Abstract
Coordinated (de)activation of heat pumps in residential heating systems can support the balance between energy production and consumption in a smart grid. Stopping the heat pump in an inappropriate operation state may result in users’ discomfort or damages to the system. In this contribution, a procedure to split domestic hot water provision (DHW) cycles from space heating cycles in the power consumption time series of heat pumps in residential heating systems is presented. The procedure bases on a support vector machine (SVM) classifier applied on statistical properties of the cycles after principal component transformation. The procedure is tested on real-world power consumption time series with 1 s resolution from three different buildings monitored over 1 year. Due to the absence of ground truth validation data, training data for the SVM algorithm are distilled from the measurement data by outlier removal and subsequent K-means classification. The dependence of energy demand for space heating (heating curve) and DHW provision on the ambient temperature are investigated for validation and agrees with Swiss building standards.
References
Olsson T (2013) Evaluation of machine learning for predicting hot water production of a heat pump. In: 4th international conference on power engineering, energy and electrical drives, Istanbul, Turkey
Summerfield A et al (2016) Detailed analysis of data from heat pumps installed via the renewable heat premium payment scheme. Technical report, February
Geidl M et al (2017) A fully operational virtual energy storage network providing flexibility for the power system. In: Proceedings of 12th IEA heat pump conference, Rotterdam, The Netherlands
Author information
Authors and Affiliations
Corresponding author
Additional information
This work was supported by the Swiss State Secretariat for Education, Research and Innovation (SERI) under Contract No. 16.0082.
Rights and permissions
About this article
Cite this article
Schuetz, P., Durrer, R., Gwerder, D. et al. Poster abstract: state of operation recognition for heat pumps from smart grid monitoring data. Comput Sci Res Dev 33, 259–261 (2018). https://doi.org/10.1007/s00450-017-0372-5
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00450-017-0372-5