Loading...
Please wait, while we are loading the content...
Similar Documents
Fast and Reliable Wildfire Smoke Detection based on Self-learning and Interactive Functionalities to provide useful Early Detection
| Content Provider | Semantic Scholar |
|---|---|
| Author | Dreibach, Joachim F. |
| Copyright Year | 2017 |
| Abstract | Every year, the risk for fast spreading Wildfires is increasing as the past Wildfire statistics proofs. Analysis show, that fire suppression effort have been effective to stop most fires by a successful initial attack. Over 97 % of wildfires kept less than 1,000 acres and 99%less than 10,000 acres (Havlina et al. 2015, Murphy et al.2013). The probability of a successful initial attack and to avoid megafires, decreases exponentially with every minute passes after ignition. So, in the same rate as the danger increases, the requirements and demands increase to detect a starting fire as quick as possible after the ignition and analyse the precise location of the fire. While in populated areas smoke plumes are reported quit early once they become visible, still the fire position is not clear (Cao,J; Boruff, B.J. & Mc.Neil, The smoke is rising but where is the fire?..). Also starting night fires will not be detected in a early stage and in this cases, a initial attack is mostly too late to be successful. Vision sensor based detection methods can be a tool to provide reliable Early Detection and precise fire information. Most of the available technology generates much to high false alerts as well as late detection while the detection range is limited to visibility. If video technology is used, smoke can be only detected when a smoke plume appears with clear density. But if a smoke plume is visible, it ́s almost late for the fire fighting resources to approach the fire place in time to perform a successful initial attack. Such parameters lead to low user acceptance and detection benefit and utilized vision detection systems are not operated for detection, but more and more for monitoring tasks. The presented detection solution describes a sensor solution in combination with the software, detecting smoke, not depending on density of a smoke plume, providing real “Early” Detection at minimized numbers of false alerts. The self-learning functionality consider changing environmental and geographic situations as well as interactive user inputs, for a continuous adapting and optimizing of the detection algorithm to generate high quality detection information to the Fire Management and the involved Resources. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://www.nfpa.org/-/media/Files/News-and-Research/Resources/Research-Foundation/Symposia/2017-SUPDET/SUPDET17-Dreibach.ashx?la=en |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
