届中国智能交通年会大会, 北京: 中国智能交通协会, 2014. QI W, LIU Q X, ZHAO N Y. Research on the prediction of passenger flow of Shenzhen Futian comprehensive transportation hub[C]. The 9th Annual China Intelligent Transportation Conference 2014, Beijing: China Intelligent Transportation Systems Association, 2014. (in Chinese)
[4]	白丽. 城市轨道交通常态与非常态短期客流预测方法研究[J]. 交通运输系统工程与信息, 2017, 17(1): 127-135. BAI L. Urban rail transit normal and abnormal short-term passenger flow forecasting method[J]. Journal of Transportation Systems Engineering and Information Technology, 2017, 17(1): 127-135. (in Chinese)
[5]	HABTEMICHAEL G F, CETIN M. Short-term traffic flow rate forecasting based on identifying similar traffic patterns[J]. Transportation Research Part C: Emerging Technologies, 2016, 66: 61-78. doi: 10.1016/j.trc.2015.08.017
[6]	谢俏, 叶红霞. 基于支持向量机的节假日进出站客流预测方法[J]. 城市轨道交通研究, 2018, 21(8): 26-29, 35. XIE Q, YE H X. Forecast for holiday passenger flow at urban rail transit station based on support vector machine model[J]. Urban Mass Transit, 2018, 21(8): 26-29, 35. (in Chinese)
[7]	戢晓峰, 孔晓丽, 陈方, 等. 基于ETC数据和A-BiLSTM神经网络的高速公路节假日短时交通流预测模型[J]. 交通信息与安全, 2023, 41(3): 166-174. doi: 10.3963/j.jssn.1674-4861.2023.03.018 JI X F, KONG X L, CHEN F, et al. A forecasting model of short-term traffic flow on expressways during holidays based on ETC data and A-BiLSTM neural network models[J]. Journal of Transport Information and Safety, 2023, 41(3): 166-174. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2023.03.018
[8]	李洪嘉, 姚红光, 李思睿, 等. 基于灰色-神经网络的虹桥综合交通枢纽客流预测[J]. 科学技术创新, 2020(36): 121-122. doi: 10.3969/j.issn.1673-1328.2020.36.053 LI H J, YAO H G, LI S R, et al. Grey-neural network based passenger flow prediction for Hongqiao integrated transportation hub measurement[J]. Scientific Technological Innovation, 2020(36): 121-122. (in Chinese) doi: 10.3969/j.issn.1673-1328.2020.36.053
[9]	PRATAP A S, AJAY T, KUMAR R D, et al. Prediction of passenger flow for north central railway region through ANN[J]. IOP Conference Series: Materials Science and Engineering, 2021, 1136(1): 7.
[10]	ZHANG Z H, HAN Y, PENG T X, et al. A comprehensive spatio-temporal model for subway passenger flow prediction[J]. ISPRS International Journal of Geo-Information, 2022, 11(6): 1.
[11]	朱兆坤, 李金宝. 多特征信息融合LSTM-RNN检测OSA方法[J]. 计算机研究与发展, 2020, 57(12): 2547-2555. ZHU Z K, LI J B. Multi-feature information fusion LSTM-RNN detection for OSA[J]. Journal of Computer Research and Development, 2020, 57(12): 2547-2555. (in Chinese)
[12]	胡洪滔. 基于LGB-LSTM-DRS的对外客运枢纽城市轨道交通客流短时预测[D]. 北京: 北京交通大学, 2020. HU H T. Short-term prediction of passenger flow of urban rail transit under external transportation hub based on LGB-LSTM-DRS model[D]. Beijing: Beijing Jiaotong University, 2020. (in Chinese)
[13]	YUE M, MA S H. LSTM-based transformer for transfer passenger flow forecasting between transportation integrated hubs in urban agglomeration[J]. Applied Sciences, 2023, 13(1): 1.
[14]	MULERIKKAL J, THANDASSERY S, REJATHALAL V, et al. Performance improvement for metro passenger flow forecast using spatio-temporal deep neural network[J]. Neural Computing and Applications, 2021, 34(2): 12.
[15]	张维, 袁绍欣, 陶建军, 等. 基于多元因素的Bi-LSTM高速公路交通流预测[J]. 计算机系统应用, 2021, 30(6): 184-190. ZHANG W, YUAN S X, TAO J J, et al. Bi-LSTM expressway traffic flow prediction based on multiple factor data[J]. Computer Systems & Application, 2021, 30(6): 184-190. (in Chinese)
[16]	吕秋霞, 钟晓情, 任雅思. 基于Bi-LSTM网络的广珠城际短期客流预测方法[J]. 五邑大学学报(自然科学版), 2022, 36(1): 0-56. LYU Q X, ZHONG X Q, REN Y S. A short-term passenger flow prediction method for Guangzhou-Zhuhai intercity railway based on Bi-LSTM network[J]. Journal of Wuyi University(Natural Science Edition), 2022, 36(1): 0-56. (in Chinese)
[17]	郭彦茹, 王家川. 面向交通行业的地理信息管理与服务平台研究[J]. 北京测绘, 2020, 34(5): 600-605. GUO Y R, WANG J C. Research on geographic information management and service platform for transportation industry[J]. Beijing Surveying and Mapping, 2020, 34(5): 600-605. (in Chinese)
[18]	TRIPATHI S, SINGH S K, LEE H K. An end-to-end breast tumour classification model using context-based patch modelling-a Bi-LSTM approach for image classification[J]. Computerized Medical Imaging and Graphics, 2021, 87: 101838.
[19]	胡周. 分布式鲸鱼优化算法的应用研究[D]. 武汉: 湖北工业大学, 2020. HU Z. Application research of distributed whale optimization algorithm[D]. Wuhan: Hubei University of Technology, 2020. (in Chinese)
[20]	孙经伟, 谷远利. 基于BO-FCM和PSO-XGBoost的城市快速路交通状态识别[J]. 交通运输研究, 2023, 9(4): 64-71. SUN J W, GU Y L. Traffic state recognition on urban expressways based on BO-FCM and PSO-XGBoost[J]. Transportation Research, 2023, 9(4): 64-71. (in Chinese)