NEUROBAT – a self - commissioned heating control system

A predictive and adaptive heating controller, using artificial neural networks to allow the adaptation of the control model to the real conditions (climate, building characteristics, user's behaviour) is described. The choice of a predictive control strategy, combined with the nonlinear modelling of the building, the user's behaviour and the weather prediction, allows the NEUROBAT controller to achieve energy savings while ensuring an optimal thermal comfort. Moreover, the commissioning time of the new controller is considerably reduced, thanks to the use of self-learning neural algorithms. In the first phase of the NEUROBAT project an innovative heating controller concept especially suited for water heating systems has been developed. The NEUROBAT heating controller has been simulated and its performance data compared to the performance data of commercial heating controllers. The promising simulation results of the NEUROBAT heating controller have been confirmed by tests on a real site, with energy gains of 13% under the condition of an optimised user comfort when compared to an advanced commercial heating controller. In a second phase of the project, an industrial prototype has been developed and comparative tests have been completed on a residential building in Basel. The qualitative performance assessment has put in evidence the advantageous self-commissioning concept (plug-and-play) of the NEUROBAT heating controller, requiring a minimal installation effort at start-up. An industrialisation phase of the project, currently going on, aims at building a commercial system based on the NEUROBAT heating control algorithm. * corresponding author: Jens Krauss Centre Suisse d’Electr. et de Microtech. SA Address: Jaquet-Droz 1, 2000 Neuchâtel Tel.: +41-32-7205546 Fax: +41-32-7205720 e-mail: [email protected] or [email protected] Centre Suisse d'Electronique et de Microtechnique SA / ESTIA Sàrl Sensors Application, Vol 2, Wiley-VCH, April 2001 : ”NEUROBAT Heating Controller” Page 2 / 25 ABSTRACT 1