摘要
高速公路短时交通流量预测对高速公路运行管理和高速公路运行效率提升具有重要意义。高速公路收费大数据具有海量性、实时性、复杂性和动态性,而传统的时间序列模型存在建模复杂、适应性差的不足。为了提高交通流量预测的精度,利用高速收费大数据,从模型识别和参数调整两方面进行了优化,提出了一种改进的ARIMA(Autoregressive Integrated Moving Average Model,自回归滑动平均模型)预测模型。通过基于真实数据的实验表明,改进后的时间序列模型有效克服了传统时间序列模型的不足,并对不同的交通状况具有较好的适应性,无论在节假日还是工作日均具有更高的预测精度。
The short-term traffic flow forecast of highway is of great significance to the operation and management of highway and the improvement of highway operation efficiency. The large data of expressway charges are massive,real-time,complex and dy. namic,while the traditional time series model has the advantages of complex modeling and poor adaptability. In order to improve the accuracy of traffic flow forecasting and use the big toll data,the improved ARIMA(Autoregressive Integrated Moving Average Mod. el)is proposed to optimize the model from the aspects of model identification and parameter adjustment. Experiments based on real data show that the improved time series model effectively overcomes the shortcomings of the traditional time series model and has good adaptability to different traffic conditions,and has higher prediction accuracy both on holiday and working day.
作者
刘艳丽
赵卓峰
丁维龙
徐扬
LIU Yanli;ZHAO Zhuofeng;DING Weilong;XU Yang(College of Computer Science,North China University of Technology,Beijing 100043;Beijing Key Laboratory on Integration and Analysis of Large-scale Stream Data,Beijing 100043;Beijing E-hualu Information Technology Co.,Ltd.,Beijing 100043)
出处
《计算机与数字工程》
2019年第5期1164-1169,1188,共7页
Computer & Digital Engineering
基金
北京市自然科学基金(编号:4162021)
国家自然科学基金(编号:61702014)资助