A Data-driven Online Identification and Control Optimization Approach applied to a Hybrid Electric Powertrain System

Abstract

AbstractWithin this contribution the optimization of the electric current flow control within a hybrid electric powertrain system and hence its power and energy management strategy is introduced. This contribution describes the control optimization . Power and energy management are methods applied to hybrid electric powertrain systems to be used in modern vehicles. Hereby the electric current flow within the powertrain, i.e. between its system components, is controlled regarding predefined control objectives such as the power or energy availability with respect to a load profile. In this contribution an online-applicable optimization method of the power and energy management strategy is introduced. Its realization contains a system identification algorithm capable of being implemented within online applications and, based on this, a suitable control optimization including a predictive approach. The control objectives to be optimized in this context are the energy consumption, the power availability, and deterioration aspects of the system and its components. The algorithm is demonstrated using the example of a fuel cell/supercap-based hybrid electric powertrain system. After a brief description of the system and operation details, the results of the identification process in relation to the real behavior as well as the optimization algorithm are presented. Applying typical driving cycles to this system it can be shown that a significant improvement of this approach compared with classical power and energy methods can be achived. The main scientific and practical question remains: what can be assumed about the future behavior of the workload (applied to the system) and how can the control be optimized for the unknown future behavior. Mathematically expressed: How exact the system behavior has to be identified (from the last measurements) in relation to the principally unknown character of the upcoming future load in relations to the different optimization goals.