Search

• We surveyed the issue of anomaly detection of electricity consumption. • We proposed a big data and machine learning based anomaly detection framework. • We analyzed the successful detection rate of two related schemes. The big data generated by Industry 4.0 is expected to increase 20-fold in the next ten years and it has raised various challenges in Industrial Wireless Sensor Networks (IWSNs). Among these challenges, detecting different types of anomalies of industrial electricity consumption in an accurate and timely manner is a priority. If not handled properly, these anomalies could lead to serious consequences, such as irregular fire and paralyzed power system components. While existing anomaly detection techniques may be efficient for old systems, they are now faced with big transmitted data. Therefore, it is important to design new methods that can detect the electricity consumption anomaly and carry out appropriate actions. In this article, we first review several existing work on anomaly detection schemes, and then introduce the system and monitoring models. Then, we present a new framework that aims to detect electricity consumption anomalies accurately and timely using sensor processing, smart meter readings, machine learning and blockchain. [ABSTRACT FROM AUTHOR]