基于粒子群和LSTM模型的变区间短时停车需求预测方法

刘东辉; 肖雪; 张珏

doi:10.3963/j.jssn.1674-4861.2021.04.010

基于粒子群和LSTM模型的变区间短时停车需求预测方法

doi: 10.3963/j.jssn.1674-4861.2021.04.010

刘东辉^1, ,,
肖雪²,
张珏³

1.
吉林警察学院信息工程系长春 130000
2.
吉林大交通学院长春 130021
3.
杭州市公安局交通警察支队杭州 310000

基金项目:

国家自然科学基金青年科学基金项目 51308249

吉林省智能交通创新团队项目 20190101023JH

详细信息

通讯作者:
刘东辉(1973—), 硕士, 教授.研究方向: 交通安全、网络安全.E-mail: 68292143@qq.com

中图分类号: U491.7
计量
- 文章访问数: 230
- HTML全文浏览量: 114
- PDF下载量: 23
- 被引次数: 0
出版历程
- 收稿日期: 2020-08-02

A Prediction Method for Short-term Parking Demands in Variable Interval Based on Particle Swarm Optimization and LSTM Model

1.
Jilin Police College Department of Information Engineering, Changchun 130000, China
2.
School of Communication, Jilin University, Changchun 130021, China
3.
Traffic Police Detachment of Hangzhou Public Security Bureau, Hangzhou 310000, China

摘要

摘要: 停车信息是智能停车诱导系统得以成功实施的关键与基础, 被广泛认为能够有效解决当前停车难问题。鉴于停车信息在解决停车问题中的重要性, 研究了基于粒子群和LSTM模型的变区间短时停车需求预测方法。为充分发挥数据在提高模型预测精度的作用, 提出了以马尔可夫生灭过程为基础概率转移模型, 将停车到达率、离开率量化车随时间变化的停车需求, 通过标定实际的停车到达率和离开率, 确定预测模型的动态预测间隔与时段; 采用LSTM网络作为基础预测模型, 并利用粒子群优化算法优化网络参数。以吉林大学南岭校区停车场为研究对象, 按工作日与非工作日分别对停车数据进行预测并与其他预测模型进行对比分析。结果表明: 提出的停车需求预测模型在工作日的预测平均绝对误差为2.53辆, 均方误差为11.89辆; 非工作日的预测平均绝对误差为2.32辆, 均方误差为10.89辆。
- 智慧停车 /
- 实时及未来时刻停车信息 /
- 短时停车需求预测 /
- 马尔可夫生灭过程 /
- LSTM
Abstract: An intelligent parking guidance system is widely considered to solve the problem of difficult parking at present, providing management, traffic participants, and parking operators with immediate and future parking information. A prediction method for short-term parking demands in variable interval based on particle swarm optimization and LSTM model is studied due to the importance of parking information. Based on the birth and death of the Markov process, the characteristics of the temporal parking demand are analyzed. It is formulated as a combination of the arrival rate and departure rate of parking calibrated by the temporal parking quantity. Dynamic prediction intervals are determined according to the calibrated arrival rate and departure rate. The improved LSTM network is used as the basic prediction model, and the network parameters are optimized by the particle swarm optimization algorithm. The parking lot in the Nanling campus of Jilin University is selected as a research object, and its parking data are predicted and compared with other prediction models. The results show that the MAE and MSE of the proposed parking demand prediction model are 2.53 vehicles and 11.89 vehicles in working days, respectively. For non-business days, the MAE is2.32 vehicles and the MSE is 10.89 vehicles. Therefore, a predictable prediction model of parking demands proposed in the work can predict the real-time and future parking demands, providing a reliable reference for management, traffic participants, and parking operators.
- smart parking /
- real-time and future parking information /
- short-term prediction of parking demand /
- birth and death of the Markov process /
- LSTM

HTML全文

图 1 停车预测时段迭代流程图

Figure 1. Iteration flow for the parking forecast period

下载: 全尺寸图片幻灯片

图 2 LSTM网络隐含层拓扑结构

Figure 2. Hidden layer topology of the LSTM network

下载: 全尺寸图片幻灯片

图 3 LSTM网络训练与预测流程图

Figure 3. Training and prediction flow of the LSTM network

下载: 全尺寸图片幻灯片

图 4 6月17日—6月23日06:00—18:00时刻场内停车数

Figure 4. Number of parking spaces inside from 6:00 to 18:00 on June 17-23

下载: 全尺寸图片幻灯片

图 5 工作日PSO-LSTM模型、LSTM模型、BP模型、小波模型预测值

Figure 5. Absolute error values of PSO-LSTM NN, LSTM NN, BP NN, and WNN on working days

