Premium
Hybrid Model Predictive Power Management of a Battery‐Supercapacitor Electric Vehicle
Author(s) -
Meyer Richard T.,
DeCarlo Raymond A.,
Pekarek Steve
Publication year - 2016
Publication title -
asian journal of control
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.769
H-Index - 53
eISSN - 1934-6093
pISSN - 1561-8625
DOI - 10.1002/asjc.1259
Subject(s) - powertrain , control theory (sociology) , battery (electricity) , electric vehicle , regenerative brake , power management , energy management , computer science , model predictive control , controller (irrigation) , hybrid vehicle , automotive engineering , power (physics) , engineering , control engineering , torque , energy (signal processing) , control (management) , mathematics , agronomy , statistics , physics , brake , quantum mechanics , artificial intelligence , biology , thermodynamics
Abstract Recently, battery powered electric vehicles (EV), such as the Tesla Model S, have reached commercialization. Future EVs will likely pair the battery with a supercapacitor to extend battery life by minimizing the effects of high power and rapidly fluctuating loads. As such, this paper investigates optimal power management of an EV powertrain with both a battery pack and a supercapacitor as energy sources. The battery‐supercapacitor‐based powertrain has four distinct controllable modes, each with a unique set of dynamics and constraints. Unique power flow expressions for the supercapacitor, vehicle motion, and drive system induction motor are presented. The overall powertrain is represented as a switched interconnected dynamical system having both differential and algebraic constraints in each mode of operation. Constrained by this model, this paper sets forth a hybrid model predictive control strategy for minimizing velocity tracking error and frictional braking (to encourage regenerative braking) while encouraging fast recharge of the supercapacitor. The optimization is performed using a relaxed representation of the control problem (termed the embedding method), collocation for discretization, and traditional nonlinear programming to compute the mode and continuous control inputs. The methodology avoids the computational complexity associated with alternative approaches such as mixed‐integer programming. The developed optimization methodology is shown to be useful as a rapid iterative prototyping tool by varying vehicle mass and electric drive maximum power to find combinations that result in satisfactory tracking performance of a trapezoidal drive profile. Performance of the originally specified powertrain and a first design iteration are compared using simulations of the EPA highway and urban drive profiles and the new European drive cycle to verify the first design iteration gives improvement.