Exploring a Direct Policy Search Framework for Multiobjective Optimization of a Microgrid Energy Management System

Abstract

With an increasing focus on integration of distributed energy resources, it is likely that microgrids will proliferate globally. Microgrid systems will be expected to achieve multiple stakeholder objectives, motivating the study of microgrid operations using a multiobjective framework. A multiobjective perspective has the potential balance the trade-offs implicit to efficient use of available resources. To address this challenge, this paper proposes a simulation based parametric approach for multiobjective optimization for microgrid energy management. The methodology generates a Pareto-approximate set of control policies, to provide a microgrid controller with diverse alternative strategies for utilizing resources to balance competing objectives. The policies also help to illustrate the complex relationships between the objectives, and the consequences of compromises across performance. The methodology is implemented on a test microgrid and the potential benefits are demonstrated with a set of illustrative case studies.