Energy Theft Detection Via Artificial Neural Networks Conference Paper uri icon


  • With the implement of advanced metering infrastructure (AMI) and the efficient communication network, modern power grids can achieve a two-way communication between utilities and their customers for electricity generation and consumption. This feature provides utilities a better view of electricity consumption, which helps utilities better schedule the power generation timely to avoid wasting resource. And the two-way communication is a fundamental element for bilateral transaction between utilities and their customers. In this way, the utilities can stimulate their customers to use electricity more economically through offering a real-time electricity rate, which can lower the total electricity rate. However, energy theft is a big obstacle for this application because it can falsify the meter readings, which can cause economical loss for both utility companies and customers. There are millions of dollars lost annually because of it. Moreover, energy theft can also cause security issues and stabilities problems to power grids. Therefore, it is very important and necessary to clear energy theft situation in power systems. With the AMI, utility companies have way more data about energy consumption of their end users than before, which brings challenge on dealing with large volume of data but also provides opportunities for energy theft detection. This paper presents a method of energy theft detection via artificial neural networks(ANNs). This method utilizes large volume of historical data to train ANN s. The trained ANNs have the ability to detect energy theft with incoming data.

name of conference

  • 2018 IEEE PES Innovative Smart Grid Technologies Conference Europe (ISGT-Europe)

published proceedings


author list (cited authors)

  • Huang, H., Liu, S., & Davis, K.

complete list of authors

  • Huang, Hao||Liu, Shan||Davis, Katherine

publication date

  • January 2018