Integration of batteries and supercapacitors (B‐SCs) is widely used to improve performance of electric vehicles (EVs). In this article, we consider the energy optimization problem of B‐SCs in EVs and propose an efficient model predictive control (MPC) algorithm for real‐time energy optimization of the hybrid energy storage system of EVs. Back propagation neural network is firstly adopted to learn the velocity prediction ability over a finite horizon by standard driving cycles. Then real‐time energy optimization of B‐SCs in EVs is formulated as the finite horizon optimal control problem by taking into account the constraints, the cost function on battery current, and the predicted velocity of the EV. Moreover, to lessen the computational burden of online solving the problem, the Pontryagin's Minimum Principle is used in a fashion of receding horizon. Compared with traditional nonlinear MPC, simulation results verify the effectiveness of the proposed MPC algorithm for real‐time energy optimization of B‐SCs in EVs.

中文翻译：

---

电动汽车电池超级电容器实时能量优化的高效模型预测控制

电池和超级电容器（B‐SC）的集成被广泛用于提高电动汽车（EV）的性能。在本文中，我们考虑了电动汽车中B-SC的能量优化问题，并提出了一种用于电动汽车混合储能系统实时能量优化的有效模型预测控制（MPC）算法。首先采用反向传播神经网络通过标准行驶周期学习有限范围内的速度预测能力。然后，通过考虑约束条件，电池电流的成本函数和EV的预测速度，将EV中B-SC的实时能量优化公式化为有限水平最优控制问题。此外，为减轻在线解决问题的计算负担，Pontryagin' s最小原则用于后退的方式。与传统的非线性MPC相比，仿真结果验证了所提出的MPC算法对于电动汽车BSC实时能量优化的有效性。