共享单车出行OD的时空注意力残差网络预测模型
曹弋(1982— ),男,辽宁大连人,博士,教授,主要从事交通运输规划与管理研究。E-mail: caoyi820619@aliyun.com |
Copy editor: 蒋树芳
收稿日期: 2023-07-17
修回日期: 2023-10-11
网络出版日期: 2024-03-31
基金资助
辽宁省社会科学规划基金项目(L22BSH003)
Spatiotemporal Attention Residual Network Prediction Model for OD of Bicycle Sharing Trips
Received date: 2023-07-17
Revised date: 2023-10-11
Online published: 2024-03-31
Supported by
Social Science Planning Fund Project of Liaoning Province(L22BSH003)
为探究共享单车出行的复杂时空规律与特性,揭示城市因素对共享单车出行OD的影响,提高OD预测精度,开展本研究。结合城市计算,考虑疫情、天气、温度、风速与节假日因素,构建共享单车出行OD的时空注意力残差网络预测模型(USTARN)。 USTARN先将共享单车OD数据通过时空特征切分捕捉单车流的时空依赖性,再结合注意力机制进行深度残差学习,最后根据城市因素学习结果调整预测结果。利用从政府数据开放平台获取的深圳市共享单车订单大数据及城市因素数据集,分析共享单车出行时空分布规律及其影响因素。将OD数据集按7:1:2划分为训练集、验证集与测试集,分别进行训练预测、模型参数自适应调整及模型验证对比实验。研究表明,USTARN模型的共享单车出行OD预测平均误差为7.68%,与不含城市计算的STARN模型及传统的CNN, BiLSTM模型相比,误差分别降低了5.93%、7.55%、6.07%,预测精度显著提高。 USTARN模型充分反映了时间、空间、疫情、天气、温度、风速等因素对共享单车出行OD的影响。研究成果对共享单车出行OD的精准预测具有理论指导意义,对该出行模式的推广并解决居民出行“最后一公里”问题具有实际应用价值。
曹弋 , 白涵文 , 王艺筱 . 共享单车出行OD的时空注意力残差网络预测模型[J]. 地球信息科学学报, 2024 , 26(3) : 556 -566 . DOI: 10.12082/dqxxkx.2024.230407
This study aims to explore the complex spatiotemporal patterns of bicycle-sharing trips, reveal the influence of urban factors on the OD of bicycle-sharing trips, and improve the accuracy of OD prediction. Combining the theory of urban computing, urban factors such as the epidemic, months, weather conditions (minimum temperature, maximum temperature, and wind speed), and whether it is a weekday along with the length information of non-motorized lanes are selected to construct a bicycle-sharing demand prediction model (USTARN) that integrates urban computing and spatiotemporal attention residual network. USTARN first captures the spatiotemporal dependence of sharing bicycle flow through spatial area division and time series slicing, then combines the attention mechanism for deep residual learning, and finally adjusts the deep residual prediction results according to the urban factor prediction results to improve the model performance. Using the big data from bicycle orders and urban factor datasets in Shenzhen obtained from the government data open platform, this study visualizes the spatiotemporal distribution patterns of bicycle-sharing trips and analyzes their influencing factors using the Python development environment. The OD data set is divided into training set, verification set, and test set in a 7: 1:2 ratio, and the model training, model parameter adaptive adjustment, and model result comparison are carried out, respectively. The results show that the average error of the USTARN model for OD prediction of bike-sharing trips is 7.68%, which is 5.93%, 7.55%, and 6.07% lower than that of the STARN model without urban computing and the traditional CNN model, which is good at data feature extraction, and the BiLSTM model, which is good at dealing with bi-directional time-series data, respectively. The USTARN model fully reflects the influence of time, space, epidemic, weather, and other factors on the OD of bike-sharing trips. Our results have theoretical guiding significance for the accurate prediction of bike-sharing trip OD, which can provide a scientific basis for urban non-motorized roadway planning and have practical application value for the promotion of bike-sharing travel mode and solving the 'last mile' problem of residents travel.
