A background reduction method based on empirical mode decomposition for tunable diode laser absorption spectroscopy system

In tunable diode laser absorption spectroscopy based trace-gas detection, the background fluctuation caused by interference fringes can significantly influence the system performance. To reduce the background fluctuation, empirical mode decomposition (EMD) is applied to process the measured signal in this paper. By analyzing the correlation coefficients between the original measured signal and each intrinsic mode function (IMF) acquired by EMD, the IMFs related to background are identified and removed. Lastly, the smoothed measurement values can be attained through a reconstruction procedure. The effectiveness of this proposed method was demonstrated by utilizing it to background and methane concentration measurements. After utilizing the proposed method, the standard deviation (STD) of the background measured in absence of absorbing gas was decreased from 1.9565 parts-per-million (ppm) to 0.1689ppm. The reduction of background fluctuation is achieved to a factor of 11. Results on the concentration measurements of 280ppm methane also indicate a precision enhancement, where the fluctuation reduced significantly with the STD decreased to 0.2537ppm. In addition, the proposed method exhibits a good performance on the running time. All the experiment results show that the proposed background reduction method has a great potential to be used in both laboratory and industrial gas detection.