[Objective] Current fire smoke detectors are susceptible to many factors that can significantly affect their accuracy, such as environmental disturbances and combustion states, resulting in false alarms. Moreover, point fire detectors inevitably cause time lags for alarms due to their optical darkrooms and insect-proof nets, delaying critical rescue time. Therefore, focusing on the limitations of point carbon monoxide (CO) detectors with absorbing gas cells, this study proposes an optical probe initial fire detection system based on tunable diode laser absorption spectroscopy (TDLAS) and laser remote sensing. Then, we present a preliminary threshold-based fire alarm algorithm.[Methods] Accordingly, this study's detection system relied on TDLAS and laser remote sensing, including wavelength modulation spectroscopy, to extract CO signals from an open-path geometry. A laser (wavelength=2331.93nm) was also used as an optical probe to replace the traditional absorption measurement chambers, achieving a CO path-integrated volume fraction measurement under complex initial fire conditions. First, we tested the signal responses of this detection system for different CO volume fraction combinations, incidence angles, and reflectances using a standard gas cell (length=0.5m) placed at the optical path and a standard reflector plate placed 4.4m from the light source to examine the limitations of the system. Then, we tested the received signal power at different distances, examined the CO released from wood pyrolysis fires, calculated the corresponding integral volume fractions, and set an appropriate threshold based on the detector's limit to verify whether this optical probe fire detector could meet the requirements.[Results] The experimental results of our limitation tests showed the following: (1) At a distance of 4.4m and target reflectance of 0.69, the theoretical detection limit was approximately 5 (μL/L)·m, and the actual detection limit was 26.75 (μL/L)·m. (2) As the reflectance decreased, the absolute value of the signal intensity also decreased, increasing the uncertainty in the measured volume fraction by a detection limit of 88.22 (μL/L)·m and reflectance of 0.07. (3) As predicted by calculations, although the absolute value of the second harmonic signal varied slightly for the 0°-10° incidence angle, the corresponding normalized absorption peak outputs were essentially the same. (4) When the distance to the target reflective surface was within 10 m, a detection limit of no less than 20 (μL/L)·m was achieved. Conversely, the standard wood pyrolysis fire tests showed that the system with a theoretical detection limit of 30 (μL/L)·m and a proposed threshold alarm value of 70 (μL/L)·m potentially triggered the initial fire alarm.[Conclusions] This study determines the feasibility of an open optical path to detect initial CO release from fire through detection limit tests and standard fire tests. Overall, the proposed method integrating TDLAS and laser remote sensing achieves the expected goal of detecting initial fire, significantly addressing the shortcomings of gas detectors using absorption cavities and point fire detectors. © 2023 Press of Tsinghua University. All rights reserved.
