Optimized control studies of a parallel hybrid electric vehicle

Abstract

This thesis addresses the development of a control scheme to maximize automobile fuel economy and battery state-of-charge (SOC) while meeting exhaust emission standards for parallel hybrid electric vehicles, which are an alternative to conventional passenger vehicles. The principle components of the drive train are a small internal combustion engine and an electric motor, both of them applying torque directly to the drive shaft for propelling the vehicle. Each component of the parallel hybrid vehicle is modeled, and throttle angle, motor current and brake torque command chosen as the control inputs. A performance index describing the total fuel and battery charge used, as well as pollutants emitted over the federal drive cycle, is defined. The problem is to find the optimal control inputs, as a function of time, that minimize the performance index under the chosen drive cycle while satisfying lower and upper bounds on the controls as well as the torque command constraint, derived from the drive cycle speed that the vehicle must follow. The problem is formulated so that optimal control theory can be used by defining the Hamiltonian of the system and deriving the Euler-Lagrange equations. Four special cases for the control bounds which are of practical importance are considered. But, because of the complicated analytical derivatives, solving the general analytical problem is not tractable. The alternate approach that is chosen is a numerical optimization method that solves the constrained optimization problem using the Recursive Quadratic Programming Method.'To evaluate various control schemes, a set of selected performance measures are studied: only SOC performance, and balanced fuel and SOC performance. Simulations under the federal drive cycle show that we achieve the design objectives while getting better results than with a simple logic controller. The optimum control results suggest that the throttle should always be kept wide open for the SOC to be maximized. This should be accomplished with Buntin's logic controller and would allow us to keep his easy control implementation while improving his performance.