华南理工大学学报(自然科学版) ›› 2022, Vol. 50 ›› Issue (9): 138-148.doi: 10.12141/j.issn.1000-565X.210771

所属专题: 2022年机械工程

• 机械工程 • 上一篇    

基于改进动态规划法的HEV多目标能量管理策略

赵克刚1 何坤阳1 黎杰2 梁志豪1 贝泾浩1 王玉龙3   

  1. 1.华南理工大学 机械与汽车工程学院,广东 广州 510640
    2.广州华工机动车检测技术有限公司,广东 广州 510640
    3.广州汽车集团股份有限公司 汽车工程研究院,广东 广州 511434
  • 收稿日期:2021-12-06 出版日期:2022-09-25 发布日期:2022-02-11
  • 通信作者: 黎杰(1964-),男,教授级高级工程师,主要从事汽车检测与控制方向的研究。 E-mail:106lj@163.cn
  • 作者简介:赵克刚(1977-),男,博士,副教授,主要从事最优控制、节能汽车和智能网联汽车研究。E-mail:kgzhao@scut.edu.cn
  • 基金资助:
    广东省自然科学基金资助项目(2020A1515010773);广东省重点领域研发计划项目(2019B090912001)

Multi-objective Energy Management Strategy of HEV Based on Improved Dynamic Programming Method

ZHAO Kegang1 HE Kunyang1 LI Jie2 LIANG Zhihao1 BEI Jinghao1 WANG Yulong3   

  1. 1.School of Mechanical & Automotive Engineering, South China University of Technology, Guangzhou 510640, Guangdong, China
    2.Guangzhou Huagong Automobile Inspection Technology Co. Ltd. , Guangzhou 510640, Guangdong, China
    3.Automotive Engineering Research Institute, Guangzhou Automobile Group Co. Ltd. , Guangzhou 511434, Guangdong, China
  • Received:2021-12-06 Online:2022-09-25 Published:2022-02-11
  • Contact: 黎杰(1964-),男,教授级高级工程师,主要从事汽车检测与控制方向的研究。 E-mail:106lj@163.cn
  • About author:赵克刚(1977-),男,博士,副教授,主要从事最优控制、节能汽车和智能网联汽车研究。E-mail:kgzhao@scut.edu.cn
  • Supported by:
    the Natural Science Foundation of Guangdong Province(2020A1515010773);the Key R&D Project of Guangdong Province(2019B090912001)

摘要:

混合动力汽车能量管理策略(EMS)优化问题是一类需要综合优化混合动力汽车多个性能指标的多目标多阶段决策问题,而传统的多目标优化算法在求解EMS这类问题时面临求解效率低、收敛性难以保证等挑战。本文结合非支配排序算法的思想,将传统的动态规划法(DP)拓展到多目标优化领域,提出了非支配排序动态规划法(NSDP)。该算法首先将行驶工况划分为多个阶段,在每个阶段中求取混合动力汽车在不同控制策略产生的累积目标值向量,并通过非支配排序算法获得当前的非支配解集以及对应的控制策略,然后利用各个阶段的非支配解集依次逆向迭代,直至获取整个行驶工况的非支配解集前沿以及对应的能量管理控制策略。在仿真实验中,分别应用加权动态规划法(WDP)和非支配排序动态规划法求解功率分流式混合动力汽车和串并联式混合动力汽车在匀加速工况的多目标能量管理策略优化问题,结果表明NSDP能够有效完成求解并保证收敛性,且求解结果在解集均匀性和求解效率方面具有显著的优势。进一步,运用NSDP求解在世界轻型车辆测试工况(WLTC)下串并联式混合动力汽车能量管理优化问题,所得非支配解集可用于分析汽车的工作特性,并能够为实际能量管理策略的制定提供可靠的参考。

关键词: 多目标优化, 能量管理策略, 改进动态规划法, 非支配排序

Abstract:

Hybrid electric vehicle Energy Management Strategy (EMS) optimization is a multi-objective and multi-stage decision-making problem that needs to comprehensively optimize several performance indicators of hybrid electric vehicles. The traditional multi-objective optimization algorithm faces challenges such as low efficiency and difficult to guarantee convergence when dealing with these problems. Combined with the idea of non-dominated sorting algorithm, this paper extended the traditional Dynamic Programming (DP) to the field of multi-objective optimization, and proposed Non-dominated Sorting Dynamic Programming (NSDP). When using this algorithm, the driving condition was divided into several stages firstly. In each stage, the cumulative target value vector generated by the hybrid electric vehicle in different control strategies was obtained, and the current non dominated solution set and the corresponding control strategy were obtained through the non dominated sorting algorithm. Then, the non dominated solution set of each stage was used for reverse iteration in turn, until the leading edge of the non dominated solution set and the corresponding energy management control strategy of the whole driving cycle were obtained. In the simulation experiment, Weighting Dynamic Programming (WDP) and Non-dominated Sorting Dynamic Programming were applied to solve the optimization problem of multi-objective energy management strategy for power split hybrid electric vehicles and series parallel hybrid electric vehicles under constant acceleration conditions. The results show that NSDP not only can effectively complete the solution and ensure convergence, but also has significant advantages in homogeneity of solution set and solving efficiency. Furthermore, NSDP was used to solve the energy management optimization problem of series parallel hybrid electric vehicles running in Worldwide Harmonized Light Duty Vehicle Test Cycle (WLTC). The non dominated solution set can be used to analyze the working characteristics of vehicles and provides a reliable reference for the formulation of actual energy management strategy.

Key words: multi-objective optimization, energy management strategy, improved dynamic programming, non-dominated sorting

中图分类号: