Non-intrusive load monitoring for existing water and natural gas metering infrastructure

Abstract

Non-intrusive load monitoring (NILM) involves measuring power transfer or resource flow upstream from several independent loads and disaggregating the information into separate loads, trading hardware installation complexity for advanced signal processing. Non-contact NILM devices avoid direct connections to the grid or network, instead leveraging physical properties of the grid or existing meter to extract load information. This thesis presents an improved model for the rotating magnetic field produced by positive displacement water meters, new methods and non-contact hardware for non-intrusively extracting flow from these and bellows-and-diaphragm natural gas meters, and signal processing techniques for reconstructing flow from measured data to enable cheap and easy retrofit of existing metering infrastructure for NILM.