下载: 全尺寸图片幻灯片

图 6 工作日PSO-LSTM模型、LSTM模型、BP模型、小波模型绝对误差值

Figure 6. The absolute error values of PSO-LSTM NN, LSTM NN, BP NN and WNN on working days

下载: 全尺寸图片幻灯片

图 7 非工作日PSO-LSTM模型、LSTM模型、BP模型、小波模型预测值

Figure 7. Predicted values of PSO-LSTM NN, LSTM NN, BP NN, and WNN on non-working days

下载: 全尺寸图片幻灯片

图 8 非工作日PSO-LSTM模型、LSTM模型、BP模型、小波模型绝对误差值

Figure 8. Absolute error values of the PSO-LSTM NN, LSTM NN, BP NN, and WNN on non-working days

下载: 全尺寸图片幻灯片

表 1 非工作日预测时间间隔

Table 1. Interval of forecast on non-working days

时段	λ	p	R²	Δt/min
06:00—10:30	2.02	-0.003	0.95	5
10:30—13:10	39.76	0.062	0.91	1
13:10—15:35	93.16	0.138	0.91	1
15:35—18:00	22.14	0.036	0.89	1

下载: 导出CSV

表 2 工作日预测时间间隔

Table 2. Interval of forecast on working days

时段	λ	p	R²	Δt/min
06:00—10:10	-0.15	-0.016	0.93	5
10:30—13:10	-22.1	-0.024	0.91	1
13:10—14:35	77.64	0.091	0.94	1
14:35—18:00	-18.14	-0.017	0.98	1

下载: 导出CSV

表 3 误差对比表

Table 3. Comparison of errors 辆

	Model	MAE	MSE
工作日	PSO-LSTM	2.53	11.89
	LSTM	19.27	13.87
	BP	14.31	316.46
	WNN	9.18	148.74
非工作日	PSO-LSTM	2.32	10.89
	LSTM	7.27	172.09
	BP	4.89	39.55
	WNN	8.04	84.87

下载: 导出CSV

参考文献(20)

