Abstract:  In order to reduce the false alarm rate and improve the detection accuracy, aiming at the suction smoke sensor system of data center cabinet, a very early fire detection cavity structure was designed. The photoelectric detection digital circuit hardware was completed, and the photoelectric signal was collected. The ensemble empirical mode decomposition (EEMD) algorithm and wavelet analysis algorithm were used respectively to filter the baseline shift and reduce the background noise.After comparing and calculating the correlation and signalto- noise ratio, the results show that the performance of the EEMD method is significantly better than the wavelet analysis method, which can effectively reduce the false alarm rate and improve the detection accuracy, and realize the "very early" detection and warning of fire. 

Key words:  , fire detection, baseline drift, data center cabinet, photoelectric signal