Most of today’s fire alarm systems use a variety of smoke detection technologies to detect fires. These include spot-type ionization and photoelectric detectors, aspirating smoke detectors and beam-type smoke detectors. In some application, planners also use beam smoke detectors, flame detectors or linear heat detectors for rapid detection.
However, some fundamental issues will always limit the detectors’ capabilities due to physical restrictions and even more in challenging environments such as harsh surroundings, like saw mills, power plants or buildings with high ceilings.
Video-based detection technologies, using intelligent video analytics and detecting the origin of a fire can be a solution in such cases as they provide very early detection.
Using video technology for fire detection
One of the main benefits of video-based fire detection (VFD) is speed and reaction time since a camera can detect a fire as soon as the smoke enters its field of view. Thus, it does not require the smoke to migrate to the sensor and triggers much faster than point-type, beam or aspiration smoke detectors. In demanding environments with high ceilings, VFD solutions can detect a starting fire within seconds while other technologies will need minutes because the smoke rises slowly.
Video-based fire detection can also cover larger areas and spaces than point-type detectors, making them a much more affordable solution for warehouses, hangars or large halls.
Operators in a monitoring center or local guards can verify the alert in the video image before triggering a fire alarm. In addition, they can pinpoint the exact location of the fire as well as involved commodities, which provides valuable time for firefighting.
VFD solutions are based on intelligent algorithms, which analyze the video image for the occurrence of flame and/or smoke. In general, there are different approaches to analyze the image like neuronal networks, contrast-based or physical algorithms.
Neuronal networks use decision trees, which are trained by labeled video footage. Depending on the trained video footage, these algorithms learn how flames or smoke look like in the video image. Neuronal networks highly depend on the used video footage. If the footage has less variation in scenery and type of fire, the algorithms may encounter difficulties.
Contrast-based smoke detection algorithms check video images for growing grey areas and the loss of contrast of the background, using the optical effects of smoke in the video image. They can only be used for smoke detection and are susceptible to illuminance changes. For good detection performance a very well structured and high-contrast background is needed.
Physical algorithms use video images to detect fires by their behavior. Flames, for example, do flicker and have distinct colors. Smoke not only covers the background, but also has a special movement behavior due to thermal buoyancy. Due to their adaption to the fire behavior, physical algorithms can minimize false alarms. Furthermore, additional algorithms can be combined with the existing ones to reduce false alarms even further.
AVIOTEC – video-based fire detection from Bosch
Video surveillance technology and the development of intelligent algorithms for image analysis as well as technologies for fire alarm systems are among the core competencies of Bosch. This expertise has led to the development of specific algorithms that allow reliable fire detection within seconds and low false alarm scenarios at the same time.
Detection based on physical smoke and flame models
Based on a unique and scientifically tested physical smoke model, AVIOTEC’s algorithms directly detect smoke at the source, reliably distinguishing between smoke and moving objects. Detection occurs within seconds, but in order to avoid costly false alarms, the solution will add a customizable verification period before it issues an alarm. While most video-based solutions need a smoke opacity of 50-65%, AVIOTEC’s patented algorithms will reliably detect smoke with an opacity of 40%, ensuring an even faster alarm notification. As AVIOTEC smoke detection relies on the movement of smoke, it requires the fire in its initiating phase to be in the camera’s field of view. It is not designed to detect ambient effects of smoke, like contrast decrease or gathering smoke below ceilings, which can easily be done with state-of-the art smoke detectors, but to speed up detection
As smoke detection uses a physical smoke model, a physical flame model based on flame color, flickering and shape underlies the detection of flames. Flame characteristics of different fires are well understood, and thus flame detection through intelligent video analysis is just as reliable as smoke detection. Using video to detect flames avoids the necessity to install and operate optical flame detectors.
Usually video-based fire detection is more sensitive to false alarms than state-of-the art fire detectors. However, compared to other video-based fire detection systems AVIOTEC is by itself highly immune against such false alarms. This robustness is backed up by an intelligent video analysis within the camera that allows to detect disturbing values such as movement, reflections or changes in lighting conditions and to offset such influences.
AVIOTEC’s smoke and flame detection capabilities are all integrated directly into the cameras, making no further analyzing equipment necessary. On top of that, once installed, AVIOTEC IP starlight 8000 cameras with their Intelligent Video Analytics (IVA) can also be used for automated surveillance tasks. They are able to detect unusual movements as well as blocked aisles or emergency exits, thus increasing safety and operational efficiency. In warehouses, AVIOTEC doubles also as an efficient prevention of theft and arson, which ranks second only to electrical problems when it comes to the causes of warehouse fires. Both fire and surveillance alarms will be transmitted via network and/or relay.
AVIOTEC IP starlight 8000 cameras cover broad areas, need little maintenance and do not require individual power supplies. With Power over Ethernet (PoE), power and video signals use the same cable, also allowing the camera to benefit from the uninterruptable power supplies (UPS) in the Ethernet switches. The entire solution thus comes with a very affordable total cost of ownership.
AVIOTEC scales well from a single camera to a networked system of distributed cameras with a central console and management system. It can relay alarms to an existing fire alarm panel or transmit them via Ethernet to the monitoring center or even a mobile device. Receiving HD quality video images in real time, gives the firefighters during their approach a good understanding of the current situation even before they arrive at the scene.
Why use AVIOTEC?
To summarize: While standard smoke and flame detection technologies work well in most environments, video-based detection with AVIOTEC offers a lot of advantages in challenging environments:
- Fast fire detection – minimizes damage and saves lives
- Fire detection at the source – responds quickly even in inversion layers with hot air pockets
- Intelligent algorithms – ensure high immunity against false alarms and precise detection
- Root cause analysis – videos can be analyzed to prevent future damages by the same matter
- PoE power supply – reduced installation costs – one cable for alarm, video and power transmission
- Parallel video surveillance and video analytics – minimizes equipment and saves costs
AVIOTEC IP starlight 8000: The first VdS approved VFD solution worldwide has also received ActivFire certification
When installing a fire alarm system or equipment in a premise it is very important that it meets the highest quality standard and is installed correctly, so that it will achieve its function and will operate as expected. This is achieved by ensuring that the equipment and the installation meets the standards as laid down, e.g. in the appropriate EN54 Specification.
Therefore, standardization of video-based fire detection is very important for this new technology. VFD is an Active Work Item in ISO, while FM3232 and UL286 standards for VFD already exist. Although video-based fire detection (VFD) offers a lot of benefits, so far there is no EN54 standard available for this technology. To close this gap in the European certification, the VdS Schadenverhütung GmbH (short VdS) has developed a sophisticated test procedure for video-based fire detection that incorporates the proven VdS guidelines 2203,”Requirements for fire protection software”, as well as the “Requirements for testing flame detectors”. The VdS is an independent, renowned institution for corporate safety and security that sets international guidelines through its synchronized set of rules. The VdS quality seal is an important investment criterion.
Based on this new VdS test procedures, VFD has been tested under harsh environmental conditions such as increased heat, cold, humidity and mechanical influences usually occurring under operating conditions. The tests followed procedures and values as defined in EN54 standards. Tests and certification by VdS stand for a reliable and fast detection of test fires like TF2, TF4, TF6 and TF8.
More recently, CSIRO has also verified and issued ACTIVFIRE certification for AVIOTEC based on the VdS test procedures and certification.