[1]	SHOUP D C. Cruising for parking[J]. Transport Policy, 2006, 13(6): 479-486. doi: 10.1016/j.tranpol.2006.05.005
[2]	CALISKAN M, BARTHELS A, SCHEUERMANN B, et al. Predicting parking lot occupancy in vehicular ad hoc networks[C]. IEEE 65^thConference on Vehicular Technology, Dublin: IEEE, 2007.
[3]	LU Rongxing, LIN Xiaodong, ZHU Haojin, et al. SPARK: Anew vanet-based smart parking scheme for large parking lots[C]. IEEE 28^thConference on Computer Communications, Janeiro, Brazil: IEEE, 2009.
[4]	BOYLES S D, TANG S, UNNIKRISHNAN A. Parking search equilibrium on a network[J]. Transportation Research Part B: Methodological, 2015, 81(2): 390-409.
[5]	CAICEDO F, BLAZQUEZ C, MIRANDA P. Prediction of parking space availability in real time[J]. Expert Systems with Applications, 2012, 39(8): 7281-7290. doi: 10.1016/j.eswa.2012.01.091
[6]	JUN Xiao, LOU Yingyan, JOSHUA Frisby. How likely am I to find parking? a practical model based frame work for predicting parking availability[J]. Transportation Research Part B: Methodological, 2018(112): 19-39.
[7]	LIU Shixu, GUAN Hongzhi, YAN Hai. Unoccupied parking space prediction of chaotic time series[C]. 10^thInternational Conference of Chinese Transportation Professionals(ICCTP), Beijing: Beijing Jiaotong University, 2010.
[8]	RAJABIOUN T, IOANNOU, P. On-street and off-street parking availability prediction using multivariate spatiotemporal models[J]. IEEE Transactions, 2015, 16(5): 2913-2924.
[9]	TIEDEMANN T, VÖGELE T, KRELL M. Concept of a data thread based parking space occupancy prediction in a berlin pilot region[C]. The 29^thAAAI Conference on Artificial Intelligence, Austin, USA: AI Magazine, 2015.
[10]	TAMRAZIAN A, QIAN Z, RAJAGOPAL R. Where is my parking spot? On-line and off-line prediction of time-varying parking occupancy[J]. Transport Research Record, 2015(5): 77-85.
[11]	VLAHOGIANNI E, KEPAPTSOGLOU K. A real-time parking prediction system for smart cities[J]. Intelligent Transport System, 2016, 20(2), 192-204. doi: 10.1080/15472450.2015.1037955
[12]	张军, 王书华, 杨正瓴. 基于粒子群优化的小波神经网络的停车位数量短时预测[J]. 计算机应用与软件, 2015, 32(11): 72-74. https://www.cnki.com.cn/Article/CJFDTOTAL-JYRJ201511017.htm ZHANG Jun, WANG Shunhua, YANG Zhengling. Short-term prediction of parking spaces in wavelet neural network based on particle swarm optimization[J]. Computer Applications & Software, 2015, 32(11): 72-74. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JYRJ201511017.htm
[13]	季彦婕, 陈晓实, 王炜. 基于小波变换和粒子群小波神经网络组合模型的有效停车泊位短时预测[J]. 吉林大学学报(工学版), 2016, 46(2): 399-405. https://www.cnki.com.cn/Article/CJFDTOTAL-JLGY201602010.htm JI Yanjie, CHEN Xiaoshi, WANG Wei. Short-term forecasting of parking space using particle swarm optimization-wavelet neural network model[J]. Journal of Jilin University(Engineering Science Edition), 2016, 46(2): 399-405. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JLGY201602010.htm
[14]	陈海鹏, 图晓航, 王玉, 等. 基于小波-ELM神经网络的短期停车泊位预测[J]. 吉林大学学报(理学版), 2017, 55(02): 388-392. https://www.cnki.com.cn/Article/CJFDTOTAL-JLDX201702036.htm CHEN Haipeng, TU Xiaohang, WANG Yu, et al. Short-term parking berth prediction based on wavelet-ELM neural network[J]. Journal of Jilin University(Science Edition), 2017, 55(2): 388-392. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JLDX201702036.htm
[15]	陈群, 晏克非, 王仁涛. 基于相空间重构及Elman网络的停车泊位数据预测[J]. 同济大学报(自然科学版), 2007, 35(5): 607-612. https://www.cnki.com.cn/Article/CJFDTOTAL-TJDZ200705007.htm CHEN Qun, YAN Kefei, WANG Renshou. Parking space information prediction based on phrase construction and elman neural network[J]. Journal of Tongji University(Natural Science Edition), 2007, 35(5): 607-612. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-TJDZ200705007.htm
[16]	蒋艳辉. 基于指数平滑法与马尔科夫链的道路交通事故预测研究[D]. 哈尔滨: 哈尔滨工业大学, 2018. JIANG Yanhui. Research of road traffic accident prediction based on exponential smoothing and Markova chain[D]. Harbin: Harbin Institute of Technology, 2018. (in Chinese)
[17]	SIMON Haykin. Neural networks and learning machines[M]. 3^rdEd. New York: Hall Press, 2009.
[18]	CAI Changchun, LIU Haolin, TAO Yuan, et al. Microgrid equivalent modeling based on long short-term memory neural network[J]. IEEE Access, 2020, 8(1): 23120-23133. http://ieeexplore.ieee.org/document/8957456/
[19]	刘衍民, 牛奔, 赵庆祯. 多目标优化问题的粒子群算法仿真研究[J]. 计算机应用研究, 2011, 28(2): 458-460. doi: 10.3969/j.issn.1001-3695.2011.02.013 LIU Yanmin, NIU Ben, ZHAO Qingzhen. Particle swarm optimizer simulation research of multi-objective optimization problems[J]. Journal of Computer Research and Development, 2011, 28(2): 458-460. (in Chinese) doi: 10.3969/j.issn.1001-3695.2011.02.013
[20]	郑金宇. 基于小波神经网络及布谷鸟算法的停车位预测方法研究[D]. 长春: 吉林大学, 2018. ZHENG Jinyu. Research on parking space prediction method based on wavelet neural network and cuckoo algorithm[D]. Changchun: Jilin University, 2018. (in Chinese)