c Springer, 2013. This is the author’s version of the work. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purpose or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the copyright holder. The definite version is published in the Proceedings of the 15th European Conference on the Applications of Evolutionary and bio-inspired Computation (EvoApplications’13), Vienna, Austria. Evolving Non-Intrusive Load Monitoring Dominik Egarter 1 , Anita Sobe 2 , and Wilfried Elmenreich 1, 3 1 Institute of Networked and Embedded Systems / Lakeside Labs Alpen-Adria-Universit¨ at Klagenfurt, Austria 2 Institut d’informatique, Universit´ e de Neuchˆ atel, Switzerland 3 Complex Systems Engineering, Universit¨ at Passau, Germany dominik.egarter@aau.at, anita.sobe@unine.ch, wilfried.elmenreich@aau.at Abstract. Non-intrusive load monitoring (NILM) identifies used appli- ances in a total power load according to their individual load character- istics. In this paper we propose an evolutionary optimization algorithm to identify appliances, which are modeled as on/off appliances. We eval- uate our proposed evolutionary optimization by simulation with Matlab, where we use a random total load and randomly generated power profiles to make a statement of the applicability of the evolutionary algorithm as optimization technique for NILM. Our results shows that the evolu- tionary approach is feasible to be used in NILM systems and can reach satisfying detection probabilities. Keywords: Evolutionary Algorithm, Knapsack Problem, Evolution, Non-Intrusive Load Monitoring, NILM 1 Introduction With the upcoming of decentralized regenerative energy sources, the amount of available energy at a particular time and, due to network capacity con- straints, location becomes dependent on the current weather situation (photo- voltaic production depends on amount of sunshine, windmill-powered plants on wind speed). One way to mitigate this issue is to provide energy storage (e. g., by batteries, pumped-storage hydropower plants, conversion to methane, etc). The other way is shaping the energy consumption at the consumer side. A typical household contains hundreds of electric appliances, whereof a few dozen are rel- evant in terms of energy consumption. In order to keep the convenience level for the customer high, we need an intelligent control system that identifies devices currently turned on and proposes minimal-invasive changes to their usage. To get this information, each relevant appliance could be equipped with a smart meter or an embedded communication and control interface able to deliver power infor- mation and characteristics [5]. Upgrading all devices in a current household this way would be painstaking and costly. An alternative approach is non-intrusive load monitoring (NILM)[8], which determines and classifies individual appliances based on characteristic load profiles. For identification only a single smart me- ter measuring the total power consumption with appropriate timely resolution is sufficient. NILM extracts features like active power, frequency etc., classifies appliances and identifies appliances by matching the measured data to a ref- erence database. Thus, the identification can be described as an optimization