Energy and power management in a series Hybrid Electric Vehicle using Selective Evolutionary Generation

This paper applies a recently developed, on-line, search-based optimization technique called Selective Evolutionary Generation Systems (SEGS) to manage battery state-of-charge and battery and generator delivered power in a Series Hybrid Electric Vehicle (SHEV). This energy and power management problem was recently tackled with a model predictive control approach that focused on improving overall powertrain operation efficiency rather than optimizing fuel consumption. However, the resultant constrained quadratic program is not easily solvable on-line in standard automotive microcontrollers, requiring explicit solutions obtained by off-line multiparametric programming solvers for implementation. In this paper, with the same motivation of maintaining SHEV operation in the high efficiency region, we apply the SEGS algorithm to SHEV control. The SEGS algorithm is not model-based but attains globally optimal behavior on-line using a probabilistic fitness distribution over a search space. It is also optimally search efficient and tunably responsive in dynamic environments despite taking only local and computationally-inexpensive decisions. Simulation results that illustrate this SEGS application are reported.