基于贝叶斯估计的多探测器火警判定方法研究

Research on multi-detector fire alarm judgment method based on Bayesian estimation

为应对感烟探测器的大量误报对消防应急响应带来的挑战,考虑目前以感烟探测器为主的火警设施误报率高且短期内难以全部更换的特点,提出基于贝叶斯估计的多探测器火警判定方法,通过多个探测器的报警时间间隔计算火源位置的后验概率分布,并提出火警真实度概念,为火警判定提供依据。结果表明:使用多探测器耦合模型时每增加1个探测器可将误报率降低约4个数量级,该方法在探测器正常、部分失效、误报的情景下均能有效判别火警。

In order to cope with the challenges caused by the large number of false alarms of smoke detectors to the fire emergency response,considering the high false alarm rate of current fire alarm facilities dominated by smoke detectors and the difficulty of all replacement in a short period of time,a multi-detector fire alarm judgment method based on the Bayesian Estimation was proposed.The posterior probability distribution of the fire source location was calculated through the alarm time intervals of multiple detectors,and the concept of fire alarm authenticity was proposed to provide a basis for the fire alarm judgment.The results showed that when the multi-detector coupling model was used,each additional detector could reduce the false alarm rate by about four orders of magnitude,and this method could effectively distinguish the fire alarms under the normal,partial failure and false alarm scenarios of detectors.