基于recurrent neural networks的网约车供需预测方法
|
摘要
以网约车订单等真实数据为数据源,结合TensorFlow深度学习框架,利用循环神经网络(recurrent neu-ral networks)方法,预测网约车在未来某时间某地点的订单需求量。提出改进LSTM RNN(长短时记忆循环神经网络)模型,经过对其优化和训练,能够有效预测网约车未来某时间某地点的供需量。对数据源进行可视化分析,排除不相关数据源干扰,以此为基础设计仿真实验。仿真实验表明,该模型的正确率比反向传播神经网络(BPNN)、回归决策树(DTR)、非线性回归支持向量机(SVR)以及随机漫步(RW)等模型高,同时,对长短间隔不同的历史数据有较好的记忆能力,在测试数据上有较强的泛化能力。
Ordered from online car as data sources,this paper used TensorFlow and recurrent neural networks to predict the supply and demand for online car at a certain point in the future. This paper presented the model of LSTM RNN,which was optimized and trained to effectively predict the supply and demand of the online car at a certain point in the future. Visual analysis of data source,help excluding uncorrelated data source,which was the basic to design simulation experiment. Simulation experiments show that the accuracy of the proposed model is higher than back propagation neural network( BPNN) and decision tree regression( DTR),nonlinear support vector regression machine( SVR) and random walk( RW),at the same time,it has the excellent memory capability of different length of historical data,and the excellent generalization capability on the test set.
Ordered from online car as data sources,this paper used TensorFlow and recurrent neural networks to predict the supply and demand for online car at a certain point in the future. This paper presented the model of LSTM RNN,which was optimized and trained to effectively predict the supply and demand of the online car at a certain point in the future. Visual analysis of data source,help excluding uncorrelated data source,which was the basic to design simulation experiment. Simulation experiments show that the accuracy of the proposed model is higher than back propagation neural network( BPNN) and decision tree regression( DTR),nonlinear support vector regression machine( SVR) and random walk( RW),at the same time,it has the excellent memory capability of different length of historical data,and the excellent generalization capability on the test set.
引文
[1] Sainath T N,Weiss R J,Wilson K W. Multichannel signal processingwith deep neural networks for automatic speech recognition[J]. IEEEACM Trans on Audio,Speech,and Language Processing,2017,25(5):965-979.
[2] Goodfellow I,Bengio Y,Courville A. Deep learning(adaptive com-putation and machine learning series)[M]. Cambridge:MIT Press,2016.
[3] Silver D,Huang A,Maddison C J,et al. Mastering the game of gowith deep neural networks and tree search[J]. Nature,2016,529:484-489.
[4] Yi H,Jung H J,Bae S. Deep neural networks for traffic flow predic-tion[C]//Proc of IEEE International Conference on Big Data andSmart Computing. Piscataway,NJ:IEEE Press,2017:328-331.
[5] Mo Y J,Kim J,Kim J K,et al. Performance of deep learning com-putation with TensorFlow software library in GPU-capable multi-corecomputing platforms[C]//Proc of the 9th International Conference onUbiquitous and Future Networks. 2017:240-242.
[6] Ichinose A,Oguchi M,Takefusa A,et al. Evaluation of distributedprocessing of Caffe framework using poor performance device[C]//Proc of IEEE International Conference on Big Data. Piscataway,NJ:IEEE Press,2016:3980-3982.
[7] Abadi M,Agarwal A,Barham P,et al. Tensor Flow:large-scale ma-chine learning on heterogeneous distributed systems[EB/OL].(2016-03-14). http://arxiv. org/abs/1603. 04467.
[8] Hinton G,Sabour S,Frosst N. Dynamic routing between capsule[C]//Proc of Conference on Neural Information Processing Systems.2017.
[9] Wongsuphasawat K,Smilkov D,Wexler J,et al. Visualizing dataflowgraphs of deep learning models in TensorFlow[J]. IEEE Trans onVisualization and Computer Graphics,2018,24(1):1-12.
