On Multi-Label Classification for Non-Intrusive Load Identification using Low Sampling Frequency Datasets

Abstract

Non-intrusive load monitoring (NILM) aims to infer information about the electric consumption of individual loads using the premises' aggregate consumption. In this work, we target supervised multi-label classification for non-intrusive load identification. We describe how we have created a new dataset from Moroccan households using a low sampling frequency. Then, we analyze the performance of three machine learning models for NILM, and investigate the impact of signal input length on performance.