STUDY ON NONLINEAR MULTI-VARIABLE-ANALYSIS METHOD AND ITS CAPABILITY OF GENERALIZATION FOR RESIDENCIAL ENERGY CONSUMPTION

• TANAKA Akio

  Jyukankyou Research Institute Inc.

• NAKAGAMI Hidetoshi

  Jyukankyou Research Institute Inc.

• ISHIHARA Osamu

  Dept. of Architecture and Civil Engineering, Kumamoto University

• SAKAI Koji

  Dept. of Architectural Eng., Oita University

Bibliographic Information

Other Title

• 住宅エネルギー消費の非線形多変量解析手法とその汎化能力の研究
• ジュウタク エネルギー ショウヒ ノ ヒセンケイ タヘンリョウ カイセキ シュホウ ト ソノ ハンカ ノウリョク ノ ケンキュウ

Abstract

This paper examines, for analysis of residential energy consumption, the neural network nonlinear multi-regression (NNW-NLR) method and the liner multi-regression (LR) method that is most frequently used and demonstrates the advantage of the NNW-NLR over the LR method in terms of the significance and the capability of generalization. The correlation for the verification data is 0.77 using the NNW-NLR equation derived from the training data, while 0.22 using the LR equation. In addition, the random number method (RNM) is proposed for variables selection in building the NNW.

Journal

• Journal of Environmental Engineering (Transactions of AIJ)

  Journal of Environmental Engineering (Transactions of AIJ) 68 (570), 67-73, 2003

  Architectural Institute of Japan

Keywords

• Housing
• Energy consumption
• Neural Networks
• Regression analysis

Details

• CRID: 1390282680177077632
• NII Article ID: 110004083996
• NII Book ID: AA11830377
• DOI: 10.3130/aije.68.67_2
• ISSN: 1881817X
• ISSN: 13480685
• NDL BIB ID: 6678954
• Web Site: https://ndlsearch.ndl.go.jp/books/R000000004-I6678954
• Web Site: https://www.jstage.jst.go.jp/article/aije/68/570/68_KJ00004078990/_pdf
• Text Lang: ja
• Data Source:
  • JaLC
  • NDL
  • Crossref
  • CiNii Articles
  • KAKEN
• Abstract License Flag: Disallowed