[10]Gallardo N,Gamez N,Rad P,et al. Autonomous decision making fora driver-less car[C]//Proc of the 12th System of Systems EngineeringConference. Piscataway,NJ:IEEE Press,2017.
[11]Gao Feng. Network traffic prediction based on neural network[C]//Proc of International Conference on Intelligent Transportation,Big Da-ta and Smart City. Piscataway,NJ:IEEE Press,2015:527-530.
[12]Chen Dawei. Research on traffic flow prediction in the big data envi-ronment based on the improved RBF neural network[J]. IEEE Transon Industrial Informatics,2017,13(4):2000-2008.
[13]Transportation Research Board,National Research Council. Highwaycapacity manual[M]. Washington DC:National Research Council,2000:113.
[14] Karlaftis M,Vlahogianni E. Statistical methods versus neural net-works in transportation research:differences,similarities and someinsights[J]. Transportation Research Part C:Emerging Tech-nologies,2011,19(3):387-399.
[15]Levin M,Tsao Y D. On forecasting freeway occupancies and volumes(abridgment)[J]. Transportation Research Record,1980(722):47-49.
[16]Yu Haiyang,Wu Zhihai,Chen Dongwei,et al. Probabilistic predic-tion of bus headway using relevance vector machine regression[J].IEEE Trans on Intelligent Transportation Systems,2017,18(7):1772-1781.
[17]Castro-Neto M,Jeong Y S,Jeong M K,et al. Online-SVR for short-term traffic flow prediction under typical and atypical traffic conditions[J]. Expert Systems with Applications,2009,6(3):6164-6173.
[18]Chan K Y,Dillon T,Chang E,et al. Prediction of short-term trafficvariables using intelligent swarm-based neural networks[J]. IEEETrans on Control Systems Technology,2013,21(1):263-274.
[19]Vlahogianni E I,Karlaftis M G,Golias J C. Optimized and meta-op-timized neural networks for short-term traffic flow prediction:a geneticapproach[J]. Transportation Research Part C:Emerging Tech-nologies,2005,13(3):211-234.
[20]Yu Xiaohan,Xiong Shengwu,Xiang Jianwen,et al. A campus trafficcongestion detecting method based on BP neural network[C]//Procof the 2nd International Symposium on Dependable Computing and In-ternet of Things. Piscataway,NJ:IEEE Press,2015.
[21]Chai Yanchong,Huang Darong,Zhao Ling. A short-term traffic flowprediction method based on wavelet analysis and neural network[C]//Proc of Chinese Control and Decision Conference. Piscataway,NJ:IEEE Press,2016:7030-1034.
[22]Chen Dawei. Research on traffic flow prediction in the big data envi-ronment based on the improved RBF neural network[J]. IEEE Transon Industrial Informatics,2017,13(4):2000-2008.
[23] Wang Peng,Zhao Gang,Yao Xingren. Applying back-propagationneural network to predict bus traffic[C]//Proc of the 12th Interna-tional Conference on Natural Computation,Fuzzy Systems and Know-ledge Discovery. Piscataway,NJ:IEEE Press,2016:752-756.
[24] Yu Wanxia,Liu Lina,Zhang Weicun. Traffic prediction methodbased on RBF neural network with improved artificial bee colony algo-rithm[C]//Proc of the 8th International Conference on IntelligentNetworks and Intelligent Systems. Piscataway,NJ:IEEE Press,2015:141-144.
[25]Wang Dong,Cao Wei,Li Jian. Deep SD:supply-demand predictionfor online car-hailing services using deep neural networks[C]//Procof the 33rd International Conference on Data Engineering. Pisca-taway,NJ:IEEE Press,2017:243-254.
[26]Hochreiter S,Schmidhuber J. Long short-term memory[J]. NeuralComputation,1997,9(8):1735-1780.
