Optimal predictive eco-driving cycles for conventional and electric cars

Djamaleddine Maamria (1) , Kristan Gillet (1) , Guillaume Colin (1) , Yann Chamaillard (1) , C Nouillant (2)

1 PRISME - Laboratoire pluridisciplinaire de recherche en ingénierie des systèmes, mécanique et énergétique
2 PSA - PSA Peugeot - Citroën

In this paper, the computation of eco-driving cycles for electric and conventional vehicles using receding horizon and optimal control is investigated. The problem is formulated as consecutive-optimization problems aiming at minimizing the vehicle energy consumption under traffic and speed constraints. The solving method is based on Dynamic Programming (DP). The impact of the look-ahead distance on the optimal speed computation is studied to find a trade-off between the optimality and the computation time. Simulation results show that in urban driving conditions, a look-ahead distance of 300m to 500m leads to a sub-optimality less than 0.6% in the energy consumption compared to the global solution. For highway driving conditions, a look-ahead distance of 1km to 2km leads to a sub-optimality less than 0.7% compared to the global solution.

Domains: Engineering Sciences [physics]; Automatic

Djamaleddine Maamria, Kristan Gillet, Guillaume Colin, Yann Chamaillard, C Nouillant. Optimal predictive eco-driving cycles for conventional and electric cars. 2018 Annual American Control Conference (ACC), Jun 2018, Milwaukee, United States. ⟨hal-01826750⟩