站点驻留时间估计和预测方法研究

针对公交车站点驻留时间存在较大变异性的特点,为了准确地预测公交车在站点的驻留时间,提高公交信号优先控制的效果,首先尝试采用概率模型来对公交车站点驻留时间进行估计和预测,但通过对多个站点的大量数据相关系数的计算和分析发现,下车乘客数和车上的乘客人数之间并不存在一定的线性关系,不完全满足概率模型的假设条件,以此为基础,采用多重线性回归并排除不显著因素,最后采用加权最小二乘回归模型来产生精确的参数估计。文中利用大量的实际调查数据,采用各种统计分析方法验证提出的站点驻留时间估计和预测模型的正确性。

Research on Bus Dwell Time Estimation and Prediction

Due to the large variability in dwell time,an accurate prediction of dwell time is difficult to achieve and there can be large differences between the forecast and actual arrival time,which affects the effect of transit signal priority control.First,bus dwell time was estimated and predicted by trying touse probabilistic model.But it was found that there is no linear relationship between the number of alighting passengers and on-board passengers through the analysis of correlation coefficient.It does not satisfy with the assumption of probabilistic models.Then the mutli-linear regression model was adopted and insignificant determinants were removed.Finally accurate model parameters estimation was produced by using weighted least square regression.The effectiveness of estimation models was verified by statistical analysis methods together with some survey data.