[27]Ma Xiaolei,Ding Chuan,Luan Sen,et al. Prioritizing influential fac-tors for freeway incident clearance time prediction using the gradientboosting decision trees method[J]. IEEE Trans on IntelligentTransportation Systems,2017,18(9):2303-2310.
[28]Zaremba W,Sutskever I,Vinyals O. Recurrent neural network regu-larization[EB/OL]. 2014-09-08.[2015-02-19]. https://arxiv.org/abs/1409. 2329v5.
[29]Schmidhuber J. Learning to control fast-weight memories:an alterna-tive to dynamic recurrent networks[J]. Neural Computation,1992,4(1):131-139.
[30]Siegelmann H T,Sontag E D. Some recent results on computing with'neural nets'[C]//Proc of the 31st IEEE Conference on Decision andControl. Piscataway,NJ:IEEE Press,1992:1476-1481.
[31] Fouladgar M,Parchami M,Elmasri R,et al. Scalable deep trafficflow neural networks for urban traffic congestion prediction[C]//Procof International Joint Conference on Neural Networks. Piscataway,NJ:IEEE Press,2017:2251-2258.
[2] Goodfellow I,Bengio Y,Courville A. Deep learning(adaptive com-putation and machine learning series)[M]. Cambridge:MIT Press,2016.
[3] Silver D,Huang A,Maddison C J,et al. Mastering the game of gowith deep neural networks and tree search[J]. Nature,2016,529:484-489.
[4] Yi H,Jung H J,Bae S. Deep neural networks for traffic flow predic-tion[C]//Proc of IEEE International Conference on Big Data andSmart Computing. Piscataway,NJ:IEEE Press,2017:328-331.
[5] Mo Y J,Kim J,Kim J K,et al. Performance of deep learning com-putation with TensorFlow software library in GPU-capable multi-corecomputing platforms[C]//Proc of the 9th International Conference onUbiquitous and Future Networks. 2017:240-242.
[6] Ichinose A,Oguchi M,Takefusa A,et al. Evaluation of distributedprocessing of Caffe framework using poor performance device[C]//Proc of IEEE International Conference on Big Data. Piscataway,NJ:IEEE Press,2016:3980-3982.
[7] Abadi M,Agarwal A,Barham P,et al. Tensor Flow:large-scale ma-chine learning on heterogeneous distributed systems[EB/OL].(2016-03-14). http://arxiv. org/abs/1603. 04467.
[8] Hinton G,Sabour S,Frosst N. Dynamic routing between capsule[C]//Proc of Conference on Neural Information Processing Systems.2017.
[9] Wongsuphasawat K,Smilkov D,Wexler J,et al. Visualizing dataflowgraphs of deep learning models in TensorFlow[J]. IEEE Trans onVisualization and Computer Graphics,2018,24(1):1-12.
[10]Gallardo N,Gamez N,Rad P,et al. Autonomous decision making fora driver-less car[C]//Proc of the 12th System of Systems EngineeringConference. Piscataway,NJ:IEEE Press,2017.
[11]Gao Feng. Network traffic prediction based on neural network[C]//Proc of International Conference on Intelligent Transportation,Big Da-ta and Smart City. Piscataway,NJ:IEEE Press,2015:527-530.
[12]Chen Dawei. Research on traffic flow prediction in the big data envi-ronment based on the improved RBF neural network[J]. IEEE Transon Industrial Informatics,2017,13(4):2000-2008.
[13]Transportation Research Board,National Research Council. Highwaycapacity manual[M]. Washington DC:National Research Council,2000:113.
[14] Karlaftis M,Vlahogianni E. Statistical methods versus neural net-works in transportation research:differences,similarities and someinsights[J]. Transportation Research Part C:Emerging Tech-nologies,2011,19(3):387-399.
[15]Levin M,Tsao Y D. On forecasting freeway occupancies and volumes(abridgment)[J]. Transportation Research Record,1980(722):47-49.
