A two-level stochastic approach to optimize the energy management strategy for fixed-route hybrid electric vehicles

Abstract

Many hybrid electric vehicle (HEV) energy management strategies are developed and evaluated under fixed driving cycles. However in the real-world driving, vehicles are very unlikely to keep running under a fixed known cycle. Instead, a lot of vehicles run on fixed routes. Unfortunately, human driving data collected on a driving simulator shows that it is very difficult to select or create a determined typical driving cycle to represent the fixed-route driving due to the uncertainties in traffic light stops and driver behaviors. This paper presents a two-level stochastic approach to optimize the energy management strategy for fixed-route HEVs. The historical data on the fixed route are utilized and a road-segment-based stochastic HEV energy consumption model is built. The higher-level energy optimization problem is solved by stochastic dynamic programming (SDP). The SDP computation uses the vehicle model and historical driving data on the fixed route and it can be conducted offline. The result of SDP is a 2-dimension lookup table of parameters for lower-level control strategy therefore this approach can be easily real-time implemented in practice. The developed stochastic approach is compared with three strategies using the data collected on the driving simulator: the optimal energy consumption by assuming all trip information is known in advance and solved via dynamic programming (DP), a determined energy management approach using typical trip data of the fixed-route driving, and a simple strategy which does not require any route data. Simulation results show that the proposed stochastic energy management strategy consumes 1.8% more energy than the optimal result after 24 trips on the fixed route and considerably outperforms the other two real-time HEV energy management strategies.