Project Advance IO – Identifying irregularities of electricity usage

Project Information

  • Partner: Advance IO
  • Students: Nozipho Hlophe (UP), Mafu Mxenge (UKZN), Mixo Ngobeni (UL)
  • Project Lead: Dhiren Seetharam, Dr. Quentin Williams
  • Project Mentors: Albert Dove
  • Year: 2017/2018

Project Description

Advance IO is a smart electricity metering project performed by iD2, a fully owned subsidiary of Advance. The project is performed for the Ekurhuleni Metropolitan Municipality. Advance provides professional services over the entire spectrum of IT development disciplines with complementary services such as IoT and related component manufacturing. The data-set consists of hourly electricity meter readings from a large number of points of connection (PoC). Each PoC represents an electricity user. The smart meter measures the watt-hours (Wh) and credits used by a PoC.

Electricity theft is a significant global problem that results in revenue loss and increased costs to paying customers, as well as a range of safety issues. The project’s primary objective is to identify irregular usage patterns and potentially detect electricity fraud.

The team defined a normal usage trend, thereafter used machine learning (long-term short memory neural networks) to identify anomalous trends from the expected electricity usage. They were able to detect anomalous trends which may be a result of electricity theft, meter bypass or a resetting meter. Measures can be put into place to mitigate theft. Furthermore, faulty equipment can also be identified from the trends and can be replaced or improved to give more accurate measurements.