Robust NILM

Robust Machine learning methods for Non-Intrusive Load Monitoring (NILM)

Recently, large-scale deployments of smart meters have sparked the interest in developing effective non-intrusive load monitoring (NILM) solutions for improving energy monitoring, awareness and reducing energy consumption in buildings. NILM uses smart meter data to infer what end-appliances running in the building and estimate their respective power consumption. This project:

  • Investigated and developed robust feature representation for appliance recognition at high frequency in NILM
  • Developed robust multi-task and multilabel learning for appliance recognition in NILM and
  • Develop end-to-end NILM and investigate its value propositions in smart-grid.