REAL-TIME PREDICTIVE CONTROL OF HYBRID FUEL CELL DRIVE TRAINS

Optimal predictive control of hybrid drive trains offers improved characteristics compared to traditional control strategies. Unfortunately that is obtained at cost of a significantly increased computational effort, which makes it hard to fulfill the real-time requirements with current vehicle control units. The paper describes a predictive control algorithm for hybrid fuel cell drive trains which allows real-time control. First, the used predictor-optimizer structure of the controller is presented, which splits the control task into the prediction of the load power and the determination of the optimal control function. After a brief description of a worst-case predictor, the design of the optimizer is presented, which is the focus of the paper. The real-time ability is obtained by an adapted modeling of the hybrid drive, which finally allows the application of linear optimization in order to solve the optimal control problem. The characteristics of the resulting control algorithm are illustrated by a simulation example.