- Author
- Kozeki, D. | Satoh, K. | Takemoto, A. | Nomura, J. | Kurio, T. | Nakanishi, S.
- Title
- Study of Fire Detection System Introducing AI Technology. Part 2. Role of Fuzzy Expert System to Reduce False Alarms.
- Coporate
- Matsushita Electric Works, Ltd., Osaka, Japan
- Journal
- Report of Research Institute of Japan, No. 71, 21-31, March 1991
- Keywords
- fire detection systems | expert systems | false alarms | artificial intelligence
- Identifiers
- Fuzzy expert systems
- Abstract
- [Abstract in English] The objective of this study is to develop a new fire detection system which is extremely reliable, i.e., far less alarming falsely, using modern Artificial Intelligence (AI) technology. In this study, the role of Fuzzy Expert System incorporated in our prototype AI fire detection system was examined to reduce false alarms. Fuzzy membership functions corresponding to flaming fires, smoldering fires, smoke of tobacco, and steam flowing out of a bathroom in a hotel guest room were defined in consistency with the rules of expert system. Experimental results showed that the Fuzzy membership functions defined in this study and the Fuzzy Expert System are useful to reduce false alarms. It was easy to identify flaming fires and steam from bathroom, but not always easy to identify smoldering fires of bed sheets with the tobacco smoke before and just when the smoke concentration reaches the value of 10%/m. However, in case of smoldering fires the fire growth rate is much smaller than in case of flaming fires. Therefore, the system can wait a little until the final decision of either fires or non-fires. By delaying the final decision a little, the system could judge either fires or non-fires clearly.