TY - GEN
T1 - Reinforced explorit on optimizing vehicle powertrains
AU - Parque, Victor
AU - Kobayashi, Masakazu
AU - Higashi, Masatake
PY - 2013/12/1
Y1 - 2013/12/1
N2 - How to build optimal vehicular powertrains? We study this question and propose an algorithm inspired by a domain-general design process. The basic idea is to interplay co-biasingly between the local approximations of discrete design and the global refinements of continuous parameters. The proposed method was evaluated to design powertrains of four types of vehicles: Series Hybrid Electric Vehicle(SHEV), Parallel Hybrid Electric Vehicle(PHEV), Fuel Cell(FC) and Electric Vehicle(EV). Simulation results show noticeable improvements on mileage per gas emissions over different study cases. To our knowledge, this is the first study aiming at designing vehicle powertrains considering the holistic point of view.
AB - How to build optimal vehicular powertrains? We study this question and propose an algorithm inspired by a domain-general design process. The basic idea is to interplay co-biasingly between the local approximations of discrete design and the global refinements of continuous parameters. The proposed method was evaluated to design powertrains of four types of vehicles: Series Hybrid Electric Vehicle(SHEV), Parallel Hybrid Electric Vehicle(PHEV), Fuel Cell(FC) and Electric Vehicle(EV). Simulation results show noticeable improvements on mileage per gas emissions over different study cases. To our knowledge, this is the first study aiming at designing vehicle powertrains considering the holistic point of view.
KW - Electric vehicle
KW - Explorit
KW - Fuel cell
KW - Parallel hybrid electricvehicle
KW - Reinforcement learning
KW - Series hybrid electric vehicle
UR - http://www.scopus.com/inward/record.url?scp=84893388591&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84893388591&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-42042-9_72
DO - 10.1007/978-3-642-42042-9_72
M3 - Conference contribution
AN - SCOPUS:84893388591
SN - 9783642420412
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 579
EP - 586
BT - Neural Information Processing - 20th International Conference, ICONIP 2013, Proceedings
T2 - 20th International Conference on Neural Information Processing, ICONIP 2013
Y2 - 3 November 2013 through 7 November 2013
ER -