期刊文献+

基于RAdam优化的DLSTM-AE交通流预测模型 被引量:1

DLSTM-AE Traffic Flow Prediction Model Based on RAdam Optimization
原文传递
导出
摘要 交通流预测在智能交通系统中起着关键的作用。然而,由于交通数据有着复杂的时间依赖性和本身的不确定性,导致预测交通流变得具有挑战性。为了进一步充分提取交通流时空性、周期性等特征,采用了一种自编码器(AE)与深度长短时记忆网络(DLSTM)相结合的组合模型(DLSTM-AE),并引入改进的适应矩估计算法(RAdam)进行模型训练。首先利用深度长短时记忆网络模型对交通流序列信息特征进行采集,并借助自动编码器结构将采集的信息压缩为一个固定维度的表示向量。然后通过解码器对该向量进行重构,实现信息的进一步挖掘。最后在模型的训练过程中,利用RAdam算法进行优化,分批次更新动量参数,缩短寻找最优解的时间,提高模型预测的时效性和精度。在高速公路交通流真实数据集上进行了仿真并与其他模型方法进行了对比。结果表明:DLSTM-AE组合模型不仅在预测结果上具有明显的优势,而且在交通流周期性方面拥有较好的曲线拟合能力,其中测试集的均方根误差值下降了约0.445~1.826,平均绝对误差值下降了约0.282~0.984,相关系数值R^(2)提高了约0.005~0.023;在周期性上,相邻周对应工作日的预测精度远高于对照组。该模型可以捕捉交通流序列中更潜在的时空性和周期性信息,可以更好地满足高速公路交通流预测的需要。 Traffic flow prediction plays a key role in intelligent transport systems. However, due to the complex time dependence and inherent uncertainty of traffic data, prediction of traffic flows becomes challenging. In order to further fully extract the spatio-temporal and periodic characteristics of traffic flow, a combination model(DLSTM-AE) combining autoencoder(AE) with deep long short-term memory(DLSTM) network is adopted, and an improved adaptive moment estimation algorithm(RAdam) is introduced for model training. First, the characteristics of traffic flow sequence information are collected by using the deep long short-term memory network model, and the collected information is compressed into a fixed dimension representation vector with help of the automatic encoder structure. Then, the vector is reconstructed by using decoder to realize further information mining. Finally, in the process of model training, the optimization is conducted by using RAdam algorithm to update the momentum parameters in batches, shorten the time to find the optimal solution, and improve the timeliness and accuracy of model prediction. The simulation on the real traffic flow data set of expressway is carried out and the result is compared with those of other model methods. The result shows that(1) DLSTM-AE combination model not only has obvious advantages in prediction result, but also has good curve fitting ability in traffic flow periodicity. The RMSE of the test set decreased by about 0.445-1.826, the MAE decreased by about 0.282-0.984, and the correlation coefficient value R^(2) increased by about 0.005-0.023.(2) In terms of periodicity, the prediction accuracy of working days corresponding to adjacent weeks is much higher than that of the control group. The model can capture more potential spatio-temporal and periodic information in traffic flow sequence. It can better meet the needs of expressway traffic flow prediction.
作者 黄艳国 周陈聪 左可飞 HUANG Yan-guo;ZHOU Chen-cong;ZUO Ke-fei(School of Electrical Engineering and Automation,Jiangxi University of Science and Technology,Ganzhou Jiangxi 341000,China)
出处 《公路交通科技》 CAS CSCD 北大核心 2023年第1期185-191,199,共8页 Journal of Highway and Transportation Research and Development
基金 国家自然科学基金项目(72061016)。
关键词 智能交通 交通流预测 深度神经网络 高速公路 RAdam算法 ITS traffic flow prediction deep neural network expressway RAdam algorithm
  • 相关文献

参考文献7

二级参考文献25

  • 1徐启华,杨瑞.支持向量机在交通流量实时预测中的应用[J].公路交通科技,2005,22(12):131-134. 被引量:21
  • 2Makr Dougherty.A Review of Neural Network Applied to Transport[J].Transportation Research,1996:151-166.
  • 3Brian L Smith,Michael J Demetsky.Traffic Flow Forecasting:Comparison of Modelling Approaches[J].Journal of Transportation Engineering,1997:261-266.
  • 4Mohammad M Hamed,Hashem R Al-masaeid,Zahi M Bani Said.Short-term Prediction of Traffic Volume in Urban Arterials[J].Journal of Transportation Engineering,1995:249-254.
  • 5Mascha Der Voort,Mark Doughert,Susan Watson.Combining Kohonen Maps with ARIMA Time Series Models to Forecast Traffic Flow[J].Transportation Research,1996:307-318.
  • 6王炜,过秀成.交通工程学.南京:东南大学出版社,2001
  • 7Vapnik V N. The nature of statistical learning theory. New York:Springer-Verlag, 1999
  • 8Shao Xiaojian,He Guoping. Lagrange support vector regression. IIMST,2005,3(1) :434-440
  • 9Boser B E, Guyon I M, Vapnik V N. A training algorithm for optimal margin classifiers. In D. Haussler, editor, 5th Annual ACM Workshop on COLT, Pittsburgh, PA,ACM Press, 1992 : 144-152
  • 10刘艳忠,邵小健,李旭宏.基于Lagrange支持向量回归机的短时交通流量预测模型的研究[J].交通与计算机,2007,25(5):46-49. 被引量:13

共引文献376

同被引文献6

引证文献1

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部