Chen Cheng-LiangChao Yung-ChengHuang Hsiao-PingJen Jing-JeiFang Ming-Dar2019-05-212019-05-21198903681653https://scholars.lib.ntu.edu.tw/handle/123456789/409724This article aims at building a real-time expert system for supporting stable and efficient ironmaking blast-furnace operations, i.e., for producing high quality and quantity pig iron with low energy cost. The knowledge base of this supporting expert system consists of experiential operating heuristics of operators and some theoretical models which include a stochastic time-series model and a coke-reduction model for predictions of hot metal temperatures and furnace heat levels, respectively. The concepts of fuzzy sets are also used supplementarily to deal with unavoidable uncertainties inherent in experiential knowledge and the commonly used linguistic descriptions about furnace heat levels. The system is composed of two parts: an abnormal conditions diagnosis system and a heat level control system. Based on sensored furnace information, the system can afford adequate judgements of furnace conditions, heat levels and their possible trends and then respond with suggested operational actions for every 30 minutes. This blast-furnace supporting expert system is currently under testing in China Steel Company.journal article2-s2.0-0024705831https://www.scopus.com/inward/record.uri?eid=2-s2.0-0024705831&partnerID=40&md5=75c324bdb03500b7036cabff0f03061f