Component sizing of a plug-in hybrid electric vehicle powertrain, Part B: coupling bee-inspired metaheuristics to ensemble of local neuro-fuzzy radial basis identifiers

Abstract

In this paper, the authors investigate the potentials of an aggregated cooperative intelligent approach to optimise the size of components of a plug-in hybrid electric vehicle (PHEV) powertrain. The intelligent model consists of a set of modular local neuro-fuzzy radial basis identifiers. These intelligent tools are finally incorporated to develop a global identifier called ensemble neuro-fuzzy radial basis network (ENFRBN). The resulted global identifier synchronously uses the local maps to predict the fuel consumption (FC) rate of a PHEV for a specific drive cycle. To do so, an experimental/simulative sampling process was performed in smart hybrid and electric vehicle system laboratory at the University of Waterloo to create a database including a set of input/output pairs. After extracting knowledge from prepared database, the authors use two well-known bee-inspired heuristic algorithms, i.e., bee algorithm (BA) and artificial bee colony (ABC) to reach a compromise on optimal size of PHEV components.