汽车工程 ›› 2023, Vol. 45 ›› Issue (7): 1200-1211.doi: 10.19562/j.chinasae.qcgc.2023.07.011

所属专题: 底盘&动力学&整车性能专题2023年

• 精选论文 • 上一篇    下一篇

基于交互式多模型卡尔曼滤波的主动悬架控制

吴骁,史文库,陈志勇()   

  1. 吉林大学,汽车仿真与控制国家重点实验室,长春 130022
  • 收稿日期:2023-01-03 修回日期:2023-02-16 出版日期:2023-07-25 发布日期:2023-07-25
  • 通讯作者: 陈志勇 E-mail:chen_zy@jlu.edu.cn
  • 基金资助:
    国家重点研发计划(2018YFB0106203)

Active Suspension Control Based on Interacting Multiple Model Kalman Filter

Xiao Wu,Wenku Shi,Zhiyong Chen()   

  1. Jilin University,State Key Laboratory of Automotive Simulation and Control,Changchun 130022
  • Received:2023-01-03 Revised:2023-02-16 Online:2023-07-25 Published:2023-07-25
  • Contact: Zhiyong Chen E-mail:chen_zy@jlu.edu.cn

摘要:

针对固定状态观测器难以保证路面自适应悬架状态观测精度的问题,本文中在交互式多模型卡尔曼滤波(IMMKF)的基础上,建立了悬架状态观测器与控制器。首先基于LQG算法与模糊控制算法建立了路面自适应主动悬架系统。结合谐波叠加法,生成A-B-D-C级空间域路面不平度模型,作为仿真系统的输入。其次以各级路面的最优LQG模型为子模型建立了3种IMMKF悬架状态观测器与控制器。仿真对比表明:14模型的IMMKF悬架状态观测器相对于普通卡尔曼滤波观测器的观测精度最大可提升98.17%,并可用于识别路面等级,并且基于14模型IMMKF的自适应主动悬架控制器的车身加速度相对于被动悬架降低了75.99%、相对于普通LQG主动悬架降低了47.16%,验证了模型的优越性。

关键词: 交互式多模型卡尔曼滤波, 模糊控制, 状态观测, 路面等级识别

Abstract:

For the problem that it is difficult for fixed state observer to ensure the accuracy of road adaptive suspension state observation, the suspension state observer and controller is established on the basis of interactive multi-model Kalman filter (IMMKF). Firstly, the road adaptive active suspension system is established based on the LQG algorithm and fuzzy control algorithm. Combined with harmonic superposition method, the A-B-D-C grade spatial domain road roughness model is generated as the input of the simulation system. Secondly, three kinds of IMMKF suspension state observer and controller are established taking the optimal LQG model of all grades of road as the sub-models. The simulation comparison shows that the observation accuracy of the 14-model IMMKF suspension state observer can be improved by 98.17% maximumly compared with the ordinary Kalman filter, and can be used to identify road grade, and the body acceleration of the adaptive active suspension controller based on the 14-model IMMKF is reduced by 75.99% compared with the passive suspension and 47.16% compared with the ordinary LQG active suspension, which verifies the superiority of the model.

Key words: interacting multiple model Kalman filter, fuzzy control, state observation, road grade recognition