48V 마일드 하이브리드 자동차의 충전유지 제어 기반 전력관리 전략과 Hamilton-Jacobi-Bellman 기법을 활용한 구현에 관한 연구

Abstract

In this thesis, an electric power management strategy is proposed which is specialized for a 48 V mild hybrid electric vehicle (HEV). Comparing to the full HEV, the 48 V mild HEV promotes price competitiveness by using low-end power components such as small-capacity and low-voltage battery, and low-power motor/ generator (M/G). This policy may deprive the mild HEV of opportunities to control an M/G at a required moment, since a small-capacity battery used to cause severe fluctuation of state-of-charge (SOC) even in the small change of stored electric power. Consequently, these operation environments do not only restrict the system functionality, but also degrades the control performance. In order to maintain sufficient electric power in a battery required for an M/G operation, this research applies the charge-sustaining strategy as a fundamental power management principle that battery SOC is managed to stay near the most efficient spot for charge and discharge. Prior to implementing the strategy, feasibilities were assessed by simulations applying the dynamic programming (DP) under two types of test scenarios. One assumes a situation that vehicle is driving while charging up a depleted battery to the target SOC to investigate influences of initial SOC on the fuel efficiency. The other assumes that permissible SOC ranges are restricted to be stayed within ± 0.2 of the initial or final SOC to order to imitate the charge-sustaining operation. The results showed that it is reasonable to control a battery SOC within narrow bounds in practical driving situation considering degradation of fuel efficiency caused by battery depletion. The proposed strategy was formulated as a nonlinear optimal regulation problem to meet two conflicting purposes simultaneously such as fuel efficiency minimization and SOC maintenance. Since the reference models are required to estimate optimality in optimal control approaches, 48 V power-net model of mild HEV was defined that includes the engine, M/G, and battery component models. The models were not only used as the reference models for optimal regulation control, but also used to analyze the conversion efficiencies to define component constraints. Analyzing the battery model, the most efficient charge and discharge points were found that enable to determine the target SOC. The nonlinear optimal regulation problem was implemented by applying discrete-time (DT) Hamilton-Jacobi-Bellman (HJB) approach. In particular, the value function in HJB was formulated as a Lyapunov function to assure the stability along with optimality. In the simulations, the fuel efficiency performances of the proposed algorithm were declined maximally by 1.3 % on UDDS and NEDC that contain urban driving environment, and 2.07 % on HWFET compared to results of the reference algorithm which applied DP approach. The results explain the proposed algorithm presents the near-optimal performances not so much different from the cases of reference algorithm. Providing that the power-net models contain precise efficiency characteristics, the proposed power management strategy can present near-optimal performances even at the practical driving environments.