汽车工程 ›› 2021, Vol. 43 ›› Issue (10): 1442-1447.doi: 10.19562/j.chinasae.qcgc.2021.10.004

• • 上一篇    下一篇

智能车辆多信号灯路口快速节能车速规划

冷江昊,孙超(),卢兵   

  1. 北京理工大学,电动车辆国家工程实验室,北京 100081
  • 收稿日期:2021-05-11 修回日期:2021-07-07 出版日期:2021-10-25 发布日期:2021-10-25
  • 通讯作者: 孙超 E-mail:chaosun@bit.edu.cn
  • 基金资助:
    广东省重点领域研发计划项目(2019B090909001);国家自然科学基金(U1964206)

Fast Energy⁃saving Speed Planning Through Multi Signal Intersections of Intelligent Vehicles

Jianghao Leng,Chao Sun(),Bing Lu   

  1. Beijing Institute of Technology,National Engineering Laboratory for Electric Vehicles,Beijing 100081
  • Received:2021-05-11 Revised:2021-07-07 Online:2021-10-25 Published:2021-10-25
  • Contact: Chao Sun E-mail:chaosun@bit.edu.cn

摘要:

现有面向多信号灯路口复杂场景的智能车辆节能车速规划方法,往往计算量大,难以实时应用。本文中提出了一种结合动态规划算法和内点法进行联合优化的智能车辆节能车速规划算法,能够基于车辆能耗模型和信号灯信息,快速计算道路节能参考车速,实现实时应用。本文中首先建立了车辆能耗模型和信号灯相位时刻数学模型,在每个信号灯路口进行时间采样,构建基于巡航车速单位距离能耗的有向无环图,并使用动态规划算法求解最优绿灯通行周期。将最优绿灯通行周期作为约束条件,建立最优控制问题并转化为非线性优化问题,使用内点法进行优化求解。动态规划求解有向无环图所得解为内点法提供近似最优的初始解,提升收敛速度。最后构建了仿真场景进行分析验证,结果表明:(1)本文提出的方法较改进的智能驾驶员模型能够有效降低能耗30%以上;(2) 与基于规则的迭代初始值生成方法相比,动态规划提供的近似最优的初始解提高计算速度40%以上。

关键词: 多信号灯路口, 车速规划, 动态规划, 内点法

Abstract:

The existing energy?saving speed planning method of intelligent vehicles for complex scenes at multi signal intersections is usually with big computational burden that cannot be implemented in real-time. This paper proposes an optimized intelligent vehicle energy?saving speed planning method combining dynamic programming and the interior point method, which can rapidly calculate the road energy?saving reference speed based on the vehicle fuel consumption model and the traffic light information, and realize real?time application. Firstly, in this paper, the vehicle fuel consumption model and the signal phase and time mathematical model are established. A weighted orientation graph based on energy consumption per unit distance at cruise speed is constructed by sampling the arrival time at each intersection. The weighted orientation graph is solved via dynamic programming to obtain the optimal period of green light passing phase. Then, taking the optimal green light passing phase as the constraint, the optimal control problem is formulated and transformed into a nonlinear optimization problem, which is solved via the interior point method. To accelerate convergence speed, the initial guess of the interior point method is set to the solution of the weighted orientation graph solved by dynamic programming. Finally, simulation is carried out for evaluation. The results show that: (1) the proposed method can reduce energy consumption by over 30% compared with the improved intelligent driver model; (2) the initial guess close to optimal solution calculated by dynamic programming improves computational efficiency by more than 40%, compared with the rule?based iterative initial guess generation method.

Key words: multi signal intersections, speed planning, dynamic programming, interior point method