STUDY ON REAL-TIME LOAD PREDICTION METHODS FOR HAVAC SYSTEM CONTROL
-
- KAWASHIMA Minoru
- Institute of Technology, Shimizu Corporation
-
- ITO Naoaki
- Dept. of Architecture, Faculty of Engineering, Tokyo Metropolitan University
Bibliographic Information
- Other Title
-
- 空調システム運転制御を対象とした実時間負荷予測手法に関する研究
- クウチョウ システム ウンテン セイギョ オ タイショウ ト シタ ジツジカン
Search this article
Abstract
First, importance of real-time load prediction and several kinds of optimization on HVAC system control are discussed. Then, six prediction models, ARIMA (Autoregressed Integrated Moving Average), EWMA (Exponential Weighted Moving Average), RLR (Recursive Linear Regression), ANN (Artificial Neural Network), KALMAN (Kalman filter) and FNN (Fuzzy Neural Network), are examined to compare their accuracy under situations where the all models use the same 3-month-long calculated load data and weather data in cooling and heating season. The results shows that the ANN model has the best prediction accuracy. It is confirmed that the ANN is a potential prediction model for practical utilization in HVAC system control.
Journal
-
- Journal of Architecture and Planning (Transactions of AIJ)
-
Journal of Architecture and Planning (Transactions of AIJ) 61 (484), 43-51, 1996
Architectural Institute of Japan
- Tweet
Keywords
Details 詳細情報について
-
- CRID
- 1390001204783086336
-
- NII Article ID
- 10004175434
- 110004654338
-
- NII Book ID
- AN10438548
-
- ISSN
- 18818161
- 13404210
-
- NDL BIB ID
- 3984815
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
- NDL
- Crossref
- CiNii Articles
-
- Abstract License Flag
- Disallowed