基于注意力机制的泊位占有率预测模型研究

doi:10.11959/j.issn.1000-436X.2020241

通信学报 ›› 2020, Vol. 41 ›› Issue (12): 182-192.doi: 10.11959/j.issn.1000-436X.2020241

基于注意力机制的泊位占有率预测模型研究

王竹荣, 薛伟, 牛亚邦, 崔颖安, 孙钦东, 黑新宏

西安理工大学计算机科学与工程学院，陕西西安 710048

修回日期:2020-10-07 出版日期:2020-12-25 发布日期:2020-12-01
作者简介:王竹荣（1966- ），男，湖南衡阳人，博士，西安理工大学副教授、硕士生导师，主要研究方向为智能计算、深度学习及优化应用等。
薛伟（1993- ），男，陕西韩城人，西安理工大学硕士生，主要研究方向为深度学习与进化计算。
牛亚邦（1993- ），男，山西吕梁人，西安理工大学硕士生，主要研究方向为深度学习。
崔颖安（1975- ），男，陕西西安人，博士，西安理工大学讲师，主要研究方向为社会化媒体大数据抽样与大数据处理。
孙钦东（1975- ），男，山东莒南人，西安理工大学教授、博士生导师，主要研究方向为智能信息处理、网络安全。
黑新宏（1976- ），男，陕西延安人，博士，西安理工大学教授、博士生导师，主要研究方向为智能系统与安全关键系统。
基金资助:
国家重点研发计划基金资助项目(2018YFB1201500);国家自然科学基金资助项目(61773313);陕西省重点研发计划基金资助项目(2017ZDXM-GY-098);陕西省教育厅重点实验室基金资助项目(17JS100)

Research on berth occupancy prediction model based on attention mechanism

Zhurong WANG, Wei XUE, Yabang NIU, Ying’an CUI, Qindong SUN, Xinhong HEI

School of Computer Science and Engineering, Xi’an University of Technology, Xi’an 710048, China

Revised:2020-10-07 Online:2020-12-25 Published:2020-12-01
Supported by:
The National Key Research and Development Program of China(2018YFB1201500);The National Natural Science Foundation of China(61773313);Shaanxi Key Research and Development Program(2017ZDXM-GY-098);Shaanxi Provincial Department of Education Key Laboratory Project(17JS100)

摘要/Abstract

摘要：

为解决泊位占有率的预测精度随步长增加而下降的问题，提出了一种基于注意力机制的泊位占有率预测模型。通过卷积神经网络获得多变量的时间模式信息作为模型的注意力机制。通过对模型训练、学习特征信息，并对相关性高的序列分配较大的学习权重，来实现解码器输出高度相关的有用特征预测目标序列。应用多个停车场数据集对模型进行测试，测试结果及对比分析表明，所提模型在步长达到 36 时对泊位占有率的预测数据能较好地估计真实值，预测精度和稳定性相比LSTM均有提高。

关键词: 时间序列预测, 泊位占有率预测, 注意力机制, 序列到序列模型

Abstract:

To solve the problem that the berth occupancy prediction accuracy decreases while the prediction step was increasing, a berth occupancy prediction model based on an attention mechanism was proposed, which was the multivariate time pattern information obtained by convolutional neural networks (CNN).The characteristic information was learned by the model training, and the sequence with higher correlation was assigned a larger learning weight, so that the highly correlated features output from the decoder could be used to predict the target sequence.Data sets of multiple parking lot were adopted to test the model.The test results show that the proposed model can estimate the real value well when the step length of berth occupancy prediction reaches 36.The prediction accuracy and stability of the model are improved compared with long short-term memory (LSTM) model.

Key words: time series prediction, berth occupancy prediction, attention mechanism, sequence-to-sequence model

中图分类号:

TP391

王竹荣, 薛伟, 牛亚邦, 崔颖安, 孙钦东, 黑新宏. 基于注意力机制的泊位占有率预测模型研究[J]. 通信学报, 2020, 41(12): 182-192.

Zhurong WANG, Wei XUE, Yabang NIU, Ying’an CUI, Qindong SUN, Xinhong HEI. Research on berth occupancy prediction model based on attention mechanism[J]. Journal on Communications, 2020, 41(12): 182-192.