[16]Yu Haiyang,Wu Zhihai,Chen Dongwei,et al. Probabilistic predic-tion of bus headway using relevance vector machine regression[J].IEEE Trans on Intelligent Transportation Systems,2017,18(7):1772-1781.
[17]Castro-Neto M,Jeong Y S,Jeong M K,et al. Online-SVR for short-term traffic flow prediction under typical and atypical traffic conditions[J]. Expert Systems with Applications,2009,6(3):6164-6173.
[18]Chan K Y,Dillon T,Chang E,et al. Prediction of short-term trafficvariables using intelligent swarm-based neural networks[J]. IEEETrans on Control Systems Technology,2013,21(1):263-274.
[19]Vlahogianni E I,Karlaftis M G,Golias J C. Optimized and meta-op-timized neural networks for short-term traffic flow prediction:a geneticapproach[J]. Transportation Research Part C:Emerging Tech-nologies,2005,13(3):211-234.
[20]Yu Xiaohan,Xiong Shengwu,Xiang Jianwen,et al. A campus trafficcongestion detecting method based on BP neural network[C]//Procof the 2nd International Symposium on Dependable Computing and In-ternet of Things. Piscataway,NJ:IEEE Press,2015.
[21]Chai Yanchong,Huang Darong,Zhao Ling. A short-term traffic flowprediction method based on wavelet analysis and neural network[C]//Proc of Chinese Control and Decision Conference. Piscataway,NJ:IEEE Press,2016:7030-1034.
[22]Chen Dawei. Research on traffic flow prediction in the big data envi-ronment based on the improved RBF neural network[J]. IEEE Transon Industrial Informatics,2017,13(4):2000-2008.
[23] Wang Peng,Zhao Gang,Yao Xingren. Applying back-propagationneural network to predict bus traffic[C]//Proc of the 12th Interna-tional Conference on Natural Computation,Fuzzy Systems and Know-ledge Discovery. Piscataway,NJ:IEEE Press,2016:752-756.
[24] Yu Wanxia,Liu Lina,Zhang Weicun. Traffic prediction methodbased on RBF neural network with improved artificial bee colony algo-rithm[C]//Proc of the 8th International Conference on IntelligentNetworks and Intelligent Systems. Piscataway,NJ:IEEE Press,2015:141-144.
[25]Wang Dong,Cao Wei,Li Jian. Deep SD:supply-demand predictionfor online car-hailing services using deep neural networks[C]//Procof the 33rd International Conference on Data Engineering. Pisca-taway,NJ:IEEE Press,2017:243-254.
[26]Hochreiter S,Schmidhuber J. Long short-term memory[J]. NeuralComputation,1997,9(8):1735-1780.
[27]Ma Xiaolei,Ding Chuan,Luan Sen,et al. Prioritizing influential fac-tors for freeway incident clearance time prediction using the gradientboosting decision trees method[J]. IEEE Trans on IntelligentTransportation Systems,2017,18(9):2303-2310.
[28]Zaremba W,Sutskever I,Vinyals O. Recurrent neural network regu-larization[EB/OL]. 2014-09-08.[2015-02-19]. https://arxiv.org/abs/1409. 2329v5.
[29]Schmidhuber J. Learning to control fast-weight memories:an alterna-tive to dynamic recurrent networks[J]. Neural Computation,1992,4(1):131-139.
[30]Siegelmann H T,Sontag E D. Some recent results on computing with'neural nets'[C]//Proc of the 31st IEEE Conference on Decision andControl. Piscataway,NJ:IEEE Press,1992:1476-1481.
[31] Fouladgar M,Parchami M,Elmasri R,et al. Scalable deep trafficflow neural networks for urban traffic congestion prediction[C]//Procof International Joint Conference on Neural Networks. Piscataway,NJ:IEEE Press,2017:2251-2258.