Current studentsExisting staffContact us
 

Prediction of indoor temperature in an institutional building

Simple item page
dc.contributor.authorAfroz, Z.
dc.contributor.authorShafiullah, G.M.
dc.contributor.authorUrmee, T.
dc.contributor.authorHiggins, G.
dc.contributor.conference9th International Conference on Applied Energy, ICAE2017 (21 Aug 2017 - 24 Aug 2017 : Cardiff, UK)
dc.contributor.editorLi, H.
dc.contributor.editorWu, J.
dc.contributor.editorYan, J.
dc.date.issued2017
dc.description.abstractThe importance of predicting building indoor temperature is inevitable to execute an effective energy management strategy in an institutional building. An accurate prediction of building indoor temperature not only contributes to improved thermal comfort conditions but also has a role in building heating and cooling energy conservation. To predict the indoor temperature accurately, Artificial Neural Network (ANN) has been used in this study because of its performance superiority to deal with the time-series data as cited in past studies. Network architecture is the most important part of ANN for predicting accurately without overfitting the data. In this study, as a part of determining the optimal network architecture, important input parameters related to the output has been sorted out first. Next, prediction models have been developed for building indoor temperature using real data. Initially, spring season of Australia was selected for data collection. During model development three different training algorithms have been used and the performance of these training algorithms has been evaluated in this study based on prediction accuracy, generalization capability and iteration time to train the algorithm. From results Lovenberg-Marquardt has been found the best-suited training algorithm for short-term prediction of indoor space temperature. Afterwards, residual analysis has been used as a technique to verify the validation result. Finally, the result has been justified by applying a similar approach to another building case and using two different weather data-sets of two different seasons: summer and winter of Australia.
dc.identifier.citationEnergy Procedia, 2017 / Li, H., Wu, J., Yan, J. (ed./s), vol.142, pp.1860-1866
dc.identifier.doi10.1016/j.egypro.2017.12.576
dc.identifier.issn1876-6102
dc.identifier.urihttps://hdl.handle.net/11541.2/40760
dc.language.isoen
dc.publisherElsevier
dc.publisher.placeNetherlands
dc.relation.fundingMurdoch University Research Scholarship (MURS)
dc.relation.ispartofseries142, 1876-6102
dc.rightsCopyright 2017 The Authors. Published by Elsevier Ltd. Under a Creative Commons license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
dc.source.urihttps://doi.org/10.1016/j.egypro.2017.12.576
dc.subjectprediction
dc.subjectindoor temperature
dc.subjectArtificial Neural Network (ANN)
dc.subjectperformance improvement
dc.titlePrediction of indoor temperature in an institutional building
dc.typeConference paper
pubs.publication-statusPublished
ror.mmsid9916917429001831

Files

Collections