Abstract
Infrared absorption spectroscopy is a widely used tool to quantify and monitor compositions of gases. The concentration information is often retrieved by fitting absorption profiles to the acquired spectra, utilizing spectroscopic databases. In complex gas matrices an expanded parameter space leads to long computation times of the fitting routines due to the increased number of spectral features that need to be computed for each iteration during the fit. This hinders the capability of real-time analysis of the gas matrix. Here, an artificial neural network (ANN) is employed for rapid prediction of gas concentrations in complex infrared absorption spectra composed of mixtures of CO and N2O. Experimental data is acquired with a mid-infrared dual frequency comb spectrometer. To circumvent the experimental collection of huge amounts of training data, the network is trained on synthetically generated spectra. The spectra are based on simulated absorption profiles making use of the HITRAN database. In addition, the spectrometer’s influence on the measured spectra is characterized and included in the synthetic training data generation. The ANN was tested on measured spectra and compared to a non-linear least squares fitting algorithm. An average evaluation time of 303 µs for a single measured spectrum was achieved. Coefficients of determination were 0.99997 for the predictions of N2O concentrations and 0.99987 for the predictions of CO concentrations, with uncertainties on the predicted concentrations between 0.04 and 0.18 ppm for 0 to 100 ppm N2O and between 0.05 and 0.18 ppm for 0 to 60 ppm CO.
Highlights
Quantitative gas analysis for the determination of the concentration of gases present in the sample has emerged as a powerful tool in a variety of applications in environmental sensing [1,2,3], medicine [4,5], or process monitoring [6]
To determine the concentrations of the gases of interest, fitting algorithms based on non-linear least squares are often employed [13,14], relying on absorption profiles provided by spectroscopic databases such as HITRAN [15]
To study the performance of the artificial neural network (ANN), namely uncertainties of the determined conTo study the performance of the ANN, namely uncertainties of the determined concencentrations, linearity, and computation time, and to be able to compare it to the results trations, linearity, and computation time, and to be able to compare it to the results obtained obtained by a fit with Equation (2), we measured absorbance spectra of 18 different mixby a fit with Equation (2), we measured absorbance spectra of 18 different mixtures of N2 O
Summary
Quantitative gas analysis for the determination of the concentration of gases present in the sample has emerged as a powerful tool in a variety of applications in environmental sensing [1,2,3], medicine [4,5], or process monitoring [6]. During training the strengths of the connections (weights) between the different nodes and layers are adjusted to minimize the deviation of the ANNs predictions to the target values, which are attached as so-called labels to the training data Training of these network structures typically requires large datasets (often in the order of hundreds of thousands of data, or more) with a high variety of concentrations and mixtures, setting a high experimental burden for the preparation of actual measurements. A quantity of 107 synthetic spectra with arbitrary concentrations between 0 to 60 ppm and 0 to 100 ppm for CO and N2 O, respectively, were generated and superimposed by a baseline with arbitrarily chosen parameters in a range, which was known from the previous investigation of the spectrometer This approach is only valid for trace gas concentrations, as higher concentrations would lead to distortions of absorption profiles, e.g., due to changes of broadening coefficients. Exemplifies potential of the proposed method for the analysis of absorption spectra within a specific spectral band
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