图/表 12

图1

图2

图3

图4

图5

表1

表2

图6

图7

图8

表3

图9

参考文献 33

[1]	ZHANG J , WANG F Y , WANG K ,et al. Data-driven intelligent transportation systems:a survey[J]. IEEE Transactions on Intelligent Transportation Systems, 2011,12(4): 1624-1639.
[2]	DONALD C S . Cruising for parking[J]. University of California Transportation Center Working Papers, 2007,13(6): 479-486.
[3]	AYALA D , WOLFSON O , XU B ,et al. Pricing of parking for congestion reduction[C]// Proceedings of the 20th International Conference on Advances in Geographic Information Systems. New York:ACM Press, 2012: 43-51.
[4]	ARORA N , COOK J , KUMAR R ,et al. Hard to park? Estimating parking difficulty at scale[C]// Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery ＆ Data Mining. New York:ACM Press, 2019: 2296-2304.
[5]	RONG Y , XU Z , YAN R ,et al. Du-parking:spatio-temporal big data tells you realtime parking availability[C]// Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery ＆Data Mining. New York:ACM Press, 2018: 646-654.
[6]	AN D . Method and system for projecting dynamic parking availability based on an ongoing survey for remote lots with high demand:US7049979[P].(2006-05-23)[2020-07-07].
[7]	CALISKAN M , BARTHELS A , SCHEUERMANN B ,et al. Predicting parking lot occupancy in vehicular ad hoc networks[C]// 2007 IEEE 65th Vehicular Technology Conference-VTC2007-Spring. Piscataway:IEEE Press, 2007: 277-281.
[8]	CHEN B , PINELLI F , SINN M ,et al. Uncertainty in urban mobility:predicting waiting times for shared bicycles and parking lots[C]// 16th International IEEE Conference on Intelligent Transportation Systems (ITSC 2013). Piscataway:IEEE Press, 2013: 53-58.
[9]	RAJABIOUN T , FOSTER B , IOANNOU P . Intelligent parking assist[C]// 21st Mediterranean Conference on Control and Automation. Piscataway:IEEE Press, 2013: 1156-1161.
[10]	KLAPPENECKER A , LEE H , WELCH J L . Finding available parking spaces made easy[J]. Ad Hoc Networks, 2014,12: 243-249.
[11]	RICHTER F , DI M S , MATTFELD D C . Temporal and spatial clustering for a parking prediction service[C]// 2014 IEEE 26th International Conference on Tools with Artificial Intelligence. Piscataway:IEEE Press, 2014: 278-282.
[12]	杨兆升, 陈晓冬 . 智能化停车诱导系统有效停车泊位数据的预测技术研究[J]. 交通运输系统工程与信息, 2003,3(4): 12-15.
	YANG Z S , CHEN X D . Research on the estimation for effective parking space of the intelligentized parking guidance system[J]. Journal of Transportation System Engineering and Information Technology, 2003,3(4): 12-15.
[13]	陈群, 晏克非, 王仁涛 ,等. 基于相空间重构及 Elman 网络的停车泊位数据预测[J]. 同济大学学报 (自然科学版), 2007,35(5): 607-611.
	CHEN Q , YAN K F , WANG R T ,et al. Parking space information prediction based on phrase construction and Elman neural network[J]. Journal of Tongji University (Natural Science), 2007,35(5): 607-611.
[14]	BLYTHE P , JI Y , GUO W ,et al. Short-term forecasting of available parking space using wavelet neural network model[J]. IET Intelligent Transport Systems, 2015,9(2): 202-209.
[15]	季彦婕, 陈晓实, 王炜 ,等. 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.
	JI Y J , CHEN X S , WANG W ,et al. Short-term forecasting of parking space using particle swarm optimization-wavelet neural network model[J]. Journal of Jilin University (Engineering and Technology Edition), 2016,46(2): 399-405.
[16]	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(2): 388-392.
	CHEN H P , TU X H , WANG Y ,et al. Short-term parking space prediction based on wavelet -ELM neural networks[J]. Journal of Jilin University(Science Edition), 2017,55(2): 388-392.
