基于模糊神经网络的短时交通流预测方法研究

为满足交通控制和诱导系统的实时性需求,减少交通拥挤状况,降低交通事故突发频率,需要对短时交通流进行预测;当前的短时交通流预测方法是采用K-近邻的非参数回归对其进行预测,预测过程中没有将预测模型中关键因素对交通流的影响进行详细的说明,导致预测结果不准确,存在短时交通流预测误差较大的问题;为此,提出一种基于模糊神经网络的短时交通流预测方法;该方法首先以历史短时交通流数据样本序列为基础,将提取的关联维数作为短时交通流的混沌特征量,然后以该特征量为依据,对短时交通流数据进行聚类,使相同的短时交通流聚合类样本比不同的交通流聚合类样本更为贴近,采用高斯过程回归对短时交通流预测模型进行建设,建设过程中利用差分方法对短时交通流预测序列进行平稳化操作之后,对短时交通流预测模型进行训练,将GPR模型引入至短时交通流预测过程中,得到交通流预测方差估计值,并确定交通流预测值置信区间,由此实现短时交通流的预测;由此实现短时交通流的预测;实验结果证明,所提方法可以准确地预测交通运输系统的实时状况,为车辆行驶的最佳路线进行了有效引导,减少了自然影响方面和人为因素对短时交通流预测结果的干扰,为交通部门对交通路况的控制管理提供了依据。

Short-term Traffic Flow Prediction Method Based on Fuzzy Neural Network Research

In order to satisfy the real time demand of traffic control and guidance system, reduce the occurrence of traffic congestion, reduce the frequency of traffic accident emergency, need to forecast the short-term traffic flow. Current short-term traffic flow prediction method is using K - nearest nonparametric regression to forecast and predict the process of no will be key factors in the prediction model of traffic flow in detail, the influence of lead to inaccurate prediction results, the problems of short-term traffic flow prediction error is bigger. For this, put forward a kind of short-term traffic flow prediction method based on fuzzy neural network. This method firstly on the basis of the history of short-term traffic flow data sample series, the extracted correlation dimension as a short-term traffic flow of the chaos characteristics, and then based on the characteristics, the clustering of the short-term traffic flow data and make the same short-term traffic flow aggregation class samples than the aggregation of different traffic flow class samples more press close to, by using the Gaussian process regression of short-term traffic flow forecasting model, using the finite difference method in the process of construction of short-term traffic flow forecasting sequences with smooth operation, after training for short-term traffic flow prediction model, introducing the Gaussian model to short-term traffic flow prediction in the process, get the traffic flow forecasting variance, and traffic flow prediction confidence interval were determined, thus realizing short-term traffic flow prediction. The realization of short-term traffic flow prediction. The experimental results show that the proposed method can accurately predict the transportation system of the real-time condition, the best way for vehicle is the effective guidance, reduces the impact on natural and human factors interference, the result of the short-term traffic flow prediction for the traffic department to provide a basis for the control of road traffic management.