Building fire alarm model based on fire source inversion according to smoke arrival time intervals

Accurate and timely triggering of a fire alarm has always been essential for building fire safety. The inversion of fire source parameters by multiple detectors to distinguish between true and false fire alarms has become a popular research topic. However, previously proposed inversion models are based on time series and are computationally intensive. Their inversion accuracy directly depends on the size of the database. Therefore, this paper proposes a fire source inversion method employing multiple smoke alarms based on Bayesian inference. A fire dynamics model along with an actual sensor response time are used to implement a fire alarm. Moreover, fire alarm authenticity is defined to distinguish false and true fire alarms, enabling early fire detection. Verification test results indicate that the proposed model can reduce the number of false alarms and realize early fire alarm detection.

Automated summary

This paper proposes a fire source inversion method employing multiple smoke alarms based on Bayesian inference to achieve accurate and timely fire alarm. A fire dynamics model and actual sensor response time are used to implement a fire alarm, and fire alarm authenticity is defined to distinguish false and true fire alarms, enabling early fire detection. Verification test results indicate that the proposed model can reduce the number of false alarms and realize early fire alarm detection.