Non-intrusive load monitoring using two-point sensors for load measurement, identification and localization

Monitoring power consumption in loads that are connected to a power bus, without the use of power measurement hardware at the input of each load, is attractive for reasons of reduced cost and hardware complexity reduction, and improved reliability. This approach to power monitoring is termed non-intrusive load monitoring (NILM). Machine learning (ML) techniques have been applied for this purpose, for instance, for learning transient or spectral signatures of loads, and using these to identify loads. However, precise load disambiguation may still pose a challenge in certain applications. We describe approaches for NILM in a ring and a bus power distribution employing two-point sensing of current, along with voltage measurement. In combination with machine learning, these approaches allow for disambiguating loads by determining load location, and can be employed for power monitoring without the requirement for individual communication or power measurement hardware.