Diagnosis and Fault Mitigation in a Microgrid Using Model Predictive Control

A microgrid is a group of distributed energy resources (energy generation and/or storage systems) prepared to meet the energy demand for which it has been designed. Microgrids can act in isolation or connected to the distribution network. This document presents a preliminary methodology to perform a Diagnosis and Online Failure Mitigation procedure in an experimental microgrid located at the laboratory of the University of Seville. By means of an analytical redundancy relationship, the model and the process are compared in each sampling period; when inconsistencies are detected, residuals are generated in order to proceed with mitigation actions. Once the specific failure has been identified, the controller may be reconfigured to achieve the mitigation of the failure. The implemented optimal control strategy has been Model Predictive Control. Therefore, the mitigation of failures is made by the joint action of the Model Predictive Controller and the Fault Diagnosis System.