Selection of the Optimal Neural Network Architecture for Next Peak Load Forecasting
-
- Onoda Takashi
- CRIEPI
-
- Ohba Eiji
- CRIEPI
Bibliographic Information
- Other Title
-
- 翌日最大電力需要予測における最適なニューラルネットワーク構成の決定法
- ヨクジツ サイ ダイ デンリョク ジュヨウ ヨソク ニ オケル サイテキ ナ
Search this article
Abstract
In engineering fields, one of the most important applications of artificial neural networks is a non-linear parametric model that can approximate any continuous input-output relation. Especially, in next day peak load forecating, the ability of approximation of artificial neural networks is very useful. However, an important but difficult problem exits in applying a neural network to next day peak load forecasting. This problem is to estimate the quality of approximation of a neural network. This quality depends on the architecture of the neural network applied, as well as on the complexity of the target relation. In other words, the pmoblem is to determine the optimal number of parameters in the neural network.<br> This paper proposes a new method for the selection of the optimal neural network architecture for next day peak load forecasting. The proposed method is based on a new information criterion. This paper also applies the proposed method to next day peak load forecasting constructed by real data and shows the comparison of error estimation of some models for peak load forecasting.
Journal
-
- IEEJ Transactions on Power and Energy
-
IEEJ Transactions on Power and Energy 118 (5), 497-504, 1998
The Institute of Electrical Engineers of Japan
- Tweet
Keywords
Details
-
- CRID
- 1390001204603700608
-
- NII Article ID
- 130006840390
- 10012645773
- 10002876200
-
- NII Book ID
- AN10136334
-
- ISSN
- 13488147
- 03854213
- http://id.crossref.org/issn/03854213
-
- NDL BIB ID
- 4469885
-
- Data Source
-
- JaLC
- NDL
- Crossref
- CiNii Articles
-
- Abstract License Flag
- Disallowed