Supervised and Unsupervised Neural Networks: Experimental Study for Anomaly Detection in Electrical Consumption
Households are responsible for more than 40% of the global electricity consumption . The analysis of this consumption to find unexpected behaviours could have a great impact on saving electricity. This research presents an experimental study of supervised and unsupervised neural networks for anomaly detection in electrical consumption. Multilayer perceptrons and autoencoders are used for each approach, respectively. In order to select the most suitable neural model in each case, there is a comparison of various architectures. The proposed methods are evaluated using real-world data from an individual home electric power usage dataset. The performance is compared with a traditional statistical procedure. Experiments show that the supervised approach has a significant improvement in anomaly detection rate. We evaluate different possible feature sets. The results demonstrate that temporal data and measures of consumption patterns such as mean, standard deviation and percentiles are necessary to achieve higher accuracy.
KeywordsAnomaly detection Neural networks Supervised Unsupervised Statistic
E. Zamora and H. Sossa would like to acknowledge CIC-IPN for the support to carry out this research. This work was economically supported by SIP-IPN (grant numbers 20180180 and 20180730) and CONACYT (grant number 65) as well as the Red Temática de CONACYT de Inteligencia Computacional Aplicada. J. García acknowledges CONACYT for the scholarship granted towards pursuing his MSc studies.
- 1.A beginner’s guide to multilayer perceptrons. deeplearning4j.org/multilayerperceptron. Accessed 25 May 2018
- 2.Araya, D.B., Grolinger, K., ElYamany, H.F., Capretz, M.A., Bitsuamlak, G.: Collective contextual anomaly detection framework for smart buildings, pp. 511–518. IEEE (2016)Google Scholar
- 3.Ashton, K.: That “Internet of Things" thing. RFiD J. 22, 97–114 (2009)Google Scholar
- 8.Dheeru, D., Taniskidou, E.K.: UCI machine learning repository (2017)Google Scholar
- 9.Gómez Chacón, I.M., et al.: Educación matemática y ciudadanía (2010)Google Scholar
- 10.Hajian-Tilaki, K.: Receiver operating characteristic (ROC) curve analysis for medical diagnostic test evaluation. Casp. J. Intern. Med. 4(2), 627 (2013)Google Scholar
- 11.IEA: World energy outlook 2011 executive summary (2011)Google Scholar
- 12.Tasfi, N.L., Higashino, W.A., Grolinger, K., Capretz, M.A.: Deep neural networks with confidence sampling for electrical anomaly detection, June 2017Google Scholar
- 15.Ouyang, Z., Sun, X., Yue, D.: Hierarchical time series feature extraction for power consumption anomaly detection. In: Li, K., Xue, Y., Cui, S., Niu, Q., Yang, Z., Luk, P. (eds.) LSMS/ICSEE-2017. CCIS, vol. 763, pp. 267–275. Springer, Singapore (2017). https://doi.org/10.1007/978-981-10-6364-0_27CrossRefGoogle Scholar
- 17.Stover, C.: Unit circleGoogle Scholar
- 18.Yijia, T., Hang, G.: Anomaly detection of power consumption based on waveform feature recognition, pp. 587–591. IEEE (2016)Google Scholar