论文部分内容阅读
城市公路隧道内任意时段交通流量的变化是个非线性的复杂过程,受诸多随机的不确定因素的影响。传统的时间序列模型多使用BP网络,但BP网络是静态网络,它只是实现一一对应的静态非线性映射关系,不适合动态系统的实时辨识,难以实施精确预测。在对城市公路隧道动态交通流分析的基础上,提出了城市公路隧道交通流量预测的动态神经网络模型,该模型基于Elman网络,具有状态记忆的功能,用Elman网络建立的时间序列模型是一个自回归滑动平均模型。它的输出不仅取决于过去和现在的输入,而且也取决于过去的输出。使用该模型仿真预测武汉首义广场隧道的交通流量,试验结果表明,该方法能够更好的提高预测精度。
The change of traffic flow at any time in urban highway tunnel is a nonlinear and complicated process, which is affected by many random uncertainties. The traditional time series model mostly uses BP network. However, BP network is a static network. It only realizes one-to-one correspondence between static and non-linear mapping and is not suitable for real-time identification of dynamic system, so it is difficult to implement accurate prediction. Based on the analysis of dynamic traffic flow in urban highway tunnels, a dynamic neural network model of urban highway tunnel traffic flow prediction is proposed. The model is based on Elman network and has the function of state memory. The time series model established by Elman network is a self- Regression sliding average model. Its output depends not only on past and present inputs, but also on past outputs. The model is used to simulate the traffic flow of Wuhan Shouyi Square tunnel. The experimental results show that this method can improve the prediction accuracy better.