最优组合预测及其在短时交通流预测中的应用

The Optimal Combination Forecast Method and Its Application in the Short-term Traffic Flow Prediction

交通流量预测结果的好坏直接关系到交通控制与诱导的效果,因此,短时交通流预测是先进交通管理信息系统中关键技术问题之一。在简要介绍最优组合预测模型基础上,以预测误差平方和最小为准则确定最优组合预测模型中的权系数,实现交通流预测模型的最优组合,并利用实际数据通过与单一时间序列模型、线性回归模型、卡尔曼滤波模型及人工神经网络模型的预测精度比较分析,验证了交通流最优组合预测模型的有效性和实用性。

The accuracy of traffic flow prediction results directly affects the traffic control and management and therefore, the short-term traffic flow prediction is the key technology for the advanced traffic management information system. On the basis of briefly introducing the optimal combination forecast model, the weight coefficient of the optimal combination forecast model is determined by the criterion of the minimum square sum of the forecasting error. It realizes that the traffic flow prediction model is optimal combination. The precision of the optimal combina- tion forecast model is compared and analyzed with the other prediction models, such as time series model, linear regression model, Kalman filtering model and artificial neural network by using the real data. It proves that the opti- mal combination forecast model of traffic flow is effective and practical.