Toward Trustworthy Machine Learning Models for Fault Detection in Energy Systems

ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation (BuildSys), pp. 280–281

Abstract

This study focuses on detecting invalid regimes of machine learning models through novelty detection algorithms. We apply them to a two-dimensional test case. Our results illustrate the impact of data noise and different hyperparameter settings.

Authors 5

1. Martin Rätz Aachen E.ON Energy Research Center Institute for Energy Efficient Buildings and Indoor Climate

   RWTH Aachen University

   Affiliation as printed

   Institute for Energy Efficient Buildings and Indoor Climate, E.ON Energy Research Center, RWTH Aachen University, Germany

2. Patrick Henkel Aachen E.ON Energy Research Center Institute for Energy Efficient Buildings and Indoor Climate

   RWTH Aachen University

   Affiliation as printed

   Institute for Energy Efficient Buildings and Indoor Climate, E.ON Energy Research Center, RWTH Aachen University, Germany

3. Phillip Stoffel Aachen E.ON Energy Research Center Institute for Energy Efficient Buildings and Indoor Climate

   RWTH Aachen University

   Affiliation as printed

   Institute for Energy Efficient Buildings and Indoor Climate, E.ON Energy Research Center, RWTH Aachen University, Germany

4. Rita Streblow Aachen E.ON Energy Research Center Institute for Energy Efficient Buildings and Indoor Climate

   RWTH Aachen University

   Affiliation as printed

   Institute for Energy Efficient Buildings and Indoor Climate, E.ON Energy Research Center, RWTH Aachen University, Germany

5. Dirk Müller Aachen E.ON Energy Research Center Institute for Energy Efficient Buildings and Indoor Climate

   RWTH Aachen University

   Affiliation as printed

   Institute for Energy Efficient Buildings and Indoor Climate, E.ON Energy Research Center, RWTH Aachen University, Germany