基于多源数据融合的公交车辆跟驰模型

Bus-following model based on multi-source data fusion

针对目前缺少公交车辆跟驰模型参数和单一数据采集方式无法获取完整车辆跟驰过程数据的问题,提出一种基于多源数据融合的公交车辆跟驰模型,采用实测数据进行参数标定和模型验证。通过移动GPS数据采集设备随车采集车辆的运动轨迹和无人机空中悬停俯拍采集公交车辆的停车间距,并进行数据融合处理得到完整过程的公交车辆跟驰行驶数据;通过引入均方误差指标来衡量实际测量值与模型仿真值之间的差异,将公交车辆跟驰模型的参数标定问题转化为一般优化求解问题;通过对实测数据进行统计分析,获取公交车辆的平均停车间距参数;采用粒子群优化算法对智能驾驶跟驰模型中的舒适制动减速度和加速度系数2个参数进行求解,得到模型的最优参数;将验证集中的实测数据与模型仿真值进行对比,对模型参数的效果进行验证,并以厦门快速公交走廊的车辆多编组运行控制为例,对跟驰模型的效果进行进一步的验证,其中头车采用速度优化模型进行控制,跟随车辆则采用上述标定的公交车辆跟驰模型进行控制。研究结果表明:公交车辆跟驰模型生成的数据曲线与实测数据曲线基本一致,跟随车辆与头车在路段上行驶和车辆进出站2个阶段的运动轨迹曲线基本一致,符合车辆跟驰模型的跟随性,进一步证明了提出标定模型参数的有效性。

Aiming at the two current problems that lack that bus-following model parameters and the inability to obtain complete bus-following process data by a single data acquisition method,a bus-following model based on multi-source data fusion was proposed,and measured data was used for parameter calibration and model verification.The bus trajectory data was collected with the mobile GPS data acquisition equipment and the parking distance of the bus was collected with the aerial hovering video of unmanned aerial vehicle.The complete process of bus-following driving data was obtained by data fusion processing.The mean square error was used to measure the difference between the actual measured value and the simulated value of the model,the parameter calibration problem of bus-following model was transformed into a general optimization problem.Through statistical analysis of the measured data,the average parking spacing parameters of buses were obtained.Particle swarm optimization algorithm was used to solve the comfortable braking deceleration and acceleration coefficient of the intelligent driver model,and the optimal parameters of the model were obtained.The measured value in the verification set was compared with the simulated value of the model to verify the effect of the model parameters.Finally,the effect of bus-following model was further verified by taking the multi-platoon operation control of bus rapid transit corridor in Xiamen as an example.The leading bus was controlled by the speed optimization model and the following buses were controlled by the bus-following model calibrated above.The results show that the simulation data curve generated by the bus-following model was basically consistent with the measured data curve.The motion trajectory curves of the following bus and the leading bus in the two stages of driving on the road section,and the bus in and out of the stop were basically consistent.It accords with the following features of bus-following model,which further proves the validity of the above-mentioned calibration model parameters.5 tabs,11 figs,29 refs.