Multistage ensemble of feedforward neural networks for prediction of heating energy consumption
Апстракт
Feedforward neural network models are created for prediction of heating energy consumption of a university campus. Actual measured data are used for training and testing the models. Multistage neural network ensemble is proposed for the possible improvement of prediction accuracy. Previously trained feed-forward neural networks are first separated into clusters, using k-means algorithm, and then the best network of each cluster is chosen as a member of the ensemble. Three different averaging methods (simple, weighted, and median) for obtaining ensemble output are applied. Besides this conventional approach, single radial basis neural network in the second level is used to aggregate the selected ensemble members. It is shown that heating energy consumption can be predicted with better accuracy by using ensemble of neural networks than using the best trained single neural network, while the best results are achieved with multistage ensemble.
Кључне речи:
neural networks / multistage ensemble / k-means clustering / heating consumption predictionИзвор:
Thermal Science, 2016, 20, 4, 1321-1331Издавач:
- Univerzitet u Beogradu - Institut za nuklearne nauke Vinča, Beograd
DOI: 10.2298/TSCI150122140J
ISSN: 0354-9836
WoS: 000382511600026
Scopus: 2-s2.0-84991710811
Колекције
Институција/група
Mašinski fakultetTY - JOUR AU - Jovanović, Radiša AU - Sretenović, Aleksandra AU - Živković, Branislav PY - 2016 UR - https://machinery.mas.bg.ac.rs/handle/123456789/2299 AB - Feedforward neural network models are created for prediction of heating energy consumption of a university campus. Actual measured data are used for training and testing the models. Multistage neural network ensemble is proposed for the possible improvement of prediction accuracy. Previously trained feed-forward neural networks are first separated into clusters, using k-means algorithm, and then the best network of each cluster is chosen as a member of the ensemble. Three different averaging methods (simple, weighted, and median) for obtaining ensemble output are applied. Besides this conventional approach, single radial basis neural network in the second level is used to aggregate the selected ensemble members. It is shown that heating energy consumption can be predicted with better accuracy by using ensemble of neural networks than using the best trained single neural network, while the best results are achieved with multistage ensemble. PB - Univerzitet u Beogradu - Institut za nuklearne nauke Vinča, Beograd T2 - Thermal Science T1 - Multistage ensemble of feedforward neural networks for prediction of heating energy consumption EP - 1331 IS - 4 SP - 1321 VL - 20 DO - 10.2298/TSCI150122140J ER -
@article{ author = "Jovanović, Radiša and Sretenović, Aleksandra and Živković, Branislav", year = "2016", abstract = "Feedforward neural network models are created for prediction of heating energy consumption of a university campus. Actual measured data are used for training and testing the models. Multistage neural network ensemble is proposed for the possible improvement of prediction accuracy. Previously trained feed-forward neural networks are first separated into clusters, using k-means algorithm, and then the best network of each cluster is chosen as a member of the ensemble. Three different averaging methods (simple, weighted, and median) for obtaining ensemble output are applied. Besides this conventional approach, single radial basis neural network in the second level is used to aggregate the selected ensemble members. It is shown that heating energy consumption can be predicted with better accuracy by using ensemble of neural networks than using the best trained single neural network, while the best results are achieved with multistage ensemble.", publisher = "Univerzitet u Beogradu - Institut za nuklearne nauke Vinča, Beograd", journal = "Thermal Science", title = "Multistage ensemble of feedforward neural networks for prediction of heating energy consumption", pages = "1331-1321", number = "4", volume = "20", doi = "10.2298/TSCI150122140J" }
Jovanović, R., Sretenović, A.,& Živković, B.. (2016). Multistage ensemble of feedforward neural networks for prediction of heating energy consumption. in Thermal Science Univerzitet u Beogradu - Institut za nuklearne nauke Vinča, Beograd., 20(4), 1321-1331. https://doi.org/10.2298/TSCI150122140J
Jovanović R, Sretenović A, Živković B. Multistage ensemble of feedforward neural networks for prediction of heating energy consumption. in Thermal Science. 2016;20(4):1321-1331. doi:10.2298/TSCI150122140J .
Jovanović, Radiša, Sretenović, Aleksandra, Živković, Branislav, "Multistage ensemble of feedforward neural networks for prediction of heating energy consumption" in Thermal Science, 20, no. 4 (2016):1321-1331, https://doi.org/10.2298/TSCI150122140J . .