In Korea, the use of fire-detection systems applying IoT technology to existing analog fire-alarm systems has increased owing to the communication technology convergence, the world's best Internet network, and the proliferation of Internet of Things (IoT). Its use can be expected to increase worldwide in the future. For IoT-based fire-detection systems to exhibit the requisite reliability (based on a low false-alarm rate), research related to the analysis of detection signals should be actively promoted and conducted. However, there has been no research activity based on actual operational data, apart from the research that has been conducted in laboratory environments. The primary reason for this state of affairs has been that the installation and use of IoT-based fire-detection systems on a large scale has been rare, worldwide. Consequently, with respect to the fire-signal characteristics of IoT-based fire-detection systems, related data in this study were obtained by investigating actual fire accident cases, using fire alarm data that occurred over a period of 5 years. Based on the signal pattern analysis results using these field data, a fuzzy logic system for recognizing fire signal patterns was developed and verified. As a result, in the actual fire accidents examined, an "alarm" condition-corresponding to the high possibility of fire among the five fire alarms-was determined 30 s before the actual fire alarm. Moreover, it was also found that approximately 80% of non-fire alarms could be reduced in the actual fire alarms that occurred at Institute K during the 5-year period examined.
Keywords: False alarm rate; IoT-based fire-detection system; Unwanted fire alarm.
© 2023 Published by Elsevier Ltd.