Browsing by Author "Xie, Shaobo"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
Item Open Access Model predictive energy management for plug-in hybrid electric vehicles considering optimal battery depth of discharge(Elsevier, 2019-02-22) Xie, Shaobo; Hu, Xiaosong; Qi, Shanwei; Tang, Xiaolin; Lang, Kun; Xin, Zongke; Brighton, JamesWhen developing an energy management strategy (EMS) including a battery aging model for plug-in hybrid electric vehicles, the trade-off between the energy consumption cost (ECC) and the equivalent battery life loss cost (EBLLC) should be considered to minimize the total cost of both and improve the life cycle value. Unlike EMSs with a lower State of Charge (SOC) boundary value given in advance, this paper proposes a model predictive control of EMS based on an optimal battery depth of discharge (DOD) for a minimum sum of ECC and EBLLC. First, the optimal DOD is identified using Pontryagin's Minimum Principle and shooting method. Then a reference SOC is constructed with the optimal DOD, and a model predictive controller (MPC) in which the conflict between the ECC and EBLC is optimized in a moving horizon is implemented. The proposed EMS is examined by real-world driving cycles under different preview horizons, and the results indicate that MPCs with a battery aging model lower the total cost by 1.65%, 1.29% and 1.38%, respectively, for three preview horizons (5, 10 and 15 s) under a city bus route of about 70 km, compared to those unaware of battery aging. Meanwhile, global optimization algorithms like the dynamic programming and Pontryagin's Minimum Principle, as well as a rule-based method, are compared with the predictive controller, in terms of computational expense and accuracy.Item Open Access Pontryagin's Minimum Principle based model predictive control of energy management for a plug-in hybrid electric bus(Elsevier, 2018-12-17) Xie, Shaobo; Hu, Xiaosong; Xin, Zongke; Brighton, James L.To improve computational efficiency of energy management strategies for plug-in hybrid electric vehicles (PHEVs), this paper proposes a stochastic model predictive controller (MPC) based on Pontryagin’s Minimum Principle (PMP), which differs from widely used dynamic programming (DP)-based predictive methods. First, short-time speed forecasting is achieved using a Markov chain model, based on real-world driving cycles. The PMP- and DP-based MPCs are compared under four preview horizons (5 s, 10 s, 15 s and 20 s), and the results show that the computational time of the DP-MPC is almost four times of that in the PMP-MPC. Moreover, the influence of predication horizon length on computational time and energy consumption is examined. Given a preview horizon of 5 s, the PMP-MPC holds a total energy consumption cost of 7.80 USD and computational time per second of 0.0130 s. When the preview horizon increases to 20 s, the total cost is 7.77 USD with the computational time per second increasing to 0.0502 s. Finally, DP, PMP, and rule-based strategies are contrasted to the PMP-MPC method, further demonstrating the promising performance and computational efficiency of the proposed methodology.