대한전기학회:학술대회논문집 (Proceedings of the KIEE Conference)
- 대한전기학회 1999년도 하계학술대회 논문집 B
- /
- Pages.776-778
- /
- 1999
신경회로망을 이용한 축열시스템의 식별기 설계
Identifier Design of Thermal Storage System Using Neural Network
- Kim, Jung-Wook (Hanyang Univ.) ;
- Lim, Hoo-Jang (Hanyang Univ.) ;
- Kim, Dong-Hun (Hanyang Univ.) ;
- Lee, Eun-Wook (Chungcheong College) ;
- Chung, Kee-Chul (Daeduk College) ;
- Yang, Hai-Won (Hanyang Univ.)
- 발행 : 1999.07.19
초록
In this paper, identifier for thermal storage system using multi-layer feedforward neural network (MFNN) is designed. It is very difficult to control thermal storage system, since thermal storage system is nonlinear and its time constant is very large. Thus, in the MFNN, delta-bar-delta algorithm for high running speed and 2-bit status input are used. Also hardware using microprocessor for identifier is developed. The experimental results indicate that the proposed method can predict temperature more accurately.
키워드