Run-time parameter selection and tuning for energy optimization algorithms

Abstract

Energy Management Systems (EMS) promise a great potential to enable the sustainable and efficient integration of distributed energy generation from renewable sources by optimization of energy flows. In this paper, we present a run-time selection and meta-evolutionary parameter tuning component for optimization algorithms in EMS and an approach for the distributed application of this component. These have been applied to an existing EMS, which uses an Evolutionary Algorithm. Evaluations of the component in realistic scenarios showreduced run-timeswith similar or even improved solution quality, while the distributed application reduces the risk of over-confidence and over-tuning.