Multi-stage ECMS considering engine dynamics and start/stop strategy for real-time application in hybrid electric vehicles

Abstract

Equivalent Consumption Minimisation Strategy (ECMS) is a popular algorithm determining the optimal blending between engine and electric motors torque on a hybrid electric vehicle. Being a realisation of the Pontryagin Minimum Principle, it condenses driving mission’s a priori information into a single factor, avoiding any need of complex predictions. This allows real-time implementation, making ECMS the only optimal control strategy that can be realistically implemented onboard real-life vehicles. Despite this advantage, ECMS is natively based on very simplified powertrain models with limited representation of dynamics, lacks transients’ impact on fuel efficiency, and misses the drivability factor on ECMS-based vehicles. This paper proposes a strategy to address these weaknesses in a hybrid electric vehicle featuring a spark ignited (SI) engine in two ways. On the one hand, the control problem is divided in three stages, each controlling a single dynamic of the powertrain with a dedicated ECMS implementation. These three stages are: torque split target by actuating engine throttle, engine ignition retardation and engine start/stop command. Motor torque requests are also coordinated within ECMS algorithms. Powertrain components feedback provides both a link between these three ECMS modules and alignment with current powertrain conditions. On the other hand, ECMS cost function minimum search algorithm considers engine dynamic response and vehicle dynamic constraints to guarantee consistent drivability. The presented strategy has been implemented as core hybrid strategy in the vehicle control unit of a production vehicle. This strategy has proven lighter and more accurate in terms of candidates’ selection than conventional ECMS implementations, plus being fully real-time capable. The algorithm has been tested in different environments such as hardware- and driver-in-the-loop and, in all cases, it controlled the torque split in a smoother manner, respecting dynamic constraints and reducing the number of start/stop events, while providing charge sustaining operation and near-optimal fuel economy.