A new AI-powered fire detection system eliminates costly false alarms by distinguishing real fire smoke from harmless aerosols like dust and cooking fumes, promising a breakthrough in fire safety technology.
False Alarm Prevention. Image Credit: Electronics and Telecommunications Research Institute (ETRI)
A group of South Korean researchers has developed an intelligent fire detection technology that drastically reduces false alarm incidents, which go off in the absence of a real fire (hereinafter "unwanted alarm"), and is on the verge of commercializing it. This technology is expected to reduce the social cost of unwanted alarms considerably.
Electronics and Telecommunications Research Institute (ETRI) announced the development of an AI sensor for unwanted alarm prevention. The sensor distinguishes between smoke caused by fire and non-fire aerosol particles by measuring particle light scattering characteristics, which vary with the wavelength of light.
The existing photoelectric smoke detector has an infrared light source and a light-sensing photodiode placed in opposite directions inside. When particles like smoke enter the detector, the photodiode captures the scattered light generated as the smoke hits the light source. If the scattered light exceeds a certain level, the alarm is activated.
However, aerosolized particles, such as dust and moisture generated by routine cooking, cigarette smoke, etc., can flow into the detector, setting off an unwanted alarm. Photoelectric detectors that detect scattered light trigger this alarm.
According to the National Fire Agency, there were 258,220 fire engine dispatches between 2021 and July 2022, and 96.6% were due to alarm malfunctions.
Alternatively, the AI sensor for unwanted alarm prevention developed by ETRI measures the distinct scattering properties of each aerosol particle based on the different wavelengths of light. Thus, it can accurately determine an actual fire outbreak.
The ETRI researchers have built a database by projecting light of different wavelengths onto aerosol particles and measuring the scattering of each particle. Combining this with artificial intelligence (AI) technology, they have developed an AI sensor for unwanted alarm prevention that distinguishes whether a particular aerosol particle is caused by fire or not before deciding to activate a fire alarm.
ETRI plans to first apply the AI sensor for unwanted alarm prevention to the aspirating smoke detectors. The aspirating smoke detector works on a similar principle to that of a photoelectric detector by drawing in air with a fan and swiftly detecting smoke. Although it detects smoke faster than a photoelectric detector, it is prone to malfunction due to dust and moisture. Therefore, it is installed and used in limited spaces such as semiconductor clean rooms and server rooms.
In particular, most of the aspirating smoke detectors currently available on the market are expensive imported products. In addition, since they lack a feature to distinguish whether a fire outbreak occurs or not if a domestic product equipped with this technology is released, it would be highly competitive in the domestic and foreign fire detector markets.
ETRI Director Kang Bok Lee of the Defense & Safety Intelligence Research Section explained, "After it is commercialized, this technology will significantly reduce the number of false alarms caused by non-fire incidents, thus reducing the cost of fire engine dispatches and wasteful use of firefighting resources that are estimated at KRW 20 billion annually."
Based on the measurement of the scattering spectrum, this technology can also be used in cosmetic, medical, environmental, and other industrial fields. ETRI is currently in discussions with relevant companies about commercializing this technology.
1) Unwanted alarm: An alarm set off when a fire detection system is triggered by factors other than those resulting from a fire outbreak, such as heat, smoke, and flame
2) Aerosol: Small particle in a solid or liquid state that is suspended in mid-air
3) Photodiode: A semiconductor component that converts optical (light) signals to electrical signals
4) Scattered Light: Light that scatters in all directions after hitting a substance or particle, changing its direction of travel
The research was conducted through the "Smoke Particle Spectrum Analysis-based Intelligent Fire Detector Development" project, which was part of the "ETRI R&D Support Project," which was funded by the Ministry of Science and ICT and the Institute of Information & Communications Technology Planning & Evaluation (IITP).