University of Hertfordshire

Performance comparison between KNN and NSGA-II algorithms as calibration approaches for building simulation models

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Standard

Performance comparison between KNN and NSGA-II algorithms as calibration approaches for building simulation models. / Basurra, S.; Jankovic, L.

Proceedings of Building Simulation & Optimization 2016. IBPSA England, 2016.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Harvard

Basurra, S & Jankovic, L 2016, Performance comparison between KNN and NSGA-II algorithms as calibration approaches for building simulation models. in Proceedings of Building Simulation & Optimization 2016. IBPSA England, Building Simulation and Optimization 2016, Newcastle, United Kingdom, 12/09/16.

APA

Basurra, S., & Jankovic, L. (2016). Performance comparison between KNN and NSGA-II algorithms as calibration approaches for building simulation models. In Proceedings of Building Simulation & Optimization 2016 IBPSA England.

Vancouver

Basurra S, Jankovic L. Performance comparison between KNN and NSGA-II algorithms as calibration approaches for building simulation models. In Proceedings of Building Simulation & Optimization 2016. IBPSA England. 2016

Author

Basurra, S. ; Jankovic, L. / Performance comparison between KNN and NSGA-II algorithms as calibration approaches for building simulation models. Proceedings of Building Simulation & Optimization 2016. IBPSA England, 2016.

Bibtex

@inproceedings{bfeca642b82b4d8d9397420f148ea0b7,
title = "Performance comparison between KNN and NSGA-II algorithms as calibration approaches for building simulation models",
abstract = "In this paper, a study of calibration methods for athermal performance model of a building ispresented. Two calibration approaches are evaluatedand compared in terms of accuracy and computationspeed. These approaches are the 푘 Nearest Neighbour(KNN) algorithm and NSGA-II algorithm.The comparison of these two approaches was basedon the simulation model of the Birmingham ZeroCarbon House, which has been under continuousmonitoring over the past five years. Data fromarchitectural drawings and site measurements wereused to build the geometry of the house. All buildingsystems, fabric, lighting and equipment werespecified to closely correspond to the actual house.The preliminary results suggest that the predictiveperformance of simulation models can be calibratedquickly and accurately using the monitoredperformance data of the real building. Automatingsuch process increases its efficiency and consistencyof the results while reducing the time and effortrequired for calibration. The results show that bothNSGA-II and KNN provide similar degree ofaccuracy in terms of the results closeness tomeasured data, but whilst the former outperforms thelatter in terms of computational speed, the latteroutperforms the former in terms of results widecoverage of solutions around the reference point,which is essential for calibration.",
author = "S. Basurra and L. Jankovic",
note = "Shadi Basurra , and Ljubomir Jankovic, ‘Performance comparison between KNN and NSGA-II algorithms as calibration approaches for building simulation models’, in BSO 2016 Proceedings. Paper presented at the 3rd IBPSA England Conference, Newcastle, September 2016. Content in the UH Research Archive is made available for personal research, educational, and non-commercial purposes only. Unless otherwise stated, all content is protected by copyright, and in the absence of an open license, permissions for further re-use should be sought from the publisher, the author, or other copyright holder.",
year = "2016",
month = "9",
day = "12",
language = "English",
booktitle = "Proceedings of Building Simulation & Optimization 2016",
publisher = "IBPSA England",

}

RIS

TY - GEN

T1 - Performance comparison between KNN and NSGA-II algorithms as calibration approaches for building simulation models

AU - Basurra, S.

AU - Jankovic, L.

N1 - Shadi Basurra , and Ljubomir Jankovic, ‘Performance comparison between KNN and NSGA-II algorithms as calibration approaches for building simulation models’, in BSO 2016 Proceedings. Paper presented at the 3rd IBPSA England Conference, Newcastle, September 2016. Content in the UH Research Archive is made available for personal research, educational, and non-commercial purposes only. Unless otherwise stated, all content is protected by copyright, and in the absence of an open license, permissions for further re-use should be sought from the publisher, the author, or other copyright holder.

PY - 2016/9/12

Y1 - 2016/9/12

N2 - In this paper, a study of calibration methods for athermal performance model of a building ispresented. Two calibration approaches are evaluatedand compared in terms of accuracy and computationspeed. These approaches are the 푘 Nearest Neighbour(KNN) algorithm and NSGA-II algorithm.The comparison of these two approaches was basedon the simulation model of the Birmingham ZeroCarbon House, which has been under continuousmonitoring over the past five years. Data fromarchitectural drawings and site measurements wereused to build the geometry of the house. All buildingsystems, fabric, lighting and equipment werespecified to closely correspond to the actual house.The preliminary results suggest that the predictiveperformance of simulation models can be calibratedquickly and accurately using the monitoredperformance data of the real building. Automatingsuch process increases its efficiency and consistencyof the results while reducing the time and effortrequired for calibration. The results show that bothNSGA-II and KNN provide similar degree ofaccuracy in terms of the results closeness tomeasured data, but whilst the former outperforms thelatter in terms of computational speed, the latteroutperforms the former in terms of results widecoverage of solutions around the reference point,which is essential for calibration.

AB - In this paper, a study of calibration methods for athermal performance model of a building ispresented. Two calibration approaches are evaluatedand compared in terms of accuracy and computationspeed. These approaches are the 푘 Nearest Neighbour(KNN) algorithm and NSGA-II algorithm.The comparison of these two approaches was basedon the simulation model of the Birmingham ZeroCarbon House, which has been under continuousmonitoring over the past five years. Data fromarchitectural drawings and site measurements wereused to build the geometry of the house. All buildingsystems, fabric, lighting and equipment werespecified to closely correspond to the actual house.The preliminary results suggest that the predictiveperformance of simulation models can be calibratedquickly and accurately using the monitoredperformance data of the real building. Automatingsuch process increases its efficiency and consistencyof the results while reducing the time and effortrequired for calibration. The results show that bothNSGA-II and KNN provide similar degree ofaccuracy in terms of the results closeness tomeasured data, but whilst the former outperforms thelatter in terms of computational speed, the latteroutperforms the former in terms of results widecoverage of solutions around the reference point,which is essential for calibration.

M3 - Conference contribution

BT - Proceedings of Building Simulation & Optimization 2016

PB - IBPSA England

ER -