IoT Load Classification and Anomaly Warning in ELV DC Pico-grids using Hierarchical Extended k-Nearest Neighbors

Remote monitoring of electrical systems has progressed beyond the need on knowing how much energy is consumed. As the maintenance procedure evolved from reactive to preventive to predictive, there is a growing demand to know what appliances reside in the circuit (classification) and a need to know if any appliance requires attention and maintenance (anomaly warning). Targeted at the increasing penetration of dc appliances and equipment in households and offices, the described low-cost solution consists of multiple distributed slave meters with a single master computer for extra low voltage dc pico-grids. The distributed slave meter acquires the current and voltage waveform from the cable of interest. It will condition the acquired data and extract 4 features per window block that will be sent to the master computer remotely over a Wi-Fi network. The solution proposed in this paper uses a hierarchical enhanced k-nearest neighbors (kNN) technique for classification and anomaly warning to trigger the attention of the user. This solution can be used as an ad hoc standalone investigation to check on a circuit when in doubt; it can be further expanded to several circuits in a building or vicinity to monitor the network. It can also be implemented as part of an Internet of Things (IoT) application. This paper presents the Hierarchical Enhanced kNN technique and its application in 3 different circuits: lightings, airconditioning and multiple load dc pico-grids. Index Terms-load classification, anomaly warning, k-nearest neighbors, extra low voltage, dc grid, artificial intelligence.