Purpose

The flexible mode transitions, multiple power sources and system uncertainty lead to challenges for mode transition control of four-wheel-drive hybrid powertrain. Therefore, the purpose of this paper is to improve dynamic performance and fuel economy in mode transition process for four-wheel-drive hybrid electric vehicles (HEVs), overcoming the influence of system uncertainty.

Design/methodology/approach

First, operation modes and transitions are analyzed and then dynamic models during mode transition process are established. Second, a robust mode transition controller based on radial basis function neural network (RBFNN) is proposed. RBFNN is designed as an uncertainty estimator to approximate lumped model uncertainty due to modeling error. Based on this estimator, a sliding mode controller (SMC) is proposed in clutch slipping phase to achieve clutch speed synchronization, despite disturbance of engine torque error, engine resistant torque and clutch torque. Finally, simulations are carried out on MATLAB/Cruise co-platform.

Findings

Compared with routine control and SMC, the proposed robust controller can achieve better performance in clutch slipping time, engine torque error, vehicle jerk and slipping work either in nominal system or perturbed system.

Originality/value

The mode transition control of four-wheel-drive HEVs is investigated, and a robust controller based on RBFNN estimation is proposed. Compared results show that the proposed controller can improve dynamic performance and fuel economy effectively in spite of the existence of uncertainty.

中文翻译：

---

基于径向基函数神经网络估计的四轮驱动混合动力汽车鲁棒模式转换控制——仿真研究

目的

灵活的模式转换、多电源和系统的不确定性给四轮驱动混合动力系统的模式转换控制带来了挑战。因此，本文旨在提高四轮驱动混合动力汽车（HEV）模式转换过程中的动态性能和燃油经济性，克服系统不确定性的影响。

设计/方法/方法

首先分析了运行模式和转换，然后建立了模式转换过程中的动态模型。其次，提出了一种基于径向基函数神经网络（RBFNN）的鲁棒模式转换控制器。RBFNN 被设计为一个不确定性估计器，用于近似由于建模错误导致的集总模型不确定性。基于该估计器，在离合器滑动阶段提出了一种滑模控制器（SMC），以实现离合器速度同步，尽管发动机扭矩误差、发动机阻力扭矩和离合器扭矩受到干扰。最后，在 MATLAB/Cruise 协同平台上进行仿真。

发现

与常规控制和 SMC 相比，所提出的鲁棒控制器在标称系统或扰动系统下都能在离合器打滑时间、发动机扭矩误差、车辆加加速度和打滑工作方面取得更好的性能。

原创性/价值

研究了四轮驱动HEV的模式转换控制，提出了一种基于RBFNN估计的鲁棒控制器。比较结果表明，尽管存在不确定性，但所提出的控制器可以有效地提高动态性能和燃油经济性。