期刊名称：Eastern-European Journal of Enterprise Technologies
印刷版ISSN：1729-3774
电子版ISSN：1729-4061
出版年度：2017
卷号：3
期号：9
页码：53-59
DOI：10.15587/1729-4061.2017.101985
语种：English
出版社：PC Technology Center
摘要：Theoretical analysis revealed that in order to create learning fire detectors, capable of adjusting to unknown conditions of application, it is expedient to consider the criterion of equality of probabilities of false detection and skipping a fire as a criterion of guaranteed fire detection. By using such detection criterion, it is possible to provide guaranteed fire detection under conditions of the absence of a priori information about statistics of the recorded data. The algorithms and structural circuits of the learning fire detectors were developed for the case of discrete and continuous data recording by sensors. Their distinguishing feature is the possibility of application under indeterminate conditions when there is no a priori information about the type of distribution laws of the recorded data, as well as their capability to adapt to previously unknown and changing application conditions and to provide guaranteed fire detection in this case. It was shown that the main limitation in the implementation of such algorithms is the need to use additional instructions from a trainer about the existence or the absence of a fire on the object. To overcome this limitation, it is proposed to apply the hypothesis about sufficient rarity of events related to a fire on the protected sites. This made it possible to use the registered information about the absence of fire as the instructions from a trainer. In this case, the resulting modified algorithm and the structural circuit of the proposed fire detector that matches it do not require instructions from a trainer and, in this sense, a detector becomes a self-learning fire detector.Results of examining the fire detectors, performed based on the example of real dynamics of the mean temperature of medium when alcohol is ignited and burned, demonstrated their high efficiency. In comparison with fire detectors that comply with the requirements of standard EN 54-5:2003, the examined self-learning fire detectors possess an essential gain in time (exceeding 170 times) of the guaranteed fire detection on the site under uncertain conditions. The ability of self-learning fire detectors to adapt to previously unknown conditions allows their application under non-stationary conditions in order to detect complex fires.
关键词：learning fire detector;guaranteed fire detection;a priori uncertainty of detection condition