表1 深圳市共享单车订单数据示例Tab. 1 Example of bicycle sharing order data in Shenzhen |
车辆编号 | 起点时间 | 起点纬度/°N | 起点经度/°E | 讫点时间 | 讫点纬度/°N | 讫点经度/°E |
---|---|---|---|---|---|---|
8c3692da54d90dbf6dc22d | 2020-12-30 13:54 | 22.523 06 | 114.040 15 | 2020-12-30 14:30 | 22.518 42 | 114.058 58 |
b502ad04cf4dc7cdbd5ba4 | 2020-12-30 13:54 | 22.522 98 | 114.040 27 | 2020-12-30 14:30 | 22.518 33 | 114.057 92 |
b5576b989742ba817cffec | 2020-12-04 0:10 | 22.522 79 | 114.039 53 | 2020-12-04 0:22 | 22.519 21 | 114.052 43 |
143ea93f0d7e0e193728b8 | 2020-12-04 0:10 | 22.522 82 | 114.039 58 | 2020-12-04 0:22 | 22.519 24 | 114.052 41 |
1bee57a32c25b9ff567ad7 | 2020-12-04 20:42 | 22.520 16 | 114.040 21 | 2020-12-04 20:47 | 22.517 22 | 114.041 52 |
表2 典型区域工作日与非工作日出行热度对比Tab. 2 Comparison of heat on weekdays and non-weekdays in typical areas |
早高峰8:00—9:00 | 工作日(4月12日) | 非工作日(4月11日) |
---|---|---|
公司企业型区域(如招商银行、万科总部等) | ![]() | ![]() |
居民住宅型区域(如深圳石厦社区) | ![]() | ![]() |
表3 共享单车OD数据集示例Tab. 3 Example of shared bike OD dataset |
时间段 | 起点经 度区域 | 起点纬 度区域 | 终点经 度区域 | 终点纬 度区域 | OD量 /辆 |
---|---|---|---|---|---|
2020-01-30 11:00 | 13 | 3 | 11 | 2 | 1 |
2020-01-30 12:00 | 2 | 3 | 14 | 5 | 2 |
2020-01-30 13:00 | 11 | 3 | 14 | 4 | 1 |
2020-01-30 14:00 | 22 | 9 | 24 | 7 | 0 |
2020-01-30 18:00 | 21 | 12 | 21 | 10 | 2 |
表4 城市因素数据预处理结果示例Tab. 4 Example of pre-processing results for urban factor datat |
时间 | 现存确诊 | 当日新增 | 最高温度 | 最低温度 | 天气 | 风速等级 | 是否工作日 |
---|---|---|---|---|---|---|---|
2020-01-30 9:00 | 0.117 52 | 1.000 00 | 0.333 33 | 0.217 39 | 0.25 | 0.090 91 | 0 |
2020-02-02 13:00 | 0.223 95 | 0.349 06 | 0.466 67 | 0.304 35 | 0.00 | 0.045 45 | 0 |
2020-02-05 13:00 | 0.335 92 | 0.042 45 | 0.433 33 | 0.391 31 | 0.25 | 0.000 00 | 1 |
2021-02-19 14:00 | 0.066 52 | 0.075 47 | 0.533 33 | 0.434 78 | 0.00 | 0.045 45 | 1 |
2021-04-03 8:00 | 0.643 02 | 0.000 00 | 0.800 00 | 0.782 61 | 0.00 | 0.000 00 | 0 |
表5 各模型误差表现Tab. 5 Error performance of each model |
模型 | MAE | RMSE | MAPE/% |
---|---|---|---|
USTARN | 16.73 | 22.87 | 7.68 |
STARN | 27.05 | 34.23 | 13.61 |
CNN | 30.82 | 40.21 | 15.23 |
BiLSTM | 27.54 | 35.75 | 13.75 |
[1] |
|
[2] |
付学梅, 隽志才. 时间维度下上海市共享单车骑行模式研究[J]. 交通运输系统工程与信息, 2020, 20(3):219-226.
[
|
[3] |
|
[4] |
|
[5] |
|
[6] |
|
[7] |
|
[8] |
张鹏飞, 翁小雄. 基于注意力机制LSTM的地铁乘客出行行为预测研究[J]. 重庆交通大学学报(自然科学版), 2022, 41(5):14-19.
[
|
[9] |
许淼, 刘宏飞, 初凯. 基于AM-LSTM模型的共享单车时空需求预测[J]. 湖南大学学报(自然科学版), 2020, 47(12):77-85.
[
|
[10] |
李福, 徐良杰, 朱然博, 等. 基于XGBoost算法的共享单车借车需求量预测[J]. 武汉理工大学学报(交通科学与工程版), 2021, 45(5):880-884.
[
|
[11] |
|
[12] |
乔少杰, 韩楠, 岳昆, 等. 基于数据场聚类的共享单车需求预测模型[J]. 软件学报, 2022, 33(4):1451-1476.
[
|
[13] |
温惠英, 张东冉. 基于Bi-LSTM模型的高速公路交通量预测[J]. 公路工程, 2019, 44(6):51-56.
[
|
[14] |
李帅, 杨柳, 赵欣卉. 基于深度学习的城市区域短时交通拥堵预测算法[J]. 科学技术与工程, 2023, 23(25):10866-10878.
[
|
[15] |
杨帆, 车向红, 王勇, 等. 城市共享单车需求的多特征CNN-BiLSTM预测方法[J]. 测绘通报, 2023(8):113-119,150.
[
|
[16] |
杨寒雨, 赵晓永, 王磊. 数据归一化方法综述[J]. 计算机工程与应用, 2023, 59(3):13-22.
[
|
[17] |
彭然, 王竞钊, 刘业浩, 等. 武汉典型城市交通因素对 COVID-19传播的影响研究[J]. 南方建筑, 2022(4):78-87.
[
|
[18] |
|
[19] |
[
|
/
〈 |
|
〉 |