moomalq / nilm-eval Goto Github PK

NILM-Eval

NILM-Eval: An evaluation framework for non-intrusive load monitoring algorithms

Project overview

NILM-Eval is a MATLAB framework that allows to evaluate non-intrusive load monitoring algorithms in different scenarios to gain a comprehensive view on their performance. NILM-Eval makes it easy to evaluate algorithms on multiple data sets, households, data granularities, time periods, and specific algorithm parameters. By encapsulating those parameters in configurations, NILM-Eval further allows the user with little effort to repeat experiments performed by others, to evaluate an algorithm on a new data set, and to fine-tune configurations to improve the performance of an algorithm in a new setting.

For more detailed information refer to the following sources:

• C. Beckel, W. Kleiminger, R. Cicchetti, T. Staake, S. Santini: The ECO Data Set and the Performance of Non-Intrusive Load Monitoring Algorithms. Proceedings of the 1st ACM International Conference on Embedded Systems for Energy-Efficient Buildings (BuildSys 2014). Memphis, TN, USA. ACM, November 2014.

• R. Cicchetti: NILM-Eval: Disaggregation of real-world electricity consumption data. Master's thesis, ETH Zurich, 2014.

• The ECO dataset: Together with Energie Thun, a Swiss energy provider, we collected the ECO data set (Electricity Consumption and Occupancy). Using NILM-Eval, we evaluated the performance of four NILM algorithms on the ECO data set.

Project team

NILM-Eval is a research activity of the Distributed Systems Group at ETH Zurich, Switzerland. It was initiated in the context of the project Smart Meter Services, in which we develop methods to analyze smart meter consumption data to offer novel services to households.

Contact

If you have questions or ideas, contact Christian Beckel.