Comparison of neural network, conditional demand analysis, and engineering approaches for modeling end-use energy consumption in the residential sector

Aydinalp-Koksal M., Ugursal V. I.

APPLIED ENERGY, cilt.85, sa.4, ss.271-296, 2008 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 85 Sayı: 4
  • Basım Tarihi: 2008
  • Doi Numarası: 10.1016/j.apenergy.2006.09.012
  • Dergi Adı: APPLIED ENERGY
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.271-296
  • Anahtar Kelimeler: residential energy consumption modeling, conditional demand analysis, neural networks modeling, BAYESIAN-APPROACH, ELECTRICITY, APPLIANCE, SPACE
  • Hacettepe Üniversitesi Adresli: Evet

Özet

This paper investigates the use of conditional demand analysis (CDA) method to model the residential end-use energy consumption at the national level. There are several studies where CDA was used to model energy consumption at the regional level; however the CDA method had not been used to model residential energy consumption at the national level. The prediction performance and the ability to characterize the residential end-use energy consumption of the CDA model are compared with those of a neural network (NN) and an engineering based model developed earlier. The comparison of the predictions of the models indicates that CDA is capable of accurately predicting the energy consumption in the residential sector as well as the other two models. The effects of socio-economic factors are estimated using the NN and the CDA models, where possible. Due to the limited number of variables the CDA model can accommodate, its capability to evaluate these effects is found to be lower than the NN model. (C) 2006 Elsevier Ltd. All rights reserved.

This paper investigates the use of conditional demand analysis (CDA) method to model the residential end-use energy consumption at the national level. There are several studies where CDA was used to model energy consumption at the regional level; however the CDA method had not been used to modelresidential energy consumption at the national level. The prediction performance and the ability to characterize the residential end-use energy consumption ofthe CDA model are compared with those of a neural network (NN) and an engineering based model developed earlier. The comparison of the predictions of themodels indicates that CDA is capable of accurately predicting the energy consumption in the residential sector as well as the other two models. The effects ofsocio-economic factors are estimated using the NN and the CDA models, where possible. Due to the limited number of variables the CDA model can accommodate, its capability to evaluate these effects is found to be lower than the NN model.