Online Optimization based Predictive Energy Management Functionality of Plug-In Hybrid Powertrain using Trajectory Planning Methods

Powertrain systems exploiting information from vehicle connectivity have widened the system boundary resulting in additional degrees-of-freedom for predictive trajectory planning. Heuristic methods based on component characteristics are currently widely used for Energy Management (EM) functionality of hybridized powertrains. Despite their better usability, increased calibration effort and sensitivity to synthetic calibration scenarios are drawbacks of such control methods. Availability of predictive data, better computing power and challenges posed by various scenarios in real driving, have led to interest in online-optimizing EM functionality. Equivalent Consumption Minimization Strategy (ECMS) approaches based on Indirect optimal control /Pontryagin Minimum principle have difficulty in handling inequality state constraints. Extensions of ECMS make use of modifications to the equivalence factor/co-state, based on prediction of driving conditions.

The proposed method uses limited time horizon prediction data to optimize engine on/off state and torque split among the energy converters using direct optimal control. Along with its ability to handle inequality constraints on the system states directly, the proposed method does not require an explicit model of additional dynamics. Further, the developed EM functionality adapts in real-time based on situation-aware prediction along with offering possibility to tune online the optimization process using heuristics on constraint-limits. These advantages along with this real-time capability and flexibility to handle change of control objectives as well as variation of control weighting reduces calibration effort. Results of the functionality shall be compared with predictive ECMS method. The functionalities developed along with their real-time capability will be demonstrated using the Combustion Engine Assist (CEA) concept.