Status and challenges of residential and industrial non-intrusive load monitoring

Non-Intrusive Load Monitoring (NILM) is the process of identification of loads from an aggregate power interface using disaggregation algorithms. This paper identifies the current status, methodologies and challenges of NILM in residential and industrial settings. NILM has advanced substantially in recent years due to improvement in algorithms and methodologies. Currently, the important challenges facing residential NILM are inaccessibility of electricity meter high sampling data, and lack of reliable high resolution datasets. For industrial NILM the identification is more challenging due to increased number of loads and the variability of equipment type, temporal patterns and industrial secrecy. From our examination of data and its use in NILM, we observe that the number of devices that can be recognized and the training period required to achiever recognition is not only a function of the algorithms but more importantly is a function of sampling rates.