The combination of batteries and supercapacitors is promising in electric vehicles context to minimize battery aging. Such a system needs an energy management strategy (EMS) that distributes energy in real-time for real driving cycles. Pontryagin's minimum principle (PMP) is widely used in adaptive forms to develop real-time optimization-based EMSs thanks to its analytical approach. This methodology leads to an off-line optimal solution which requires an extra adaptive mechanism for real-time applications. In this paper, a simplification of the PMP method is proposed to avoid the adaptation mechanism in real-time. This new EMS is compared to well-known conventional strategies by simulation. Furthermore, experimental results are provided to assess the real-time operation of the proposed EMS. Simulation and experimental results prove the advantages of the proposed approach by a reduction up to 50% of the batteries rms current on a real-world driving cycle compared to a battery-only EV.

Evaluating performances of real-time strategies for hybrid energy storage systems (HESSs) of electric vehicles (EVs) always requires optimal energy management strategies (EMSs) as offline benchmarks. Dynamic programming (DP) is well-known due to its ability to obtain global optimal solutions based on the numerical searching technique. Nevertheless, DP accuracy depends on the numericalization fineness. Analytical optimal control methods, typically Pontryagin’s minimum principle (PMP), are also frequently used as effective counterparts. However, solving optimal control problems based on these methods often depends on the complexity and the characteristic of the system model; basically, it is sophisticated since there is no general way to solve the issue. This article proposes an alternative approach of using PMP to develop an optimal EMS for battery/supercapacitor HESSs. The novel strategy is based on formulating the problem in terms of power and energy, which forms a state-constrained optimal control problem. PMP is then applied with a penalty function, in which the inequality state constraints are reformulated to deduce a new state-unconstrained problem. The proposed optimal EMS is hundreds of times faster than DP with better results. Moreover, the optimal solution is piecewise constant that could give significant insights to develop real-time strategies in future studies.

To minimize battery aging of electric vehicles (EVs), it is paramount to manage efficiently their energy consumption. An energy management strategy (EMS) has recently been developed, where batteries and supercapacitors (SCs) are coordinated as a function of the driving mode which is determined manually by the driver. In the present article, we improve the EMS by developing a driving mode predictor (DMP) that determines automatically in real-time the driving mode simply from the speed history of the vehicle. The DMP is designed using supervised learning (SL), a branch of machine learning (ML). A strength of our approach is that it is applicable to predict the driving modes of any EV. To predict the driving modes during a trip made by a given EV, the DMP needs to follow the speed evolution of the EV and know its maximum reachable speed. The integration of the DMP into the EMS results in an enhanced EMS called DMP-based EMS that determines automatically in real-time the power to take from each source of energy of the EV as a function of its speed history. The results obtained from real driving cycles confirm the prediction quality of the DMP and the energy efficiency of the proposed DMP-based EMS.

The necessity of rejecting the external disturbance and model uncertainty is vital for control design to achieve high performance and robustness of the system. Disturbance cancelation techniques are, therefore, of interest for both academic and industrial research and development. Among them, disturbance observer based control (DOBC) and active disturbance rejection control (ADRC) are widely applied, thanks to their high performance and effectiveness in synthesis and realization. However, two knowledge gaps vis-à-vis the dynamical delay and the unification of these techniques exist in literature. This article proposes a study on the unification of DOBC and ADRC for the application of speed control of permanent magnet synchronous motor. The explicit formulas of the estimated disturbance are carried out and the dynamical delays of the observations are, therefore, straightforwardly deduced. By addressing these transparent delay formulas, the unification of DOBC and ADRC is verified. The identical performance of two methods in rejecting both external disturbance and model uncertainty is demonstrated by simulation and experimental validations.