A deep learning approach for anomaly detection and prediction in power consumption data
- PDF / 5,895,332 Bytes
- 19 Pages / 547.044 x 736.903 pts Page_size
- 92 Downloads / 219 Views
SHORT COMMUNICATION
A deep learning approach for anomaly detection and prediction in power consumption data C. Chahla · H. Snoussi · L. Merghem · M. Esseghir
Received: 14 March 2019 / Accepted: 13 July 2020 © Springer Nature B.V. 2020
Abstract Anomaly detection in power consumption data can be very useful to building managers. It allows them to detect unexpected power consumption values, identify unusual behaviors, and foresee uncommon events. This paper proposes a novel unsupervised approach to detect anomalies in power consumption data. We combine the clustering-based methods with the prediction-based ones to learn typical behavior scenarios and to predict the power consumption of the next hour. These scenarios are explored by applying the K-means algorithm on 24 different K-means groups representing the 24 h of the day. This is based on the assumption that identical daily consumption behavior can appear repeatedly due to users’ living habits. In order to detect the anomaly 1 h before its occurrence, a Long Short-Term Memory (LSTM ) has been trained to predict the next power consumption value. This predicted value with some earlier data values are concatenated into a vector then compared with the learned typical scenarios. We used AutoEncoders to detect anomalous days in general and this
C. Chahla () · H. Snoussi University of Technology of Troyes, Institute Charles Delaunay-LM2S, 12 Rue Marie Curie, Troyes, France e-mail: [email protected] L. Merghem · M. Esseghir University of Technology of Troyes, Institute Charles Delaunay-ERA, Troyes, France
novel method to specify at what time the anomaly has occurred. Our approach not only detectss anomalies in off-line mode but also allows real-time detection on live data streams. Keywords Anomaly detection · K-means · LSTM · Auto-Encoders · Power consumption
Introduction Anomalies in data are patterns that do not match the well-defined notion of normal behavior (Chandola et al. 2009). Anomaly detection in power consumption data can be very effective, making energy-efficient home improvements as well as saving cost. Residential and commercial buildings consume 60% of the world’s total amount of electricity (UNEP 2017). Lighting, heating, and cooling constitute the biggest part of energy consumption. But lighting all alone is the most important energy usage (Energy 2010). Techniques utilized for reducing the energy waste nowadays, for example, rely on motion detector for each source of light switching them on and off. In general, the electric power demand is highly affected by weather conditions. In summer, the growth of power demand on the consumer side can be correlated with cooling demands. Similarly, in winter, power consumption increases because of heating needs. The power consumption varies also during
Energy Efficiency
the same weather between weekends and weekdays. Thus, anomaly patterns would differ from weekdays to weekends. In fact, anomalies can be classified into three categories (Chandola et al. 2009): (1) Point anomalies: when an individual poi
Data Loading...
