复杂环境四轮独立驱动电动汽车多目标层级预测控制

Many automobile makers will stop producing and selling conventional fossil fuel powered vehicles from 2019 on, which means electric vehicles will be used more widely. However, due to the inherent issues of electric vehicles such as insufficient power and short mileage per charge, safety is not guaranteed in complex driving scenarios. In this project, four-wheel independently actuated electric vehicles that are relatively more advantageous in dealing with complex driving scenarios are considered. Based on model predictive control, a hierarchical predictive controller will be designed taking into account multi-source environmental factors such as road slopes, curvature, speed limits, varying tire road friction coefficients, dynamic obstacles, and battery output power constraints etc. Specifically, a new vehicle dynamics model integrating environmental factors will be designed at the upper level. The model will then be used as the prediction model to solve the vehicle coordinated longitudinal and lateral motion control problem which will provide the optimal reference longitudinal traction forces and steering angle. At the lower level, a powertrain energy consumption model based on the motor characteristics and the slip loss characteristics of the tire will first be designed. Then, the optimal distribution strategy of torques and steering angles will be proposed for the distributed motor-drive system of four-wheel independently actuated electric vehicles. The hierarchical controller will further enhance the driving performance with respect to safety, power satisfaction, stability, energy consumption and comfort of four-wheel independently actuated electric vehicles in complex driving scenarios. The four-wheel independently actuated electric vehicle is a typical complex over-actuated system itself. The involvement of multi-source environmental factors and multiple possible conflicting performance indices makes the control problem both practically significant and scientifically challenging.

从2019年起各大汽车厂商将纷纷停产或停售传统燃油车，电动汽车将迎来更广泛的市场化应用。然而，电动汽车固有的动力不足和巡航里程较短等问题可能会给复杂环境中行驶带来一定安全隐患。本项目研究复杂环境中具有较好行驶性能的四轮独立驱动电动汽车，基于预测控制理论，设计考虑如道路坡度、曲率、限速、路面附着系数、动态障碍物、电池功率约束等多源复杂环境因素的层级预测控制器。上层建立考虑环境因素的车辆动力学模型，解决车辆横纵向运动控制问题并得到最优纵向力和转角；下层建立基于电机能效和轮胎滑移损耗特性的驱动系统能耗模型，设计电动汽车四轮独立驱动分布式电机系统转矩和转角的最优分配策略。层级控制器进一步加强四轮独立驱动电动汽车在复杂环境中安全、动力、稳定、节能和舒适等多目标行驶性能。由于四轮独立驱动电动汽车本身是一个复杂的过驱动控制系统，环境因素复杂多源性和系统多性能指标使得该控制问题兼具实际意义与科学挑战性。

目前世界各国已积极谋划彻底禁售燃油汽车，项目执行期间，由于疫情影响，全球汽车销量出现下降。然而，电动汽车汽车却迎来逆势增长，2022年电动汽车销量已占比10%，国内占比更是高达20%以上，汽车市场全面电动化已是大势所趋。然而，与传统燃油车相比，电动汽车所带来的能量管理和巡航里程较短等问题阻碍了电动汽车更广泛的市场化应用，也可能会给电动汽车在复杂环境中行驶带来一定安全隐患。本项目围绕车辆在智能运动-能量消耗-安全行驶一体化多目标预测控制，主要开展了如下三方面研究内容：（1）研究了电动汽车运动-动力一体化建模与预测控制方法，充分考虑电动汽车电池系统和运动系统之间的耦合做出更高效、节能的决策，并在车辆运行典型工况下对车辆运动性能和能量消耗进行了详细分析；（2）考虑车辆非线性动力学，设计考虑车辆运动特性的平滑路径，研究了复杂路况下车辆轨迹规划、智能避障与预测控制的安全行驶运动控制策略，并进行了大量仿真对比实验；（3）研究了一种智能车辆多目标预测控制框架下的上层动态调度与派遣方法，充分考虑车辆行驶距离、行驶速度与能量消耗之间的关系，建立时效性、能耗性等的多目标优化控制问题和解决方案，所提技术可以直接应用于车辆速度轨迹规划与能耗优化。三方面研究内容形成了智能车辆调度派遣-轨迹规划-运动控制-能耗分析的“上下分层、层间互馈”的多目标层级预测控制完整方案。所提出的技术方案和研究结果对车辆电动化和智能化提供全面的支撑，也为复杂环境下智能运动体全系统一体化建模与控制理论积累了关键技术。