论文部分内容阅读
针对短时交通流变化周期性与随机性的特点,提出了新的混合预测模型,包含非参数回归模型与BP神经网络模型2种单项模型。非参数回归模型利用相关历史交通流数据,通过数据库匹配操作,确定预测结果,以充分体现交通流的周期稳定性。采用3层BP神经网络模型反映交通流的动态与非线性特点。采用模糊控制算法确定各单项模型的权重,并按不同权重有效组合成新的混合模型。采用西安市某路段30d的交通流量数据验证混合模型的预测效果。试验结果表明:该混合模型的平均相对误差为1.26%,最大相对误差为3.53%,其预测精度明显高于单项模型单独预测时的精度,能较准确地反映交通流真实情况。
Aiming at the periodicity and randomness of short-term traffic flow, a new hybrid forecasting model is proposed, which includes two models of non-parametric regression model and BP neural network model. The non-parametric regression model uses the relevant historical traffic flow data to determine the prediction results through the database matching operation in order to fully reflect the periodic stability of the traffic flow. A 3-layer BP neural network model is used to reflect the dynamic and nonlinear characteristics of traffic flow. The fuzzy control algorithm is used to determine the weight of each individual model and effectively combine them into new mixed models according to different weights. The traffic flow data of a section of Xi’an for 30 days is used to verify the prediction of the hybrid model. The experimental results show that the average relative error of the hybrid model is 1.26% and the maximum relative error is 3.53%. The prediction accuracy of the hybrid model is obviously higher than that of the single model alone, which can accurately reflect the real situation of traffic flow.