短时交通流预测中的特征选择算法研究
|
摘要
短时交通流预测是智能交通系统的重要基础,其精度直接影响到交通控制和诱导的效果.对于交通流预测中的非参数回归方法,其中一个重要的问题是状态向量的选取.本文提出基于ReliefF和Delta Test的特征选择算法来对特征向量进行选择.首先使用ReliefF算法根据特征和类别的相关性对状态向量进行快速初步筛选,加快算法的执行效率.接下来以Delta Test为性能指标,使用遗传算法对状态分量的权重进行进一步优选.最后通过基于实际数据的算例,对本文方法优选的状态向量与时间序列状态向量,简单时空关联向量进行了对比.结果表明,本文的方法在一般交通状态条件下和突变交通状态下都具有较好的性能.
The short-term traffic flow prediction is an important foundation of intelligent transportation system,and its performance significantly influences the effectiveness of traffic control and guidance. For the nonparametric regression method in traffic flow prediction, one of the most important problems is state vectors selection. This paper presents a feature selection algorithm based on ReliefF and Delta Test to select feature vectors. Firstly, ReliefF algorithm is used to filter the state vectors according to the correlation between features and classes to speed up the efficiency of the algorithm. Then using Delta Test as performance index, genetic algorithm is used to optimize the weight of state component. Finally, the state vector selected by this method is compared with the state vector of time series and the simple Spatial-temporal correlation vector. Numerical results show that the proposed method outperforms the other two usually used methods under both general and abrupt traffic conditions.
The short-term traffic flow prediction is an important foundation of intelligent transportation system,and its performance significantly influences the effectiveness of traffic control and guidance. For the nonparametric regression method in traffic flow prediction, one of the most important problems is state vectors selection. This paper presents a feature selection algorithm based on ReliefF and Delta Test to select feature vectors. Firstly, ReliefF algorithm is used to filter the state vectors according to the correlation between features and classes to speed up the efficiency of the algorithm. Then using Delta Test as performance index, genetic algorithm is used to optimize the weight of state component. Finally, the state vector selected by this method is compared with the state vector of time series and the simple Spatial-temporal correlation vector. Numerical results show that the proposed method outperforms the other two usually used methods under both general and abrupt traffic conditions.
引文
[1] CLARK S. Traffic prediction using multivariate nonparametric regression[J]. Journal of Transportation Engineering, 2003, 129(2):161-168.
[2] POLSON N G, SOKOLOV V O. Deep learning for shortterm traffic flow prediction[J]. Transportation Research Part C:Emerging Technologies, 2017(9):1-17.
[3] ZHENG Z, SU D. Short-term traffic volume forecasting:A k-nearest neighbor approach enhanced by constrained linearly sewing principle component algorithm[J]. Transportation Research Part C Emerging Technologies, 2014(43):143-157.
[4] WU Y, TAN H, QIN L, et al. A hybrid deep learning based traffic flow prediction method and its understanding[J]. Transportation Research Part C:Emerging Technologies, 2018(90):166-180.
[5] VLAHOGIANNI E I, KARLAFTIS M G, GOLIAS J C.Spatio-temporal short-term urban traffic volume forecasting using genetically optimized modular networks[J]. Computer-Aided Civil and Infrastructure Engineering, 2010, 22(5):317-325.
[6] ZHU J Z, CAO J X, ZHU Y, et al. Traffic volume forecasting based on radial basis function neural network with the consideration of traffic flows at the adjacent intersections[J]. Transportation Research Part C, 2014, 47(2):139-154.
[7] YANG S. On feature selection for traffic congestion prediction[J]. Transportation Research Part C Emerging Technologies, 2013, 26(1):160-169.
[8] SUN S, ZHANG C. The selective random subspace predictor for traffic flow forecasting[J]. IEEE Transactions on Intelligent Transportation Systems,2007, 8(2):367-373.
[9] ALJAWARNEH S, ALDWAIRI M, YASSEIN M B.Anomaly-based intrusion detection system through feature selection analysis and building hybrid efficient model[J]. Journal of Computational Science, 2018(25):152-160.
[10] KIRA K, RENDELL L A. The feature selection problem:Traditional methods and a new algorithm[C]. Tenth National Conference on Artificial Intelligence. AAAI Press, 1992:129-134.
[11] KONONENKO I. Estimating attributes:Analysis and extensions of RELIEF[C]. European Conference on Machine Learning. Springer, Berlin, Heidelberg, 1994:171-182.
[12] PI H, PETERSON C. Finding the embedding dimension and variable dependencies in time series[J]. Neural Computation, 1993, 6(3):509-520.
[13] EIROLA E, LIITI?INEN E, LENDASSE A, et al. Using the delta test for variable selection[C]. ESANN, 2008:25-30.
[14] KHOSRAVI A, MAZLOUMI E, NAHAVANDI S, et al.A genetic algorithm-based method for improving quality of travel time prediction intervals[J]. Transportation Research Part C, 2011, 19(6):1364-1376.
[2] POLSON N G, SOKOLOV V O. Deep learning for shortterm traffic flow prediction[J]. Transportation Research Part C:Emerging Technologies, 2017(9):1-17.
[3] ZHENG Z, SU D. Short-term traffic volume forecasting:A k-nearest neighbor approach enhanced by constrained linearly sewing principle component algorithm[J]. Transportation Research Part C Emerging Technologies, 2014(43):143-157.
[4] WU Y, TAN H, QIN L, et al. A hybrid deep learning based traffic flow prediction method and its understanding[J]. Transportation Research Part C:Emerging Technologies, 2018(90):166-180.
[5] VLAHOGIANNI E I, KARLAFTIS M G, GOLIAS J C.Spatio-temporal short-term urban traffic volume forecasting using genetically optimized modular networks[J]. Computer-Aided Civil and Infrastructure Engineering, 2010, 22(5):317-325.
[6] ZHU J Z, CAO J X, ZHU Y, et al. Traffic volume forecasting based on radial basis function neural network with the consideration of traffic flows at the adjacent intersections[J]. Transportation Research Part C, 2014, 47(2):139-154.
[7] YANG S. On feature selection for traffic congestion prediction[J]. Transportation Research Part C Emerging Technologies, 2013, 26(1):160-169.
[8] SUN S, ZHANG C. The selective random subspace predictor for traffic flow forecasting[J]. IEEE Transactions on Intelligent Transportation Systems,2007, 8(2):367-373.
[9] ALJAWARNEH S, ALDWAIRI M, YASSEIN M B.Anomaly-based intrusion detection system through feature selection analysis and building hybrid efficient model[J]. Journal of Computational Science, 2018(25):152-160.
[10] KIRA K, RENDELL L A. The feature selection problem:Traditional methods and a new algorithm[C]. Tenth National Conference on Artificial Intelligence. AAAI Press, 1992:129-134.
[11] KONONENKO I. Estimating attributes:Analysis and extensions of RELIEF[C]. European Conference on Machine Learning. Springer, Berlin, Heidelberg, 1994:171-182.
[12] PI H, PETERSON C. Finding the embedding dimension and variable dependencies in time series[J]. Neural Computation, 1993, 6(3):509-520.
[13] EIROLA E, LIITI?INEN E, LENDASSE A, et al. Using the delta test for variable selection[C]. ESANN, 2008:25-30.
[14] KHOSRAVI A, MAZLOUMI E, NAHAVANDI S, et al.A genetic algorithm-based method for improving quality of travel time prediction intervals[J]. Transportation Research Part C, 2011, 19(6):1364-1376.