公交运行时间预测及仿真

Prediction and Simulation of Bus Operating Time

研究公共车辆运行时间准确预测问题,公交运行时间受多种因素的影响,具有非线性和不确定性的变化规律,很难定量确定各因素对预测的影响程度。提出采用神经网络和证据理论融合的预测方法,利用神经网络的非线性特征和证据理论的不确定性融合特点,可以解决上述问题。首先利用RBF神经网络的预测控制方法对公交运行时间进行预测,然后应用D-S证据理论对不断预测的误差分析结果进行实时融合和修正。以哈尔滨市某一公交线路的站点运行时间预测为例进行仿真分析,结果表明,采用神经网络和证据理论的融合方法较单一方法比较,能够显著提高预测精度。

The prediction of bus operating time was studied in this paper,as various factors can interfere the time during bus operation,and have characters such as nonlinear and uncertainty,it is very difficult to express it by establishing models.Therefore,a fusing Neural Network and a D-S evidence method have been put forward in this paper, which has the characters such as nonlinearity and uncertainty.At first,we usied RBF neural network for prediction, then fused the error curve based on D-S evidence theory timely.Take the bus operation time in Harbin for example to simulate it,the result shows that this method can promote the perdition accuracy.