[17]	韩印, 郑喆, 赵靖 ,等. 基于灰色—小波神经网络的有效泊位预测[J]. 交通运输系统工程与信息, 2017,17(5): 60-67.
	HAN Y , ZHENG Z , ZHAO J ,et al. Forecasting of effective parking space based on grey-distributed wavelet neural network model[J]. Journal of Transportation Systems Engineering and Information Technology, 2017,17(5): 60-67.
[18]	GONG G J , AN X N , NAWARAJ K M ,et al. Research on short-term load prediction based on seq2seq model[J]. Energies, 2019,12(16):3199 (1-18).
[19]	RANGAPURAM S S , SEEGER M W , GASTHAUS J ,et al. Deep state space models for time series forecasting[C]// Advances in Neural Information Processing Systems. Massachusetts:MIT Press, 2018: 7785-7794.
[20]	RAN X D , SHAN Z G , FANG Y F ,et al. An LSTM-based method with attention mechanism for travel time prediction[J]. Sensors, 2019,19(4): 1-22.
[21]	FAN C , ZHANG Y , PAN Y ,et al. Multi-horizon time series forecasting with temporal attention learning[C]// Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery ＆ Data Mining. New York:ACM Press, 2019: 2527-2535.
[22]	LAI G , CHANG W C , YANG Y ,et al. Modeling long-and short-term temporal patterns with deep neural networks[C]// The 41st International ACM SIGIR Conference on Research ＆ Development in Information Retrieval. New York:ACM Press, 2018: 95-104.
[23]	SHIH S Y , SUN F K , LEE H . Temporal pattern attention for multivariate time series forecasting[J]. Machine Learning, 2019,108(8-9): 1421-1441.
[24]	范晓诗, 雷英杰, 路艳丽 ,等. 基于DTW的长期直觉模糊时间序列预测模型[J]. 通信学报, 2016,37(8): 95-104.
	FAN X S , LEI Y J , LU Y L ,et al. Long-term intuitionistic fuzzy time series forecasting model based on DTW[J]. Journal on Communications, 2016,37(8): 95-104.
[25]	王亚男, 雷英杰, 雷阳 ,等. 高阶直觉模糊时间序列预测模型[J]. 通信学报, 2016,37(5): 115-124.
	WANG Y N , LEI Y J , LEI Y ,et al. High order intuitionistic fuzzy time series forecasting model[J]. Journal on Communications, 2016,37(5): 115-124
[26]	VASWANI A , SHAZEER N , PARMAR N ,et al. Attention is all you need[C]// Advances in Neural Information Processing Systems. Massachusetts:MIT Press, 2017: 5998-6008.
[27]	LI S , JIN X , XUAN Y ,et al. Enhancing the locality and breaking the memory bottleneck of transformer on time series forecasting[C]// Advances in Neural Information Processing Systems. Massachusetts:MIT Press, 2019: 5244-5254.
[28]	李琳辉, 周彬, 连静 ,等. 基于社会注意力机制的行人轨迹预测方法研究[J]. 通信学报, 2020,41(6): 175-183.
	LI L H , ZHOU B , LIAN J ,et al. Research on pedestrian trajectory prediction method based on social attention mechanism[J]. Journal on Communications, 2020,41(6): 175-183.
[29]	何坚, 朱喆, 王伟东 ,等. 多维时空因果关系学习的停车泊位占用率预测技术[J]. 北京工业大学学报, 2020,46(1): 17-23.
	HE J , ZHU Z , WANG W D ,et al. Saliency fusion of multispectral images with non-uniform texture and complex surface defects[J]. Journal of Beijing University of Technology, 2020,46(1): 17-23.
[30]	梅振宇, 章伟 . 基于复杂性测度的泊位占有率序列动力学分析[J]. 浙江大学学报(工学版), 2018,52(4): 727-734,743.
	MEI Z Y , ZHANG W . Dynamic analysis of parking space occupancy series based on complexity measurement[J]. Journal of Zhejiang University(Engineering Science), 2018,52(4): 727-734,743.
[31]	HUANG C G , HUANG H Z , LI Y F . A bidirectional LSTM prognostics method under multiple operational conditions[J]. IEEE Transactions on Industrial Electronics, 2019,66(11): 8792-8802.
[32]	HOCHREITER S , SCHMIDHUBER J . Long short-term memory[J]. Neural Computation, 1997,9(8): 1735-1780.
[33]	GRAVES A . Supervised sequence labelling with recurrent neural networks[M]. Berlin: Springer, 2012.

