Article ID: | iaor1992775 |
Country: | Japan |
Volume: | 35 |
Issue: | 8 |
Start Page Number: | 461 |
End Page Number: | 466 |
Publication Date: | Aug 1990 |
Journal: | Communications of the Operations Research Society of Japan |
Authors: | Nakamori Yoshiteru, Kaneda Mayumi, Nomura Junji, Kurio Takashi, Nakanishi Shinji |
Keywords: | artificial intelligence: expert systems |
In this paper, the authors propose a fire-judgement expert system applying fuzzy theory, using an analog sensor for smoke, temperature, and CO gas. The purposes of the intelligent fire warning system are ‘early detection of fire’ and ‘decrease of false alarms’. This system infers what happens from several kinds of sensed data, using judgement rules based on expert knowledge. Moreover, in order to tune these rules the authors try to apply a fuzzy clustering method to the experimental sensed data. On the other hand, as for how to inform the inferring process, this system displays fuzzy-certainty factors of typical fire/non-fire phenomena such as fire, tobacco smoke, and steam, until making a final decision. The authors have done various experiments using fire, tobacco smoke, and steam, and in all cases, the system made a satisfactory decision. [In Japanese.]