Energy management strategy for battery/supercapacitor hybrid electric city bus based on driving pattern recognition

Abstract

The rapid development of cloud techniques like Vehicle-to-Cloud (V2C) makes it possible to gather more information and develop computationally efficient energy management systems (EMS) for electric vehicles. This paper proposes a novel EMS with low computational cost targeting hybrid battery/ultracapacitor electric buses to reduce energy consumption and battery life degradation. In the offline training process, by applying the K-means clustering method with 10 selected features, 16 typical driving conditions are classified. For each driving condition, dynamic programming is employed offline to generate global optimal results, which are then used in control rule extraction for online operation. During the online operation, the proposed EMS executes the designed driving pattern recognition algorithm with V2C assistance to select optimal control rules. The simulation results indicate that the proposed EMS effectively decreases the battery degradation and energy consumption cost by 13.89%, compared with the conventional rule-based strategy. In addition, it is shown that V2C assistance leads to a 6.81% lower cost. Besides, the robustness of the proposed EMS is validated by testing the EMS with highly randomized input with uncertainties up to 15% and long duration of V2C data packet loss up to 10 s.