The horizon in predictive energy storage control

Energy saving storage control in power conversion systems is considered. Savings are achieved by operating the power converters in efficient operating points, meeting the desired power output with an energy storage device. Predictions of the future desired output and conditions are used, with the largest gain for an infinite prediction horizon. This is not realistic predictions for the far future are unreliable. So, it is of interest to know the performance degradation due to a finite horizon. This paper presents a guideline for the horizon length, using an interpretation of the characteristics of the optimal solution for a stylized problem that can give performance close to the infinite horizon one. Characteristics of the active constraints determine the desired horizon. When avoiding active constraints covering the full horizon, the horizon can be reduced to the smallest possible one.