参数	参数值
训练集数据量/个	1 302
迭代次数/次	1 000
学习率	0.001
隐含层节点数/个	64
输入数据长度	18
测试集数据量/个	326
输出步长	36
遗忘率	0.75
输入特征数/个	4

停车场	LSTM （步长=10）	sequence to sequence （步长=36）
BHMBCCMKT01	0.008 6	0.010
BHMBCCPST01	0.024	0.021
BHMBCCSNH01	0.037	0.035
BHMBCCTHL01	0.028	0.030
BHMBRCBRG01	0.033	0.036
BHMBRCBRG02	0.078	0.059
BHMBRCBRG03	0.033	0.029
BHMEURBRD01	0.025	0.021
BHMEURBRD02	0.012	0.012
BHMMBMMBX01	0.030	0.027
BHMNCPHST01	0.044	0.040
BHMNCPLDH01	0.022	0.015
BHMNCPNHS01	0.019	0.012
BHMNCPNST01	0.015	0.016
BHMNCPPLS01	0.019	0.017
BHMNCPRAN01	0.018	0.019
Broad Street	0.021	0.015
Bull Ring	0.035	0.035
NIA Car Parks	0.015	0.013
NIA South	0.014	0.011
Others-CCCPS105a	0.032	0.027
Others-CCCPS119a	0.018	0.016
Others-CCCPS133	0.015	0.013
Others-CCCPS135a	0.049	0.040
Others-CCCPS202	0.041	0.034
Others-CCCPS8	0.035	0.029
Others-CCCPS98	0.021	0.021
Shopping	0.028	0.029
中位数（median）	0.025	0.021
平均值（mean）	0.028	0.024
最大值（max）	0.078	0.059
最小值（min）	0.008 6	0.010
标准偏差（std）	0.015	0.011

基于注意力机制的泊位占有率预测模型研究

Research on berth occupancy prediction model based on attention mechanism

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 12

参考文献 33

相关文章 7

Metrics

推荐阅读 0

预测步	真实值/个	有注意力机制预测值/个	误差/个	无注意力机制预测值/个	误差/个
1	181	172	9	95	86
2	197	192	5	171	26
3	196	200	4	172	24
4	200	197	3	172	28
5	203	198	5	172	31
6	196	197	1	173	23
7	186	182	4	172	14
8	175	167	8	169	6
9	149	155	6	164	15
10	131	129	2	134	3
11	105	123	18	109	4
12	11	22	11	90	79
13	14	21	7	15	1
14	26	21	5	16	10
15	39	39	0	34	5
16	63	65	2	54	9
17	87	85	2	83	4
18	113	119	6	105	8
19	126	128	2	132	6
20	148	143	5	136	12
21	161	167	6	154	7
22	172	170	2	161	11
23	177	178	1	166	11
24	170	171	1	166	4
25	165	162	3	160	5
26	149	145	4	155	6
27	142	128	14	135	7
28	106	127	21	125	19
29	95	86	9	90	5
30	11	39	28	82	71
31	11	22	11	15	4
32	22	18	4	13	9
33	39	35	4	28	11
34	83	74	9	58	25
35	120	115	5	120	0
36	194	178	16	144	50
平均	—	—	6.75	—	17.75

[1]	冯海林, 张潇, 刘同存. 融合评论文本特征和评分图卷积表示的推荐模型[J]. 通信学报, 2022, 43(3): 164-171.
[2]	仲美玉, 吴培良, 窦燕, 刘毅, 孔令富. 基于中文语义-音韵信息的语音识别文本校对模型[J]. 通信学报, 2022, 43(11): 65-79.
[3]	陈卓, 朱淼, 杜军威. 基于多视角图神经网络的欺诈检测算法[J]. 通信学报, 2022, 43(11): 225-232.
[4]	王洪雁, 袁海. 基于骨骼及表观特征融合的动作识别方法[J]. 通信学报, 2022, 43(1): 138-148.
[5]	赵晓娟, 贾焰, 李爱平, 陈恺. 基于层级注意力机制的链接预测模型研究[J]. 通信学报, 2021, 42(3): 36-44.
[6]	郭璠, 张泳祥, 唐琎, 李伟清. YOLOv3-A：基于注意力机制的交通标志检测网络[J]. 通信学报, 2021, 42(1): 87-99.
[7]	李琳辉,周彬,连静,周雅夫. 基于社会注意力机制的行人轨迹预测方法研究[J]. 通信学报, 2020, 41(6): 175-183.