基于混合粒子滤波的高速公路交通参数自适应估计方法

A Method of Exporessway Traffic Parameter Self-adaptive Estimation Based on Hybrid Particle Filtering

为提高高速公路交通参数的估计准确度,在宏观交通流模型和状态空间模型的基础上,基于贝叶斯理论,提出了一种基于混合粒子滤波的交通参数估计方法。考虑到估计结果对模型参数变化的敏感性,避免采用预设固定模型参数对估计准确度的影响,通过建立自由流速度与饱和度之间的变化关系,提出了交通状态影响下的模型参数自适应调整策略。仿真结果表明:基于混合粒子滤波的交通参数估计准确度要明显高于卡尔曼滤波估计,在正常和事故场景下,能够快速识别交通量和速度较明显的波动,表现出了更强的稳定性;交通状态影响下的模型参数自适应调整策略会明显提高交通参数估计准确度,在发生事故情况下,也可达到较好的估计效果。

In order to improve the accuracy of exporessway traffic parameter estimation, a traffic parameter estimation method based on HPF is proposed based on macroscopic traffic flow model and state-space model according to Bayesian theory. Considering the sensitivity of the estimation result to the model parameters and avoiding the effect of using preset fixed model parameters on the estimation accuracy, a self-adaptive adjustment strategy for model parameters under the influence of traffic state is proposed through establishing the relationship between free-flow speed and saturation. The simulation result shows that （1） the traffic parameter estimation accuracy based on HPF is obviously higher than that of Ka^man filtering, it is able to quickly identify the obvious fluctuations of traffic volume and speed under normal and accident traffic conditions, which reflects a higher stability; （2） the self-adaptive adjustment strategy of model parameter under the influence of traffic conditions can obviously improve the estimation accuracy and achieve better estimation effect even in case of accident.