|
1. A method of predicting an occurrence of an electrical arc between at least one of a plurality of blades and a socket among a meter installation of a plurality of smart meters, each smart meter comprising the plurality of blades adapted to be inserted into the socket, the method comprising: receiving historical data over a predefined time period from the plurality of smart meters and prior occurrences of electrical arcs for each of the plurality of smart meters, the historical data previously stored and comprising input components for each smart meter, the input components comprising: an amount of daily consumption of electricity for the meter; historical meter events for the meter; all service orders associated with the meter; meter manufacturer data; and historical weather temperature data; pre-processing the historical data to generate electrical arc features, the pre-processing comprising transforming the historical weather temperature data into a number of heating degree days and a number of cooling degree days during a predefined time period; training a first machine learning algorithm using the electrical arc features; receiving, as prediction input data for a first smart meter, one or more of the input components corresponding to the first smart meter; using the trained first machine learning algorithm and the prediction input data to predict a future occurrence of the electrical arc between at least one of the plurality of blades and the socket for the first smart meter, the predict the future occurrence of the electrical arc based on at least on the prediction input data for the first smart meter; and based on the prediction, automatically disabling the first smart meter.
|
