论文部分内容阅读
为了在预测中体现短时交通流序列的变化趋势,提高预测精度,对短时交通流序列变化特征进行研究,将其细分为上升趋势、平稳波动趋势、下降趋势,并运用转折突变法求解变化趋势参数。将变化趋势参数作为一个输入变量,建立径向基(RBF)神经网络预测模型。选取实际交通流序列,计算其变化趋势参数,并对交通量进行预测。研究结果表明:预测结果的最大相对误差为5.72%,小于未引入变化趋势参数预测的最大相对误差;提出的变化趋势参数能体现交通流变化趋势,引入该参数能提高预测精度,该预测方法可行。
In order to reflect the changing trend of short-term traffic flow sequence and improve prediction accuracy in the forecasting, the characteristics of short-term traffic flow sequence change are studied and divided into upward trend, steady fluctuation trend and downward trend, and the method of turning catastrophe Change trend parameters. Taking the change trend parameter as an input variable, a radial basis (RBF) neural network prediction model was established. Select the actual traffic flow sequence, calculate the trend of its parameters, and predict the traffic volume. The results show that the maximum relative error of the prediction result is 5.72%, which is smaller than the maximum relative error predicted by the parameter without changing trend. The proposed trend parameter can reflect the trend of traffic flow, and the introduction of this parameter can improve the prediction accuracy. .