Influence of Architecture Design on the Performance and Fuel Efficiency of Hydraulic Hybrid Transmissions

Hydraulic hybrids are a proven and effective alternative to electric hybrids for increasing the fuel efficiency of on-road vehicles. To further the state-of-the-art this work investigates how architecture design influences the performance, fuel efficiency, and controllability of hydraulic hybrid transmissions. To that end a novel neural network based power management controller was proposed and investigated for conventional hydraulic hybrids. This control scheme trained a neural network to generalize the globally optimal, though non-implementable, state trajectories generated by dynamic programming. Once trained the neural network was used for online prediction of a transmission’s optimal state trajectory during untrained cycles forming the basis of an implementable controller. During hardware-in-the-loop (HIL) testing the proposed control strategy improved fuel efficiency by up to 25.5% when compared with baseline approaches. To further improve performance and fuel efficiency a novel transmission architecture termed a Blended Hydraulic Hybrid was proposed and investigated. This novel architecture improves on existing hydraulic hybrids by partially decoupling power transmission from energy storage while simultaneously providing means to recouple the systems when advantageous. Optimal control studies showed the proposed architecture improved fuel efficiency over both baseline mechanical and conventional hydraulic hybrid transmissions. Effective system level and supervisory control schemes were also proposed for the blended hybrid. In order to investigate the concept’s feasibility a blended hybrid transmission was constructed and successfully tested on a HIL transmission dynamometer. Finally to investigate controllability and driver perception an SUV was retrofitted with a blended hybrid transmission. Successful on-road vehicle testing showcased the potential of this novel hybrid architecture as a viable alternative to more conventional electric hybrids in the transportation sector.

Name thesis:degree_name

Doctor of Philosophy (PhD)

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Agricultural and Biological Engineering

Year

2015

Author dc:creator

• Sprengel, Michael Wayne

Contributors dc:contributor

• Monika Ivantysynova
• Dennis Buckmaster
• John Starkey
• Andrea Vacca

• Blended Hybrid
• Dynamic Programming
• Hardware-in-the-Loop
• Hydraulic Hybrid
• Neural Network
• Power Managment

Repository record dc:identifier

https://docs.lib.purdue.edu/open_access_dissertations/1317

OAI identifier oai:identifier

oai:docs.lib.purdue.edu:open_access_dissertations-2533