Enhancement of an Adaptive HEV Operating Strategy Using Machine Learning Algorithms

For vehicle manufacturers as well as for many drivers saving fuel has been a popular issue. In order to maximize the potential of the consumption-savings, optimization of operating strategy of vehicles, particularly of HEV (HEV: hybrid electric vehicle.), becomes an increasingly important task. To enhance the current rule-based operating strategy of HEV, an adaptive heuristic operating strategy has been developed, which identifies driving patterns and chooses the best parameter set for the control strategy from a database. However, this strategy does not fit to the driving behavior of individual drivers. Therefore, a further knowledge-based approach that independently optimizes the operating strategy has been developed using of multigene symbolic regression by utilizing supervised machine learning. The investigation showed that a knowledge-based approach is able to save about 18.3 % CO2 and fuel compared to a basic heuristic strategy.