基于四分位数对路段速度的预测优化

Optimization of road-segment speed prediction based on tetra-quantile

使用出租车GPS数据作为基础,采用更加合理有效的路段速度作为交通状态参数,分析路网划分后的路段速度时间序列,利用四分位数特性优化算法,提高预测模型的合理性和准确性,并通过真实历史数据验证方法的可靠性.从带有随机性和不确定性的交通流变化中,通过分析找出其中的规律性,以预测未来几个时段的交通流变化.结果表明四分位法既体现出了路段速度的变化趋势,同时削弱了极端值和异常值的影响,能够展现出合理的交通状态变化过程,并且其计算简便,为大规模数据处理有效节省了计算资源.对计算结果的曲线拟合证明了四分位法处理路段速度的可靠性,对交通状态预测具有重要意义.

Taking taxi GPS data as basis and logical and valid road-segment speed as traffic state parameter,the time series of road-segment speed on meshed road net is analyzed and the tetra-quantile characteristic optimization algorithm is used to improve the rationality and accuracy of the prediction model and verify the reliability of this method with real historical data.By means of analysis,the regularity of traffic flow variation is found from its randomness and uncertainty to predict it in oncoming couple of time-intervals.The result shows that the tetra-quantile method will embody the trend of road-segment speed as well as weaken the influence of its extreme and abnormal values,showing a reasonable traffic state variation process.Besides,its calculation will be simple,saving computing resources effectively for large-scale data processing.The fitted road-segment speed curves will successfully verify that the road-segment processing with tetra-quantile method is reliable and significant for traffic state prediction.